19660004870_Skipwith_Magnetic_Tape_Recording 19660004870 Skipwith Magnetic Tape Recording
User Manual: Pdf 19660004870_Skipwith_Magnetic_Tape_Recording
Open the PDF directly: View PDF 
.
Page Count: 337
| Download | |
| Open PDF In Browser | View PDF |
•
https:llntrs.nasa.gov/search.jsp?R=19660004870 2018-05-05T01 :47:47+00:00Z
CAS t.s
tiL ~
COpy
NASA SP.5038
TECHNOLOGY SURVEY
-Technology Utilization DiviSion-
MAGNETIC
T APE RECORDING
I\I~~
"
......
.\.
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION
NASA SP-5038
TECHNOLOGY
SURVEY
Technology
Utilization
Division
MAGNETIC
TAPE RECORDING
By Skipwith W. Athey, Ph. D.
(Prepared Under Contract NASw - 945)
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION
Washington, D. C.
January 1966
NOTICE
This document was prepared under the sponsorship of the
NatIOnal Aeronautics and Space AdmmistratlOn Neither
the United States Government nor any person actmg on
behalf of the Umted States Government assumes any liability resultmg from the use of the mformatlOn contamed
in thil;; document, or warrants that such use will be free
from privately owned rights
For Sale by the Superintendent of Documents, U S. Government
Printing Office, Washington, D.C., 20402 - Price $1 25
Foreword
The Administrator of the National Aeronautics and Space
Administration has established a technology utIlization program for "the rapid dissemination of information . . . on
technological developments ... which appear to be useful for
general mdustrial applIcatIOn." From a variety of sources,
mcluding NASA Research Centers and NASA contractors,
space-related technology IS collected and screened; apd that
whIch has potential industrial use is made generally available.
InformatIon from the nation's space program is thus made
available to American industry, including the latest developments in materials, processes, products, techniques, management systems, and analytical and design procedures.
Magnetic tape recorder technology dIffers somewhat from
the technology of some of the other fields that are being covered.
The magnetic tape recorder is a highly developed device and
ItS detailed technology IS hidden below a thick layer of practical engineering design. It is therefore difficult to extract
technology per se from the tape recorder field, nor would this
be very useful under the objectives of the survey program.
Hence, this partIcular survey does not deal to as great extent
as some others with specific details of technology but broadly
covers two aspects of the applied technology in the development of which NASA has participated.
One area of technology which IS specifically NASA sponsored is that which has led to the development of miniature
severe-environment tape recorders for satellite and space probe
use. In this area NASA has directly sponsored innovation,
emphasizing reliability If not new concepts. The biggest dollar impact of NASA work on tape recorder technology has
been as a major customer for commercial ground-based tape
recorders for telemetry data acquisition and related purposes.
It would be impossible to extract NASA's specific contributions
to data acquisition and data reduction technology from the
mass of application lore and knowhow which has been built
III
IV
FOREWORD
up in this area. As a major customer, however, NASA has
had a strong indirect influence on the recorder development
that has taken place through purely commercial channels.
This survey, therefore, diScusses the entire range of recorder
technology with emphasis on the miniature high environment
recorder and work in which NASA may be seen to have had a
major influence.
THE DIRECTOR, Technology Utilization Division
National Aeronautics and Space Administration
Acknowledgments
Ideally, a survey is complete and precise; tlllS one is neither
as complete nor as precise as I would have liked to make it.
Some of its shortcomings can be charged to the difficulty of
obtaining information in time to include it. The "cut-oft' date"
of the information is uneven; the average cut-off date was
toward the end of 1964 although some 19_65 i:nformation
is mcluded. Other faults, however, involving sms both of
commission and omission, are chargeable only to the particular approach I have taken and for which I must accept full
responsibility.
Thanks are naturally due to the NASA contractors and subcontractors who have cheerfully supplied information.
Equal thanks are due to those recorder manufacturers who,
although not specifically suppliers of new recorder technology
under contract to NASA, halve given the writer much information about their techniques.
The following organizations have been of considerable help
in supplying information in general terms: Ampex Corporation, Mincom Division of the Minnesota Mining and Manufacturing Company, Consolidated Electrodynamics Corporation,
Sangamo Electric Company, Genisco Data, Borg-Warner Controls, Tech-Center Division of Cook Electric Company, College
HIll Industries, Leach Corporation, Precision Instrument Company, Radio Corporation of America, Astro-science Corporation, Lockheed Electronics Company, Ralph M. Parsons Company, Raymond Engineering Laboratory, Inc., and of course,
the Jet Propulsion Laboratory of the California Institute of
Technology .
In the course of performmg this Survey, I visited Goddard
Space Flight Center, Lewis Research Center, the George C.
Marshall Space'Flight Center, the Manned Spacecraft Center,
Langley Research Center, Wallops Station, and the Flight
Research Center. The cooperation and assistance, particularly
of the Technology UtilIzation Offices of these Centers was esv
VI
ACKNOWLEDGMENTS
sential to the accomplishment of this task. It is perhaps appropriate to smgle out Mr. Pleasant T. Cole, manager of the
Recording Techniques Group at Goddard Space Flight Center,
whose orIginal persistence led to the preparation of this Survey;
Mr. John Warden of the Patent Office at Jet Propulsion Laboratory, who somehow managed to obtain contractor data in
quantities well beyond my expectations from the Jet Propulsion
Laboratory files, and Mr. Leonard Ault of the Congressional
Relations Office of Goddard Space Flight Center for hIS tireless
efforts in digging out other mformation from NASA files.
SKIPWITH
W.
ATHEY
CONTENTS
Page
CHAPTER 1. INTRODUCTION
CHAPTER 2 THE FIELD OF MAGNETIC RECORDING..
Entertainment Recording
Instrumentation Recordings
Transverse Recording .
Digital Recording
Aubome Recording .
The Scope of This Survey
Grollnd-Based Recorders
Flight Recorders
1
5
5
7
9
10
12
12
13
16
CHAPTER 3 THE ELEMENTS OF THE TAPE RECORDER
19
19
22
24
The Recording Medium
Recording and Reproducing Transducers
The Tape Moving Mechanism
Eledronics
Summary
27
29
CHAPTER 4. RECORDING METHODS.
31
33
Linear or Analog Recording .
FM Recording
Special FM Recording Systems.
Puls¢ Duration Modulation
Puls¢ Amplitude Modulation .
Puls¢ Code Modulation (PCM)
Puls¢ Recording Waveforms
Oth¢r Recording Methods
35
41
42
46
48
52
55
CHAPTER 5. HEAD-TAPE INTERACTION
57
65
74
76
Head-Tape Geometry-Biased Recording
Head-Tape Geometry-Pulse Recording
Future Requifements
vn
VITI
CONTENTS
Page
CHAPTER 6. TAPE MOVING SYSTEMS
Configurations
The Open Loop
Tape Velocity at Capstan and Head
Closed-Loop Recorders
Maintainmg Tension in the Closed Loop
Zero-Loop Configurations
Recorders of Unusual Configuration
Tape Reeling Mechanisms
Reeling Motors
Brake Tension Control
Servo Reeling Devices
Reeling Irregularities
Tape Housekeeping
Rewind, Fast-Forward, and Search
Data Reduction Functions
Transverse Recorders
Unusual Dnve Schemes
CHAPTER 7 DISTURBANCES TO TAPE MOTION
77
18
80
94
95
96
102
103
105
106
111
114
116
118
120
121
122
126
127
R~~
1~
Instantaneous Time Error
Flutter Sources
Flutter Redudlon
Disturbance Compensation Techniques
129
1 30
1 32
1 36
CHAPTER 8 TAPE RECORDER HEADS-STRUCTURE
AND FUNCTION
Head Types
Head Materials
Head Strudure
Head PreciSion
Erase Heads
CHAPTER 9 MAGNETIC TAPE
Magnetic Properties of Tape
Tape Base Materials
Tape Bmders
Tape Manufadurin9 Procedures
Other Tape Properties
SpeCial Tapes
CHAPTER 10 RECORDER ELECTRONICS
Signal Electronics-Dynamic Range
Signal Equalization
Implementation of Recording Fundlons
Other Signal Problems
Pulse Techniques
Control Eledronlcs
141
141
143
144
1 52
153
155
156
158
159
160
162
164
167
167
171
112
114
115
176
CONTENTS
IX
Pa.ge
CHAPTER 11 TAPE RECORDER MECHANICAL COMPONENTS
Modular Construction of Flight Recorders.
Motors
Bearings
Clutches
Brakes
GUides
Drive Belts
Vibration Isolation.
Torquing Springs .
CHAPTER 12. METHODS OF TESTING AND EVALUATING TAPE AND RECORDERS
Distortion Testing
Signal-to-Noise Ratio Testing
Flutter-Testing
Time-Displacement Error Testing
Dropout Testing
Tape Testing
Environmental Tests
CHAPTER 13. MINIATURE HIGH-ENVIRONMENT RECORDERS
Reel-to-Reel, Record-Only Recorders
Reel-to-Reel Recorders, Responsive to Playback Commands
The Endless-Loop Recorder
Unusual Recorder Formats
Miniature Transverse-Scan Recorders
Drive Systems for Miniature Recorders
Special Problems of Space Probe Recorders
CHAPTER 14 COMPLETE RECORDING SYSTEMS
The Interplanetary Monitoring Platform (IMP)
OGO-A
Nimbus
Central Data Processmg FaCility
CHAPTER 15. UTILIZATION FACTORS IN MAGNETIC
RECORDING
APPENDIX
References
Index
Author Index
181
183
187
191
196
198
198
199
200
202
203
203
205
207
211
212
213
213
215
217
224
238
256
263
264
270
271
272
281
289
292
301
307
311
317
325
CHAPTER 1
Introduction
NASA's contribution to tape recorder technology may appear to
lie in such glamorous devices as the tape recorders of the Tiros and
Nimbus satellites which -receive signals carrying weather info~ation
during an orbit around the earth and then transmit those signals
back to the ground when the satellite is in view of ground stations.
These quite important recording devices are, however, only useful as
members of a complex hierarchy of devices all of which participate in
the ultimate process of obtaining information from NASA's efforts.
Their existence and utility is based on the existence of many more
prosaic recordmg devices which have been involved in the
development of such spectacular units.
Satellite recorders must operate without maintenance, must show
great reliability, and must consistently deIrver performance closely
related to the needs of the program in whICh they are used. Such recorders must be conservatively desIgned and exhaustively tested. They
invariably sacrifice peak of state-of-the-art performance for reliability
and the ability to eXlst m unfriendly environments. The development
of such recorders represents the combination of considerable effort by
many manufacturers and government laboratories, but the average
home hi-fidelity fan can purchase for $150.00 in a local furniture store
a recorder which appears to have considerably higher performance
specifications.' The differences in environment and reliability are
responsible, of course, for this peculiar situation.
The output of one of the speciahzed satellite or space probe recorders will invariably be recorded on the ground by a commercial instrumentation recorder of conservative design but of considerably higher
performance than the device in the satellite. The output of this
ground recorder will be reproduced and analyzed agam and again
through the use of further conventional commercial recorders.
In the development of almost every smgle piece of NASA hardware of any degree of complexity beyond a single bolt and nut, con1
2
MAGNETIC TAPE
RECORDING
ventional or commercial magnetic recorders have been involved.
Quantities of data have been collected from tests in the laboratory, at
ground test sites, in static firing tests, in ballistic vehicle firings, and
in prelIminary orbital shots leading to the development of the complete booster/satellite combinations now in use. The use of magnetic
recording in the development process far outweighs the amount of
recording done either on the ground or in flight in satellite or space
probe programs.
It is as a major user, directly and indirectly, of such test recording
procedures, that NASA has made its real contribution to tape recorder
technology. As a most demanding customer with almost lmlimited
requirements in magnetic recording, NASA, its contractors, its subcontractors, and its sub-sub-contractors have represented a great part
of the market influence which has encouraged the development of
recorders of improved performance by commercial manufacturers.
In this role lies NASA's greatest importance as a supporter of tape
recorder technology.
It would be impossible to extract and present III an organized manner the mass of knowhow that NASA and its contractors have
obtained about the use of magnetIC recording. Occasionally a specific development can be traced to the needs of a particular NASA
program but, as often as not, other government programs, directly
or indirectly, have had sufficiently silllilar requirements that it is difficult to separate the NASA influence from that of other organizations. It was to deal somehow with this massive NASA influence that
the format of the present Technology Survey was evolved.
The entire range of current tape recorder technology is presented
here in survey and outline form. Except in isolated cases, no attempt
is made to say that NASA has been responsible for this or that aspect
of the commercial recorders currently available. It is beheved that
the technology for the acquisition of which NASA IS at least in part
responsible can best be made available to industry by presenting a
broad view of the current state of the art. Both to the user and to the
potential desIgner of sophisticated magnetic recording equipment such
a survey should provide some assistance. It is hoped that the user will
be made sufficiently aware of the capabilIties and limitations of magnetIC recording that he can make proper system decisions in his utilizatIOn of magnetic recording. To the designer this report may not seem
as sophisticated a document as it will to the user, but it may, by
emphasizing the user's problems and interest, guide in some small
measure the development of new recording instruments. In specific
mstances, the breadth of the survey will make it possible to present
material not adequately presented preVIously in either periodical or
INTRODUCTION
3
textbook literature. Although not exhaustive, the coverage has been
designed to be complete enough to give warning where problems of
development or extension of recording capabIlities may exist.
The purpose of magnetic recording IS t~ receive electrical signals of
a wide variety and to record and store those signals in a form which
allows their accurate reproduction at a later time. The ideal recorder
is one for which the only difference between the electrical signal received by the recorder and that reproduced by the reproducer is the
gross time delay between recording and reproduction. All the characteristics of the electrical signal should be preserved perfectly.
To describe the accuracy of reproduction, the same terms may be
used that are used to describe the accuracy of transmission of an
electrical signal through some transmIssion means. Typically, such
transmIssion is described in terms of signal-to-noise ratio, bandwidth,
and- distOrtion-of various-types;-iridiiding botli-phase-and amplitude.
Of relatively small importance in most transmission systems is any
disturbance of the time scale of the signal, although such disturbances
are encountered in sky wave radio transmission and VHF transmission
subject to variable diffraction. Time scale disturbances are, however,
inherent in any recording process.
The state of the recordmg art can thus be described in terms of the
standard electrical characteristics of the transmission system, to which
must be added specification of disturbances of the time scale of the
input signal. As in most complex systems, the various electrical performance characteristics as well as the time scale disturbance are interrelated in the typical recording system. Likewise, almost every
element involved m the recording part of the process and the mechanism transporting the recording medium has some effect on all the
characteristics by which the performance of the overall system is
described.
In addition to the effect of the recording and reproducing process
on the signals handled as in a transmission system, the storage process
has another dimension. This dimension is the volume of storage
medium required to preserve a signal or signals of particular performance characteristics and lasting for a given length of time. This
storage volume parameter is supplementary to the "nonsignal" parameters which recording shares with other transmission systems. Such
nonsignal parameters include volume, power consumption, and physical dimensions as well as weight.
The presentation of the recording art is made for several purposes,
including advising the reader what he may find available in the way of
4
MAGNETIC TAPE
RF)CORDING
caprubilities of solving his recording problems as well as the tutorial
purpose of establishing the design principles which have led to this
current state of the art. To accomplish both these purposes, the survey
presentation is organized at several levels. Initially the interrelationship between the elements of the process and the effects on recording
performance will be described in general terms. The several elements
of the recording process will be separated and characterized in gen~ral terms. Under each of these elements the basic design principles
will be presented, and in this process, the effect of the elements on
performance will be analyzed. Where necessary, separate analyses
will be presented of the relationship between process elements which
cannot actually be separated in such a straightforward manner. At
the conclusion of thIS synthetic process a final presentation of complex
complete recording systems will be made, and these complex systems
will be analyzed to show design philosophy, informatIOn flow, and
the complex tradeoffs necessary between the various elements of a
complete recording system.
CHAPTER 2
The Field
of Magnetic Recording
It would be neIther practical nor appropriate to include in this
survey, in detail, every current application of_magnetic recorders-impractical because the field is too wide, inappropriate because this
NASA-supported study should not range too far from NASA's specific interests in recorders. To place NASA's areas of greatest concern
in perspective, this initial section will, however, contain a brief description of the entire broad field of recording applications. In light
of the background so presented the selection of the parts to be covered
in detail on the basis of N AHA.'s interests and uses will then be
discussed.
Covering in detail only selected types of recorders and recorder
applications will fortunately not result in the omission of any important technology. The field' of magnetic recording contains such
internal relationships that all important aspects of recorder technology
will necessarily be covered in descrihing the relatively limited area in
which NASA is interested. There will be, however, one deliberate
omission-rapid start/stop mechanisms for tape transports to be used
as memories for digital computers will not be covered, on the hasis that
their technology is in no way NASA-inspired.
ENTERTAINMENT RECORDING
Magnetic recording first made a name for itself in the field of sound
entertainment (broadcasting and disk recording). Wire recorders
had extensive use as "electronic notebooks" in W orId War II, and crude
metal-tape recorders were used for such special applications as 1minute voice recorders for speech training in the 40's. Not until
modern tape made of plastic and oxide became available and ac bias
made improved signal-to-noise ratio and fidelity possible did magnetic recording have any impact on broadcastmg and professional
sound recording. The reusability of magnetic tape, the high signal5
6
MAGNETIC TAPE RECORDING
to-noise ratio it provided (originally seeming almost limitless), and
the opportunity it gave for simple editing made tape appear to be the
answer to nearly every sound recording problem. Although disillusion set in after the initial enthusiasm and few of the potentials of
tape proved to be as great as originally predicted, the net utility and
performance level of the magnetic recorder has continued to increase
over the years. It h'as not displaced the disk record and may very
well never do so, but It is now essential to the modern field of high - '
fidelity disk recording. Although the tape signal-to-noise ratio has
proved not to be limitless and, about 3 or 4 years ago, appeared to be
the factor setting an absolute upper limit to the utility of magnetic
tape for sound, recent developments have lowered the noise barriers
once more and even greater signal-to-noise ratio is now available.
The development of the use of magnetic recording for sound
naturally branched into two paths. One path led to the home recorder
used by the enthusiast to record sounds generated by him or his musical (or other) friends or to record sound radio broadcasts. With the
increased availability of such recorder/reproducers in the home, the
market for commercially prerecorded tape has developed until it is
now a major factor in the United States entertainment market. A
wide range of recorders for such application is available, from extremely crude battery-operated units costing a few dollars to equipment which, at least in its nominal specifications, has performance
equivalent to that of professional recorders. Except for the crudest of
"uch recorders, they all follow the same mechanical scheme. They
dIffer in such sometimes controversial refinements as the use of torque
motors for supply and takeup reeling rather than less expensive
clutches or brakes of nominally poorer performance. The current
amateur recorder is, on casual examination, clearly related in mechanical scheme to the first post-World-War-II professional audio recorders; it may not, however, approach even the earliest models in
refinement and performance.
The other main path of development of the entertainment sound
recorder was dIrected to professional use. Except for minor variations, this development has consisted of refinement of baSICally the
same recorder that was introduced immedIately following World War
II. Those early recorders were used for recordmg radio broadcasts
for later scheduling, and for recording the master tapes from which
disk records would later be made. Such recorders could be used for
editing their own tapes and the science and art of editing tape by
physically cutting and splicing it reached a high degree of refinement.
Within this general field, specialization has naturally taken place;
semiportable equipment IS available for broadcast recording in the
TaE FIELD OF MAGNETIC RECORDING
7
field and bulky equipment of extremely high performance has been
developed for production of master tapes in the studio.
The first magnetic sound recorders of 1946-47 were close copies of
the German broadcast equipment developed during World War II
(Hansell [1945]) (Ranger [1947]) (Lindsay and Stolaroff [1948]).
They delivered qujte satisfactory tape moving performance for
applicatIons where the human ear was the judge of the smoothness
of tape motion and there has therefore been little encouragement for
improvement in sound recorder mechanical design. The parallel
development of magnetic recording for sound in motion pictures used
the same mechanisms as were already in use for motion picture
optIcal sound tracks, and such mechanisms are in use to this day.
The relatively stiff sprocketed film is simply coated with magnetic
material and placed in basically the same mechanisms that were developed -for optical recording -(Miller: [1947]). Attempts have
been made to apply this motion-picture based technique of sprocketed
tape handling to instrumentation recording WIthout much success.
Interestingly enough, a part of the motion picture recordmg technique
was carried over later to instrumentation recorders but in such a form
that, although the parent equipment and its descendants resemble each
other physically, they do not really operate on the same principles.
INSTRUMENTATION RECORDINGS
It was soon obvious that, by simply runnmg the tape faster, sound
recorders could be adapted for the recording of scientific information
in analog form. For scientific use one does not usually know what
the spectrum of the signal to be recorded will be, so it is necessary to
expect any signal amplitude at any frequency. The high-frequency
preemphasis which contributed heavily to the signal-to-noise ratlO of
sound recording was therefore not usable for scientific applications.
The signal-to-noise ratio achieved by early instrumentatlOn recorders
was naturally much worse than would have been expected by scaling
up from the sound recorders which preceded them.
By modern standards, the first instrumentation recorders were rather
crude. The signal-to-noise ratios were marginal for many applicatiOns and they had a lot of flutter. ThIS flutter included some at relatively low frequencies which might have bothered the audio user and,
with higher bandwidths, the high-frequency flutter originating in
tape scrape and in the vibration from unsupported tape also became
important. These early recorders, nevertheless, made possible the
acquisition of data which previously had been inaccessible and they
were hailed with enthusiasm. Certain of their shortcomings were,
perhaps, inadequately evaluated. The broademng of the spectrum
178~0-6f>-2
8
MAGNETIC
TAPE RECORDING
of a desired signal by the broadband flutter that was invariably present somehow did not seem important to many early users, and many
early recordings may reproduce data signals which resemble the
Hutter characteristics of the recorder more closely than they do the
characteristics of the instrument from which the data was acquired.
Only recently has the subtle damage which this broadening of spectrum
can do been fully understood and an intensive effort made to elIminate
it (Ratner [1965]).
It was not long before the instrumentation user demanded recording
equipment with better signal-to-noise ratio and with dc response. FM
recording on the tape was then introduced to supplement direct or
analog recording. Even with the relatively high flutter of early
instrumentation recorders, FM recording offered advantages in signalto-noise ratio and permitted obtaming dc response. Broadband flutter
was, of course, the major limitation on the utility of FM recordmg,
and flutter compensation techniques were soon devised (Peshel
[1957'] ) . These techmques removed a good deal of the noise produced
by the flutter but left the initial flutter in the recovered data. It is
only quite recently that this data flutter has received major attention.
Flutter reduction has been the mam mechanical design goal of instrumentation recorder designers. The original open-loop tape handling
system borrowed from audio recording soon began to give place to
so-called "tight-loop" tape drives which minimized the amount of
unsupported tape and the consequent high-frequency flutter problems
and provided improved isolation of the uniform tape motion from
external disturbances (Schoebel [1957]).
As the use of instrumentation recording broadened, Instrumentation
engineers adapted new kinds of modulation to their systems for telemetering data. They became interested therefore in recording these
new kinds of modulation. The tape recorder had to be able to deal
with pulse amplitude modulation, pulse width modulation, and pulse
position modulation as well as frequency modulation. More recently,
pulse code modulation has acquired great importance for telemetering
precise data, and more sophisticated pulse schemes such as pulse
frequency modulation and single-sideband frequency modulation have
come into use. The characteristics of instrumentation recorders had
to be modified to deal adequately WIth such modulation schemes, and
the instrumentation recorder has gradually become a complex modular
assembly made up of a basic tape moving mechanism plus an almost
lImitless number of plug-in units to adapt the recorder to the many
types of signals it must handle.
THE FIELD OF
MAGNETIC
RECORDING
9
TRANSVERSE RECORDING
While in the mIddle 1950's sound recording for entertainment was
developing in a straightforward nonspectacular way and the performance of instrumentation recorders was likewise slowly improving, the
rotary-head recorder, developed to record entertainment television
signals, suddenly- appeared on the scene and made available greatly
increa.c;ed recording capability (Ginsburg [195'7] ). ThIs recorder used
transverse recording paths across a WIde tape in order to obtain hIgh
head/tape speed without correspondingly high longitudinal tape
speed. Because the signal recovered from the relatively narrow track
used in this recorder was nonuniform in amplitude, a frequency modulation scheme was used. This modulation system violated all the normal rules of FM and could almost be proved by engineering calculations to be impossible (Anderson [1957]). However, the particular
form of signal distortion-wliicnit produced did little-damage totne visual effect of a television picture reproduced from such a modulation
scheme. Since this visual effect was the ultimate criterion for evaluating this recording technique in its mitial application, this "impossible" recording mode was quite successful. It made possible recordings with a bandwidth of greater than four megacycles and the
instrumentation engineer soon attempted to apply this recorder to his
requirements. In these more severe applications, the peculiar transfer
characteristic of the "video recorder," quite satisfactory for television
use, made it a relatively poor analog recorder but quite a good pulse
recorder.
Abo~t the time that the video recorder became available for instrumentation use, interest developed in so-called predetection recording
(Klokow and Kortman [1960]). In predetection recording a data
signal is intercepted in the telemetry receiver IF before it has been
passed through the final demodulator and is heterodyned down into a
band which can be recorded directly on an instrumentation recorder.
In simplest terms this scheme has the advantage that the instrumentation engineer gets a "second chance" in his choice of the demodulation
mode with which he will recover his final data signal. When the
received signal is marginal in quality it gives him an opportunity to
derive the optimum amount of information by optimizing his second
detector. Since frequency modulation is almost universally employed
for the RF transmission link of telemetered data, the predetection
recording is made from an FM carrier. The signal-to-noise ratio
improvement of FM over AM then takes place after pla,yback and the
requirements on signal-to-noise ratio in the recorder are therefore not
severe. Relatively low signal-to-noise ratios in predetection recorders
are quite useful. More recently, the predetection technique has been
10
MAGNETIC TAPE
RECORDING
applied where no existing recorder has the necessary characteristics to
record the post-detection signal, i.e., in the case of single sideband
frequency modulation of data onto the multiplex input to the telemetering transmitter.
Longitudinal tape speeds have been pushed up to 120 inches per
second and the number of cycles recordable per inch has been greatly
increased, and predetection recording is now done with both the rotary
head and longitudinal recorder (Riley [1962]). The difficulty of
eliminating the tIme base instability in the rotary-head recorder, resulting from switching its rotatmg heads four times per head drum turn,
has somewhat delayed its acceptance for predetection recording. The
linear recorder, now able at 120 inches per second to record a 1%
megacycle bandwidth, has almost monopolized the predetection field.
The wider bandwidth of the rotary-head recorder is essential, however,
for some predetection applications, and with newly developed timestabilization and slow-switching techniques it now provides the best
overall time stability of any recorder (-+-25 nsec) (Ampex [1964]).
DIGITAL RECORDING
The term "digital recording" is used here, in a somewhat inaccurate
sense which it has acquired through extensive use, to cover magnetic
recording in equipment peripheral to digital computers. "Digital
recording" in the more general sense, i.e., the recording and reproduction of pulses which represent numbers in the typical digital computer format, is widely used in instrumentation recording. However,
the digital recorder, in the sense of this particular subtopic, provides
mass memory for a digital computer.
Since the digital computer operates on the prinCIple that all the
data with which it deals is in the form of binary numbers, every
binary digit of each number has almost equal importance for the
accuracy of the overall results of the computation. Although errorchecking and error-correcting techniques are available, the data stored
m the memories of a computer must be essentIally perfectly accurate
if the computer IS to be successful. When magnetic tape recording
is used for the mass memory functIOn, it is required to conform to this
rigid standard of accuracy. The design of a tape recorder for association with a computer must obviously be extremely conservative; the
density with which digital computer data is recorded has lagged by
5 or 10 to 1 below the density used for other recording applications.
When the need for extensive mass memory for computers first became apparent, magnetic recording on drums and disks soon was
ttpplied to this service, and continues to be so applied to this day.
The drum or disk produces a continuous flow of data, but any par-
THE
FIELD
OF MAGNETIC
RECORDING
11
ticular piece of data is not accessible at any particular instant; it
becomes accessible when the drum or dIsk has turned. Computer
systems and static electronic submemories were designed to complement drum and disk memories and deal with this inherent accessIbility
delay. Part of the requirements of such rotary memories were carried over to initial applications of tape memories. The tape memory
of the recorder of a modern digital computer is typically used to
"dump" relatively large masses of data either into a static electronic
memory or into a subsidiary drum or disk memory, and, similarly, to
accept a batch of data from one of the submemories in a relatively
sporadic operation. An essential requirement of a computer tape
transport is therefore the ability to start and stop very rapidly on
command. A modern computer may operate at a bit rate from about
100 kilobits per second up to a megabit per second or higher. A delay
in- starting-of-a millisecon-d or two--on-the part-of a tape transport
represents the passage of a long time in computer operation. Digital
tape recorders for computer use therefore typically start m between
ljz and 10 milliseconds and stop in about the same length of time. The
rapid stop is essential to effiCIent utilization of the tape area, since a
longer stop period means that a larger area of the tape is not available
for recording data.
Fast start/stop transport mechanisms for digital computers usually
use storage columns in which the tape is retamed by vacuum or, in
some transports, by an array of rollers mounted on light movable
spring-loaded arms which store a few feet of tape. These storage
mechanisms provide a supply of tape to the heads and capstans while
the relatively massive reels are being accelerated or take up the tape
while they are being decelerated.
Several tracks (usually 7 to 16) are recorded across the width of
digital computer tapes and, as the density of the recording on the tape
goes higher, it becomes more difficult to maintain the proper time
relationship between data spread across the various tracks. Fixed
misalignment between head and tape produces "static skew," and other
than pedect guiding of the tape as it passes the head results in "dynamic skew." Both effects damage the tIme relationships between
the data on the several tracks. The reduction of dynamic skew remains a major problem in computer tape transports. The problem
is so severe that formats are often designed to avoid insofar as possible
requiring association of data distributed across the tape, and to favor
distributing the data in a given "word" serially along an individual
track. Complex local storage buffers have been designed to permit
the data from various tracks to be stored locally at the relatively
irregular rate at which it may arrive from the far-from-flutter-free
12
MAGNETIC TAPE RECORDING
tape transport so that, at the other end of the buffer, the local computer
clock can move the data out at the precise rate the computer requires
(Gabor [1960]).
AIRBORNE RECORDING
The term "airborne recording" is used here to cover the applIcation
of all those recorders, usually rather small and light, which must perform on aircraft, missiles, or satellites. Such recorders must operate
unattended with great reliability and survive rather unfriendly environment. They generally have a balance between design factors
quite different from those of equipment used in friendly environments
on the ground. The term "airborne" is often not quite accurate because in quite a few applications recorders of this class do not fly in a
vehicle of some sort but must meet all of the other requirements typical
of "flying" conditions. The term is chosen for convenience rather
than accuracy.
It was initially extremely difficult to provide any kmd of reliable
performance in unfriendly environments and early airborne recorders
were extremely crude. But recorders for such programs now as Nimbus, OGD and OSO are quite impressive performers even by groundbased standards. Such superior performance is, however, the exception, and a deliberate choice is often made to minimize the performance
requirements placed on the airborne recorder at the cost of placing
more severe requirements on the ground recorder which will receive
the playback from the airborne unit. Compensating means sometimes
are also provided for correcting errors produced in the airborne unit
to produce an overall data transmission link of a quality impossible to
achieve otherwise.
THE SCOPE OF THIS SURVEY
The primary development of tape recorder technology for NASA
use has been in the area of the small airborne high-reliability recorder.
At the same time, NASA has been a major customer of ground-based
equipment and NASA engineers and contractors have devised elaborate systems for sophisticated application of such ground-based
equipment. In a deliberate and arbitrary way the scope of this
survey is limited to these two fields.
By definition, therefore, the "high-fidelity" audio recorder is eliminated, but where audio recording is associated with Instrumentation
recording In the form of a voice monitor or a cue track it is included,
at least by reference. That a bio-medical recorder for use in the
Gemini program happens also to carry a voice track does not elIminate
It from the survey. Nor are certain instrumentation recorders which
can also be used for audio service eliminated on this arbitrary basis.
THE FIELD
OF MAGNETIC RECORDING
13
The dIgital recorder, In the sense of the fast-start/stop recorder for
computer peripheral use, is specifically eliminated as being beyond
the scope of the survey. Most of the technology of such recorders,
except for the fast-start/stop mechanism itself is, however, covered.
The magnetIC recorders covered in this survey include, naturally,
not only the recorders developed specifically for NASA's uses but
also commercially avaIlable units which NASA has purchased. These
commercially available units include "off the shelf" items in the case
of large ground-based recorders as well as airborne recorders developed
for other services which have been purchased and used by NASA. In
this discussion of recorder types no distinctIOn will be made between
those developed for and those simply purchased by NASA. The
classificatIOn of recorders will be based on technical characteristics
rather than sponsorship.
Recorders employed-by NASA can 'be divided Into two groups
roughly on the basis of size and weight. One group of recorders is
Intended to be installed in a more or less fixed position on the ground.
Such recorders can be large and heavy and are usually provided with
fairly friendly environments during operation. The other large group
of recorders for NASA's applications is made up of those which are
air or space-borne. Such recorders are subjected to unfriendly environments in both operation and nonoperating modes, must be small
and light, and must use very little power. Recorders of this second
class are not accessible for maintenance or for changing the recording
medium. They therefore must be extremely reliable and must provide
operating modes which use and reuse the recording medium very
effectively.
GROUND-BASED RECORDERS
The products of six manufacturers dominate the field of groundbased recorders currently purchased and installed. These recorders
have basic similarities and differ only in certain performance figures
and in the flexibility with which they may be applied to different tasks.
Mechanically, such recorders are typically reel-to-reel devices,
usually taking a full 14-inch diameter reel, and employ a closed-loop
tape metering system (chapter 6). They all employ some form of
tension servo designed to regulate the tension at the entrance (and
sometimes the exit) of the closed loop. These tension servos may
themselves be of the (electrical) open- or closed-loop type and range
from those which determine tension by measuring differential supply
pressure in an air-lubricated turn around post to those which shine
a. light past the tape reel onto a photocell to determine how much tape
remains on the reel.
14
MAGNETIC
TAPE
RECORDING
Typically these machines use torque motors for takeup and for
supply reel holdback. Some use mechanical hrakes for starting and
stopping but others use dynamic braking of the torque motors themselves for dealing with transient condItions. These latter usually
employ some sort of solenoid-operated ''brute force" dog brake to lock
the reels when the recorder power is shut off.
These machines may also be divided into those which do and do not
use differential capstans. Those not using differential capstans use
a single capstan for defining both the exit and the entrance of the
closed metering loop. These machines depend on the maintenance of
entrance and exit tensions to assure tension withm the closed loop but,
as discussed in chapter 6, they are not alone in requiring this condition. The differential-capstan machines are further divided into two
groups. One type employs a "two-diameter" capstan (described
later), with separate pinch rollers causing the tape to touch the larger
or the smaller diameter of the capstan in order effectively to meter
more tape out of the closed loop than is metered in. The other type
of differential-capstan machine employs either two capstans of different diameter driven at the same speed or capstans of the same
diameter driven at slightly different speeds to accomplish the same end.
Most of these machines provide tape lifting facilities of one kind
or another so that the tape does not run across the heads when it is
being moved rapidly forward or rewound. These may either 'be literal
tape lifters which operate to move the tape away from the heads or
may accomplish the same result by moving the heads away from the
tape. The implication of the provision of this feature is that much
shuttling back and forth of the tape is often involved, as it is indeed
for certain applications of such recorders. In a tracking station a
recorder may simply be used to record an original tape which is then
taken off the machine without rewinding since the station procedures
are usually based on minimum local tape handling and minimum use
of the machines which do the essential initial recording. Auxiliary
to the work of such recorders, sometimes in the station, and more often
in data reduction centers, much shuttling back and forth, rewinding
and dubbing of tapes takes place and the tape lifters are useful in
these applications.
Currently available ground machines provide wide variation in
their flutter and time dIsplacement error performance. The (absolute) time displacement error varies at present with commercial machines from plus or minus a quarter millisecond to plus or minus half
It microsecond at a tape speed of 120 inches per second. Two machines
may have similar flutter performance alt.hough they differ to this
degree in time displacement error performance. The 10w-time-dIs-
THE FIELD OF MAGNETIC RECORDING
15
placement-error machines have improved low-frequency flutter but
the high-frequency flutter follows about the same pattern as in other
recorders. The combined flutter of the low-time-error machines therefore is SImilar to that of more conventional machines. The goal in all
recorders tends to be limitation of the amount of unsupported tape in
the vicinity of the heads, since this unsupported tape is generally
believed-to be the source of the high-frequency flutter -which-for many
wideband applications is the significant flutter.
Such machines invariably are fitted with speed-control servos. This
control mechanism can be a rather straightforward device which makes
a record of the precise frequency of the local power at the time that the
recording is made so that the machine can be locked to the local power
, on playback. It may also permit locking a recorder reference tone to
a local crystal OSCIllator at the reproduce point WIthin a half-microsecond on playback. In accomplishing this-wide range of-speed control, the machines use both direct and alternating current motors; at
one time one manufacturer used a number of very small dc motors
to minimize the rotatmg mass, and in another case, a printed circuit
motor is used for the same reason. In general, dc motors seem to be
preferred for tighter speed control.
Starting and stoppmg such recorders is an important problem and
significant differences exist between the various models in the way in
which they treat the tape during such transient conditions. The
modern recorder usually starts in a rather complex way, often bringmg the supply and takeup reels up to speed before the pressure roller
clamps the tape against the capstan in order to mmimize starting
transients. (Some manufacturers emphasize that they do not do
this.) When the start-stop controlling mechanism fails or is misadjusted, errors may be produced in the recordings and the tape
damaged may be beyond repair. The user's choice between recorders
often is made on the basis of how well the individual unit deals with
the start-stop condition rather than on some of the numbers in the
overall specification.
Although many tape recorders still in use on the ground have vacuum-tube electronics, all those currently supplied for the more sophisticated services are entirely solid state. Occasionally tubes may be
found in the servo motor-drive amplifiers of a machine which is otherwise solid state.
As described later (chapter 4), there are many different record
and playback "modes," almost everyone of whICh requires its individual electronics assembly. Broadband FM, direct record, PAM,
PDM, PCM, and predetection recording modes are all encountered in
modern installations. In tracking stations it is common for each of
16
MAGNETIC TAPE RECORDING
these modes to be provided for almost every track of multitrack recorders since such tracking stations have to deal with the output of a
wide range of satellites with a correspondingly wide range of recording ~equirements. In data reduction statIOns, more limited flexibility may be provided, although in some central data handling
Installations the full range of normal modes is required and is extended by special playback modes required by peculiar recordIng
conditions. This latter condition is particularly true where data is
recovered by playback from an airborne unit and is to be reduced on
the ground. It is a major feature in such recorders for most of their
applications that they be easily and logically convert~d from one
mode of operation to another and that several tracks on the same tape
can successfully be used simultaneously In different modes.
Ground-based recorders are subjected to routine preventIve maintenance and to continuous calibration and checkout procedures. Important features of such recorders are long head wear and uniformity
and stabIlity of calibration. Difficulty in providIng these features may
not be crippling, however, since in a particular situation an expensive
and elaborate maintenance procedure may be worth while if it makes
available an otherwise unavailable special recording mode. Typically
such machines operate in rooms in which human beings are reasonably
comfortable and, although they may be subjected to dampness or dust,
the environment is usually similar to that of the laboratory In which
the equipment was developed.
FLIGHT RECORDERS
TypIcally the flight recorder is small, light, and uses very little
power. These characteristics outweigh the importance of most of
the electrical performance features which are significant in groundbased equipment. The exact specifications of a spaceborne recorder
usually are determined by deciding how much the miniature unit can
be simplified at the price of elaborate methods of processing its output on the ground. Flutter, for example, may be accepted in a flight
recorder with the specific intent of using elaborate flutter compensation at some later process in the reduction of the data from the
recorder.
The tape in a flight recorder may be transported from reel-to-reel
or supplied from and taken up into an endless loop. Although the
reel-to-reel configuration has fewer uncertain mechanical problems, it
usually limits the playback modes available and requires relatively
complex control techniques. Most endless-loop recorders are so constructed that continuous slippage takes place between many layers of
tape in the tape pack. Reel-to-reel recorders often have been used
THE FIELD
OF MAGNETIC
RECORDING
17
where an endless-loop machme was more logical because of the potential unreliability of the loop device.
Flutter in flight recorders may be acceptable when it is up to 10 or
20 times as severe as in ground-based equipment. Either providing
enough mass for straightforward flutter reduction techniques or a refined enough mechanical filter or servo control mechanism usually
adds too much undesirable -weight or complexity for adequate reliability in unmaintained equipment. One percent peak-to-peak flutter
is perfectly acceptable for a flight recorder. The usually severe environment of the flight recorder also requires that such items as pressure rollers or resilient elements be used with care because of problems
of failure of the specialized materials involved. Many flight recorders
even eliminate the capstan pressure roller entirely by providing a
large wrap around the capstan.
__There_are_two_major_classesofspacebornerecorders. One class is
used in satellites where the typical application IS to obtain data during
an entire earth orbit and to return it to a receiving ground station during the relatively short time when the satellite is in radio view of the
ground station. These recorders therefore have relatively slow record
and fast playback characteristics. The other class of recorder, used
mainly for deep space probes, operates in exactly the opposite way.
Data is recorded in real time at a fairly conventional rate but is played
back at an extremely slow rate because of the bandwidth limitations of
transmission over interplanetary distances. The high ratio of record
to playback speeds raises both electronic and mechanical problems in
this application. For some space applications, where the playback
speed may be one one-thousandth of the record speed, the problem is
particularly severe.
The most important single characteristic that distinguishes the flight
recorder from one based on the ground, in the current period of relatively large boosters and somewhat relaxed weight and power requirements for such machmes, is the extreme reliability required.
A recorder that is to be used for space or satellite probe application
must usually have an unattended failure-free life of at least 1 year.
The problems of mechanical reliability under these conditions have
proven to be the most severe to overcome.
A secondary class of flight recorders includes those which do not play
back on command to produce signals to be telemetered to 2. ground station but in which the tape itself, along with the recorder, is recovered.
These fall roughly into two groups. In studying reentry conditions
t.he recorder travels with a reentry test object and records what happens during the severe accelerations and decelerations of reentry.
Such a recorder must be able to wit.hstand a severe mechanical en-
18
MAGNETIC TAPE RECORDING
vironment and survive so that the tape may be recovered. A related
application is one in which transmission is blacked out by flame attenuation of boost~rs or separating rockets during part of a research
investigation. To obtain the data that normally would be transmitted
via radio waves to the ground during this blackout condition, a recorder records continuously and reproduces continuously, the reproduced signal having a fixed delay relative to the recOrded signal. By
recording both the original and delayed signals on the ground, a
flame-attenuation blackout of 10 to 50 seconds would not cause any
data to be lost. The recorder and tape do not, of course, have to be
recovered, but the construction must be as rugged as that of a recoverable unit.
Another group of recorders producing ,recoverable tape is associated
with manned space flights. These recorders have to survive a somewhat rugged environment but, since they travel with a man, they are
usually treated little worse than the man. They must be reliable and
have long playing times to produce archival records over an exrended
mission of information otherwise transmitted directly. Beyond this,
the requirements on these units are not too severe.
CHAPTER 3
The Elements of the Tape Recorder
Certain common elements are basic to the operation of every magnetic -tape recorder. These are a re~ording medium,' recording and
reproducing transducers, a mechamsm for moving the medium past
the transducers, and electronic devices which process the input and
output signals (and sometimes control the tape-moving mechanism).
Each element affects the performance characteristics of the complete
recorder, and the influence of each element interacts wIth that of the
others.
THE RECORDING MEDIUM
Although several forms of magnetic recording media are currently in use, one form monopolizes most applications. This medium
consists of a thin plastie backing or base on which is coated microscopic magnetic oxide particles dispersed in and bound to the base
by a thermoplastic or partially thermosettrng binder. The action of
this form of medium will be discussed in detail (chapter 9) ; that of
the less common forms will be reviewed briefly in relationship to the
rather specialized applications to which these forms have been applied.
Other media of some importance include metallic nickel-cobalt layers
electroplated or electroless-deposited on a "base metal" carrier, usually
of phosphor bronze or similar material, and metallic coatings of the
same general type placed on a plastic base.
When the oxide-particle recording medium moves past the magnetic
field of the recording head, each of the very large number of particles
in the magnetic coating is somewhat differently affected by the recording field. When a particular section of the medIUm has moved away
from the record field, remanent magnetization is left in the medium.
Just as the magnetic material is made up of many different particles,
the remanent magnetization is made up of the magnetic effect of a
large number of individually magnetized particles. When this com-
19
20
MAGNETIC TAPE RECORDING
posite remanent field is passed over the intercepting gap of the reproducing head, each magnetized particle has its individual effect in
inducing flux into the reproduce head and, hence, signals into the
recorder output. The net process is thus one of transmitting the
signal from the input of the recorder to the output of the reproducer
in the form of the integrated influences of a very large number of
individual particles. That the coating is particulate rather than continuous is thus important to the recording process and the total number of particles is a significant operating parameter. Interaction
between the large number of individual particles also makes the operation of the medium complex.
In most information transmission systems a certain number of
samples (electrons, film grains, magnetic particles) proportional to
the average instantaneous value of a signal is transmitted. The number of such samples received is subject to a statistical random variation around the average to an extent dependent on the number of samples involved. This is an elaborate way of saying that in the magnetic
recording system, as in any other discrete-sample (electron, grain,
particle) system, the fundamental signal-to-noise ratio of the received
signal depends on the number of samples transmItted, or in this case,
on the number of magnetic particles involved (Schade [1948]). This
signal-to-noise ratio is roughly proportional to the square root of the
number of particles. There are, of course, other sources of noise beside
thlS basiC one (Mee [1964]), but in general, the more particles in the
magnetic medium, the better the signal-to-noise ratio.
An important qualification to that last statement is: "all other
thmgs being equal." When particle size changes, almost all the other
magnetic properties of the particle change. The total remanent flux
and the magnetic stabIlity of the medium decrease if one simply
changes particle size without changing anything else (same total
volume of magnetic material). The art of making the modern magnetIc particle dispersion includes, among other things: (1) obtaining
enough total remanent magnetism; (2) having as large a number of
particles as possible; (3) holding the particles mechanically firmly
on the base; (4) retaining other important magnetic characterishcs
such as "squareness ratio" and high saturation magnetization; and
( 5) making the particle dispersion perfectly umform (see chapter 9) .
The tools available to the magnetic medium (tape) designer mclude
making minor changes in the chemical (and hence magnetic) properties of the particles as well as changing their size. The exact nature
of the oxides used by the various tape manuIacturer3 IS probably
guarded by them more carefully than any other trade secret. It will
THE ELEMENTS OF THE ToAPE RECORDER
21
not be possible to discuss other than the crudest outlines of the influences at work in this field because most knowledge is maintained on
a completely proprietary basis.
In any group of very fine particles the same agttating influences
which lead to the familiar Brownian Movement are at work. There is
thus a certain thermal energy contained in the fine magnetic particles
of a tape. If the particles become smaller, the amount of thermal
energy of each particle eventually becomes greater than its magnetic
energy. Under these circumstances although a particle can be magnetized and aligned in a magnetic field, it immediately reverts to
random alignment when the field is removed (superparamagnetism)
(Mee [1964]). With partICles this fine, the tape would not be able
to retain any remanent magnetism even at room temperature. Before
reaching this extreme state of instability, there are intermediate states
for intermediate particle sizes where only a slight increase in temperature or a slIght amount of mechanical work (bending around a
mechanical guide) will affect the remanent magnetism of a recorded
tape, a situation which is not tolerable for precislOn recording and
reproducing.
For proper recording and reproductlOn to take place, the magnetic
medium must come into close contact with the record and reproduce
transducers. This means that the surface of the tape must be extremely smooth and the actual magnetic material must not be shielded
from the transducers by a perceptible layer of binder. At the same
time, the tape must be mechanically strong enough to retain the magnetic materIal in position when subject to frictional movement across
the head. The smoothness of the tape and head surfaces also controls
the uniformity of speed with which the mass of magnetic particles is
moved past the transducer. If either surface is rough, the medium
will fail to contact the head and the relative movement of the medlUm
past the head will tend to be irregular. Therefore surface properties
are also SIgnificant in determming the signal-to-noise ratio.
Action affectmg the signal-to-noise ratio thus occurs at the surface
of the tape, or more exactly, the surface at which the tape and the
transducers interact. WIthin the tape an analog to surface smoothness, that is, the uniformity of dispersion of the magnetic particles,
similarly affects signal-to-noise ratio. The gaps of the recording and
reproducing transducers deliberately introduce sharp discontinuities
m the external effects of the magnetic properties of these transducers.
At these points of sharp discontinuity the detailed structure of the
medium mteracts with the transducers. N onulllformity in the detailed structure of the tape, that is, of the uniformity of dispersion and,
hence, microscopic uniformity of magnetic properties, is examined
22
MAGNETIC TAPE
RECORDING
in the recording and reproducing process. Hence, although made of
particles the tape must, in a sense, act as if it were perfectly uniform
for additional noise not to be created in the record and reproduce
process.
The factors discussed so far seem concentrated on the signal-tonOIse ratio. Suppose, now, that the thickness of the medium were
increased. Assuming their size is maintained the same, the number
of particles involved in recording would also be increased. From
what was said above this would appear to increase the signal-to-noise
ratio. It would do so, but only for long wavelengths, that is, for
signals reqUIring low recording resolution. To record with high
resolutIOn, that is, with good mechanical definition on the tape, the
influences of the recording and reproducing transducers must be
limited to short distances along the direction of tape motion. As we
will see elsewhere, increasing the distance of portions of the medium
from the transducer decreases the definition with which the transducer
is able to specify the record and reproduce points for those portions
of the medium (Eldridge [1960]). This happens in the parts of the
thick coating far from the transducer. Therefore, for good performance at both short and long wavelengths, the tape coating must be as
thin as possible and still produce a large enough reproduce signal.
Desirable characteristics of the medium are, then, that It have as
large a number of partIcles per unit volume in the coating as possihle,
that the coatmg be physically as smooth as possible, that the dISpersion of particles within the coating be very uniform, and that the
magnetic properties of the material be such that as thin a coatmg as
possible can be used for a given signal. The importance of these properties of the medIUm is only slightly influenced by the properties of
other parts of the record system.
RECORDING AND REPRODUCING TRANSDUCERS
Except for very special applications, most magnetic tape recordmg
and reproducing transducers in use today are of one single desIgn.
The recording transducer or head consists of a closed magnetic circuit,
usually made of laminated high-permeability metal, in which a gap of
controlled dimensions is provided. On this magnetic circuit are
wound coils for inducing a magnetomotive force in the circuit. A
typical recording head is made of laminations roughly of the shape of
the letter "Ot placed together tip to tIp to produce a structure which
usually looks lIke an 0 with flattened sides and a rounded top and
bottom. One of the two gaps in the structure IS made as small as
pOSSIble and the other is made of controlled dimensions. TIllS assembly produces a {Tinging flux at the controlled gap which has elements
along the directIOn of tape movement past the gap and hence is able
THE ELEMENTS OF THE T.APE RECORDER
23
to pl'oduce a longitudinal remanent magnetization in the tape. The
physical forms of such record heads may vary widely but the essential
elements will be the same (chapter 8).
The typical reproduce head of the modern tape recorder is sensitive
to rate of change of flux or dcp/dt. The magnetic and electrical structure of such a head is esssentially identlcal.to that of the record head.
The operatmg gap in the reproduce head, however, is usually conSIderably smaller than that m the record head, and the windings of the
coil are designed to match the input Impedance of reproducmg amplifiers rather than the output impedance of the recording drivers.
When the gap of such a head encounters the flux pattern recorded on
the tape, flux passes into the magnetic circuit and change III that flux
induces voltage in windings, hence dcp/dt.
SpeCIal reproduce heads are sometimes used. They may, for example, be sensitive to the value of the flux rather than to its rate of
change. These are particularly useful where the tape moves so slowly
on reproductIOn that very lIttle voltage is induced by a change in flux.
Such flux-sensitive heads include those in which the presence of the
flux from the tape m the front gap modulates the reluctance of a partiaUy-saturated magnetic circuit supplied with an external flux source
so as to change the total amount of flux in the circuit. The external
flux applied is usually alternating and the output signal is thus a
carrier modulated by the tape flux. Many variations of this scheme
have been used. Another common flux-sensitive head is made by introducing a piece of semiconductor material particularly sensitive to
the RaIl effect into either the front or the back gap of the head. These
specialized forms of heads will be considered later.
An important characteristic of the common heads just descrIbed
is that they define sharply the fringing flux field at the point where
they contact the tape. This requires extreme precision and accuracy
of mechanical construction. Typically, the mating faces of head gaps
are lapped to optical finishes as are the rounded surfaces of the front
of the head which contaots the tape. The head structure is usually
made of a ferrite or of metal laminations from 2 to 6 IDlls in thICkness.
Ferrites are particularly hard and, if they are properly constructed of
proper materIals, have a very long wear life. Ferrites also become
particularly useful at relatively high frequencies where losses in most
magnetic metals increase (above 2 megacycles per second).
The lines defining the sides of the gap of a preCISIOn head must be
perfectly parallel and perfectly straight to produce good resolution
and maximum utIlization of the recording medium. ThIS accuracy
must be accomplished with magnetic metals which tend to be rather
soft and to gall. Techniques of constructIOn of such precIsion heads
78lH128 0---65---3
24
MAGNETIC TAPE
RECORDING
have been greatly refined in the last few years; heads of almost any desired mechanical characteristics can now be made, although this was
not true 10 years ago.
The accuracy with which a head interacts with the tape is determined largely by the accuracy of construction and the finish of the
front of the magnetic head. The head also must not vibrate and must
be stiff enough so that it does not, by moving, become a part of the
problems of maintaining uniform relative head-tape motion.
THE TAPE MOVING MECHANISM
The function of the tape moving mechanism is to move the magnetic medium in a smooth and uniform manner past the recording and
reproducing transducers. The designer's problems include both preventing irregularities of motion originating within the mechanism itself and protectmg the tape motion from external influences which
may tend to force the mechanism to move in an irregular manner. To
perform these functions properly the tape moving process is usually
separated into two steps. One step is that of supplymg the tape from
a supply reel and windmg it up on a takeup reel, referred to here as
"Tape Reeling." The second step is that of metenng the movement of
the tape past the transducers with umformity, referred to as "Tape
Metering." The design of the complete mechanism involves designing as smooth a metermg mechanism as poSSI'ble and providing a reeling mechanism which protects the movement within the metering
mechanism from disturbing mfluences originating in the supply and
takeup process or outside the recorder proper.
All parts of the tape moving mechamsm, with emphasis on those
charged with metering the tape, must be designed to minimize the
typical irregulanties of mechanical devices. Such irregularities are
nonuniformity of rotation from tooth effects in gears, motion jerks
from splices m drive belts, nonuniformity of torque with rotation in
bearings, out-of-roundness or runout in rotating members, and chatter
or stick-slip problems everywhere in the mechanism. In addition,
llonuniformity in drive from the prime mover, originatmg in cogging
in alternating current motors and commutator effects in direct current
motors, must be minimized.
In the takeup and supply mechanism, lumped together in the "reeling" function, other sources of irregularIty of motion are encountered.
Typically, elements of the reeling mechanism move more slowly and
have lower rotational rates than those withm the metering mechanism, and they often include frictional devices such as clutches or
brakes. The supply and takeup reels are often not part of the recorder but are brought to the recorder after a history of handling
THE ELEMENTS OF THE ~APE RECORDER
25
which is not under the recorder designer's control. Reels which have
become warped or bent can seriously influence the operation of the
recorder.
lib many flight recorders s an endless loop replaces the supply and
takeup reel. Such devices have entirely different motion irregularity
problems, resulting primarily from the requirement that the layers of
tape slide continuously past each other without chatter or hang-up.
Errors in any of the elements ofthe tape moving mechanism will, of
course, produce irregularities in the time scale of the reproduced
signal. In effect, such irregularities add a peculiar kind of noise to the
overall transmission. The significance of this noise will be discussed
in detail later. In addition to the obvious first order effeot of the
signal not coming out at the expected time, there are second order effects of modulation of the time scale which produce spurious signals
particularly difficul,t to separate from the desired SIgnals.
A particular point of interaction between the medium and the moving mechanism is the generation of high-frequency random speed variation at the point of friction between the medium and the transducer.
Wherever the tape is not continuously supported and, at one end or
the other of its unsupported section, passes with friction over some
portion of the mechanism, any lack of smoothness of the tape or of the
device over WhICh it passes will tend to shock-excite the unsupported
tape into both lateral and longitudinal vibration. The effect of this
vibration is to produce a type of random speed variation with a broad
irregular spectrum which creates a kind of noise particularly difficult
to deal with. It is thus essential that the mechanism designer make
all parts which contact the tape as smooth as possible and minimize the
amount of unsupported tape in the tape path he chooses. Air lubrication to support the tape completely free of frictional movement is a
powerful technique for dealing with this frictional excitation problem.
In the simplest tape recorder mechanisms the metering mechanism
may be driven by a governor-controlled or synchronous motor and
straightforward techniques may be used to provide the varying
speed-torque characteristics necessary for the supply and takeup reels.
In high performance recorders, however, some sort of servo control of
the tape speed is almost universal. At this point there is interaction
between the electronic elements of the recorder and the mechanical
elements. For the most sophisticated servo speed controls, interaction
takes place between the medium and the electronic and mechanical
elements of the recorder. (This speed· control interaction is supplementary to the fundamental mechanism/medium/electronic interaction taking place at the transducer.)
26
MAGNETIC TAPE RECORDING
Speed control servos range from simple to complex. The simplest
servo control is used in connection with the synchronous motor drive
of the recorder and involves the recording of the frequency of the supply mains power by modulating it onto a carrier of the tape. This
carrier is played back while reproducing and the power frequ\ncy
derived therefrom is compared with the frequency of the reproducing
power source. Compensation is then provided to lock the two power
systems together. This type of servo guarantees that the average speed
of the recorder and reproducer are identical and therefore that there
will be no gross time error (Davies [1954]). It makes no compensation, however, for instantaneous speed variations. More sophisticated
servos have recently come into use in which an approximately 100 kilocycle SIgnal is recorded at the same time as the desired information
signal. This 100 kc signal is played back in reproduction and is compared with a local crystal oscillator-generated signal. The instantaneous error between the timing of these two signals operates in a very
fast-acting servo loop so as to minimize the time displacement between
the recorded 100 kc signal and the locally generated signal (Schulze
[1962] ). Within the last 5 years, the capabilities of such precision
servos have progressed from providing a gross time displacement of
plus or minus 5 microseconds at the normal maximum tape speed to
current time displacement accuracy of a half-microsecond.
The discussion so far has been limited to the characteristics and
performance effects of the tape moving mechanisms of essentially
isolated recorders. Only indirect reference has been made to such
external disturbing influences as variation in the supply power, voltage or frequency, mechanical shock and vibratIOn. Large groundbased recorders do in practice operate in an environment which is
essenHally isolated. However, in many important applications, recorders are subject to extremely hostile enVIronments. The tape moving mechanism then must be able to perform its basic function without
being seriously affected by such an environment.
Such a tape moving mechanism must not be seriously affected by
wide temperature variation or thermal shock and also must continue
to move the tape smoothly under severe mechanical shock and vibration. The design of such mechanisms often involves avoiding design
solutions which would be satisfactory if the environment were not
severe but which may magnify vibration effects or react catastrophicn.lly to shock. Specific design factors will be dIscussed in greater
detail later (chapter 11). The severe environment mechanical design decisions may, however, be characterized by citing a few specifics.
A formal decision is often necessary between shock mounting and
hard mounting the entire recorder, depending on the nature of the
THE ELEMENTS OF THE ~APE RECORDER
27
vibration environment and the structure of the device in which the
recorder is mounted. The design of a capstan bearing assembly, a
relatively straightforward process for a ground-based recorder, involves for a shock-environment recorder discarding many potentially
useful ball bearings because they cannot survive shock; reevaluating
bearing loads and load life, and perhaps discarding whole system
designs because the recorder must also operate in a vacuum with its
lubrication limitations.
Although vibration and temperature problems may cause difficulties
with electronics it is in the mechanical elements of the recorder that the
severe environment raises the most difficult design problems.
ELECTRONICS
The electronics of a magnetic tape recorder must conform generally
only to the electronic requirements of other typical electronic equipment for similar environments. In only a few specific areas is specialized electronic performance required for the recorder. These include
equalization and mput amplifier signal-to-noise ratio.
The recording and reproducing process usually involves laying down
a magnetic pattern on the tape, the intensity of which is roughly proportional to the intensity of the current passing through the recording
head. A reproducing device sensitive to the flux in the tape would
then approximately reproduce, except for secondary effects, an output
signal identical to that applied to the recording head. The most commonly used reproduce transducer is, however, a device sensitive not to
flux but to rate of change of flux. The output signal therefore tends
to be the time derivative of the record signal. The signal emerging
from the reproduce transducer must thus be integrated in order to
recbnstruct the recorded signal in the output. In the linear (analog)
recording mode, and in many nonlmear recording modes, this integration of the signal takes place in the electronics. It is often accompanied by the post-emphasis or equalization needed to compensate for
wavelength (and hence frequency) signal intensity losses occurring in
recording and reproduction as well as to remove the effect of preemphasis applied to improve signal-to-noise ratio.
Both recording and reproduction are scanning processes m which
the mechanical characteristics of the transducers interact with those of
the medium to produce typical wavelength-sensitive responses. As
in all scanning processes, both recorded and reproduced signals tend
to decrease in amplitude Wlth decreasing wavelength on the tape.
Again, as in all scanning processes, there are actual nulls in the system
response where the effective gap dimensions of transducers bear certain relationships to the wavelengths of signals recorded on the tape
(Westmijze [1953]). Equalization to compensate for these mechani-
28
MAGNETIC TAPE
RECORDING
cal effects must accompany the integrating process if the signal is to
be restored to its original form.
Any equalIzation process equalizes the noise generated both in the
tape and in the early stages of the electronic equipment along with
the desired signal. Magnetic recorders therefore have rather complex
and nonuniform output noise spectra. Combined with the degradation
of signal-to-noise ratio produced by the nonuniform spectra is the
normal tendency, as the storage capacity of the medium is pushed
farther and farther, for the signals reproduced to become smaller and
smaller. The net result is that the input amplifier which handles the
signal from a reproducing transducer in a magnetic recorder must
deal with a very severe signal-to-noise problem. In recorders operating at the limIts of the state of the art, the noise capabIlity of the input
amplifier rather than the noise of the recording medIUm may determine
the system noise performance. Many a recorder manufacturer has
been embarrassed when he found he could not take advantage of an
improvement in the recording medIUm because the nOIse m the mput
circuits of his reproducing eqUIpment masked the improvement in the
characteristics of the medium.
The interrelationship between the medIUm nOIse and system noise IS
quite complex and will be discussed later (chapters 5, 9, 10). Where
precise control of the movement of the tape is obtained WIth a servomechanism this relationship is further complicated. Relatively fastacting servos have recently been introduced to magnetic recorders to
('ontrol the motion of the tape, particularly on reproduction, by attempting to minimize electrically the time error in a reference signal
recorded on the tape. The degree to WhICh this time error can be
reduced is a measure of how much the overall performance of a recording system which depends not only on amplitude but also on time
accuracy can be Improved. Studies have recently shown that there is
a practical and theoretical maximum to the degree to which thIS improvement can be accomplished (Develet [1964]). This limit on
improvement is typical of many such situations in which error compensating schemes, for which the Improvement available has no theoretical limit, reach a practical limit when noise generated either in
the original system or in the correcting system succeeds in confusing
the compensation process.
Although electronic elements are essentIal to the transducing of signals relative to magnetic tape, the magnetic recording process is basically a mechanical one. The complete process generates an electrical
signal which depends on the combined effects of mechanically dispersed separate partIcles which contribute, dependmg upon their mechamcal position and the smoothness of their mechanical movement,
THE ELEMENTS OF THE TAPE RECORDER
29
to the magnetic flux intercepted by a mechamcal gap in a reproducing
transducer. Therefore, except for improvement in input amplifiers,
t.o which improvement there is a theoretical upper limit, progress in
magnetic recorders is made prImarily through improvement in mechanical elements.
SUMMARY
This brief discussion attempts to relate the various elements of the
recording system to show the way in which their qualities interact
with each other. The organization of the analytlC portion of this
Technology Survey is based on treating each of the individual elements
and their interactions more deeply. For this reason several chapters
are labeled wit.h the detailed elements which make up the recorder and
reproducer. To these element chapters are added certain chapters/on
interaction. Specifically, the head/tape mteraction process cannot be
discussed $dequately by considering the head and the tape alone and
therefore the two are discussed in a combmed section. Irregular tape
motion in the form of flutter and time dIsplacement error is related to
both the tlLpe reeling and the tape metering mechanism as well as in
part to hell-d/tape interaction. Therefore a separate section analyzes
tape motion irregularity. The section immediately following this one
is entitled "Recording Methods" and deals with the various ways in
which the basic magnetic recording process can be used to produce
overall system-transmission to optimize certain qualities. This section
clearly partakes of elements of all the other sections. There is also
a brief an!l-lysis of the storage capabilIty of magnetic recorders and a
discussion of the degree to which the theoretical storage ~pability is
currently realized in practical recorders. Beside the analysis of recorder elements and how they effect recorder performance, several sections have been added to make the Survey complete. One of these is
on "Methods of Testing and Evaluating Recorders" and one is on
"Complete Recordmg Systems." This last section deals with the methods which are used by systems engineers to recognize the limitations
of each portion of the system and to desIgn the system so as to mimmize
t.he effect of these limitations. A third section partakes of both analysis and system specIfication m describing the problems and current
solutions peculiar to the airborne miniature tape ,recorder in which
NASA's specific development efforts have been concentrated.
Page intentionally left blank
Page intentionally left blank
CHAPTER 4
Recording Methods
Modern magnetic tape---a dIspersIOn of magnetIc oxide particles
earried on a plastic hae:kmg-was first put into widespread use at the
end of World War II. Some tape recorders used methods of recording with this new type of tape which had been devIsed when the
available magnetic recordmg media were quite different. Recordmg
methods better adapted for wire or for tape rolled from magnetic
metals were thus used in some early tape recorders (Begun [1937]).
The recorded magnetIc field was often perpendicular to the surface
of the tape, or was induced to lie in the plane of the tape by placing
recording head pole PIeceS on opposite sides of the tape. Such older
recording techniques have essentially disappeared from practical use.
"Longitudinal recording," that is, with the remanent flux directed
along the direction of tape motion, induced by fringmg flux surrounding a gap in a closed magnetIC structure in contact with one side
of the medium, is almost universally used. Some research and advanced development work is now being done in other than longitudinal
recording and in ways of achIeving longitudinal recording by other
t han the currently used means ('Mee [1964]).
What is now substituted for the orIginal variety of methods of impressing the magnetic impulses on the tape is a variety of ways in
which the mformation to be record~d is processed before it is placed on
the tape and is reprocessed after reproduction. This processing involves either recording the mput signal directly or converting it to
pulse, FM or other form in a "coding scheme."
The almost universally employed frmgmg recording field is produced by a precisely controlled gap placed in a magnetically soft
structure. By some means, the informatIon to be recorded is caused
to vary the magnetomotive force applied to the magnetIc structure and
t he fringing field surrounding the gap in the structure correspondingly
varIes the magnetic intensity operating on the magnetic material of
31
32
MAGNETIO TAPE REOORDING
the tape. As longitudinal recording densities have increased and
methods like those of magnetic tape recording have been applied to
drum and disc computer files, methods of recording other than by the
use of a -fringing field have been investigated. No great strides have
been made in thIS area but some current work in the use of so-called
"horseshoe heads," as yet unpublished, shows promise.
The various recording methods (coding schemes) used with magnetic tape fall broadly into two categories. One method, which might
be referred to as the "linear" method, operates on the basis that the
transfer characteristic of magnetic tape can be made to be approximately linear. By transfer characteristic is meant the relationship
between the instantaneously applied signal magnetomotive force and
the remanent magnetism in the tape. This is the basic mechanism used
for audio recording which gave tape recording its practical start during W orId War II. Linear recording is achieved by mixing the
recorded signal with a high frequency ibias. In the absence of the bias
the transfer characteristic of magnetic tape is far from linear. By
a somewhat related mechanism called the use of "dc bias," a similar
linear effect can be produced. There is, however, a vast philosophical
difference between the ways in which dc bias linearization and ac bias
linearization take place.
Much of the fundamental or semi fundamental research and advanced development work applied' to the magnetic recording process
has been concerned with explaining and thereby, hopefully, mastering the ac bias process of linearization (Spratt [1964]). Many incomplete and even erroneous explanations of this process have been widely
accepted and publicized primarily because relatively few people have
gone into the subject in any depth. It is now correct to say, however,
that the linearization process is fairly well understood, although the
precise mechamsm cannot be pinpointed in detail. The process is well
enough understood to use a current statement of its mechanism as a tool
for evaluating development of heads, tapes, and head-tape interaction
devices.
The linear recording process was referred to above as being "approximately linear." Of the entire scope of variation of linearized remanent magnetism from maximum positive to maximum negative, only
the center one-third in amplitude is linear enough to be considered essentially free of distortion. The center linear portion is bounded by
two non-linear portions of a shape which is apparently inherent in the
ac biased tape recording process. The distortion introduced in audio
recording by the shape of these curved end portions of the transfer
characteristic is relatively accepta:ble to the human ear. Tape record-
RECORDING ME'l1HODS
33
ing, therefore, acquired an InItial reputation of bemg a "linear process
which overloads gracefully." In a way, this is unfortunate because
the graceful overload is a term properly descrIptive only of audio
recording. In teclmlCal recording the intermodulation distortion
caused by the overload can be disastrcus.
Any method of usmg magnetic tape which depends on the linearity
of the so-called linear portion, and also on limiting the signal excursions to the linear portion, necessarily limits the degree of utilIzation
of the tape. For many applIcations, not very much of the linear portion is adequately linear enough, and other recording modes have had
to be devised. These recording modes do not in general depend on
linear propertIes of the tape but solely on its ability to indicate that
it is magnetized in one direction or another or not magnetized at all.
Such recording modes, to which the term "pulse recording" can
conveniently be applIed, depend only to a secondary degree on linearity
of transfer characteristi~ of the tape. Some current sophisticated uses
of pulse recording do require that the tape transfer characteristic be
under control, but for the purpose of a general review of recording
modes, pulse recording can be considered to depend only on a "plus,
minus, or zero" signal from the tape.
Pulse recording on tape can be dIvided into two mam mechamsms.
One of these carries information in the variation in pulse start and
stop times. The other depends only on the presence, absence or polarity of the pulse within a more or less fixed time pattern. In the first
group of time dependent pulse recording methods are pulse duration
modulation (PDM), pulse positIOn modulation (PPM), and pulse
frequency modulation (PFM); both of the latter are variants of
PDM. The non-t,ime-dependent pulse recording methods are based
on digital pulse modulation such as is employed in memory devices for
computers. 'With thIS type of modulation, the recorder is asked only
to deal with pulses, the number, position and polarity of which carry
the total sIgnal. The relationshIp of the recorder to a signal already
in digital form is in this case clearcut. Where the signal is in analog
form, it must be encoded by a relatively complex scheme before it. can
be recorded in pulse form.
LINEAR OR ANALOG RECORDING
The essence of the analog recordmg process is that it can be considered to be essentially linear. The action of the high-frequency
bias which makes this lmear operation possible is treated in detail
under "Head-tape Interaction." One applIcation note IS, however,
appropriate to discuss here since It deals with the actual degree of
linearity achieved.
34
MAGNETIC TAPE RECORDING
The audIO recordist and many instrumentation recording engineers
will usually quote a fairly definite figure for the difference in the
output between a signal which causes one percent third harmonic
distortion in an analog recording system and a signal which produces
three percent third harmonic distortion. The signal which represents
full saturation of the tape is also usually quoted as having a fixed
relationship to the one percent signal. There are differences between
tapes, although they are surprisingly small, in the values quoted for
these numbers. It is safe to say that the shape of the semilinear transfer characterIstic is roughly the same for all tapes. The way of
arriving at the signal levels representing various amounts of distortion
bears comment here, however.
If one operates an analog recorder by adjusting level and bias so as
to produce normal operatIOn with an output signal containing one
percent thIrd harmonic dIstortion, this signal output level can be
called "zero leveL" If the input signal level is then increased until the
output signal dIstortion reaches three percent third harmonic, the
level change is almost invariably 6 db, or 2 to 1 in voltage. When the
input level is increased until the output level essentially stops increasing, saturatIOn is considered to be reached (the output level sometimes
decreases with increased level where the higher level may produce tape
erasure). This output level is often quoted to be 13% or 14% db
above the one percent third harmonic output level. This figure must,
however, be used with considerable care, since it depends on a peculiarity of the instrumentation typIcally used to measure the phenomenon.
If one operates a tape recorder so as to get an actual instantaneous
mputloutput plot of the transfer characteristic, one finds, as stated
above, that the linear part of the characteristic extends only over onethird of the total positIve-to-negative saturation characteristic. Measurements show that if the one percent distortion point is considered to
be reached with a signal swing from plus unity to minus unity in
output voltage, the saturation point will represent a swing from plus
three to minus three units approximately. This represents, on an
absolute basis, about 9 or 9% db as measured in the la:boratory. The
difference between the 9% db and the 13% or 14% db figure results
from the use of the typical linear full wave rectifier meter in measuring
the higher figure. The meter is not "concerned" by the fact that the
output wave has changed from the sine wave for which it is normally
calibrated to read into a square wave. As saturation increases and the
corners of the sine wave are filled out to become a square wave, the
meter sImply interprets this filling out as increased output. WIth the
t.ypical non-sinusOIdal signal WIth whICh the instrumentation recorder
deals, this misinterpretation by the meter can be quite serious.
RECORDING
ME'IlHODS
35
Further consideratIOn of the linear recording process is postponed
for discussion under "Head-tape Interaction." The nonlinear recording modes will be discussed in approximately historical order, starting
with the FM mode and proceeding through such older pulse modes
as the PDM and PWM to PCM variations thereon. PAM (pulse
amplitude modulation) WIll be discussed in this section out its requirement of recording linearity WIll be referred to the appropriate later
section of the Survey.
FM RECORDING
Shortly after the introduction of magnetic recording for scientific
recordmg of analog signals it became apparent that there were many
analog recording applications for which a simple extension of audio
recording techniques was not adequate. This was particularly true
for signals containing dc components and for signals for which the
signal-to-noise ratIO of existing analog recorders was inadequate. To
deal with these two recording problems FM recording was the first
new recording mode to be introduced.
In the current form of FM recording the input signal frequency
modulates an oscillator, the output of which is recorded on the tape.
Typically, saturation recording is used on the tape, that IS, the FM
carrier is applied either as a rectangular or sme wave signal directly
to the recordmg head at such an amplItude as to saturate the tape
either in one direction or another. The function of the tape is thus
limited to storing the times of the axis crossings of the FM signal.
A standardIzed set of FM values and relationships has been in use
for many years. The center carrier frequently is usually %0 the
number of mches per second at which the transport is operating m
kc, i.e, 13.5 kc for 15 inches per second. Recently these carner frequenCIes have been doubled WIth consequent improvement in system
bandwidth. Typically the FM carrIer IS modulated 40 percent and,
up until recently, the bandwidth in kc has always been % the tape
speed in mches per second; for example, 2.5 kc for 15 inches per
second. Systems WIth twice and even four tImes these bandwidths have
been introduced withm the last few years. More recently, FM systems
WIth bandwidths up to 400 kc for recorders operating at 120 inches
per second have appeared to complement the 1% megacycle bandwidth
analog recordmg capabIlity (table 4-1).
36
MAGNETIC TAPE RECORDING
TABLE
4-l.-FM Recoriling Bandwidths
Standard Response
Tape speed IpS
Bandwidth
Carner
120
60
30
15
0--20, 000 cps
0--10, 000 cps
0-- 5,000 cps
0-- 2,500 cps
0-- 1,250 cps
625 cps
0-312 cps
0--
108 Kc
54 Kc
27 Kc
13.5 Kc
675 Kc
338 Kc
1. 68 Kc
7~
3%
1%
Extended Response
Tape speed IpS
120
60
30
15
7~
3%
1%
Bandwidth
0--40, 000
0--20, 000
0--10, 000
0-- 5,000
0-- 2,500
0-- 1,250
625
0--
cps
cps
cps
cps
cps
cps
cps
Carner
216 Kc
108 Kc
54 Kc
27 Kc
135 Kc
675 Kc
338 Kc
Double Extended Response
Tape speed IpS
120
60
30
15
7~
3%
1%
Bandwidth
0--80, 000
0--40, 000
0--20,000
0--10, 000
0-- 5,000
0-- 2,500
0-- 1,250
cps
cps
cps
cps
cps
cps
cps
CarrIer
432 Kc
216 Kc
108 Kc
54 Kc
27 Kc
13 5 Kc
675 Kc
Wldeband
Tape speed IpS
BandwIdth
120
400 Kc
CarrIer
900 Kc
RECORDING
METHODS
37
The FM recording system currently used usually offers 10 to 15 db
better signal-to-noise ratio than the corresponding analog signal-tonoise ratio obtained at a given transport speed. These Illcreased
signal-to-noise ratIOS are purchased dIrectly at a loss in recorded bandwidth, of course. The noise III the FM record system is limited almost
entirely to the FM noise_produced by flutter in_the recorder. When an
FM record system is operated WIth proper limiting, the output is insensitive to the amplitude modulation produced by the recorder noise and
instantaneous speed variation or flutter produces the only important
noise. For many FM recording applications a steady carrier of 100 kc
is recorded at the same time as the signal on the same or an adjacent
track. By playing back this carrier through a separate discriminator,
a signal representing the flutter; noise is obtamed. This flutter signal
can be applied to the data signal discriminator in such a way as to compensate for much of the flutter noise. Improvements in signal-to-nOlse
ratio of 6 to 12 db are achieved by this compensation technique. Although this compensation reduces the noise from flutter, it does not
remove the flutter from the recorded data signal (chapter 7).
There is some confusion in the present usage of the terms "narrow
band FM" and "wide band FM" depending on the application for
which such recording is applied. For the purposes of this discussion,
the terms shown m the table will be used to refer to "narrow bandwidth FM," "extended bandwidth FM," and "double extended bandwidth FM." "Wide band FM" will be used to refer to the 200 and 400
kc systems used with the 1% megacycle analog recorder. The term
"narrow band FM" has occasionally been applied to the analog recording of FM/FM telemetered signals such as those following the standard
IRIG FM/FM channel scheme. Reference to such recording will be
made here only as a special case of analog recording. The frequency
mUltiplexing of the IRIG multichannel scheme would not operate were
the rest of the system nonlinear. Unless the recording is in the linear
analog mode, interchannel modulation would destroy the utility of the
telemetering system (IRIG [1960]).
A typical FM-mode electronic assembly provided as an accessory to
a ground-based im;trumentation recorder uses a voltage controlled
multi vibrator oscillator to generate the FM signal. Much proprietary SOphIsticatlOn goes mto these veo's and there are dozens of different models on the market. Multiloop positive and negative feedback
IS used to stabilize these veo's. Since the FM recording mode system
is expected to carry dc, the veo's and their associated amplifiers are
usually chopper-stabilized. The output of an FM record electronic
unit is a rectangular wave usually designed to have such current and
voltage levels as to be applicable directly to a record head. Some
38
MAGNETIC TAPE
RECORDING
manufacturers, however, provide a head driver which is basically a
relatively high powered linear amphfier marehed precisely to the head
which can accept a variety of input signals which are not in proper
form for direct application to the head.
The FM recording mode may be used in a satellite, space probe, or
high environment recorder because it optimizes some recorder transfer
charactel'lstics that are not necessarily those for which the groundbased recorder uset purchases an FM system. Such flight recorder
electronics are therefore optimized for lightness, low power, and high
reliability in severe environments, usually at the cost of the typical
stability refinement just mentioned.
The FM playback electronics unit accepts a signal either directly
from the head or from a head preamplifier which is used for all recording modes. There is still some variety in the types of discriminator
used in FM playback units commercially supplied but the majority are
of the pulse-counter type or the more elaborate phase-locked-loop type.
A typical FM playback unit would receive input signals which are
successively negative and positive spikes representing the axis crossings of the recordmg rectangular wave. After amplification these
spikes are often converted to unidirectional form and applied to a oneshot multivibrator or delay line device which produces from each spike
a pulse of standard length. When this pulse is severely clipped, it
becomes a constant-energy or constant-charge pulse. In the simplest
discriminator, this train of constant energy pulses is applied to an
averaging circuit which may not be more complicated than a resistor/
condenser combination, but usually is that combination followed by a
filter of fairly complex design. The filter is low pass, designed to
cut off just above the useful data bandwidth, to maintam uniform
phase and amplitude characteristics up to the cutoff and to attenuate
as rapidly as possible beyond cutoff so as to eliminate the carrier
represented by the pulse train from the output. Typically a maximally-flat amplItude response or maximally-flat time delay response
(optimum phase equalization) option is provided in the filter, often
with a swireh.
The performance of such a very simple system depends on heavy
limiting in the record unit to assure that the axis crossings which are
applied to the tape are defined as sharply as possible. This must be
followed after reproductIOn from the tape by equally severe limiting
to eliminate the effects of signal amplItude variation inherent in the
tape recording process. The degree to which the "constant energy"
pulses are in fact constant energy is also dependent on stabilized time
determining elements in the multi vibrator and heavy and accurate
limiting of the pulse amplitude. For a flight recorder much of the
RECORDING
ME'l'HODS
39
refinement may be eliminated in the interest of low power and low
weight with the result that the system may be slightly tape amplitude
variation sensitive but, of course, much less so than an anaJog system.
The phase-locked-loop discriminator has been extensively discussed
in the literature (as has, of course, the simple discriminator just described) and no elaborate description of It will be undertaken here
(Gilchriest [1957]). It will be enough to describe the general principles on which it operates. In the phase-locked-loop discriminator,
the shaped and limited input signal is compared in a phase detector
with the signal from a local voltage-controlled osClllator. The output
of the phase detector IS used to vary the frequency of the local oscillator
so as to keep it locked in phase with the mcoming signal. The control
voltage from the phase detector to the voltage controlled oscillator is
the output signal. This is clearly a more complex device than the
simple averaging detector described before and its use has therefore
been limited to sophisticated FM recording reqUlrements. Its great
advantage is that it may be made extremely stable and may be made to
give a useful output signal over a wider range of signal-to-noise ratios
than the simple pulse counter discriminator. To a certain extent it
also is easier to add a flutter; compensation signal to such a
discriminator.
'
The phase-locked-loop discriminator IS used almost universally in
ground station telemetry equipment to interpret the output of FM/FM
telemetry systems (McRae and Sharla-Nielsen [1958] ). Such a telemetry discriminator invariably has an input for a flutter compensation
signal. This flutter compensation technique is currently more widely
applied to the telemetry discriminator than it is to the FM record system itself. In other words, an FM/FM telemetry signal which has
been analog-recorded will usually have a 100 kc reference signal
recorded along with it and this reference signal will be used to eliminate flutter noise from the output of the individual telemetry
discriminator.
Flutter compensation is considered m more detall m chapter 7 j
certain aspects of this compensation are mentioned here because they
relate to the performance of FM recording as a coding scheme.
The FM flutter compensation techmque is very useful but it is
f'harply llmited in the improvement that it can provide, in much the
same way that the desirable features of FM recording are limited as
a higher level of performance is sought. For example, the typical
flutter compensation scheme operates by deriving m a reference discrimmator, as explain_ed above, a signal proportional to the flutter
noise generated in the recorder. The flutter noise signal is delayed, as
it passes through the reference discrimmator, relative to the data sig788-028 <>-69---4
40
MAGNETIC TAPE RECORDING
nal; the data signal must be correspondingly delayed if accurate compensation is to be achieved. For a wide band system, achieving this
delay without distorting the prime data can be very dIfficult (Ott
[1962]). It is also appropriate to anticipate chapter 7 by noting
that straightforward FM flutter compensation removes only flutter
noise and does not remove flutter from the data. Only very elaborate
correction schemes can make this next step of improvement beyond
the simple compensation.
The video recorder, primarily designed for broadcast use, introduced
a new series of FM recording techniques to the magnetic recording
field (Ginsburg [1957]). The video recorder appears at first examination to operate in an FM mode which violates every rule of FM
t.ransmission. The carrier frequency is usually about. 4 megacycles,
the modulating signal has a bandwidth of 4lh or 5 megacycles, and
the maximum FM deviation is considerably less than this bandwidth.
Part of t.he FM spectrum so generated is cut off or "folded over" at
zero frequency and another part of it is suppressed by the limited
bandwidth of the recorder. This peculiar modulation scheme might
never have been tried were t.he particular applIcation for which it was
first attempted not itself peculiar in that the overall result was to be
judged as a TV picture by t.he human eye. Obviously, this recording
scheme inherently produces hIgh distortion, but the particular kind
of distortion involved appears to be such as to be easily tolerated by
the eye viewing a television signal so treated. More recently, sophisticated versions of this modulation scheme have been developed in
which the violations of normal FM recording relationships are largely
compensated for. Such video-type recording systems are now able
to produce relatively low dIstortion and quite impressive bandwidths.
A rotary-head recorder can also be used for predetection recording,
that IS, for recording a signal directly from a receiver IF before it
has entered the second detector. The received signal in this case is
always frequency modulated, and is heterodyned down from the IF
frequency to one which fits the recorder bandwidth. It is not passed
through the FM modulator and demodulators used when putting an
analog signal through the recorder, but is recorded direct. The video
recorder, because it uses a rotary head rather than the fixed head of
other instrument.ation recorders, has flutter and time displacement
error characteristics quite different from those of the longitudinal
recorder. The FM noise characteristics likewise differ. The process
of developing adequately stable rotary-head recorders for sophisticated broadcast use including color television has reached the point
where it is now possible by the use of variable electrical delay lines
to reduce the total time instability to -+-25 nanoseconds. These tech-
RECORDING
MET-HODS
41
mques are now available for making extremely stable predetection
recordings (Klokow and Kortman [1960]) (Ampex [1964]).
SPECIAL FM RECORDING SYSTEMS
Analog recording of telemetered data on ground recorders for the
space effort has been concentrated on FM/FM telemetry. Because
of the standardization. offered by the IRIG telemetry agreement, there
has been a tendency to limit data transmissIOn to the characteristics
and bandwidths of the IRIG standard bands. Recently, however"
many users have found the rigid limits of these bands, with their
constant percentage deviation and hence their carefully ordered increase in bandwidth from one end of the standard band to the other,
unnecessarily confining. New FM/FM standards have therefore
arbitrarily been used by some organizations and an alternate constantbandwidth system is being considered by IRIG to supplement the
previous constant-percentage-bandwidth system.
Many of the telemetry ground equipment manufacturers also offer
nonstandard constant-bandwidth systems despite the pressure to use
the IRIG standards. A further development of the full utilization
of FM/FM, not for long-distance telemetry but simply for convenient test-data handlmg, is the introduction by some manufacturers of
complex multiplexing systems to provide many more channels than
either the present or the proposed IRIG standards do. Such equipments consist basically of modular units that permit putting 75 or
more uniform FM data channels of 1 to 2 kcps bandwidth on the
direct record tracks of an instrumentation recorder operating at a
moderate speed (Martin [1963]).
There is a continual ebb and flow of interest in the telemetering
commumty in the various methods of telemetry and, correspondmgly,
mterest in the recording problems these forms raise. FM/FM telemetry is still considered perfectly satisfactory by some groups domg
quite complex work. Other groups have abandoned it almost completely for PCM, and still others use mIXed systems. Each group
maintains a position that is largely based on the way in which its
partICular field has developed. The recorders and auxiliary equipment available in the commerCial market have characteristics that
have been arrived at by the interaction of the characterIstics of previously available eqUIpment and the changing needs of users. It IS
therefore qUIte impOSSIble to predict the future direction of development of telemetry recording equipment. Analog recording can
probably be expected toadecline m importance with a corresponding
mcrease in importance of PCM and wideband FM, recorded either
directly or in a predetection mode.
42
MAGNETIC TAPE
RECORDING
PULSE DURATION MODULATION
Pulse duration modulation (PDM) was introduced to the magnetIc
tape recording field about the same time as was FM modulation. In
pulse duration modulation, the modulating signal is caused to vary
the time of occurrence of the leading edge, the trailing or both edges
of a train of equal-amplitude pulses forming a pulse carrier (fig. 4-1).
--,
I
I
I
I
I
--,
I
I
I
I
I
L __ L--'-_-J
I
I
I
I
I
IL __
I
I
I
I
I
I
I
I
I
IL __
--,
I
I
I
I
I
TIMED
r--
r--
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
-....I..,---J __ .J
r--
I
I
I
I
I
I
I
I
__ JI
I
I
I
I
I
I
__ .JI
I
TIME - -
b
rI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
_J
L_
-,I
-,
I
I
I
I
I
I
I
I
I
I
I
-,
rI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
_J
L_
TIME-c
(after Black "Modulatwn Theory")
}'IGURE
4-1-Pulse-duration-modulation (PDM) wave forms
(a) Leadmg edge of pulse fixed, trailing edge modulated, (b) trailIng edge
of pulfi;e fixed, leading edge modulated, (c) both edges of pulse modulated
symmetrically around center
This modulation system does not require a linear amplitude characteristic in the recording process, which IS concerned only with establIshing with accuracy the start and stop times of the pulses.
With PDM, during modulation the short-term area average changes
and hence the ac axis moves. In fact, one method of extracting the
mformation from PDM pulses is simply to determine the average
value of the signal by a filtering process. Because the axis moves,
the transmission system must handle the movement with fidelity.
ThIS means that the auxiliary equipment, 'including all the signal
circuits within the recorder proper, must be linear if the system is
to be linear overall. The tape Itself, because it is concerned on,ly
RECORDING
ME'IlHODS
43
with pulse times, is relieved of amplitude linearity. Particularly in
early use of magnetic recording for scientific data collection, the price
of ltuxiliary circuit linearity was gladly paid for the advantage of
eliminating the need for tape linearity.
There are many ways in which duration modulated pulses may be
produced. ~ FIgure 4-:2 shows two ways of accomplishing this for
pulses in which the trailing edge alone is modulated. The conversion
technique shown at a is known as uniform sampling because the successive pulse durations are proportIOnal to values of the signal at uniformly-spaced SIgnal sampling times. The procedure here is to sample
the signal at uniform intervals and convert each sample to a steady
value which remains until the time of the next sample, at which time
the steady value changes to that associated with the new sample. This
gives a peculiar kind of PAM (pulse-amplitude-modulated) wave as
shown. This PAM wave is then added to a sawtooth generated by the
same synchronizer as drives the original signal sampling process. The
combined signals are then passed through a slicer which is a device
with the property that its output is zero when its input is below a
certain value and is a maximum when its input is above that value. In
effect, it takes a slice of the wave at the point marked "slicer level"
and converts it into pulses which are positive when the input signal is
above that value and zero when it is below. The result is a pulse-durahon-modulated wave. The production of such a uniform sampling
wave is somewhat complex. A somewhat simpler technique for derivmg the PDM wave is shown in b of figure 4-2 and is known as natural
sampling. The sawtooth is simply added to the signal wave and passed
through a slicer as before. The output is PDM, but differs in form
from the PDM derived by the first method because the lengths of the
sampling intervals are dependent upon the values of the signal. This
I~ a much simpler signal to generate but it is much more complex to
regenerate at the end of the transmission system. The usual practice
is to regenerate it as if it were a uniform-samplmg signal and to accept
the distortion produced by the nonuniform sampling times.
The purpose in describing in such detail the mechanics of PDM
sIgnal generatIOn is to emphasIze one of the significant characteristics
of pulse transmission systems. This characteristic IS that one almost
lllvariably knows some specific thing about the pulse arrival time
which often makes it possible to improve the overall time stability
of the transmIssion system. This knowledge, for example, may be
t hat the starting times of the pulses are exactly evenly spaced. With
tlus knowledge a system can be devised for recreating at the far end
of the system pulses WIth exactly the same startmg times and lengths
44
MAGNETIO TAPE REOORDING
/",--- r:'"
« _
•
1
TlME---',
~--
--.-~--:f-':'"
....
_--,
SIGNAL
PAM
SAWTOOTH
SWEEPWAVE
-
- - - - - SLICING LEVEL
COMBINED WAVES
JLJlJL
PDM
TlME ___
a
SAWTOOTH AND
SIGNAL COMBINED
nJlJL
PDM
(after Black "Modulation Theory")
FIGURE
4-2.-Methods of generating PDM.
(a) Uniform sampling, (b) natural samplmg
as those that we-re fed into the system. Thus the time instability of
the recoidermay be removed. The same function can be performed for
pulse amplitude modulation and for pulse code modulation. In the
latter case, not only are the times of the pulses known but also the amplitudes in that the information is contained only in whether the pulse
is present or not or, for some more elaborate coding schemes, whether
the pulse has the value "one" or "zero."
The usual PDM recording practice is to convert the rise at the beginning of a PDM pulse and the fall at its conclusion to either bidirec-
RECORDING
ME'lIHODS
45
tional or unidirectional spike signals for recording on the tape. The
circuits immediately associated with the tape recordmg process itself
need deal only with these spikes and therefore need not have dc
response. A "boxcar" circuit or bIstable multIv~brator or, for the most
modern equipment, one of the more sophisticated modern alternates to
these circuits, is used to convert the spikes back into a tram of
[o=~~-=I.~I.
--- -- -- ---,
l!'_~_~_~_'!._~_~
x x
x
~
1010100
o
I
0
I
0
1
x
x
x
x
x
x x
x x
x
x
x
Xx)(
0
0000010
I
1
1
1
1
I
4-6.-Bit-serial PCM on tape.
FIGURE
The word at the left is the first transmitted and is -followed, after a space, by
that indicated in the line with it at the right as made up of 7 X's representing
arbitrary digits. To conserve tape, adjacent tracks spread across the tape are
used for successive words transmitted, the array being arranged so that there
Is an essenthllly steady fiow of wo.rds onto the tape In this example the space
between the last word entered on the 7th track and the second word entered on the
1st track is such as to give the same word-start spacing as the inter-word spacing
Since the words on each track are considered indindually, only track to track
timing erro.rs greater than a full word length could confuse such a recording
system.
L'WOflO
y
1
a
I 0
1
0
1
0
0
1
: 1
1
0
0
1
0
1
1
0
0
1
I
10
0
1
0
1 0
1
0
0
1
:0
0
0
1
0
0
1
0
0
0
: 1
10
1
0
0
0
1
1
1
0
0
0
1
0
o
0
1
II
0
1
0
0
1
0
1
0
1
1
0
: 1
1
1
1
1
a
0
1
1
1
1
01
1
a
0
0
0
0
0
0
1 0
0
0
0
1:=1
5
----l
b
a
FIGURE
c
d
4-7.-Bit-paraUel PCM recordlDg.
(a) Successive words are recorded in strips across the tape width for, this
case, perfect time alignment between tracks (the words for this figure are the
same as for figure 4-6), (b) III this case there is severe IDter-track jitter which
makes it difficult to see even visually which bits belong to which words, (c) ID
this case the bits are spread out farther along the tape and the displacement is
systematic such as would be caused by skew, S represents the word to word
spacing needed with this much skew if no confusion is to exist between individual bits of successive words, (d) the situation of (c) but with close word
spacing where interference occurs between the words and no simple technique
will untangle the words III the presence of this much skew.
52
MAGNETIC TAPE RECORDING
An IRIG standard now exists for recorders designed to receive
parallel PCM recording (IRIG [1960]). It is a relatively conservative standard because of the difficulties of achieving close packmg
density. Many tape recorders are currently available which meet this
standard and which operate satisfactorily at longitudinal pulse packing densities of 1,000 bits per inch. These recorders operate in thIS
way without any compensating schemes and without correcting possible errors in bit arrival times; they achieve correct operation simply
through maintenance of close mechanical tolerances and uniform tape
motion. There are, however, methods available for packing bit parallel PCM much more densely than even with current techniques and also
for achieving equivalent high density through the use of bit serial
PCM.
PULSE RECORDING WAVEFORMS
The simplest pulse waveform conceptually is the return-to-zero or
RZ pulse. This might consist of a recordmg current (and hence recorded flux) waveform such as shown in figure 4-8a. A "one" is
represented by a positive going pulse and a "zero" by a negative going
pulse.
This particular waveform is used in elementary equipment but its
inefficient use of bandwidth has restricted its use. It is a redundant
coding system since there are two flux transitions per bit of information required.
A slightly more efficient waveform is one which remains at either the
positive or the negative maximum excursion possible III the absence of
a signal. A "one" in this system is represented by an isolated pulse
from one extreme of signal to the other or, in the case of tape, from
saturation in one direction to saturation in the other direction. At the
end of the pulse the original value is resumed and two flux transitions
are required for each "one." A "zero" is represented by the absence of
a pulse. This is called "return-to-bias" or "RB" recording (fig. 4-8b).
In NRZ coding, redundancy is sharply reduced by assigning, at the
most, a single transition to each separate bit. The information derived from this coding scheme in a recorder IS whether the flux changed
or not. There are several ways in which these two basic pieces of information can be used to give a binary coding scheme. In "straight"
NRZ, shown in figure 4-8c, a flux change indicates that the symbol
following a given symbol is different from the first symbol. Thus the
change from zero to one is indicated by a flux change. When a series
of ones follow each other, no flux change is indicated, but when the
last one IS followed by a zero, there is another flux change. The socalled NRZI coding, which is shown in figure 4-8d, indicates a one by
RECORDING
o
0
o
o
000
a
RZ
L -_ _ _ _
b---...I
C
53
METHODS
----,LSlSl____;----
nJ,...-----
d
L -_ _
FIGURE
RB
NRZ
NRZI
4-8.-Digital recordmg wavefonns.
(a) In RZ recording of the fonn shown here, the binary bits of the word given
above the waveform are indIcated by downward pulses for O's and upward pulses
for l's, (b) for "return-to-bias" or RB recordmg in the fonn shown here, a pulse
is transmitted for a one and no pulse for a zero, (c) for "non-return-to-zero" or
NRZ recording the signal value does not change unless there is a change m the
value of the digit, that is, the horizontal line at the left indicates that the flrst
2 digits are the same as each other, the next downward flux change mdicates
that the succeedmg digit differs from the precedmg one, as with the next 4 flux
changes since there is a steady alternation of l's and O's (sometimes called NRZ
(change), (d) in NRZI recording a flux change represents a 1 and no flux change
11 0 (sometimes called NRZ (mark».
a flux change and a zero by no flux change. The primary advantage of
. NRZI over NRZ is that there is a one-to-one relationship between the
signal and the symbol. If there is an error, only the symbol for whIch
the error is made is lost. In straight NRZ, if a flux transition were
missed, all successive symbols would be exactly opposite to what they
should be until synchronism was somehow regained. These straightforward coding processes have the difficulty that there may be long
"floats" in which a particular data combination produces extremely
long signal pulses. Low-frequency phenomena in the recorder may
therefore affect the accuracy of data contaming such long floats.
The NRZ process uses at a maximum one flux reversal per bit of information and on the average uses less than this number. Where there
are, for NRZ, long periods of identical symbols, or, for NRZI, long
strings of zeros, the coding IS very efficient in that fewer flux transItions are used. It is essential for this efficient coding to work that the
timing of the flux transitions or their absence be accurately known.
So called "phase modulation" or "Manchester" coding techniques
have been introduced to elimmate some of the problems of providing
54
MAGNETIC TAPE RECORDING
accurate time information in interpreting binary symbols (Hoagland
[1963]). Phase shift or phase modulation coding is also called selfclocking. It is so called because there is an Identifiable pulse signal
in each "cell" in which a binary bit may be found. A signal may therefore be derived from each cell to keep the system in synchronism or to
provide a clock. Figure 4-9a shows such a coding scheme. In this
particular version of phase shift coding, the zero is represented by a
downward flux transition in the middle of the bit cell and a one by an
upward transitIOn in the cell. In order to maintain this relationship
it is necessary for additional flux transitions to be introduced from time
to time when two identical transitions must follow each other. In
other words, an extra downward transition must be provided between successive ones and an extra upward transition between successive zeros. There is, however, a derivable output spike, either positive
or negative, in the center of each bit cell and this output spike can be
used to synchronize a local clocking scheme. In this system it is possible, with proper design, for the rate of data flow (recorder tape
speed) to be quite irregular and still to detect properly the piece of inI
I
10
010
0
I
I
1
I
01
1
1
1
1
1
I
I
I
1
I
MANCHESTER
1
bl
I
I
I
I
1
DIPHASE
tl t! t! t: t! t! t! t! t
1
I
I
I
I
I
I
I
FIGURE 4-9.-Phase-shift or phase-modulation digital recording wave forms.
(a) The "Manchester" form of phase-shift coding designates a zero with a
downward flux change in the middle of the "bit cell" and a one with a corresponding upward flux change, (·b) this phase·shift coding, known as "modified diphase," indIcates a zero by having no flux change durmg the "bit cell" and a one
by havmg such a flux change, decoding being accomplished by examining the
successive values of the wave form at the arrows, which are %, of the way
through the "bit -cell," III determining whether the values are the same for successive examinations (a one) or different (a zero), a flux change always being
provided between successive "bit cells" for timmg.
REOORDING
METHODS
55
formation contained ifl each binary cell. Many versions of this coding
have been devIsed with various advantages for particular applIcations.
Some of these schemes optimIze the simpli~ity of creating the code
and others optimize the accuracy of detecting the output (Gabor
[1960]) .
A speCIfic version of phase shift coding described as "modified q.iphase" has been specifically applied to the Gemini onboard PCM recorder. This coding scheme carries the information in whether there
is a phase transitIon within the "bit cell" or not. As illustrated in
figure 4-9b, thIS particular code gives a phase transition within the bit
cell for a one and no such transItion for a zero. There is always a flux
transition at the start of each bit cell from which the clocking scheme
can be operated. Detection of the data III this code can be performed
- by examining-the instantaneous signal value three-quarters of the way
through the bIt cell, as shown by the arrows in the figure. If the signal
value changes between any successive pair of examinations, the digit
in the bit cell observed is a zero; if the value remains the same, the
digit is a one.
Basically the phase shift modulation coding scheme can be seen to
have two fundamental frequencies. One of these has the length of two
bit cells for a full cycle and the other has the length of a single bit cell.
In a simplified way, one might say that the system runs either at one
or the other of these two frequencies. In practice, of course, the shift
between the two waveforms is irregular and unpredictable, and the
result is that a broad spectrum is generated. Basically, however, the
highest frequency present in a phase shift modulation coding process
is twice that in NRZ. The price of the extra bandwidth IS gladly paid
in many applications for the improved accuracy with which the data
is derived.
OTHER RECORDING METHODS
Tho several recording modes described in this section do not by any
means exhaust the possibilities that have been employed. Only a few
other modes have, however, had any extensive use. Perhaps the most
important of these is what is called "Carrier Erase."
In carrier erase, the tape is prerecorded WIth a continuous tone signal of high amplitude and of frequency well above the intelligence to
be recorded. The recorder sImply dc-erases the carrIer in proportion to
the intensity of the information signal, which is applied directly to the
record head. This technique has the advantage of extreme simplicIt.y
788-0280---«>---5
56
MAGNETIC TAPE
RECORDING
in the recorder and not much complexity in the reproducer, since the
signal is obtained in the form of a modulated carrier which is relatively easy to demodulate. Although of low recording density, the
simpliCIty and fair linearity of the process have led to its extensive use
in very small) severe-service recorders.
CHAPTER 5
Head-Tape Interaction
A convenient way to analyze the difficult subject of interaction bet~~enhead _and tape is t~ postuJ.ate an _ele~ent~ry idealized _tape
recording process and then to examine in some detail the performance
of this process and the factors influencing that performance. As the
examination proceeds, it will become apparent what practical changes
are advisable in making and reproducing the recording, first, to produce useful results, and then to refine those results. By this procedure,
the complex interaction between the elements involved can be developed
with a minimum of repetitive analysis.
For the first test recording, let the recorder consist of a tape moving
mechanism and a recording head of the type commonly used in modern
recorders, i.e., one which produces the recording flux as a fringing field
surrounding the recording gap. This recording head is driven from
an appropriate linear amplifier which applies test signals directly to
the recording head. The tape is smoothly driven at one of the currently standard speeds. After making a test recording and rewinding
the tape, it should be possible to reproduce the recorded signals. The
recording head could be used for reproduction, but it is more instructive to use a modern reproduce head with slightly different characteristics from those of the record head. The output of this head is
connected to another linear amplifier uniform in frequency response.
Two things would be obvious on examining the reproduced output:
First, the output would be badly distorted, mostly with odd harmonics,
to a degree almost independent of the input signal level. Second, the
frequency response of the system would be nonuniform, being roughly
"humped," peaking somewhere at the geometric average of the lowest
frequency and the highest frequency that the recorder could reproduce.
The signal-to-noise ratio would not appear -to be too bad, but, since no
signal could be reproduced without distortion, there would be no reference signal .level for which the signal-to-noise ratio measurement
57
58
MAGNETIC TAPE R1!)cORDING
could be specified. Consider now why the hypothetical recorder would
operate in the manner just described.
In the first place, the transfer characteristic of the head-tape combination, i.e., a plot of instantaneous input voltage versus instantaneous output voltage, is a severely curved line when an alternating current signal is applied directly to the recording head. The curvature
is symmetrical around zero voltage and the distortion produced by the
curvature therefore consists entirely of odd harmonics. (See fig. 5-1.)
Br
RECORDED
INDUCTION
ON TAPE
RECORDING
SIGNAL CURRENT
FIGURE
5-1.-Unblased magnetic recordmg.
With the distorted portion of the normal tape transfer characteristIc between
A and B bemg used the rerorded waveform and resulting output are sharply
dIstorted.
This transfer characteristic is often likened to two initial magnetization curves of the magnetic recording medium placed back to back.
Although it resembles these curves, the mechanism involved is quite
different from that involved in forming the initial magnetization
curves.
Having chosen a particular tape speed, the relationship of applied
frequency and recorded wavelength is now determined. If there were
a means for determining the strength of the flux recorded in the tape,
an examination would show that placing a current in the recording
head constant at all test frequencies would produce flux in the tape
approximately constant at all wavelengths. When this constant-flux
recording was reproduced we would find, at least at low frequencies,
that the output from the reproducing amplIfier was proportional to
59
HEAD-TAPE INTERACTION
frequency. This would result from the dependence of the reproducing
head output, not on the amplitude of the flux, but on the rate of change
of flux. The classic 6 db per octave rise of a differentiated signal thus
occurs since, in effect, the input signal has been differentiated in the
reproducing process.
In a real recorder/reproducer, however, this 6 db per octave rise
would not be maintained as higher frequencies (shorter wavelengths)
were approached and the response would eventually flatten out and
turn down very sharply. The response would eventually go to zero
at a frequency which could be calculated to be that at which the reproducing head gap was effectively one wavelength long (fig. 5-2).
(In I
WAVELENGTH -
01
60
~
001
0001
~
50
D
"0
-
40
..... ~
I
~ 30
Q.
~
V
=>
o
20
10
,....-
1/
o
/
1,\
\
\
~""
FREQUENCY ___
FIGURE
5-2.-Elementary recorder frequency response.
To improve the rather discouraging performance of this tape recorder, which perhaps was the situation that Valdemar Poulsen faced
when he invented the first tape recorder at the turn of the century, distortion might be attacked first. The rather unpleasant transfer characteristic of the figure suggests that adding a fixed magnetization to
that of the signal so as to shift the operating point of the recorder up
from zero to a fairly straight part of the transfer characteristic might
help. This would reduce the size of the reproduced signal since less
of the total "flux swing" would be used and the noise output of the
recorder would increase but the transfer characteristic would straighten
out quite well (fig. 5-3). Many modern dictation recorders and even
some semiprofessional audio ones use this so called "dc bias" mode of
operation. Assuming for the moment that the better linearity of the
60
MAGNETIC TAPE RECORDING
dc-biased signal is satisfactory, the next concern might be with the very
poor frequency response.
Br
----------~~4-rT---f---------H
RECORDING
SIGNAL CURRENT
I---I-oc
BIAS
FIGURE 5-3 -Recording with dc bias.
The useful part of the characteristic from A to B is shorter than tor the unbiased
condition of figure 5-1, but the output IS undistorted.
Since the signal is differentiated in reproduction, an integrating
equalizer, that is, one with a 6 db per octave downward slope with
frequency, could be inserted in the reproducing channel. This would
straighten out quite successfully the low frequency portion of the frequency response curve, but the flattening out and droop at high frequencies would be intensified. Some kind of high frequency boost
would now be needed; a resonant or feedback RC equalizer could supply enough high frequency gain in the reproducer output to flatten
out the response of the complete recording system over a reasonable
frequency range (fig. 5-4). The equalization would increase the high
frequency noise in the output just as it increased the high frequency
signal. The overall signal-to-noise ratio would thus be degraded by
the equalization. The final signal-to-noise ratio achieved thus depends on the frequency,range the recorder attempts to cover.
The test device would then be a usable magnetic recorder; it is probable that half a million recorders usmg the same design principles
have been commercially sold for amateur and office dictation use within
the last 10 years. Such It device is still rather crude and many techniques are now available to improve the performance by several orders
of magnitude. First, the dc bias could be removed and ac bias could
be added. That is, a steady ac signal of about ten times the maximum
signal amplitude could be added to the signal in the recording head.
61
HEAD-TAPE INTERACTION
01
60
(In)
WAVELENGTH001
0001
I
50
.a 40
~
I
~
30
Q.
-
f-- ,-
/'
~.
-- ../
I-
/'
::>
20
..,-
10
/
~r
~
o
o
b
r--
i'~
V
,
1--"\
1-
I,
\
~
..... ~
V
~
1\
\
FREQUENCY - - -
FIGURE 5-4.-Equalization of elementary recorder.
(a) Basic unequalized frequency response, (b) simplified results of, on the
left, mtegratlon and equalization to remove normal differentiatlOn of d~/dt playback head, and on the right, of a peaked equalizer to remove the effects of
scanning losses.
It would be found that this would improve the signal and reduce the
noise of the reproduced recording. (This biasing technique was the
development step that made modern recording possible.) Second, the
reproducing head gap could be made as narrow as possible so as to
reduce the wavelength and hence raise the frequency at which the output response goes to zero. The recorder would then r~uire less high
frequency reproduce equalization and would handle an increased
bandwidth.
The record current would still be constant with frequency. If
nothing were known about the energy spectrum of the input signal
(the usual situation with scientific recording), this record current
relationship could not be changed. If, however, something were
known about the spectrum of the input signal, as is, for instance, the
case in audio recording for entertainment purposes, certain frequencies
could safely be preemphasized on the basis that the input signal would
not contain significant high-intensity components at these frequencies
and there would be little possibility of overloading the tape. A
decision was made, early in the development of audio recording, based
on the assumption that there was little high-frequency energy in a
typical audio signal, to preemphasize high frequencies. Having preemphasized these frequencies, they can be deemphasized in reproducing. That is, less high-frequency reproduce equalization would be
62
MAGNETIC TAPE RECORDING
used, less high-frequency noise would appear in the output, and the
signal-to-noise ratio would be improved. A little low frequency preemphasis could also be used, but high-frequency preemphasis would
have the maximum effect. The recorder as now described is essentially
that used for both professional and nonprofessional audio recording.
The technique of deciding the amount of preemphasis from the
knowledge of the spectral characteristics of the input signal has been
described in glowing terms here. In theory this is what is done with
an audio recorder; in practice, a study of the actual spectral distribution almost proves that it cannot successfully be done. In other words,
a scientific investigation of the spectrum is unable to account for the
amount of preemphasis that can be used without harm. The most
recent change in audio preemphasis/deemphasis, introduced with a
view to improving signal-to-noise ratio, came about as the consequence
of some work which showed that the spectrum distribution of music
would not permit such preemphasis! (McKnight [1959 J) Nevertheless, the new equalization is quite successful and is now almost universally employed for "master" recording (original recording as the basis
for mass reproduction of disc or tape copies for home use) .
To record a very wide band of frequencies, say twice that of the
initial recorder, tape can be run twice as fast and all equalizers and
similar devices correspondingly scaled up in frequency. This process
can be continued, scaling up many times in bandwidth from the initial
audio range. By increasing tape speed, higher-frequency rooording
can take place at the same wavelength. There is a point at which,
however, frequency, per se, becomes a limiting factor in certain of the
rooorder elements. Eddy current and hysteresis losses begin to limit
the usefulness of record and reproduce heads and the heads must be
made of Imaterials designed for use at these higher frequencies.
As the abilities of the basic tape recorder are pushed higher and
higher, second order effects begin to be important and the action of
the recorder must be analyzed in greater and greater detail. For
example, further investigation would show that there are several complex frequency-sensitive losses which add to the basic reproducing
wavelength losses (McKnight [1960]). These losses produce perceptible effects even for the elementary recorder, but become more important for more sophisticated recorders. There is a loss of signal as the
portion of tape involved is placed farther and farther away from both
the record and reproduce head. This "spacing loss," as it is called, is a
function of recorded. wavelength. Scanning loss, discussed above,
comes mainly from the finite size of the reproduce heads although
some characteristics of the record head gaps also affect wavelength
response. A widely misunderstood phenomenon called demagnetiza-
HEAD-TAPE INTERACTION
63
tion loss is also found. It results from a tendency of the tiny permanent magnets placed in the tape on recording to demagnetize themselves, this effect being greater as the magnets get shorter and the
wavelength correspondingly shorter. There is another loss which is a
functIOn of the thickness of the magnetic material on the tape. This
is a form of spacing 1'oss, that is, of loss dependent on spacing of the
material away from the head; thIS spacing differs for different depths
in the tape, so that only part of the tape is effectively in contact with
the head. The phenomenon has been studied recently in great detail,
particularly for digItal recording (Eldridge [1960]). The analytic
process of separating the effects of all these losses has been discussed
extensively in the literature (McKmght [1960]) (Wallace [1951]).
The basic process by which ac bias causes the record characteristic
to become linear is often described as being quite mysterious. The
current understanding- of this-phenomenon is-not, however,-quite so
uncertain as thIS typical reaction implIes. A more accurate descrIption
IS:
1. The process of anhysteretic magnetization is generally con-
sidered to be an adequate description of the gross effects of ac
bias as the magnetic tape passes through the biased recording
fields. An essentially anhysteretic magnetization process takes
place when a section of tape is brought up to the record head
and passes beyond it. As it approaches the head the bias
and signal field get progressively stronger to a peak level and
then, as the tape section leaves the head, they gradually fall off.
This is precisely the process involved in anhysteretic magnetization where a sample of magnetic material is placed in a field
varying just as described.
2. The process by which anhysteretic magnetization takes place is
not fully understood at present, although the characteristics
produced are faIrly well understood to be dependent on the
interaction between the magnetic fields of the partIcles rather
than on properties of the magnetic particles themselves.
(Schwantke [1958] ) (Daniel and Wohlfarth [1962] )
(Eldridge [1961]).
3. The detailed process which goes on as the tape approaches the
head, passes it, and passes on is not fully understood because
the actual magnetic fields present have not been adequately explored. For example, the magnetic field to which the tape is
subjected is directed along the surface of the tape in one direction as the tape approaches the head and IS along the surface
of the tape in the other direction as the tape leaves the head.
At some point in the process the field direction must rotate; just
64
MAGNETIC TAPE RECORDING
how this takes place and how it affects a set of dispersed magnetic particles IS not clearly worked out as yet (Mee [1964]).
If the emphasis is placed on recordings where the time of the starting and stopping of pulses contains the significant information, such
characteristics as lmearity and frequency response are not of primary
importance. The same tape movmg mechanism can be used but the
current applied to the record head can be chosen to give as much
magnetIzation in the tape as is possible to produce tape saturation.
To indicate the start of a pulse this saturating current can be turned
on or the current direction can be instantaneously reversed. Alternately, a positive pulse can be recorded by turning on a pOSItive saturating current and turning it off at the end of the pulse, allowing the
current to fall then to zero. A negative pulse can be recorded by
turning on a saturating current in the opposite direction and the end
of the pulse indicated by turning this pulse off. This is called "RZ"
or "return-to-zero" recording. Alternately, the current can be turned
on in one direction or another to indicate the start or stop of a pulse.
This is called "NRZ" or "non-return-to-zero" recording. NRZ recording, because It is more important, will be analyzed in further
detail.
If this recording IS reproduced with a flat linear amplifier and a
conventional reproduce head, a positive output pulse will appear for
each positive going recording current transition and corresponding
record flux transition; a negative pulse will appear for each negative
current transition and flux transition. A satisfactory pulse recorder
seemsrthus to have been provided without very much effort. The elements descrIbed are actually the only essential ones in a modern pulse
recorder, but for efficient pulse recording, refinement is required.
HIgh pulse recordmg density and sharply defined reproduced pulses
are needed, and the mechamsm involved in achieving the proper denSIty and sIgnal level in such recording must be analyzed.
When the pulse current through the recording head is turned on or
turned off instantaneously, it effectively maps the external flux field of
the recording head onto the tape (Eldridge r1960]). (This is the subject of a later analysis.) When this mapped field passes the reproduce
head, it produces an output signal essentially proportional to the
instantaneous rate of change of dIfference in the amount of flux passing
into the two halves of the reproduce head, one on each side of the
reproduce gap. Operating on this difference, it is'found that the
ability of the reproduce head to define the position of the pulse on the
tape IS three to six times as poor as that of the record head in placing
it there. This is a fundamental relationship, independent of detailed
HEAD-TAPE INTERACTION
65
head design. To improve the pulse packing density, attention, therefore, should be directed primarIly to the reproduce head.
The technique of employing the recorder simply to carry information on pulse starting and stopping times opens up its use for many
recording modes quite dIfferent from the linear analog mode first discussed for the basic test recorder and first used in audio recording.
These different modes, such as PDM, PCM, FM and other modulations,
permit optimizing various perfonnance parameters of the complete
recording process and providing best fits between a recorder and the
rest of a system in which it IS used. This subject is discussed in detail
under "Recordmg Methods."
By introducing the initIal crude tape recorder and gradually refining it for various applIcations the basic processes of interaction
between tape and head have been discussed in a systematic way. For
a g~Ileral review, _thIS outline treats the subject in adequate depth.
Certain parts of the head-tape interactIOn process whIch have detailed
importance will now be discussed more fully.
HEAD-TAPE GEOMETRY-BIASED RECORDING
The geometrical elements of the head and tape which enter significantly into the record/reproduce process are the shape of the head pole
pieces, the dimensions of the head gap, the spacing between the head
and the recording medium, and the depth of the recording medium.
Each of these geometrical factors has some effect on the response at
short and long wavelengths. These wavelength effects are some of
the tape recording factors about which our current understanding is
most complete.
The most critical dimension of the record and reproduce gap is its
length; the depth of the gap and size of the corresponding rear gap
in the head structure are also important in detenninmg head performance. The ultimate high-frequency (short-wavelength) limitation to recorder response is the length of the reproduce gap. In an
mdirect way, the length of the record gap may also influence this limit
but the primary effect lies in the reproduce gap. There is always one
critical tape wavelength at which the total amount of flux in one direction and the other included within the gap, and therefore affecting the
head, is independent of the position of the gap along the tape. At this
particular wavelength no output signal is produced. This is an
example of a scanning process which typlCaUy has a null at one particular wavelength. As with other scanning processes, at wavelengths
shorter than the wavelength of the first null there is still some response
from the head, and at even shorter wavelengths, a succession of addItional nulls. The characterIstic shape of response from such a scan-
66
MAGNETIC TAPE
R:E>CORDING
ning structure is a sin x/x function and is encountered in any process
in which waveforms are scanned by a structure WhICh integrates all
portIOns of the wave included within the structure length. Daniel
and Axon (Daniel and Axon [1953]) have analyzed the experimental
fact that a measurement of the so-called gap nulls and the shape of
the scannmg effect for magnetic recording does not precisely follow the
sin x/x formula. They postulate certain other interrelated phenomenon, giving a more exact function for the response. They include
possible mechanisms for the observed fact that the effective length of
the gap, judged from the position of the response nulls, is longer than
the physical gap (fig. 5-5).
10~---------.----------------,----------------,
O~--------------
___
iii 10
~
U)
g
20
_
__
J
""
~
__
..
~
______________
01
~
~
4OL-________
~
.....
30
_ _ _ __J
10
GAP LENGTH/WAVELENGTH
100
(after Mee)
FIGURE
5-5 -Reproducing gap-loss functions
(a) Simple sm x/x formula, (b) more complete formula (see text).
A longer gap includes more flux withm the gap pole pieces and
therefore produces a greater long-wavelength output. However, the
response of such a long gap will fall off faster with decreased wavelengths and at any particular wavelength it is Impossible to say without calculation whether widening the gap will increase or decrease the
response. The choice of the reproduce gap length is therefore an
important one related to the total bandwidth WhICh the recorder is
designed to cover. The discussion so far has referred to wavelengths
HEAD-TAPE INTERACTION
67
in purely sinusoidal terms. The Daniel and Axon analysis and the
sin x/x function do refer to sinusoidal functions; the same sort of gap
('ffeet, however, occurs in pulse recordIng and determines the digital
packIng denslty achievable (Eldridge [1960]).
The surface of the typical recordIng and reproducing head facing
the tape forms a gradual smooth curve. The tape is stretched over
this curve so as to contact the head over a specific length of tape. The
radius of the curve and its relationship to the rest of the geometry of
tape movement In the recorder determines the manner in whIch the
tape approaches and leaves the head pole pieces. This rate of approach and the length of head-tape contact affects the response of the
recorder at wavelengths of the order of the length of this contact.
When the frequency of the SIgnal recorded on the tape is low enough
so that the recorded wavelength exceeds the total length of head-tape
contact, the amount of flux which succeeds in entering the pole structure depends on the relationship between the contact length and the
wavelength. The result of this dependence is that there IS a scanning
effect producing variations in response at these extremely long wavelengths quite analogous to the gap-length phenomenon occurring at
much shorter wavelengths. The low-frequency scanning effect does
not produce nulls in response as sharp as those of the hIgh-frequency
gap scanning phenomenon but produces response variations of a few
db. These variations in response are sometimes called "head bumps."
The shape of the pole pieces and hence the rate at which the tape
approaches and leaves the head affects the amplitude of the head
bumps. At extremely long wavelengths the head-bump phenomenon
may in some cases, by setting the lower frequency limit, determine the
number of octaves that can be covered at a given speed by a particular
recorder In the direct recording mode.
The depth of the gap, although not as fundamental as its length,
IS an Important parameter of reproduce head design. Gap depth is
naturally related to the useful life of the head, in that any head wear
resulting from abrasion from contInued passage of tape reduces the
gap depth. The fraction of the fringing flux from the tape which
IS "collected" by the pole pieces and induced to cross the gap in the
process of being introduced into the magnetic circuit of the head
depends on the reluctance of the gap compared to the reluctance of the
rest of the head magnetic circuit. As the gap becomes extremely short,
~ven though the reluctance of the rest of the magnetic circuit, constructed of soft magnetic material, is very low, the reluctance of the
gap may be small enough to approach that reluctance if the gap is
also deep.
The magnetics of the process that goes on in the reproduction of
68
MAGNETIC TAPE RECORDING
magnetic tape recording is easiest to analyze by analogy to the corresponding recording action. Calculation of the amount of flux
which crosses the gap on record, and hence the amount of flux involved
in the fringing field, involves the magnetomotive forces induced in
t he magnetic circuit. The source of magnetomotive force is, of course,
the current in the record coil and the total magnetomotive force rise
is usually considered to be developed between the two ends of the
coil. The magnetomotive force drop in the magnetic circuit with a
relatively long gap almost always appears across the gap itself. This
magnetomotive force across the gap is the driving force for the
fringing flux which actually links the tape and does the recording.
On reproduction, the flux fringing out from the tape, driven by
the remanent permanent magnetomotive force originally developed
by the recording fields, appears across the gap and the effective magnetomotive force across the gap drives flux through the pickup coils.
If the gap is so small that the reluctance of the soft magnetic circuit
and of the gap are about the same, much of the magnetomotive force
drop appears along the magnetic circuit and is not avail ruble for
inducing fluX: and hence voltage, in the pickup coil. One way of
looking at these gap dimension phenomena is to consider that the
actual gap has a shunting effect on a theoretical gap of infinite reluctance. This shunting effect, and the effect of wear on gap depth,
make the choice of the gap depth important in the design of a magnetic head. A deep gap will wear for a long time but will have high
shunting and hence less output, all other things being equal.
It IS generally accepted that the factor which determines the recording resolution of the record gap is not the gap length but the sharpness
of the gap trailmg edge. This is an indirect way of stating that the
rate at which the recording flux falls off at the trailing edge of the
gap determines the precision with which the record head locates on
the tape the point at which a particular signal is recorded. The gap
length has a secondary effect on the recording phenomenon. This
secondary effect is related to the head-tape spacing and the recording
medium depth. The practical effect of recording gap length is to
influence the relationship between optimum bias for a particular test
frequency and the tape thickness for linearIzed recording (McKnight
[1961]). For pulse recording, the record gap length has little effect
except on the amount .of record current necessary to produce tape
saturation. In any case, the record gap is always considerably longer
than the reproduce gap.
The spacing between the head and the tape and the thickness of
the active coating on the tape combine to form the third factor of
head-tape geometry which influences the record/reproduce process.
69
HEAD-TAPE INTERACTION
The same basic process is involved in determining the effect of the
thickness of the medium and of the head-tape spacing. The effect of
these geometric factors can be analyzed by inspecting an infinItely
thin layer of medium and determining how the spacing of this layer
from the record or reproduce gap affects the signals transduced by
the gap.
Figure 5-6 shows the fields to which an elementary (infinitely thin)
layer of tape is subjected as it moves past a recording gap at various
distances from the gap. The parameter ylg, the ratio of the distance
from the surface of the head to the gap length, obviously controls
sharply the intensity of the maximum field to which the tape is subjected. Figure 5-7 shows the way in which the field intensity along
the direction of tape movement in the center of the gap varies with
spacing from the head surface. Assuming for the moment that the
gap-length is 1.0 mil and the tape coating is 0.3 mil thick (rather
t.ypical values), and that the coating hes actually in contact with the
surface of the head, it is apparent that the intensity of the field in the
center of the gap for the nearest layer of the tape is about 1.5 times that
for the layer farthest from the head. For the field at the edge of the
gap the ratio of near-layer intensity to far-layer mtensity is almost
infinite. It is thus quite obvious that the recording field intensity is
sharply mfluenced by spacing.
H
(after Dumker)
FIGURE 5-6.-Fields surrounding the recording gap
70
MAGNETIC TAPE RECORDING
IO __----~----r-----r_----._--~
8o
X
•
x
o
10
y/g
(after Mee)
Fl/dt" reproducing head
FIGURE
higher frequencIes, for which the effective permeability of the magnetic material of the head may be low, invalidate the statement that
most of the magnetomotive force appears across the gap. It is more
accurate to say that a controlled amount of the total magnetomotive
force appears across the gap (chapter 5).
The two baSIC kinds of reproduce heads are the so-called "d/dt"
and flux-sensitive heads. The d/dt head is almost identical III construction to the record head just described and "obtains its information" about the flux in the magnetic tape by determining the voltage
induced in the reproduce head winding as the flux intercepted by the
gap and passed through the head changes with time (fig. 8-2). The
output of such a head is effectively the time derivative of the flux
intercepted. The r.emanent flux induced in the tape by the record
head is almost linearly proportional to the record current. The overall transfer characteristic of a record/reproduce system using the
d/dt head is therefore differentiated. In a particular record/reproduce system, such differentiation may be a disadvantage or an
advantage.
'When the rate of movement of tape is extremely slow, the d/dt
output is correspondmgly low. For such appbcatIOns, heads of a dIfferent baSIC operating mode are often used. These are so-called fluxsensItive heads, of whIch there are two basic types. In one type,
the flux mtercepted by the head from the tape is introduced in the
same magnetlc CIrcuit as the flux induced by an aUXIliary signal
appbed to some auxIliary windmgs. A pickup winding detects
changes induced m the aUXIliary signal as the tape flux modifies the
magnetic characterIstIcs of the nonlinear magnetIC cirCUIt used. This
IS the so-called "modulator head" (fig. 8-3). A varIation of thIS type
is one in whICh the tape flux is caused to vary the reluctance of a
TAPE RECORDER HEADS
143
magnetic path where the reluctance is externally measured by some
method other than the carrIer technique described.
The second type of flux-senSItIve head employs the so-called "Hall
effect." The Hall effect occurs when a steady current and a magnetic
flux are applied orthogonally to a material exhIbIt.ing the Hall phenOlJlenon. _The Hall effect occursjn_most of the common semiconductors, and IS especially strong in gallium arsenide. The "Hall voltage"
appears across electrodes orthogonal to the current and flux directIOns
and IS proportlonal to the product of the flux intenSIty and the steady
current intensity. Although very simple and quite effective in many
applications, Hall effect heads suffer from thermal sensitivity and
many other complicatmg properties which have been dISCUssed in the
literature (Stein [1961a]).
MAGNETIC CORE
MAGNETIC
\._,"~_M_A_GN_E--
METHODS
OF TESTING
AND EVALUATING
205
value, that is, one varying instantaneously from plus 1 volt to minus
1 volt, has an average value of 0.636 volt and an rms value of 0.707
volt. All the rectifier meter "knows" is that a signal with an average
of 0.636 volt has been applied to it, and it cannot distinguish well
between a square wave or a sine wave having the same average value.
If a square wave which varies from plus 0.636 volt to minus 0.636
volt, with perfectly sharp transitions, is applied to such a meter, it
will produce the same deflection as a sine wave of 1 volt peak value.
If calibrated to read the rms or effective value of an applied sine
wave, this meter WIll indIcate 0.707 volt whether the I-volt-peak sine
wave or the 0.636-volt square wave is applied.
If one were to measure the output of an overdriven tape recorder
with a peak-reading meter, it would be discovered that the peak value
of the square wave produced by the overloading of the tape would
be about 9.5 db hIgher than the peak value of a sine wave which was
reproduced with 1 percent hannonic distortion. The typical fullwave linear rectifier meter would, however, interpret thIS level difference about 4 db higher (20 loglo 1.0/0.636). Care must therefore
be used in interpreting the manufacturer's saturation specification
for the use of a recorder in applications requiring wavefonns other
than sinusoida1.
Intermodulation is seldom specified for tape recorders despite its
importance. It can be tested in any of the conventional methods,
subject to the limitations mentioned above which origmate in the
equalization.
SIGNAL-TO·NOISE RATIO TESTING
The peculiar equalization necessary to produce uniform response in
a tape recorder is not only responsible for peculiarities in distortion
response but in more important peculiarities in noise response. Because of the large amount of post-equalization reqmred, the noise spectrum out of the tape recorder tends to have the same shape as that of
equalization. This is the result of an approximately uniform noise
output from the tape being equalized along with the signal by the
equalizers (chapter 10). With scientific recorders it is a typical practice to specify several different noise levels in order to guide the user
toward proper applIcation of the recorder. One method of presenting
the noise level of the tape recorder is shown in figure 12-1. The frequency response of the recorder for some reference input level, such as
that for 1 percent output distortion for the distortion reference frequency, is plotted, centered around "0 db." The nOIse over the entire
recorder bandwidth is measured with an averaging meter and this noise
level is indicated by drawing a horizontal line, the ends of which are
located at the upper and lower transmission limits of the filter with
206
MAGNETIC TAPE RECORDING
+20
+10
IP-......
l\
!.
o
''\'
CD
9
...
-10
...J
~ -20
...J
I-
~ -30
I-
C
:::;)
~
0
0-40
E
f-
-50
--
F
-
-60
100
IK
10K
lOOK
FREQUENCY-CPS
(after aft Ampem Corp. drawing)
FIGURE 12-1.-Tape recorder frequency response and noise display.
In this method of presenting the recorder performance, A represents the saturation output signal available from the tape (related to pulse or FM signal
levels), B represents a plot of the 1% disto.rtlOn frequency response, over which
is laid a cross hatch indicating the specification limits, C represents the noise
level observed in a band at the output of the recorder which has filtered limits
represented by the ends of the line C. Similarly for D and E and F the vertical
position of the line and the horizontal location of its end indicates the noise
level and band limits for the particular measurement, respectively.
which this noise was measured. The vertical position of the line indicates the level of the noise signal relative to the reference signal level.
Other lines are drawn for narrower noise-filter bands and typically
these include a whole series of half-octave or third-octave filters.
(The half-octave and third-octave filter values roughly plot out the
shape of the post-equalization of the recorder.) This detailed noise
specification is essential to the user of the recorder who wishes to "dig"
a particular kind of signal out of the nonnal recorder noise. If he is
shown the way in which the signal-to-noise ratio for a particular
narrow band of frequencies varies over the recorder pass band, he is
better able to decide how to apply the recorder than if simply given a
single broadband noise value.
It should be emphasized that the exact way in which the noise band,
of any width, is filtered is most important in determining the actual
signal-to-noise figure that can be quoted, particularly for the verywide-band recorder (Ratner [1965]).
METHODS
OF TESTING AND
EVALUATING
207
{;
As of the moment of writing this survey, it is difficult to say whether
tape or equipment is the major limitation of recorder noise. It is
standard practice to attempt to make high-performance recorder electronics provide a noise level somewhat better than that determined by
the tape itself. As tapes have been improved, the manufacturer has
been forced to improve his input system. The state of the art is currently such that it is becoming increasingly more difficult to achieve
this electronic design goal. Signal-to-noise ratio has been discussed
in considerable detail in chapter 10. It is appropriate to add here only
that it ~s often a practice to indicate a system noise and an equipment
noise. The system noise is that measured with erased tape running
through the entire system and the equipment noise is that measured
with the equipment operating as close to normal as possible but without tape contacting the reproduce head. Typically, the equipment
noise is measured with tape moving but isolated from the head by a
spacer so that such noises as hum from taklmp and-supply or capstan
motors are included in the overall noise measurement.
Many ingenious schemes have been proposed for improving the relative signal-to-noise ratio of the tape recorder by such techniques as
automatic switching between channels as a function of applied level.
Although for specific applications such techniques may be useful, they
are not considered to lie within the scope of this survey.
FLUTTER TESTING
In the simplest terms one may say that flutter is measured with a
flutter bridge. The simplest flutter bridges are derived from the
deviceS qesigned for disk- and film-recording measurements in theatrical and other audio systems. These often consist of a Wien bridge
selective circuit tuned to discriminate against the particular test frequency used and calIbrated to indicate the amount of average deviation
from this frequency as the test signal is fed through the bridge. A
simpler form of the flutter meter used with scientific recorders is now
preferred as a more sophisticated audio flutter meter.
Such a flutter meter is basically an FM detector. A test signal is
recorded on the tape and is then reproduced through this FM detector.
If no flutter were present, there would be no signal output from the
detector. Flutter appears as the output signal from the detector and
may be measured in many ways. Flutter is sometimes recorded on a
graphic recorder particularly when detailed analysis of the flutter performance is being made in the process of recorder development. Flutter may alsq be averaged and calibrated to indicate as a r~ figure.
The more critical flutter meters present single- or two-sided peak-toflutter data.
208
MAGNETIC TAPE RECORDING
~
As has been discussed in cha.pt.er 7, flutter occurs at a wide ra.nge
of frequencies. In the specificatIOn of flutter it is essential that the
bandwidth of the flutter components included in the measurement be
noted.
An audio flutter-measuring device of good quality may use a test
frequency of 3,000 cycles per second and include all flutter frequencies
up to 250 cycles per second. According to commonly accepted practice, in this band up to 250 cycles per second, all those flutter
frequencies are included which are perceptible to the ear. For a scientific recorder, however, such a narrow band of flutter frequencies does
not describe recorder performa.nce adequately. The recorder tha.t is
to be used for FM recording, particula.rly for WIde band FM recording, is limited in its FM signal-to-noise ra.tio by Its flutter performance.
If, for example, an FM band 10 kilocycles wide is to be recorded in
actual use of the recorder, no flutter measurement of this recorder
would be adequate unless it included flutter frequencies up to the same
10 kilocycles per second. It is, therefore, necessary to make a much
wider band flutter measurement on such a recorder than on an audio
recorder.
The test frequency obviously must be much higher than 3 kilocycles to measure a 10 kc bandwidth and typically this test frequency
is 10, 50, or 100 kilocycles per second. It is also a practice to present
the results of the flutter measurement by passmg the output of the FM
detector through a varia.ble ba.ndwidth filter and to plot a level of
output from this filter as a. function of the filter ba.ndwidth. The data
are then labeled "cumulative flutter components up to a frequency
of-," where that frequency is the hOrIzontal aXIS (fig. 12-2). As a
more conservative means of presenting flutter, peak-to-peak values
¥ 100
II
4
~ 90
f:!
¥
4
It
80
j~
~ 60
~40 l..--- I--
1&.1
30
~
20
5
10
~
0
>
::E
u
10
33/4
/
70
Ii 50
II
1718
/'"
-
:.- :-
,...."
/
V
~2
l..- ~ ....
~ I.- j..to
100
I--
15
30
J,...-'
~ ~ I-"'"
60
V
i""""
IK
10K
HIGH CUTOFF FREQUENCY. CPS
FIGURE
12-2.-Cumnlative flutter presentatIon, with a fixed lower band llmit
and a variable lugh cutoff frequency
METHODS
OF TESTING
AND
EVALUATING
209
are usually given. (Current flutter performance of a high class instrumentation ground-based recorder may be 0.3% peak-to-peak with
a bandwidth of 10 kIlocycles per second.)
Although the plot of the measured peak-to-peak flutter of a particular high performance recorder may display a few irregularities at the
lower filter bandwidths, it is usual for a smooth, slight rising curve
of flutter versus filter bandwldth to be presented for high-frequency
filter cutoffs. The flutter produced above about 1 kilocycle per second
is almost all "hash" produced by shock exertation of the tape from
scraping and friction. The smoothness of the flutter curve so presented often masks flutter problems which may be important for certain applications, particularly in FM recording.
The most exacting flutter performance measurement is made by
searching the output of an FM detector with a narrow-band filter.
Typically a fixed-bandwidth filter is traversed through the frequency
region from zero to approximately 10 kcps or higher. Often this
traverse is done by an automatic graphic recorder. An automatic
traverse with a one-third octave bandwidth filter is not possible with
any commercially available filter since it would require continuous
filter adjustment. However, different parts of the spectrum can be
examined with different fixed bandwidths to approximate the same
results (fig. 12--3).
It is the usual practice to reduce such a detailed flutter spectrum
measurement to the flutter per cycle basis. This means that the results
are presented as if a I-cycle-wide filter had been traversed through
the flutter spectrum. Traversmg with such a narrow filter would be
a tedious and inaccurate process. With a 10-cycle filter, however,
the search can be done quite easily~ven more so for a 50- or 100-cycle
filter. The total flutter energy intercepted in a lO-cycle band can
usually be assumed to be 10 times as great as that mtercepted in a
1 cycle band and in the same proportion for other bandwidths. If
the scale of the flutter spectrum is thus modified to take care of the
ratio of bandwidths, 10-, 50-, or lOO-cycle data can be presented as 1
cycle data.
Such a detailed flutter spectrum is most useful for FM recording
of complex signals, the spectrum of which must be searched in later
analysis. As noted above, the typical flutter spectrum is fairly smooth
with pronounced spikes Ql' local irregularities at particular frequencies. These spikes often appear at the vibration frequency when a
recorder is tested under vlhration. Otherwise, they occur at frequencies related to once-around reels, capstans, capstan idlers, guide rollers,
etc. Although the broadband flutter, and hence FM noise performance, of a recorder may be quite adequate for some applications, the
..,~ 10-~
....
....
zUJ
rlAr
~ 10-6
~
UJ
0:
~ 10- 7
o
III
Z
~
::l 10-8
10- 9
J!1
"
10
!\
~~
20
30
~
..
r-~
\\0 J' ~
40
50
60
'M ""1
70
80
90
"
~: ~~~
v
\I
100
200
300
400
500
600
,
~ ~
\~
700
800
900
1000
FREQUENCY (CpS)
FIGURE l2-3.-Typical instrumentation reeorderjreprodueer flutter spectrum. The flgure is traced from an original plotted
by a graphic recorder-the calibration was adjusted for the di1ferent bandwidths used above and below 100 cps to give
results in flutter per cycle.
METHODS
OF TESTING
AND EVALUATING
211
use of such a recorder for detailed subsequent spectrum analysis may
be severely limited if large flutter spikes are present. It is therefore
sometimes a practice to specify the l-cycle flutter spectrum for an
application in which detailed analysis will later be required.
As a practical matter the usual high-performance flutter measurement is made with a telemetry discriminator and the crystal oscillator
source. The IRIG specification, for example, calls out the discriminator made by a particular manufacturer, without comment, as being the proper instrument to use for routine flutter measurements.
Ratner (Ratner [1965]) points out that the practice of specifying
broadband flutter in a lO-kcps band is not very useful with 400-kcps
FM systems now in use.
TIME-DISPLACEMENT ERROR TESTING
Time-displ{l.OOlP,ent error is n()w be.ing ~ut~ely tested as an important recorder characteristic. Time-displacement error is closely related to flutter since in the simplest sense it is proportional to the time
integral of flutter (chapter 7). In the information-theoretical sense,
time-displacement error may be the measurement more fundamentally related to the ability of the tape recorder to transmit information.
The time-displacement error is a direct measure of a contribution
to the total uncertainty of the value of the recorded variable at a
particular time. With a given signal-to-noise ratio the value of a
particular variable at a particular time is known to a certain level
of accuracy, as limited by the noise. At the same time, the total time
displacement error is a measure of the uncertainty with which it can
be stated at what time the variable had the specified value.
Two kinds of time displacement error testing are normally used.
The most common is the so-called pulse-to-pulse test which is related
not only to the more fundamental considera.tions noted above, but to
the utility of the recorder for certain classes of pulse recording. Performance for PDM, PFM, a.nd PCM recording is evaluated with this
kind of pulse-to-pulse measurement.
The pulse-to-pulse measurement is made by synchronizing a good
oscilloscope with each pulse reproduced by a recorder. A crystalcontrolled pulse signal is then fed into the recorder and the oscilloscope set to trigger the start of its sweep as each pulse arrives. The
actual pulse itself next in line is seen at the far end of the oscilloscope
trace. Jitter along the base line of the pulse at the end of the line
is a measure of the stability of the pulse-to-pulse spacing or relative
time displacement error. No specific method for recording this is
usually available although a time exposure of the oscilloscope display
gives a better measure of the total pulse-to-pulse error than visual
observa.tion.
212
MAGNETIC TAPE RECORDING
Visual methods are also used for absolute time-displacement error
measurement. A standard method of making thIS measurement consists of recording a crystal-controlled square wave directly on the
tape and reproducing this square wave to be displayed on an oscilloscope. The oscilloscope horizontal sweep is triggered from the same
crystal oscillator as was used for recording. In the oscilloscope display, the jitter along the base line of the transitions of the square wave
is a direct measure of the time-displacement error between the local
crystal and the reproduced square wave. Depending On the performance of the recorder, a wide range of display techniques may need to
be used; time-displacement error of this class in instrumentation recorders varies from -+-0.2 microsecond to ±0.25 millisecond.
DROPOUT TESTING
Dropout testing is essential in the production of digital magnetic
tape and its pre-evaluation for use. No tape can be sold today for
digital use unless it has been 100 percent tested for some particular
dropout performance. The cost of (digital) computer tape is often
related directly to the packing density for which it has passed 100
peroent dropout free test. For many applications in instrumentation
recording such dropout testing is also important.
A dropout test consists of recording a pulse or other signal on a
tape and reproducing this signal while observing the continuity of
the reproduced level. Automatic dropout testing involves signal level
detectors which indicate a dropout when the signal drops below a
particular value. An extremely elruborate dropout tester has been
built by one firm in which several different decreases of signal level
are separately measured, if the signal drops 3 db at one point, this is
noted as a "3 db-dropout"; if it drops as low as 6 db below normal
this is noted as a "6 db-dropout," and so on for several different decreases in signal level. Although this device has been proposed as a
useful and sophisticated tool for placing instrumentation and dIgital
tapes in order of quality for dropouts, its primary utility is to the
manufacturer of tape who is interested in observing the level of performance of his tape-manufacturing process. For the tape user, a
dropout is usually an error, and he is interested in specifying one
particular level of dropout which he can barely tolerate and is interested only in whether the tape achieves no dropouts as deep as
that level or not.
The user is also concerned with the presence or absence of dropouts
as a function of recording density. The computer tape manufacturer,
therefore, tests his tape first for dropouts at a high recording density.
For example, he may test first at 800 pulse bits per inch. If the tape
METHODS
OF TESTING
AND EVALUATING
213
passes this test, it is put into the "800-bit-per-inch" bin and sold at the
highest possible price. Tape which fails this test may then be tested
at 556 bits per inch and, if it passes, placed in the "556~bit" bin. This
may progress on down through one or two intermediate stages until
the tape either passes or fails a 200 hit per inch test. If it fails this
test it is no longer considered useful for computer tape and is often
sold to the unfortunate instrumentation tape buyer.
So little tape passes the more sophIsticated of these tests and even
some of the less sophistIcated ones that the inspection process involved
in computer tape manufacture also involves means for repairing
dropouts. It IS a typical practice when a dropout is found by the
automatic tester to have an operator attempt to polish out the nodule
or tape defect which caused the dropout. From observing the level
of activity of such inspectors in a tape manufacturing plant, one
concludes they are responsihle for the salability of mal!Y reels of
otherwise unusable computer tape.
TAPE TESTING
The testing of tape is now a fairly standardized operation and there
are military specifications for tape of various kinds. Such tests include the measurement of the physical and mechanical properties of
the tape, a simple measure of its output level, short-wavelength performance, and signal-to-noise ratio mcluding both dc and ac noise.
These tests have the value of being standardized and of having commercial validity because they are agreed to as a formal specification.
The substance of these tests is best obtained from the appropriate
milItary or NASA procurement specIfication.
For many applications these standardIzed tests do not go far enough.
For example, lubrICation of the tape is all-important for endless-loop
service. No standardized tests are availabl~ for tap~ lubrication
but it is certain that the user of an endless-loop recorder requires
some kind of test to estrublish whether the tape can be used in his
recorder at all or not. In the same way, for many applications, the
actual life of the tape must be tested, particularly under extreme environments. No standardized tests have been worked out for these
special cases,
ENVIRONMENTAL TESTS
Tape recorders and tape for satellite and space probe use must pass
severe environmental tests. Typical vihration, shock, temperature,
humidity, and the rest of the space environment tests can be applied
to the recorder. These need differ m no fundamental way from any
other environmental tests. Exceptions may, however, need to be
214
MAGNETIC
TAPE RECORDING
taken because the tape itself cannot pass a really severe heat test.
Specialized ways of applying heat tests and of attempting to assure
recorder reliability without the full application of the tests is often,
therefore, necessary.
Other tests which recorders can pass if carefully desIgned but which
require careful design are those of shock and vibration. Recorder
VIbration testing and isolation is discussed in chapter 11.
CHAPTER 13
Miniature High-Environment Recorders
As a valuable means of documentation, the tape recorder has accompanied every step of development of space technology. The la.rge,
ground-based instrumentation recorder has partiCllpated in the process
of developing space technology from the beginning and will continue
to do so WIth steady refinement from its "pre-space" form. Tape
recorders small and rugged enough to accompany experimental vehicles did not, however, exist when the space program started. Such
recorders had to he developed from the crudest of beginnings to meet
the Increasing requirements of progressively more sophisticated
programs.
The first rugged mmature recorders were intended to ride right
along with test rockets, the resulting recordings being recovered after
the test and reproduced on ground equipment. The recorder was
required only to withstand the test environment and then simply to
allow the record to survive the abrupt return to earth. In later
stages, rocket sled recorders had to continue to record during shocks
comparable to those encountered in a crash landing.
As the technology progressed, recorders which could survive insertion into orbit
and then play bltCk data to the ground on command were needed.
Ruggedness, SUl"Vlval and performance were no longer enough; the
recorder also had to use very little power and t.o be small and light
in order to be included in the payload of the tiny booster capacities
initially available.
The technology was inadequate 00 the tasks of the initial attempts.
The flight recorder acquired at one time a very bad reputation in the
space program because recorder failure at a crucial moment could
cause the l~ of most of the time and energy that went into a large
experimental project. This temporary reputation, combined with all
the other problems of developing a new device, resulted in the subsequent Bight recorder development program being best described as
215
216
MAGNETIO
TAPE
REOORDING
conservative. If one considers the situation where a large part of
the utility of a smgle sateUite launching, which may cost many million dollars, may depend on data which is recorded on relatively
fragile, normally heat-sensitive magnetic tape, the conservatism seems
more than justified.
NASA has purcha,sed many types of recorders a,s have other government agencies. NASA has also sponsored, directly or through subcontractors, the development of recorders to meet requirements where
no commercially avwilable unit met the need. This development has
been concentrated in the field of flight recorders for severe environments. As a major customer for ground-based recorders and the
various other eqUlpments involved in data acquisition and reduction,
NASA has naturally influenced the development of this equipment
cla,ss as well, as would any major customer. Direct NASA sponsorship of new recorder technology, however, has been concentrated on
the fl'ight recorder.
The historian of the art of high-environment recording encounters
certain difficulties because of the nature of the space program. Relatively small quantities have been needed of each specific recorder developed for a particular application. There may have been a couple
of engineering models, a prototype or two, and a certruin number of
flIght models, some of which were actually flown and 'Others of which
were employed entirely in preflight testing. Detailed instruction
books and the usual amount of documentary ma.terial generated in
the process of producing a production device is often lacking because
of the small quantities involved. More documentation could very
well have been done for some of these units, but the speed with which
the space program has moved forward has militated against it. The
result is that there is very little formal documentation many specific
recorders, some of which included features which were genuine innovations. In the following sections, an attempt 18 made to include
those recorders for which the best documentatIOn is available, emphasizing the innovative features. To assure a,s complete a review
of the art as possible, many recorders developed for other services of
the government are included, a,s are recorders which are proprietary
developments of manufacturers.
Despite the impres:;ive performance capability demonstrated by
flight recorders currently being applied to the most advanced space
programs, some of the older and less sophisticated "workhorses" are
still in use. They should therefore be included in any general review
of the development of the present state of the art of high-environment
recorders. To provide thIS historical function and at the same time
to describe the current state of the art, the next sections of this chapter
will consider in order:
or
MINIATURE
HIGH-ENVIRONMENT RECORDERS
217
(1) the reel-to-reel recorder used with ground-based playback
from physically-recovered tape,
(2) the reel-to-reel recorder commanded in flight to play back
its recorded data, and
(3) the endless-loop recorder.
The unusual recorder formats required for special tape drives will
then be discussed, as will transverse-scan flight recorders. Finally,
certain mechanical design problems peculiar to miniature recorders
will be considered.
REEL-TO-REEL, RECORD-ONLY RECORDERS
One of the earliest forms of miniature high-environment recorder to
be developed and used is shown. in figures 13-1,2,3 and 4. Recorders
of this type were developed and supplied by several companies including Leach Corporation, Borg-Warner Controls, Astro-Science,
and Cook Electric. This recorder usually held 50 to 100 feet of
half-inch or I-inch tape in a reel-to-reel configuration. It is no
criticism of the device to say that it was "brute force;" the entire
objective was to get a mechanically rugged device into as small a
space as possible. In some of the early sounding rockets in which it
was used, not more than a 3- or "4-inch diameter tube was available
for all the instrumentation and the recorder had to share this space
with other devices. The flutter was often high and the carrier-erase
technique of recording (chapter 4) was widely used since this permitted dc response and required minimum record electronics complication without the flutter-sensitivity of FM.
The initial tape metering and tensioning mechanism was extremely
elementary. Some recorders used capstans with rubber pinch rollers
in a conventional manner with mechanical drag on the supply reel
and a slipping clutch on the takeup reel. The design concentrated
entirely on getting some kind of a record under extremely high impact; the actual amount of power used was often not too important.
Particularly with the short record time required, the design could be
somewhat inefficient and still be satisfactory. When longer record
times were demanded and the recorders were used in more sophisticated systems the efficiency and hence the method of tape drive had
to be reexamined.
The following slightly edited statement by the chief engineer of
one of the pioneer manufacturers of this class of recorder is selfexplanatory in its coverage of the development of current capability:
"The original machine was puck-driven by a dc motor and was used
for carrier-erase recording only. Since the original development,
a gear drive has 'been substituted for the puck drive and an ac motor
218
MAGNETIC TAPE RECORDING
(photo courle8fl Leach, OOrp.)
FIGURE
13-1.-The Leach Model MTR--862 miniature high-environment recorder.
(ph~o
]j~IGURE
courle8fl ABtro-Science Oorp.)
13-2.-The Astro-Science Model TR-1875 miniature high-environment
recorder.
The tape path can be deduced from the parts of the tape visible as well as
positions of the two reels on the extreme lower left and right with the pinch
roller visible in the center left. This view shows early-1900's electronic construction in the rear of the mechanism.
MINIATURE
HIGH-ENVIRONMENT REOORDERS
219
(a)
(b)
(photos rour.tesy Cook Eleotric Co. )
FIGURE 13-3.-Front (a) and rear (b) views of tbe Cook Model MR-51 bighenvironment tape transport.
Note in (a) tbat the takeup reel is surrounded by a rugged housing to protect
tbe recorded data. Note also, in (b) tbe gear-tootb-tonewheel speed-sensing
mechanism. The tape path is clear in this unit as are the pressure pads for
maintaining head/tape contact.
220
MAGNETIC TAPE RECORDING
(photo courte8Y Borg-Warner (Jontrol8)
FIGURE 13--4.-The Borg-Warner Model R-10l miniature high-environment recorder (the photograph is of an early version of this recorder which bas been
renumbered) .
The small space for electronics shown in this photograph probably indicates
that this particular unit took advantage of the simplicity of carrier-erase
record'ing technique to limit the amount of plectronics necessary.
has been employed to improve the speed regulation and environmental immunity of the machine. The simplicity of the machine is
what makes its high-environment characteristics possible. Basically,
it has a single capstan with a permanently engaged rubber pinch
roller. The supply reel is held by means of a friction brake and
the take-up reel is driven by means of a rubber-covered roller driving
the perimeter of the reel. Slippage occurs in the center drive for
the rubber-covered roller. The tape path is simply across the head
and through the pinch roller onto the take-up reel.
"Environmental test data have proved that it can start and run at
400 G's sustained acceleration with less than 3 percent peak-to-peak
flutter. The machine will run at 600 G's sustained acceleration if
started at a lower level with the same flutter characteristics. Shock
tests have been performed up to 610 G's for 3 milliseconds. In this
case, the peak-to-peak flutter does not exceed 2 percent. Higher shock
levels up to 1,200 G's for shorter durations have been performed. In
these tests, a stutter is observed with a recovery time approximately
50 milliseconds. The machine operates through extremely high vibration environment with a random input to the shaker of 0.8 G2 per
cps."
MINIATURE
HIGH-ENVIRONMENT RECORDERS
221
The development of this basic type of recorder was carried much
further. Larger reels, more sophisticated. tape moving mechanism,
and more conventional electronics were added to its ca.pahilities.
Units of this general class, using 8ibout one-inch-wide tape moving
from reel-to-reel in a very straightforward array, eventually acquired
differential capstans and quite sa.tisfactory ta.pe motion. Such recorders were available essentially as commercial items fairly early in
the high-environment testing business and were supplied by several
organizations. An example of a typical recorder of somewhat larger
capacity than the original cylindrical units developed before the move
to differential capstans is shown in figure 13-5. It is essentially a
small IRIG-standard instrumentation recorder for severe environment.
Quite a few straightforward tape recorders were developed for such
record-only service as rocket sled data collecting a.nd ejection seat
testing, where size was relatively less important. These recorders
differed radically in format and appearance from the miniature cylindrical type just discussed. A typical example is shown in figure 13-6.
They use 1-inch-wide tape, operate between 10 and 120 inches per
second, and usually record on from 8 to 15 data tracks, with rms flutter
below 1 percent. They a.re driven by shunt-wound dc motors, often
governor-controlled, and usually weigh between 9 and 20 pounds. A
feature of these recorders is that the tape, as it is recorded, passes
into a cassette (shown on the right in the figure) of very rugged construction. The idea is that, even though the recorder mechanism
itself may not survive the impact, the tape in the cassette will do so.
Recorders were also designed to allow the record to survive reentry
tests by being ejected just before impact of the reentry vehicle itself
so that the recorder could be recovered after not quite as rugged an
impact as that suffered by the test vehicle proper. Such recorders
often actually recorded under rela.tively mild environments. One
version of such a. device is shown in figure 13-7. It is 51h inches in
diameter by about 7 inches long when encased, and weighs 5.3 pounds.
It holds 900 feet of quarter-inch tape, operating at 45 inches a second
for 3.6 minutes. In one particular application for this unit, IRIG
FM subcarriers are recomed and flutter compensation can be applied
to improve the overall flutter performance.
Clearly an offspring of the original reel-to-reel rugged recorder,
but very specialized and subjected only to about the same environment as an astronaut is a recorder recently developed for bio-medical
use in the Gemini spacecraft. It is shown in figure 13-8. The over.all dimensions are 9 x 6.5 x 1.7 inches and the weight 4 pounds plus a
half pound for tape. It records seven cha.nnels in direct mode for
222
MAGNETIC TAPE RECORDING
(a)
(b)
(phOt08 courte8Y L each Corp.)
FIGURE
13-5.-The Leach Model MTR-l200 high-environment recorder.
(a) A closeup view of the reel and head structure, (b) an exploded view of
the entire recorder showing another version of modular electronic elements.
MINIATURE
HIGH-ENVIRONMENT RECORDERS
223
(photo courtesy Oook Electric 00.)
FIGURE 13-6.-The Oook Electric MR-31E high-environment tape transport.
Note the rugged construction of the cassette, upper right, designed to protect
the recorded tape no matter what happens to the recorder.
100 hours using a tape capacity of 880 feet, operating at 0.0293 inch
per second. Such data as electrocardiogram and electroencephalogram signals, blood pressure, temperature, respiration, and galvanic
skin response are recorded by this device. The use of a direct record
mode at this very low speed is relatively unusual.
The playback of the data from this recorder is done through a
special preamplifier designed to deal with the phase problems existing
at the very low fmquencies involved in bio-medical recording. The
tape is played back after recovery from the capsule (along with the
astronauts) at 16 times its record speed or at 0.4688 inch per second.
It is copied onto another tape, also operating at 0.4688 ips in an FM
mode to preserve low frequencies and dc response. The first copy is
then played back at 16 times its record speed or at 71h inches per second through a conventional FM-mode pla.yback amplifier for final
data reduction. Thus a 256-to-l speed increase is achieved.
224
MAGNETIC
TAPE
RECORDING
(photo courtesy Cook Electric Co.)
FIGURE
13-7.-The Cook Model DR-25--2 recording system designed for recoverable-capsule service (see text) .
The recorded-tape cassette of rugged construction is visible to the right of
the center line.
REEL-TO-REEL RECORDERS, RESPONSIVE TO PLAYBACK COMMANDS
The recorder shown in figure 13-9, designed for use on the Mariner A
program, is at the same time one of the last stages in the development
of the recorder type discussed above and an early example of the
recorder group of the current heading. This recorder was developed
to replace one similar to that shown in figure 13-5, but with capacity
to accept playback command. It records and plays back at the same
speed.
The recorder carries 80 to 90 feet of I-inch tape which passes a
14-channel record and playback head driven by a differential-capstan
scheme at 15 ips. The two capstans each have rubber pinch rollers
and the downstream capstan (the one on the takeup side) is driven
at 2 percent faster surface speed than the upstream capstan. The
pinch roller pressures are so adjusted that the upstream capstan is
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
225
(photo courtesy Cook Eleotric Co.)
FIGURE
13--8.-Tbe Cook Model DR--3OC--7 biomedical rerording system.
This is the Gemini biomedical recorder. Note the single head with its thick
pressure pad (right of upper center) and the extremely long thin motor (above
and to right of head).
the metering element and the tape moves essentially at the speed of the
surface of this capstan. The downstream capstan is designed to slip
but in so doing to maintain a controlled tension in the tape.
The drive motor for this unit is a 6,000 rpm, 400-cycle hysteresis
synchronous one requiring 3112 watts and operating at an efficiency
of approximately 40 percent. In the development of these transports
it. was discovered that conventional motor design could not approach
this efficiency by as much as 2 to 1 and a specialized design was required.
The capstan is driven at 600 rpm through a two-stage reduction
using Mylar belts. The capstan pulls the tape off the supply reel and
the supply reel drive is supplied by a Mylar belt to the takeup reel
through a clutch differential. The belting and pulleys are so arranged that the takeup reel is always slightly overdriven. This, of
course, requires that a maximum amount of slippage 'b e dealt with.
That is, the full difference in speed between the supply and takeup
reel is consumed in the clutch. Driving in this fashion avoids the
use of two separate clutches, one acting as a brake on the supply reel
and the other acting as a clutch on the takeup reel, with the consequent potential unreliability. One of the interesting aspects of this
recorder is that, with indirect drive of the takeup, a complex belting
system, and a motor with an efficiency of 40 percent, it still can be accelerated or reversed in 1 second.
226
MAGNETIC
TAPE
RECORDING
(photo from a NASA report)
FIGURE
13-9.-A reel-to-reel tape transport designed for the Mariner A program.
This recorder, although never flown, was given at least preliminary
vibration tests and survived them. It is interesting to note that there
was no means of relieving the pressure between the. pressure roller
and the capstan and it was therefore possible for a flat to develop on
the pressure roller. The probability of the elastomer surviving a
temperature cycle and a long storage period with a continuously applied pressure seems not very great.
A basic problem of a reel-to-reel recorder that must record and
playback on command is that it must not only be able to reverse itself,
but must he provided with various safety mechanisms to be sure that
the tape is, in fact, stopped or reversed before the end is pulled off
the reel. On many occasions the author of a paper who is busy
apologizing for some shortcomings of an endless-loop recorder justifies the use of such a recorder on the basis of its avoidance of the
complications of control of a reel-to-reel device. Given the control
complications, the recorder engineer immediately attempts to put as
much storage capacity into the recorder as possible to justify them.
The Mariner A recorder described above was therefore probably the
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
227
only reel-to-reel recorder subject to ground command which held as
little tape as 75 feet.
It does not seem possible or profitable to attempt to follow in detail
the development of the current reel-to-reel ground commandable recorder. There are many such recorders currently available in a relatively fixed format, plus a few somewhat unusual ones for special
applications. The discussion of this section will be limited to a description of representative recorders of this class.
The most widely-used reel-to-reel flight recorder has the reels
mounted co-axially, uses differential capstans, substitutes tape wrap
for the pressure roller at the capstan and maintains reel tension by
means of N egator springs. Examples of such recorders are shown
in figures 13-10, 11, and 12. The Negator spring-tensioning system
is implemented differently by each manufacturer and sometimes varied
by the individual manufactur,e r for different applications. A representative tensioning system is shown in the photograph figure 13-13
and in the diagram figure 13-14.
(photo courtes1/ Leach Corp. )
FIGURE IS-IO.-The Leach Model MTR-2IOO coaxial-reel flight recorder.
228
MAGNETIC TAPE RECORDING
(photo courtesy Precision Instrument Co.)
FIGURE 13-11.- Tbe PI type PS--S03T coaxial-reel flight recorder.
Many of these recorders operate at one speed for record and another
playback. Some of the more subtle aspects of achieving this
dual-speed operation are discussed in a later section. Two rather
straightforward ways of achieving dual-speed operation are shown
in figures 13-15, 16, and 17.
Figure 13-15 shows a capstan-drive system operated at two speeds
by a single motor operating always at the same speed. It involves
the use of a magnetic clutch to perform the speed. shift. (It may be
assumed that this is for a unit which records relatively slowly and
then plays back rapidly.) In the slow-record mode, the motor. drives
the shaft pulley and clutch element combination labeled "High Speed
Input" through a plastic belt speed. reduction. This jackshaft is
belted to another intermediate shaft which carries the flywheel and
which, in turn, is belted to another jackshaft which is so labeled in
the drawing. The second jackshaft turns somewhat slower than the
flywheel and, in turn, the pulley labeled "Low Speed. Input" is driyen
at a still lower speed by an a.dditional plastic belt. When the magnetic clutch is not activated (this is the condition shown in the drawing) , the low speed input drives by clutch friction the shaft labeled
"Clutch Output." This shaft is belted in turn to one of the two capstans which is connected by still another belt to the second capstan
for differential action. When the clutch is activated, the cross-hatched
fo~
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
229
(photo courtesy Ralph M. Parsons Co.)
FIGURE 13-12.-The Parsons Model AlR-940 coaxial·reel flight recorder.
This particular version of this recorder is currently being supplied to the
N.AJSA Flight Research Center at Edwards Air Force Base and is representative
of the large number of variations on this basic format manufactured by the
Parsons Company.
(photo courtesy Ralph M. Parsons Co.)
FIGURE 13-13.-Negator springs as employed to provide holdback and takeup
torque in a coaxial-reel recorder.
The "S" configuration of the springs can be S('en, as can the gears needed to
transmit the spring torque to an output shaft.
230
MAGNETIC TAPE RECORDING
(drawing oourtesy Ralph M. Parsons 00.)
FIGURE
13--14.-Exploded mechanical schematic of the Negator tension system
shown in figure 13--13.
This configuration of springs forces the two reels to rotate in opposite
directions.
(from a Leach Oorp. drawing)
13--15.-A solenoid-clutch-operated speed-change mechanism for a differential-capstan flight recorder driven by a eingle motor (see text).
FIGURE
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
231
member, which is mounted by a bellows to the output shaft of the
clutch, is drawn away from the low-speed input point and presses
against the high-speed input. The clutch output then is driven at
the higher speed.
The mode chosen when the clutch is activated is the high-speed
mode because the high-speed. mode obviously lasts for shorter length
of time. The clutch current is therefore needed only for a short time.
The particular form of clutch shown desirably does not have any
movi,ng magnetic elements, The flywheel is used as an intermediate
element in the low-speed drive where its filtering action will be more
effective because the filter action is more important at the low speed,
This explanation and drawing is somewhat over-simplified and diagrammatic but is intended to illustrate here the principle of the active
clutch for speed change.
An alternate method of obtaining two-speed operation which involves two motors but no mechanical clutches is illustrated in figures
13-16 and 17, which were provided by the manufacturer. In this operation, the speed of the capstan A is dependent on the algebraic sum
of the speeds of the two possible drive pulleys Band C. For lowspeed operation the starting friction of the high-speed motor multiplied by the belt reduction from that motor to pulley C essentially
locks C in a fixed position. The low speed motor then drives Band
B in turn drives the capstan A. With the control logic switch in the
other position, pulley B is essentially locked and the speed of A is
dependent on the speed of pulley C. In this particular device, as
shown in the photograph of figure 13-17, there is also a reduction ratio
within the differential proper. The nearer large pulley carries the
pivots for the small pulleys and the entire structure rotates as one.
The far pulley is locked to the smaller diameter of the two pulleys on
the center shaft. If B, the far pulley, is locked in position and C
is caused to rotate, the small idler pulleys on the outside of C rotate
around their own pivots as well as the whole assembly rotating. The
motion of the small pulleys is transmitted to the larger center shaft
pulley according to the reduction ratio between the two pulley-pairs
involved. This larger center pulley drives the capstan. The assembly then provides, if B is allowed to rotate, an output speed which
is the alegbraic sum of Band C but modified hy the small pulley
diameters involved.
Many versions of the co-axial reel-to-reel recorder are availllible
commercially and they have been applied for many different services.
The general physical arrangement of these recorders is apparent from
the photographs. Although the developers of these units emphasize
individual novel features in recommending their use there is a certain
'188-0080-60----16
232
MAGNETIC TAPE RECORDING
TAPE
BELT
REDUCTION
CAPSTAN
BELT
REDUCTION
(drawing courtesy Ralph M. Parsons Co. )
13-16.-A two-motor two-speed flight recorder drive system using a
differential to connect the separate drives to the capstan without clutches (see
text) .
FIGURE
(photo courtesy Ralph M. Parsons Co.)
13-17.-A plastic-belt-driven differential capstan drive for a flight recorder, such as used in the system of flgure 13-16 and as described in the text.
]j'IGURE
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
233
similarity between the various designs which will be described by reference to figure 13-12. (This figure is chosen not to recommend or
criticize its particular design but because the photograph shows the
tape path more clearly.) It will be noted that the tape is led around
two sides of the case from its exit from the reel to its engagement with
the capstan. This is done to minimize the twist imparted to the tape
in transferring from between the reels located at two different levels.
Although the twist cannot be avoided, the unequal stress it places on
the tape is minimized by spreading it out over a long distance. Another common feature visible in this recorder is that the path of the
tape around the capstan is so arranged as to provide almost a 270 0
wrap. (The capstan is the lighter-colored roller nearer the camera,
just to the left of the head.) The same kind of wrap is provided at
the far capstan but it appears from the photograph that the actual
wrap is accomplished with an unflanged roller, while the side guiding
of the tape is being handled by a flanged roller some distance from
the capstan. It will be further observed that the case of this recorder
is quite ruggedly constructed and has provision for gasket sealing
against the environment.
The electronics is also of modular construction which is typical of
such devices. In figure 13-10 a somewhat lighter construction of
recorder is shown where a somewhat different technique is employed
to minimize the amount of tape twist.
It must be realized that there are many forms of these recorders
and that every manufacturer individually modifies his principal design for different applications. However, the general functional and
structural concept appears to be quite sound, which explains the wide
use of this general format.
In addition to the co-axial recorders just discu!'Eed., which have
been very widely applied, it seems appropriate to d~ribe in some
detail some specific NASA-sponsored recorders which have been developed for particular programs. Three such' recorders will therefore be described in the following paragraphs.
A coaxial-reel recorder of fairly standard fonnat but exhibiting
some unusual features was recently developed for the Gemini program. This recorder is. designed to receive two channels of PCM
data from on-board systems of the Gemini capsule, to record this
data for 4 hours at 1% inches per second and to play it back on command at 22 times t.hat speed or 41.25 inches per secohd. The playback
occupies 10.9 minutes.
The general arrangement of t.he recorder can be seen from figure
13-18. This recorder uses quarter-inch tape wound on reels of relatively low inside/outside diameter rat.io but adaptable to direct play-
234
MAGNETIC
TAPE RECORDING
(photo courtesy Radio Corp. 01 America)
FIGURE
13-18.-The Gemini PCM recorder.
The two 90° tape twists are arranged to provide edge-guiding forces (see
text) .
back on standard NAB reel recorders. The tape tension is maintained
by a series of four N egator springs operating between two reels which
rotate in opposite directions. With the low inside/ outside diameter
ratio of the reels and opposite directions of rotation, the addition of
a single flywheel brought the residual rotational inertia of this unit
to a very low value.
The tape path contains two rather sharp 90° twists as can be seen
in the front of the picture. Although this produces the nonunifonn
stresses in the tape which have been discussed elsewhere, this recorder
is SO designed that the lateral aligning forces, by which the tape in
being twisted attempts to regain its original fonn, are used to provide
a positive edge guiding influence and are claimed to reduce the skew
to an extremely low value.
Another interesting feature of this recorder, about which no detailed
data is aV3Jilable, is the use of the speed-change mechanism. The two
speeds are achieved through two different plastic belt mechanisms
connecting the motor to the capstan at different reduction ratios. The
shift between the two belts at the capstan is achieved by what is
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
235
described by the manufacturer as a "hysteresis speed changer." This
device, which appears to operate somewhat like a hysteresis synchronous motor, energizes one of two stationary coils to transmit the
motor power to a selected belt transmission system.
This recorder employs a modified diphase recording technique to
achieve a high packing density and self-clocking action. The packing
density is 2,730 bits per inch and the modulation scheme guarantees
one flux change per bit cell. A one is represented by a change of flux
in the center of the cell and a zero by the absence of such change. The
reproduce scheme involyes having a precision one shot mult.ivibrator
time an interrogation of the flux value three-quarters of a bit cell after
the start of the cell. If the flux value is the same as at the last inquiry
the bit under examination is a one; if it differs, the bit is a zero (Katz
[1964]).
This particular self-clocking mode appears to have certain advantages for relative high packing density. The self-clocking feature
is used to derive a control signal for operating a phase-locked-loop
playback scheme. In t.his system the data is esSIe!ltially read out by
a multivibrator oscillator in a phase-locked-loop which is locked to
the clocking impulses in a somewhat loose manner. The particular
playback scheme provides an output which is smooth in time jitter
or short-term jitter but follows long-term speed variations so as to
guarantee accurate data, although leaving some irregularities in output bit rate.
Perhaps the most sophisticated reel-to-reel commandable recorder
flown so far is that employed to record the output of the A VCS
(Advanced Video Camera System) and HRIR (High Resolution
Infrared) systems aboad the Nimbus satellite. Two different forms
of this particular recorder are used for these two applications but
the two are almost identical physically in the t."''O cases. The basic
unit is shown in figure 13-19 (Burt, Clurman and Wu [1963]).
The following description, adapted from that. of the designers of
t.his recorder, is a general review of its design techniques. The tape
is stored on and exchanged between two parallel co-axial reels approximately % of an inch apart. As the tape leaves one reel, it passes
around a series of four rollers and enters the second reel. The axes
of two of the rollers are inclined at slight angles to the reel axis in
order to lead the tape out of the plane of one reel and into the plane
of the second. These angles are computed so that if all components
are perfect t.he tape will track perfectly. To correct for any unavoidable small errors, however, two of the rollers are slightly crowned to
provide a restoring action for any small lateral displacements of the
tape.
236
MAGNETIC TAPE
RECORDING
(photo courle8fl Radio Corp.
01 AmerictJ)
FIGURE 13-19.-The Nimbus AVeS and HRIR tape transport.
The tape passes around one of the rollers twice-<>nce upon leaving
one reel and again upon leaving the second reel. This roller is belt
driven by the motor and serves as the tape drive capstan. The capstan has an effective tape wrap of nearly 360 0 • The double contact
of the tape with the capstan constitutes in effect a closed-loop system;
this tends to cancel out at the capstan disturbing torques due to some
low-level transients in the tape tension. The tension from the tape
outside of the closed loop is provided by Negator springs which torque
the two reels in opposite directions. The presence of the steady torque
from the springs permits the large wrap around the capstan to develop enough fr.iction to avoid the use of pressure rollers. It is interesting to note that whereas for 1,200 feet of tape one of the reels
turns about 750 times, the relative number of turns between the reels
in this case which is the number of turns the Negators have to deal
with is only 50, or %5 of that value.
Since two speeds are required in this recorder, although different
speeds for the two applications, a planetary belt reduction scheme is
used similar in principle if not in execution to the one discussed above.
For the A ves system this recorder runs at the same speed on record
and playback. For, the HRIR system the record speed is ¥I6 of the
playback speed. The record speed is 30 ips, as is the playback speed
for the A YeS, and the HRIR record speed is 1% ips.
MINIATURE
HIGH -ENVIRONMENT
RECORDERS
237
This recorder is quite good in flutter performance, delivering about
0.02 percent rms between 0.5 and 30 cycles per second and 0.10 percent
rms between de and 5,000 cycles per second. Despite these rather
impressive flutter figures, as noted in chapter 14, it is necessary to
compensate for picture displacement in the high-resolution vidicon
pictures handled by this recorder.
In AVeS service, a 60 kcps-bandwidth signal is recorded in FM
mode. The actual subcarrier deviation is from 73 to 120 kcps, indicating that the modulation scheme resembles that used for rotaryhead video recorders. The HRIR data has a bandwidth of 5 kcps
and is also recorded in FM mode. Erasure for the Aves is accomplished with no power consumption through the use of a 3-permanentmagnet de erase scheme. This method is not usable for the HRIR
because the latter carries two tracks, recorded in opposite directions,
and the unused track would be ruined by permanent magnet erase.
The recorder used for OGO for orbital data storage had to be of
reel-to-reel format because of the long storage time required (12
hours). This recorder, which actually is the predecessor of the Nimbus unit described above, uses the same general physical layout ("Weintraub, D'Amanda and Resek [1964]).
Two forms of the same recorder are used in OGO. The first for
EGO (the Eccentric Orbiting Observatory) is the one which requires
the long storage time. The highly eccentric orbit of this satellite has
an apogee of 60,000 miles and a perigee of 400 miles. The second
OGO, known as POGO (for "Polar Orbiting Geophysical Observatory"), is scheduled for later launch with an apogee of 570 miles and
a perigee of 160 miles. In EGO the data may be collected for a
continuous 12-hour interval and played back in 1l1-/2 minutes. In
the POGO mission, data is collected for 3 hours and played back in
51h minutes. In either case, a total storage capacity of 4.3 X 10 7 bits
is provided.
The OGO recorder has an effective tape packing density of 3,375
bits per linear inch but it achieves this by providing 9 tracks with
an individual track density of only 375 bits per inch. It is designed
for a very good error rate, between 1 X 10 5 to 5 X 106 bits per error.
For EGO the record speed is 0.296 ips for 12.2 hours and the playback speed is 18.96 ips for 11.4 minutes. For POGO the record speed
is 1.18 ips for 3.05 hours and the playback speed is 37.9 ips for 5.7
minutes. In each case the recorder is responsible for conversion from
serial input to parallel recording and from parallel recording back
to serial output.
An interesting difference between this and the Nimbus recorder is
that the two reels travel in the same direction and the residual angular
238
MAGNETIC TAPE RECORDING
momentum is reduced by a necessarily larger flywheel than needed
for Nimbus.
One extreme was perhaps reached in the coaxial-reel commandable
flight recorder with the lOB-bit unit developed for the Mariner program. This unit carries 1,800 feet of 1-inch tape and records on
seven data tracks and one sync track at 24 ips. The gross data input
rate is 83,333 bits per second, and this can be played back at 84, 168,
336, or 672 bits per second. The record-playback speed ratio reaches
a maximum of 1,000-to-1, and the slowest playback speed is 0.024 ips.
The particular recorder developed for this program uses many of the
techniques discussed later under the Mariner 106-bit and 1Q7-bit endless-loop recorders. The lOB-bit unit also had to provide a phaselocked-loop playback scheme to operate at four different speeds. In
_ this recorder, serious consideration had to be given to the effects of
sterilization temperatures as high as 145 0 C for the first time.
Figure 13-20 is a photograph of the prototype lOB-bit recorder.
(photo from a NASA report)
13-20.-The recorder designed to store 10" bits of information, developed
by Raymond Engineering Laboratory Inc. for the Jet Propulsion Laboratory
Mariner program. One reel is visible; the other is located in a symmetrical
posiltion on the other side of the mounting plate.
FIGURE
THE ENDLESS-LOOP RECORDER
Mechanisms for supplying strips of material from a reel-like pack
and returning it to the same pack have existed for many years. Endless-loop motion picture projectors have long been available, using
a storage geometry which resembles closely that of the current endless-
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
239
loop flight recorder. For some time, endless-loop tape recorder cartridges have been commercially available from several manufacturers.
These commercial cartridges are used primarily for "point of sale"
message repeater devices in retail stores or for storing commercial
announcements for radio stations. Although most such commercial
tape packs hold only a few feet of tape, some have a capacity that
reaches or even exceeds that of current endless-loop flight recorders.
The existence of these commercial endless-loop tape cartridges and
their successful use for many years did not produce a cartridge technology that could be relied on to provide reliable endless-loop flight
recorders. This is simply because the commercial tape cartridge does
not have to have the level of reliability of the flight recorder cartridge
and because it is not subjected to any kind of severe environment.
The commercial and flight recorder cartridges are identical in principle. The tape is wound in each case on a one-sided reel of the sort
that would be called in the motion picture industry a "flange." The
tape is supplied from the center of the pack by being pulled out between the inner layer of the pack and the hub of the flange or reel.
After passage through the recording mechanism it is wound-up on
the outside of the pack. With a simple flat flange as described, the
hub of the flange follows the tape as it is pulled off and hence rotates
at a speed which is the linear velocity of the tape divided by the
circumference of the hub. The hub-flange combination is therefore
over-driven as far as the rest of the tape pack is concerned since
every other part of the pack is larger in diameter than the center.
This overdriving force is that which winds the tape up on the outside of the pack.
The same geometry guarantees that every layer of the tape pack
is moving at a different speed from its inside or outside neighbors.
Since the tape is pulled out at a certain linear speed at the inside and
wound up at the same linear speed at the outside, the linear tape speed
is the same at all points. However, as the diameter of the ind'ividual
layers increases gradually from the tnnennost to t.he outermost layer,
the rotational velocity of each layer must correspondingly decrease.
This relationship is responsible for the continuous slip.
With the geometry just described there is a holdback force opposing
the pulling of the tape from the inside of the capstan, made up of
whatever holdback force in the external mechanism opposes the winding up of the tape on the outside transmitted through the interlayer
friction of the pack to the inside layer. If the friction between layers
is high, these forces will be transmitted freely through the pack.
However, if the interlayer friction, which is certain to be nonuniform
from layer to layer and probably to be of the stick-slip type, is too
240
MAGNETIC TAPE
RECORDING
high, the pack will probably jam. Any practical application of such a
pack thus requires that there be interlayer lubrication. (No one has
ever been able to reduce the interlayer friction to the point where it is
too low for tape windup.) The supply and takeup forces produced
by and working on the pack are thus created by a complex relationship
between the rotating friction of the supporting flange and hub and the
tape interlayer friction, which in turn is based on the coefficient of friction between the layers and the interlayer forces which depend on the
tightness of the pack.
For the commercial application of this principle, development was
largely by frustration. Endless-loop tape packs always have had a
reputation for jamming more easily than any other kind of tape
recording equipment. Gradually a body of practical know-how on
tape lubrication and pack geometry has grown up. The same process
had to take place in the development of the endless-loop flight recorder,
but had to proceed by a somewhat more systematic route.
If the ratio of the inside to the outside diameters of an endless-loop
tape pack is low, that is if the ratio approximate6 unity, the relative
'velocity between layers is also low. Put simply, if n is the ratio of
the outside diameter to the inside diameter and there are m layers,
the interlayer velocity is the mth root of n multiplied by the velocity
of the outside of the pack. Early recorders (Project Vanguard, for
example) used only 75 feet of tape in a very slim pack. The interlayer problems were relatively small for this recorder. As the tape
length requirements grew, to 200, 300, 600 and now 1,200 feet, in as
compact. a recorder as possible, the interlayer problems have grown
as well. In the earliest endless-loop flight recorders, the tape pack
performed the normal supply and takeup ·functions in much the same
way that these must be performed for an open-loop recorder. A
single capstan was used which pulled the tape forward over the heads
against the holdback force of the tape beinf;r pulled off the center of
the pack. There is invariably a holdback force at the center, no matter
what. the bulk of the pack does, because of the friction involved in
extracting the inner layer from the space between the next layer of
tape and the hub. This holdback force, however, is relatively irregular, and as the pack size grows, the irregularity grows. Thus,
as larger tape packs and better tape moving performance were demanded, a basic change from the ori~nal open-loop tape-metering
configuration was required. The change was to the differential capstan, a configuration which placed the burden of maintaining tension
across the heads entirely on two capstans, the up-stream one rotating
slightly faster than the down-stream one. The pressure of the roller
holding the tape against the up-stream or supply capstan is adjusted
MINIATURE
HIGH-ENVIRONMEN'Jl RECORDERS
241
to a force great enough to assure that this capstan does the actual
metering of the tape and orercomes as far as possible irregular holdback forces in the pack. The down-stream capstan, being slightly
over-driven, slips continuously, with somewhat less roller pressure, to
maintain the tension across the heads.
With a solid hub-flange combination, there is continuous overdrive
of the flange relative to each layer of the tape in the pack, since the
hub rotational rate is dependent on the minimum or inner diameter.
Friction between the edges of the tape and the flange provides a force
tending continuously to wind the tape up. This force is one of those
involved in providing take-up of the entering section of tape on the
outside of the pack. As packs increased in size the friction between
the flange and the rest of the pack increased to the point where ballbearing rollers were substituted for the flange in the interest of reducing the total motor. load. When rollers were substituted, much of the
wind up force, present with the solid flange, disappeared, since there
is no positive drive from the hub to actuate the rollers. Although
not a problem in medium-sized cartridges, for larger cartridge packs
it was found necessary to put steps in the diameter of the rollers so
as to guarantee that there would be a certain amount of overdrive to
the outer layers of the tape (Stark [1964]).
Obviously, the interlayer friction problem requires that the tape
have good lubricating properties and the difficulty of obtaining these
properties has been one of the leading forces slowing down the development of the endless-loop recorder. The mechanical interrelationships in such a device are discouragingly complex. For example,
unless the interlayer frjction is low enough the pack will jam up and
the recorder will simply not work; the major tape takeup force, however, is generated by the interpack friction and the two influences
must somehow be reconciled.
Inherent in any such recorder is a twisting and warping of the tape
as it leaves the center of the pack. The effect of this distortion on the
tape path must be prevented from causing flutter at the head. Endless-loop recorders developed recently for the Aeronomy Group at
Goddard Space Flight Center, have therefore provided a shallow
groove, as wide as the tape, in the capstan pressure roller to improve
the guiding of the tape at this point. Work is currently under way
in the Aeronomy Group to study the "peel off" geometry in detail
and to provide positive guiding of the tape during this entire operation.
In addition to "jamming up" of the pack, a basic tape handling
problem with endless-loop recorders is that under certain conditions
a "loose loop" may be formed. This means that with marginal takeup
forces severe mechanical environment stresses may cause a loose loop
242
MAGNETIC TAPE RECORDING
of tape to appear between the end of the useful tape path and the
takeup function in the pack. The tape used in such r.ecorders tends
to shrink on heating; the amount of tape which is placed in the pack
must therefore be so adjusted that the pack will neither jam from
over-shortening of the tape nor throw a loose loop. The pack must
be somewhat loose to control intrapack friction and yet means must
be provided fo:r taking up possible excess tape. For example, adjustment of the diameters of the tapered pack-support rollers can control
the takeup relationship.
An endless-loop recorder may be required to ope~ate during the
launch phases of a space operation and this means it will be required
to continue to operate under shock and vibration, as well as perhaps
at severe temperature extremes. Under some circumstances this may
be an ad vantage; some endless-loop machines throw a loop wn1ess they
are operated during launch. Recently an endless-loop recorder with
a particularly large tape pack could not be flight-qualified until snubbers were designed to maintain the pack in position in the non-operating mode during launch. Without the snubbers, a loose loop appeared
as the pack "shook down" during the launch vibration.
A general treatment of the static and dynamic characteristics of
the endless-loop tape pack presents ext.raoroinary analytic problems.
Certain rather simple relationships have been established between the
parameters of the pack but any att.empt to construct a mat.hematical
model has resulted in conclusions largely inapplicable to the practical
tape pack (General Kinetics [1963]).
Quite a few modifications of the basic circular pack in the endlessloop recorder have been tried to minimize the intrapack friction. One
particular concept which has been thoroughly investigated on paper
but does not seem to have found its way into much flight equipment
is that of the so-called "square-loop" tape pack. Figure 13-21 shows
the basic difference between a square-loop and a circular-loop array.
The idea behind the square loop is simply that the only friction between the layers of a square loop configuration occurs at the corners.
If the size of the square can be increased and the radius of the corners
left the same, the length of tape may be increased almost without
limit without increasing the amount of friction. No friction occurs
between layers during the straight travel between the corners because
the linear speed of the tape in all parallel paths is the same. Therefore, for example, if one has a 3-inch diameter circular tape pack one
can, as it were, cut the pack apar.t into four quadrants and insert
lengths of straight tape between those quadrants. In theory this decreases the amount of friction for the amount of tape stored. In
practice the large add'i.tion to size resulting from this configuration
MINIATURE
HIGH-ENVIRON:&IENT
RECORDERS
243
o
b
FIGURE 13-21.-Endless-loop tape configurations.
The compact circular tape pack of (a) can be "cut apart" and straight segments of tape added thereto to produce the larger storage arrays of (b) and (c)
without adding interlayer friction (see text).
and the greater difficulties of controlling the tape over the straight
portion compared to the geometric simplicity of dealing with it in a
circular pack have prevented extensive use of this concept. The
inertia-compensated recorder and one other reentry recorder discussed
elsewhere use modified versions of this particular configuration. In
these two recorders the loop is not square but oval and there are simply
two straight portions between semicircular ends (fig. 13-21c). In
the application for which these recorders are designed other characteristics necessarily reduce the effectiveness of the reduction in friction
and in most cases the internal friction for. the guiding necessary to
produce the oval loop configuration loses much of the advantage of
the reduced friction.
The many problems discussed above may make it appear that the
endless-loop recorder could not successfully meet the challenging
requirements of flight application. It is a tribute to the skill,
resourcefulness and persistence of the engineers who have undertaken
the development of such recorders that the low-to-medium-capacity
endless-loop recorder can now be described as a very reliable device.
One has only to note the long and useful lives of many such units in
unattended space application, as described in the following pages, to
be convinced of this accomplishment. 'Vith the current level of technical effort directed toward increasing the tape capacity of the endless-loop recorder, it can be anticipated that it soon will be possible
to remove the modifier "low-to-medium-capacity" from the sentence
above.
The endless-loop recorder has specific application to the satellite
and space probe field because it permits operating continuously without having to provide the complex control modes involved in reversing
a reel-to-reel recorder. One of the first of these units to fly was that on
Score I which carried 35 feet of tape in an endless-loop and recorded
244
MAGNETIC TAPE RECORDING
at 3% inches per second for 4 minutes. Its major purpose was to
repeat voice and teletype messages on ground command. The Vanguard II recorder carried a 75-foot continuous loop of tape and operated on interrogation from the ground station to repeat data on
cloud brightness.
The tape capacity, number of tracks, and general performance level
of the endless-loop recorder have gradually increased. The Tiros
satellite contained a recorder employing a 200-foot endless tape loop
for storing the data for the low-resolution radiometer. This was in
addition to the reel-to-reel recorders used for the weather pictures
themselves. The Orbiting Solar Observatory which has appeared in
two forms, OSO-1 and OSO-2, has in each case employed a somewhat larger endless-loop recorder. A 285-foot loop was used in OSO1 to store analog FM data and the same length of tape was used in
OSO-2 to store PCM data. A somewhat smaller recorder, following the modular scheme of construction discussed in chapter 11, was
developed for the UK-2 international satellite. Other recorders of
this class are used to store spacecraft situation and grid data aboard
Nimbus. Recorders of 106 _ and 107 -bit capacity have also been built
for the Mariner program.
The Tiros endless-loop radiometer recorder has successfully operated for more than 2 years in orbit. The design of this unit is interesting in being an intermediate between the simple recorders of the
Vanguard series and the sophisticated units currently being produced.
The overall mechanical drive scheme of the recorder is shown in figure
13-22. It has several interesting features.
In this recorder the holdback force was provided entirely by the
opposition of the tape to being pulled out from the center of the endless-loop cartridge, that is, it was an open-loop transport. Two motors were used, a de motor for the high-speed playback mode, and a
two-phase, 137.5 cps ac record motor operating at 4,125 rpm. It was
felt that the dc motor could be used despite the brush problems in the
vacuum because being the high-speed unit it operated for a relatively
short time (Falwell, Stark and White [1963]).
An interesting feature of the two-speed drive is that the record motor operates during the playback operation. The overrunning spring
clutch shown in the drawing allows the playback motor to drive the
capstan while essentially ignoring the relatively slow rotation produced by the record motor. One of the gear-driven switches, labelled
the "Time-Sharing Switch," is driven at 256 rpm by a plastic belt
and through gear reduction provides a switching function between
various input signals, sampling each of the signals for approximately
6 seconds. The other gear-driven switch is that for playback timing.
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
245
RECORD.4 INCH/SEC
TIME-SHARING
r::---~~::::--':'------'
i~~ ~
I
I
iI
TWO-PHASE
137.5 CPS
RECORD MOTOR
4125 RPM
2.721l
\,
TAPE REEl
do
PLAYBACK
MOTOR WITH
MAGNETIC
SHIELDING
REOUCTION 5000 rpm
Q
FLYWHEEl
(from a N ABA report)
~'IGeRE
13-22.-l\1eehanicaJ schematic of the drive mechanism of the Tiros infrarro data recorder.
The function of this switch is to ensure that the playback cycle only
continues for the length of the endless loop and that the recorder is
switched ba,ck into the record mode when the entire tape has been
played back. The function of the switch is rather simple since it
counts the number of revolutions of the playback motor and then
operates the mode reversing switch.
An interesting method has been used to remove playback jitter in
this recorder. Since one of the required recorder functions is to
convert the input serial information to parallel for recording and
then to convert it back from parallel to serial, a storage buffer is needed
for playback. In the jitter-removal system, the reproduced word
is placed in paranel in a storage register in a conventional manner.
If the storage register is not full at the point when the next serial
word should be read out of it, a series of zeros is automatically transmitted until a signal is received indicating that the register is full.
The word is then read out of the register while another word is being
fed into the second layer of the register. The series of zeros and
the output word are both timed by a local crystal clock and the data
246
MAGNETIC
TAPE
RECORDING
rate is therefore perfectly uniform. The cost of this method of compensating for recorder jitter is that the transmitted data word varies
in length from 9 to 15 bits rather than simply having the same length
as the input data \Yord (Townsend, Feinberg, and Lesko [19631).
The dc motor used for playback is servo-controlled through the
signal generated by a tachometer ac generator placed on its output
shaft. This somewhat unusual arrangement was felt to give a better
weight-efficiency ratio than was obtainable with the synchronous
motors then available. A photograph of this recorder is sho'n1 in
figure 13-23. A rather similar recorder, although mounted on a
square rather than a round structure and differing in details, is also
used aboard the Nimbus weather satellite, primarily to store PCM
orbital data on spacecraft. sensors and grid data.
An endless-loop recorder which represents rather clearly the state
of the art at the time of its development is that used on the UK-2
satellite. This was the first recorder in which the concept of modu-
(photo courtesy Raymond Engineering Laboratory)
FIGURE 13-23.-Tbe Tiros I infrared recorder.
MINIATURE
247
HIGH-ENVIRONMENT RECORDERS
.Jarization was formally used. The development of this recorder
through breadboard, engineering test model and prototype was undertaken in-house at the Goddard Space Flight Center.
The recorder is an endless-loop unit holding about 300 feet of
quarter-inch-wide tape. In record mode it operates at 0.15 ips and in
playback at 12 ips. The two-speed. ope.ration is achieved by means
of a single motor which rotates in one direction for record and in the
other direction for playback. The speed reduction for the two modes
is accomplished by a technique shown in figure 13-24. Two overrunning clutches are involved in the speed change technique. In
record mode, the motor drives the intermediate shaft but the overrunning clutch on that shaft is disengaged and the belt between that
shaft and one of the capstans therefore does not transmit any power.
MOTOR
R = RECORD
P = PLAYBACK
(from a N A8A report)
13-24.-Mechanical schematic of method of obtainmg two speed drive
in the UK-2 recorder by reversing the direction of a. single motor (see
text).
FIGURE
The intermediate shaft drives the rim of a Mylar-covered wheel
mounted on the auxiliary shaft next to it at a reduced speed. The
belt from this shaft then drives the second capstan through an overrunning clutch. The direction of operation of that clutch is such that
in the record mode this belt drives the capstan positively. The two
capstans are belted together. When the motor is reversed, the overrunning clutch on the intermediate shaft engages and the belt. from
788--Q28 0--65--17
248
MAGNETIC TAPE RECORDING
that shaft to the capstan then supplies power to it and, through the
interconnecting belt, to the other capstan. The direction of rotation
of the overrunning cluteh on the second capstan shaft is such that it
is then disengaged from the record auxiliary shaft.
The hysteresis synchronous motor is driven by a 100 cycle per second
square wave which is generated by countdown from a 400 cps tuningfork-controlled oscillator. The frequency and hence the speed provided is claimed to be accurate within 0.1 percent.
A remote-controlled endles.'l-loop recorder requires a timing device
to deliver a message that all that has been recorded has been played
back so that the recorder can be restored to record mode. In this
unit the timing mechanism is a solid-state oscillator operating at
approximately 7 cycles per second. When playback is desired, the
command rec~iver, triggered from the ground, supplies a pulse to the
playback timer relay which switches the record-playback circuit into
playback and turns on the 7 -cycle oscillator which then feeds a 1,000pulse divider. At the end of a thousand pulses, or approximately
140 seconds, an output pulse is sent by the pulse divider to the recordplayback mode relay, returning the recorder to the record mode.
Endless-loop recorders for use in deep-space probes have very low
power consumption and slow playback speeds. For example, a breadboard l()6-bit PCM storage recorder for use in the Mariner programs
has a total power input of 0.425 watts with a record speed of 3.6 inches
per second and a playback speed of 0.03 inch per second. To play
back the entire loop of tape at this low speed requires 33 hours.
The rigid timing required for successful data transmission over
interplanetary distances requires that space-probe recorders be locked
in playback bit rate to an internal spacecraft clock. This recorder
therefore employs a phase-locked-loop servo to drive the playback
motor. Typically, the output of the phase-locked-loop oscillator in
such a servo is a square wave and a hysteresis-synchronous motor of
maximum efficiency (low damping) may jitter with this input waveform. Any ripple in the output of the phase-locked-loop driving amplifier will cause further motor jitter. A special gated-integrator
technique was developed for this recorder to minimize the ripple. To
insure a satisfactorily-uniform output bit rate, an additional lockedoscillator clocking system had to be used. In this scheme the output
bits are put into a one-bit-deep register out of which they are clocked
by a local VCO. This VCO is locked to the somewhat jittery bit rate
coming off the tape through a filter which assures that the oscillator
rate will not jitter fast but may change at a relatively slow rate. This
matching between high-speed input jitter and low output data rate
MINIATURE
HIGH-ENVIRONMENT RECORDERS
249
change is accomplished with storage only one bit deep. A photograph
of this recorder is shown in figure 13-25.
Figure 13-26 shows a lOT-bit version of the lO6-bit Mariner recorder
made by the same contractor. It differed from the 106-bit unit in holding 700 instead of 300 feet of tape and in using four rather than two
data tracks. Mylar "pinch belts" are used to maintain contact between
tape and capstan.
The flight recorder used in the successful Mariner IV mission which,
in July, 1965, sent to earth the first closeup pictures of Mars, combined
(photo courtesy Raymond Engitneering La/)oratm-y)
FIGURE
13-25.-The 1oo-bit storage-capacity recorder built by Raymond Engineering Laboratory for the Mariner program.
features of the 106-bit and lOT-bit breadboards. This unit had a
record speed of 12.84 ips and a playback speed of 0.01 ips, giving a
playback data rate of 8.33 bits per second with a recording density of
833 bits per inch. Two tracks were used on a 330-foot loop of lf2" tape,
giving a usable storage capacity of 5.2 X 106 bits. A two-capstan drive
system with polyester pinch belts was used. This recorder lay dormant
in space for the more than seven months of the flight to Mars and was
then successfully turned on to record 21 digitally-encoded 240,000-bit
pictures. It then operated in playback mode for eighteen days to
transmit the entire picture series to earth twice.
The Orbiting Solar Observatory (OSO) program has used continuous loop recorders for orbital data storage. The first of these operated
successfully for 11 weeks for a total of 1,000 hours of information
recording during 1,380 orbits on OSO-l (fig. 13-27). This first OSO
recorder handled analog FM data recording at 0.60 ips for 95 minutes
on 285 feet of quarter-inch tape. Playback was accomplished in 5.2
250
MAGNETIC TAPE RECORDING
(trom a N ABA report)
FIGURE
13--26.-The Mariner program 107-bit endl_ess-Ioop recorder, built by
Raymond Engineering Laboratory Inc.
The paths of the elastic "pinch-belts" which substitute for the pressure roller
in providing tape-capstan friction are clearly apparent in this picture.
(photo CO'Urtesy Raymond Engineering Laboratory)
FIGURE
13--27.-The OSO-l (8-16) Orbiting Solar Observatory endless-loop
analog recorder.
MINIATURE
HIGH-ENVIRONMENT RECORDERS
251
minutes at a tape speed of 11.0 ips. This recorder was 7 inches in
diameter and 3 inches high, weighing 5 pounds.
A second version of this recorder similar in mechanical characteristics but manufactured by another organization was used to record
PCM data for OS0-2.
Stark has described an endless-loop recorder with very large tape
capacity (1,200 feet), which was recently developed by the Aeronomy
Group at Goddard Space Flight Center (Stark [1964]) (fig. 13-28).
(N A.BA. photograph)
FlGUBB 13-28.-An experimental l,200-foot capacity endless-loop flight recorder
developed by the Aeronomy Group at Goddard Space Flight Center.
The endless-loop recorders described immediately above are all of
the so-called classic type. The tape is in the form of a flat circular
pack and the supply is from the inside of the pack, which causes the
pack as a whole to rotate, and the takeup is on the outside of the pack.
NASA has carried out and sponsored a considerable amount of work
on this particular class of endless-loop structure. There are, however,
other means of storing an endless-loop of tape in compact form. One
of these, shown in figure 13-29, stores the tape in two levels in the
standard format to conserve space. The tape is wound onto the outside of one pack and is pulled from the center of that pack to be wound
on the outside of the other pack. It then is drawn from the center of
the second pack.
A very unusual endless-loop construction is that shown in figure
13-30. The tape is strung back and forth in a series of loops and the
entire array of loops is wound around a drum. The tape is carried out
of the pack by a belt similar to the Cobelt (see below), and the entire
252
MAGNETIC TAPE RECORDING
(photo courtesy Borg-Warner Controls)
FIGURE
13-29.-The Borg-Warner R302 endless-loop recorder which stores the
tape loop at two levels to provide improved tape handling.
structure is rotated by an enveloping wide belt. Short endless-loop
recorders have also been built with the tape simply random-stored,
that is, left loose in a bin, as shown in figure 13-31. This is not, of
course, a compact construction but is a very simple one. In this recorder a belt is also used to carry the tape in and out of the bin.
A somewhat unusual endless-loop recorder developed for reentry
studies is shown in figure 13-32. This top view emphasizes the extremely rugged construction and the unusual loose tape pack which is
essentially the conventional endless-loop configuration expanded from
a circle to an oval. An elruborate procedure is followed in getting the
tape from the inside to the outside of the storage loop to avoid the
twist stretching of the edges of the tape discussed elsewhere. The tape
irave]s counterclockwise around the outside of the pack and on the
lower side of the pack, that is, toward the bottom of the picture, passes
just below the center of the picture around a roller. This roller guides
it around a very similar roller somewhat above it and to the left and
the tape is then carried around an unusually-shaped fixed guide. This
guide is clearly of a form which translates the tape from one plane to
another in such a way that no part of it is stressed more than any
other. The price of the very heavy friction encountered at this guide
MINI A TURE
HIGH-ENVIRONl\IEN'I1
RECORDERS
253
TAPE EDGE
ClAMPING
BElT
ADJUSTMENT
ROLLERS
MOUNTING
SCREWS
TAPE LOADING
DRI V E ROllER
(drawing courtesy Ralph M. Parsons Company)
FIGURE
13-30.-The Parsons Model CRB-80 continuous rotary-bin recorder/reproducer (see text) .
is apparently willingly paid for the smooth transition this technique
provides. The first roller referred to above is mounted on a pivoted
arm and is apparently spring-loaded to take care of stretch of the
pack. The arm rotation is also damped by a plastic belt pressed against
its outer surface.
The manufacturer of this device stresses the advantages of this
particular tape configuration and guiding technique 'for avoiding
bindup between the layers of the tape under severe acceleration. This
unit is understood to survive 100 G's acceleration without binding and
to be operable with somewhat reduced performance at 80 G's acceleration continuously. A particular version of this unit operates at 30
inches a second and reaches a maximum delay time of 40 seconds,
which means a storage of 100 feet of tape.
An interesting point is made by the manufacturer in discussing the
use of this recorder for short-term time delay for bridging the reentry
flame attenuation period. The life of the recorder in active service
254
MAGNETIC TAPE RECORDING
is dependent on the tape life and this tape life depends on the number
of cycles rather than any actual number of hours. For example, a 40second-delay loop is circulated 900 times during a 10-hour period while
a 20-second loop circulates 1,800 times in the same period. The manu-
(photo c011.rte8Y Ralph M. Par80ns 00.)
FIGURE
13-31.-The Parsons CLR--225 "bin" recorder.
This technique of storing a relatively short loop of tape in random coils in a
fiat bin has been widely used.
(photo courtesy Litton Ind.)
FIGURE
13-32.-The Westrex Model RA-l683-B reentry recorder/reproducer.
MINIATURE
HIGH-ENVIRON MEN'll
RECORDERS
255
facturer points out that a 30-second tape loop operating continuously
for a period of 30 hours exhibits serious loss of high frequency
response. No tape apparently now available will exceed this performance. The statement is made that 1,500 to 1,800 cycles per operation may be considered the tape life limit for acceptable high frequency response. A 30-second tape loop therefore has a life of
approximately 12 to 15 hours within specifications.
An orbital tape recorder operating in a 90-minute orbit and interrogated on each pass goes through 16 double cycles, that is, one of
record and one of playback, every day. There are orbital recorders
which have operated for a year or more under these circumstances,
representing a total number of tape passes of 6,000 or more. It is
not intended as a criticism of the reentry unit discussed above to comment that the additional tape strain and friction which is introduced
in the interest of making this machine survive an extremely severe
shock and vibration environment increases the tape friction and hence
shortens the tape life.
The basic technique used to provide the compensation of the socalled inertia-compensated recorder shown in figure 13-33, is to fill
the recorder operating volume with .a fluid of approximately the same
specific gravity as the tape. This recorder also has other interesting
features. The tape is in an oval loop endless-loop configuration, some
(photo courtesu Oollege Hill Ind,. Division)
FIGURE
13-33.-The College Hill Industries inertia-compensated recorderl
reproducer Model 005.
The mechanism by which the tape travels from the inside to the outside of
the oval loop is not easily followed in this photograph but careful examination
will show tbe corrugations on the outside of tbe capstan (lower right) resulting
from the slitting technique and also the sharp radius on each head preceding the
active gap as described in the text.
256
MAGNETIC TAPE RECORDING
of the difficulties of which are discussed elsewhere. The presence of
the inertia compensating fluid surrounding the tape can, to a certain
extent, improve the lubrication of the tape layers but the hydrodynamics of the situation are far from simple and it appears that some
of the tape bind complications that have occasionally arisen result
from the presence of this liquid.
In order to operate a tape recorder, as it were, under a liquid, it, is
necessary to take certain precautions to insure that the recorder operates in a normal manner. The typical pinch roller and capstan construction is somewhat reduced in effectiveness by the tendency of the
inertia-compensating liquid to lubricate the contact between these
elements. A specialized technique is used in this particular recorder
to overcome this difficulty. The actual capstan driving the tape is
of rubber and the tape is pressed against it by a metal roller. To
insure friction between the capstan and the tape even in the presence
of the fluid, slits are placed in the surface of the capstan at an angle
to the center line of the tape. These slits apparently act as squeegee
pumps to clean out the fluid so that good adhesion takes place between
the capstan and the tape.
Another somewhat unusual aspect of this recorder is that the tape is
operated in a Moebius strip. The ta.pe is coated on both sides and the
center of the tape passes through the mechanism twice before the same
point on the recording surface is reached for the second time. This
extends the recording time at the cost of the possibility of printthrough from tape layer to tape layer. However, the presence of the
compensating fluid probably minimizes the importance of this printthrough since the separation between the layers is maintained by the
fluid.
The same fluid film which tends to reduce interlayer friction and
print-through in the pack also tends to keep the tape away from the
heads. In order to reduce the thickness of the layer at the record and
reproduce point the tape is passed around a sharp radius (0.004 inch)
under a pressure of 10 to 15 grams just before it encounters the head
gap. This reduces the thickness of the residual fluid film to 1 micron,
which provides satisfactory high-frequency response for the applications considered (Monopoli [1962]).
UNUSUAL RECORDER FORMATS
Several unusual drive mechanisms have been developed for and applied to flight recorders to overcome some of the fundamental problems of such recorders. Although the general scheme of the recorders
using these novel drives resembles closely the schemes of recorders discussed under the three classifications covered above, the drive mecha-
MINIATURE
HIGH-ENVIRONMEN'Il RECORDERS
257
nisms change the design philosophy of these recorders enough to make
it advisable to consider them separately. This section is therefore
devoted to the "Cobelt" and "Iso-elastic" drive schemes.
The Cobelt drive scheme was first applied to a recorder designed for
use on a rocket sled. This recorder had several features designed to
permit it to operate satisfactorily under heavy vibration and accelerations up to several hundred G's. In this recorder, shown in
figure 13-34, no conventional reels are used. Instead the recorder is
constructed very solidly of a "sandwich" of heavy metal plates which
are assem'bled rigidly on both sides of precision spacers only 0.001 inch
thicker than the tape width. The tape, instead of being supported
between reel sides, is handled by the blocks of metal which form the
body of the recorder. When the recorder is assembled, the entire tape
guide function is carried out by these side plates. For withstanding
shock, this construction is much superior to one using a reel of any
kind since a reel side must necessarily be relativ.ely flimsy.
In order to maintain contact of the tape with the head, the Cobelt
drive is used here. This, as can be seen in the photograph, consists of
an auxiliary plastic belt which presses the tape toward the heads.
This belt also provides the drive normally supplied by the capstan and
pinch roller. The tape is thus both pulled along by friction with the
(photo courtesy Genisco Data)
FIGURE
13-34.-The Genisco Data Model 10-110 tape transport.
The Co belt return path can be seen directly above the four heads, as can the
way in which the massive base plate is hollowed out to support the rolls of tape.
The depressed portion of the structure is actually as deep as the full tape width
and the housing is completed by laying a flat plate over the structure seen here.
258
MAGNETIC TAPE RECORDING
Cobelt and pressed by it against the heads. The only disadvantage of
use of the Cobelt drive outside of this particular equipment is that the
lateral guidance of the tape, unless it is severely constrained as it is in
this recorder, is affected by the straightness of the driving belt. There
is a tendency for the tape to wander to follow the belt and the technique
therefore seems limited at the moment to use where it is as precisely
guided as in this particular recorder.
The "Iso-elastic" drive technique, which is illustrated in figure 13-35,
was introduced to improve the tape drive and reeling functions by
combining them in a unique way.
(0)
(SEAMLESS MYLAR)
I b)
L
MOTOR
4-3
4-1
BELT
UNMOUNTED
(UNSTRESSED)
BELT MOUNTED
(STRESSED)
NOT ROTATING
BELT MOUNTED
AND TRANSMITTING
TORQUE
(from aNABA report)
FIGURE
13-35.-Schematic representation of the principle of the Iso-elastic tape
drive.
The principle on which this drive operates is clearly described in this drawing.
The reel packs are driven at a constant velocity at their periphery by
a polyester belt. Tape tension is determined by the difference in the
surface speeds of the two driven capstans which transfer motion to a
so-called Iso-belt under a nonslip condition. The tension in the tape
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
259
results from the physical properties of the Iso-belt and is not dependent on the transfer of forces through dynamic friction as in more conventional systems. A brief description of some recorders developed
around this principle follows, adapted from JPL-SPS 37-25 Vol IV
p.217-227.
A series of tape recorders, primarily designed for space probes, was
developed to prototype form using this drive system. At the time of
initiating this development program, a series of specific requirements
existed at the Jet Propulsion Laboratory for space probe recorders.
These are shown in table 13-1. The tape transport mechanism which
was developed to meet the general requirements of the 108 bit recorderl
reproducer of table 13-1 is shown in figure 13-36. The major elements
defining the specific features of its operation are (1) the Iso-elastic
belt system, (2) a specialized differential drive, (3) tape guides and
heads, (4) the supporting structure, and (5) reliability considerations.
The Iso-elastic belt is a seamless polyester belt which rides directly
on the outer layer of the tape on each reel pack. The belt is driven by
two capstans, a fast capstan and a slow capstan. Sufficient pretension
is applied to the belt by the tension idlers that a nonslip condition is
maintained between the capstan surface and the belt. A schematic
representation of the mechanics of the Iso-elastic principle is shown
in figure 13-35. The effective operating tension Tx' results from the
fact that the magnetic tape is inelastic when compared to the elastic
properties of the drive belt. Therefore the peripheral speeds of the
tape pack "are, in effect, synchronized by the inelastic coupling of the
tape itself between them regardless of the relative diameter of either.
FIGURE 1~6.-A
(from a NASA report)
10' bit-capacity developmental spaceprobe recorder employing
the Iso-elastic drive principle.
260
TABLE
MAGNETIC TAPE RECORDING
13-1.-Tape Recorder-Reprod1Wer General Requirements
108 Bit recorder-reproducer
Storage capacity, information bits _______ _
Total volume, including electronics ______ _
Weight, exclusive of electronics __________ _
Power consumption, 400 cycles __________ _
Sterilization __________________________ _
Two-speed operation ___________________ _
Bit error probability ___________________ _
>10
8 bits (NRZ)
10 x 14 x 3 in.
<101b
<4w
125 0 C
50:1 ratio
10-5
107 Bit recorder-reproducer, two-motor drive
Storage capacity, information bits _______ _
Total volume, including volume for
electronics __________________________ _
Tapewidth ___________________________ _
Bit packing density for lOS bit error
probability __________________________ _
N umber of channels ___________________ _
High speed ___________________________ _
Lowspeed ____________________________ _
Speed change mechanism _______________ _
Weight, e~clusive of electronics __________ _
Flutter, hlgh-speed ____________________ _
Stop-start distance, low-speed ___________ _
Power consumption, motors only ________ _
Hermetic sealing ______________________ _
107 bits (NRZ)
6 x 6 x 2% in.
y. in.
1,000 bits/in.
4
12 in./sec
0.5 in./sec
Mylar-belt differential
<31b
<10%
<0.15 in.
<6w
0.1 atm residual, 2 yr
10 7 Bit recorder-reproducer, single-motor drive
Storage capacity, information bits _______ _
Package size, including volume for electronics _____________________________ _
Tape width ___________________________ _
Bit packing" density for lOS bit error
probability _________________________ _
Number of channels ___________________ _
High speed ___________________________ _
Lowspeed ____________________________ _
Speed change mechanism _______________ _
Weight, exclusive of electronics __________ _
Flutter, high-speed ____________________ _
Stop-start distance, low-speed ___________ _
Power consumption ____________________ _
Hermetic sealing ______________________ _
107 bits (NRZ)
6 x 6 x 2% in.
y. in.
1,000 bits/in.
4-8
15 in./sec
1.5 in./sec
Drive frequency change to
motor
<31b
<2%
<0.1 in.
<2w
0.1 atm residual, 2 yr
In the practical case of the recorder, initial tension is supplied by the
tension idler. This is required to account for the change in length of
the Iso-elastic belt because of the change in geometry resulting from
the variable amount of tape on each pack as a consequence of operation.
This assembly and those described later were designed on the basis of
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
261
a resultant zero spring rate of the assembly so that the initial tension
would be held constant.
The special differential mechanism reierred to above is one in which
two motors drive two elements of a differential at slightly differing
speeds. The differential so operates that when the motors are driving
the differential pulleys in the same direction the output speed is the
average of the sum of the two pulley speeds. 'Yhen the motors are
driving in opposite directions the output velocity is the average of
the difference of the two speeds.
The Iso-elastic belt structure displaces the tape transversely relatively little, but in order to guarantee extremely good guidance so that
low amplitude modulation from lateral motion of the tape will occur,
a long trough guiding system is used in this recorder. The overall
guide length is approximately 9 inches.
It was discovered in analyzing the vibration sensitivity of tape packs
of the size required for this machine in a system of reasonable stiffness
that the resonant frequency of the tape-mass-supporting-structure combination was well within the range of vibration frequencies to which
the overall mechanism would be subjected. The concept was therefore
introduced of providing support for the entire recorder assembly not
only at its borders but also at the reel hub shafts to increase the effective rigidity of the overall system.
A 107 -bit recorder meeting another set of requirements in the table
was developed with the view to providing a much smaller package.
The general arrangement of this recorder is shown in figure 13-37.
The same type of differential was used in this mechanism as in the
lOB-bit recorder, but because of size problems, the differential itself
had to be reduced considerably in diameter. This meant that the
spider belt within the differential might exceed its life cycle. Although several attempts were made to substitute other materials for
the Mylar when it was discovered that a very short failure-free life
was predictaJble, it was necessary to estimate that this recorder could
not successfully operate with the differential concepts originally proposed over the necessary active life cycle.
Another 107-bit recorder was therefore developed in an attempt to
simplify construction on the basis of what had been learned in the first
lOT-bit unit. In this recorder, a single motor driving the capstan peelley through a single polyester belt was substituted for the differential.
The two speeds were achieved by running the motor at 12,000 rpm
with a 400 cycle per second power supply and at 1,200 rpm with a 40
cycle per second power supply. This recorder (fig. 13-38) appears,
as a prototype, to meet most of the requirements indicated in Table
13-1 and promises a useful set of flight hardware in the future.
262
MAGNETIC
TAPE
RECORDING
(from a NASA report)
FIGURE
13-37.-A. 10' bit-capacity spaceprobe recorder employing two motors and
a plastic-belt differential drive (see text).
(from a N ABA report)
r'IGURE
7
13-38.-The improved 10 bit-capacity spaceprobe recorder employing a
Single motor drive.
MINIATURE
HIGH-ENVIRONMENT
RECORDERS
263
MINIATURE TRANSVERSE-SCAN RECORDERS
For dealing with megacycle bandwidths and massive data storage,
the rotary-head video-type recorder has been proposed for flight servine. The mechanical complexity and short head and tape life of this
class of recorder did not originally hold out much hope of adapting it
to an application requiring such high reliability and ruggedness.
However, in 1961 a %-cubic-foot, 4O-pound rotary-head recorder was
d.esigned specifically for flight service. This unit had a bandwidth of
over 4 megacycles and recorded for 20 minutes. Although this recorder has not flown it was an interesting demonstration of the feasibility of this technique. More recently, two separate programs have
been undertaken by NASA to investigate the potentiality of the
rotary-head recorder for two-speed operation. A group at the Flight
Research Center is investigating the high-speed record, low-speed
playback capabilities of the helical scan industrial television recorder
to see if two-speed operation can be achieved for this unit. Recently
an extensive investigation of many aspects of rotary-head recorder
application was completed by Ampex Corporation for Goddard Space
Flight Center.
The Ampex study covered a wide range of experiments and can only
be summarized in the briefest outline here. The broadly stated objectives were:
1. To evaluate existing or proposed magnetic tapes usable for
this class of recorder and to determine whether tape life could
be adequately defined or predicted,
2. To evaluate the factors which influence head life and to attempt
to define head life quantitatively,
3. To evaluate and propose methods for achieving various time
base-expansions (up to 50/1 expansion), and
4. To evaluate generally methods of applying rotary-head recorders to space.
Many results of the investigation were encouraging. It was found
that tape usually showed initial dropouts which disappeared after a
few passes, then settled down to a uniform number of fixed dropouts
over a given life and then suddenly developed a sharply increasing
number of dropouts as the oxide binder apparently failed mechanically
and the end of tape life was indicated by clogging of the head.
"Good" tape apparently can be selected by automatic dropout testing,
which divides tape samples into groups which differ by an order of
magnitude in error or dropout rate, the lower rates shown in this test
being well correlated with long life. The study indicates that current
tape should survive 600 passes under 1'OO1-to-1'OO1 condit.ions and 3,000
'188--008 ~5---18
264
MAGNETIC TAPE RECORDING
passes for loop service. The lower reel-to-reel figure apparently results from damage in end-of-reel handling.
Head life factors are extremely complex, and the study showed that
several suspected influences had no actual effect, such as damage from
a "shock wave" phenomenon propagated ahead of the head tip as it
indented the tape. The conclusions included confirmation that bidirectional operation, which seems almost unavoidable in space application of such recorders, is not an optimum operating condition. This
results from another conclusion, which is that the tip shape of the
head wears to a true circular arc inclined toward the direction of
travel, and the head therefore operates peculiarly when the direction is
reversed. Head wear goes down very rapidly with speed, as could be
expected, with, for example, an e{(pected number of feet of tape
traversed before wearout eight times as great at 30 ips as at 1,500 ips.
A test bed recorder operated so well at high-time-expansion ratios
in these tests that the following prediction is made of achievable expansion performance: Signal-to-noise ratio, 36 db (video SIN) total
time displacement error, 15 microseconds, potential time expansion
ratio with usable SIN ratio, 150/1, and shortest wavelength recoverable, 85 microinches (12,000 cycles per inch).
Although the breadboard recorder delivered at the conclusion of
this study is designed only for further investigation of influences and
design £actors for this class of recorder and not as a first step for
rotary recorders into space, the conclusion can probably be drawn that
the step is inevitable in due time.
DRIVE SYSTEMS FOR MINIATURE RECORDERS
Although there are situations in which a high-environment recorder has few limitations on its size, weight, and power consumption,
advanced technology needs to be brought to bear on the recorder drive
problem when these limitations are present. This discussion is directed to the situation where maximum efficiency, minimum weight
and maximum reliability must be achieved.
Many of the characteristics of motors used for miniature recorders
and the considerations involved in motor choice are discussed in chapter 11. The discussion in this section is limited to specialized matters
such as the choice as to whether one motor or two should be used and
how fast the motor should be designed to rotate. In the Lockheed
report referred to above, the point is made that the hysteresis-synchronous motor has great advantages of speed, accuracy, reliability
and low noise generation, although the low efficiency of these motors is
a strong disadvantage. They are partiCUlarly susceptible to the fact
that when attempts are made to increase their efficiency, the damping
against jitter oscillation becomes very poor. As the report points out,
MINIATURE
HIGH-ENVIRONMEN'!1
RECORDERS
265
a deliberate choice to accept some jitter is often made in the interest of
efficiency.
In applying the hysteresis synchronous motor, the operating speed
and hence the number of poles and the supply frequency must be
established. Although the motor is usually expected to be more efficient in every way the faster it runs, there is a limit to usable motor
speed because the higher the speed the greater the burden on the speed
reduction unit. In figure 13-39, motor efficiency is plotted versus
rotational speed for a typical hysteresis-synchronous motor, where the
frictional losses are assumed to lie largely outside the motor. In figure
13-40, the output torque of the motor as well as the torque of a typical
duplex-pair bearing, which represents the frictional losses within the
motor, is plotted against motor speed. The difference between these
curves, which is the relative net power, is also plotted; it is interesting
to note that it is higher for lower speeds. A significant comment is
made in the report apropros of this unusual result, that, although
optimum efficiencies can be obtained at speeds less than 2,000 rpm,
peak efficiency shifts toward higher speeds as the power is increased.
Two thousand rpm therefore appears an optimum motor speed. Having chosen a speed one must somehow standardize the supply
frequency.
60
~
30
V
1000
,.......
~
~
3000
5000
DESIGN MOTOR SPEED (rpm)
7000
(after Lockheed [1962])
FIGURE
13-39.-Motor efficiency versus design motor speed for a typical O.3-watt
synchronous motor (typical spacecraft record power level).
Many such motors have been designed for 400 cps because of their
history of use in aircraft systems. Under normal circumstances, 400
cps produces much too fast a motor for spacecraft service but it is
desirable to use a frequency which is derived from 400 cycles. One
hundred cps is about as low a frequency as can be used without requiring oversized inductive components in the power supply, and also
266
MAGNETIC
TAPE
RECORDING
1.00 ,.....---..---r-----,.------r---r-----,
0.08
I-l.-~..=_--+---+---+_-~--__l
·7 0.061---~1 --3001_--+----1-----1----11.5
2
...
~
f:
~
1.4
..,
~
~
~
O.04I----t--~:__-+-=-.j.;;;;;;;:=:J--_i 1.3 ~
...
1.2 ~
0.02 f---......",;...--+--+--=-C::--~f----11.1 ...
>
1.0
o L-_......L_ _....L_ _...l...-_ _L-_........J'--_---l 8.9
1000
3000
5000
~
...n::
...J
7000
OESIGN MOTOR SPEED
(after Lockheed
[196~])
FIGURE 13-40.-Torque and power l'elationships in the motor of figure 13-39.
(a) Output torque of equivalent frictionless motor, (b) available output
torque with typical bearings, (c) torque requirement of a typical duplex-pair
bearing set, (d) relative net power output.
permits using a relatively small number of poles for a 2,OOO-rpm
motor.
A primary decision which must be made in the spacecraft recorder
is how it shall be operated at two different speeds. There are several
ways in which two-speed operation can be achieved. These include
running a single motor at two different speeds with a fixed reduction,
using a single motor and selecting different reduction ratios with
solenoid-operated clutches, running the motor in one direction for one
speed and in the other direction for the other speed with a differential
to change speeds, and providing two independent motor-speed-reduction systems. A table comparing the single and double motor approach is reproduced here from the report (table 13-2). In the report
from which this data came, the two-motor drive system is recommended for the particular class of recorders emphasized in the report but it is not necessarily the universal best choice. The schematic
perspective drawings of figures 13-41 and 13-42 show the differences
between typical single- or two-motor spacecraft recorder drive. Other
systems for providing two-motor operation are discussed under specific recorders in previous sections of this chapter.
A preliminary feasibility and breadboard study of an illtermittentmotion digital tape recorder/reproducer was sponsored by the Jet
MINIATURE
TABLE
HIGH-ENVIRONMENT
RECORDERS
267
13-2.-0omparison of Single- and Double-Motor Drives for
Olass I Spacecraft Recorders
Type
Advantages
Single-motor __ Less weight.
Smaller plan size.
Lower cost.
Double-motor _ Low-speed motor designed for purpose
allows greater
efficiency.
Smaller depth.
Easily modularized
because of split drive.
All-belt drive.
Fewer parts, less complicated.
Disadvantages
Greater depth required.
Compromise motor design
increases power required
at low speed.
More complicated.
Transmission requiring
pressure-run drive for
reversal.
Harder to modularize because of interdependence
of record and reproduce
drives.
Additional weight of motor.
Larger size.
Higher cost (additional
motor).
Propulsion Laboratory during 1963. The interest in this program
originated in a requirement in typical interplanetary space probes for
recording the output of such devices as ion chambers and GeigerMuller tubes which have extremely large dynamic ranges (from 1
count per hour to 100,000 counts per second), operate asynchronously
and, despite the wide dynamic range, have a low average data rate
when the output is converted to digital form (maximum of 20 bits
per second) . These requirements suggested the desirability of a device
capable of recording, uniquely, one bit at a time, placing such recorded bits on a magnetic tape in a regular fashion and having the
capability of synchronous read out. To achieve these ends, the following basic systems considerations were established:
1. The recorder-reproducer is to record and playback only upon
receipt of a command pulse.
2. The maximum stepping rate is to be 200 steps per second or
greater.
3. Size and physical configuration shall be such as to store 105
bits.
4. No standby power shall be supplied to the recorder-reproducer.
268
MAGNETIC TAPE RECORDING
RECORD-REPRODUCE
MOTOR
CAPSTANS
LOCATED
AT
©
II<
®
(from a NASA report)
FIGURE
13-41.-Mechanical arrangement of a typical single-motor spacecraft
recorder.
This is a perspective representation of essentially the same elements shown
in figure 13-24. The reference letters are appropriate to the report in which
this drawing first appeared.
5. Ultimate life expectancy of the recorder-reproducer is to be
about 108 steps. (Storer [1963]).
A successful breadboard model of this unit was completed. A general view is shown in figure 13-43. A special quarter-inch tape, perforated with standard 8-mm motion picture film sprocket holes
and silicon-lubricated on the oxide side, was used. The reels, in this
case, were torqued by a Negator spring in a stacked-reel configuration.
The magnetic actuator visible in the photograph required refined and
careful design. It consists basically of an armature which swings
back and forth through a 6° arc between two pole pieces of highpermeability metal. A permanent magnet is inserted in the magnetic
circuit so that the armature will be held against whichever pole piece
MINIATURE
HIGH-ENVIRON:&IEN'l1
269
RECORDERS
REPRODUCE
MOTOR
CAPSTANS LOCATED
AT
®
&
®
(from, a NASA r eport)
FIGURE 13-42.-Mechanical arrangement of a typical two-motor spacecraft recorder.
This drive method limits the number of overrunning clutches required to one.
(from
FIGURE
a NASA report)
13--43.-Bread!board intermittent-motion spacecraft recording mechanism
(see
te~t).
270
MAGNETIC TAPE RECORDING
it has contacted last. The motion is transmitted to the tape from the
actuator through a small sprocket engaging the tape perforations.
This study, although very preliminary, does establish the feasibility
of the use of such an intermittent recording mechanism. The breadboard was tested with a loop of tape at a rate of 200 steps per second
for a continuous period of 5 days before failure. This amounted to
over 90 million accumulated steps and failure was due simply to wearout of the armature pivot shaft..
This mechanism gives hope of making proper recordings of many
intermittent phenomena which, up to now, have been left. unrecorded
because of the continuous drain that normal recording techniques
would place on the spacecraft power supply.
SPECIAL PROBLEMS OF SPACE PROBE RECORDERS
In addition to the normal flight recorder problems which are covered
above, there are some peculiar to space probes.
Any tape recorder that must undergo a sterilization at high temperature for interplanetary use encounters a whole new series of environmental stresses which have not been significant in any previous
development. Just how to make recorders survive this process is
not, as yet, well known. Tape, naturally, is the weakest link in the
recording process as far as withstanding heat is concerned and tape
temperature problems are discussed in chapter 9. The same temperature problems occur with plastic drive belts, and many other parts
of the recorder may be damaged by heat. For example, the head is
almost always constructed partially of plastic and few plastics will
withstand sterilization temperatures. The result is distortion of the
head and resulting poor head performance. At the moment, this
entire subject is under intensive study, but as of the time of the survey,
there are no "guarant.eed methods" for assuring that a flight recorder
will survive the temperature cycle.
Because the space probe must transmit its data over such long distances, data rates reproduced from the onboard recorder are always
very low. A problem which must always be solved in any tape
recorder which plays back at such low speed is that of dealing with the
extremely small playba.ck voltages obtained from a conventional dcp/dt
playback head. Typically, at 0.03 ips, playback voltages do not exceed 100 microvolts. However, although the voltages are tiny, the
bandwidth needed is correspondingly low, and surprisingly, without
any very spectacular development, most contractors have had little
difficulty in providing conventional playback amplifiers for these
signals. Flux sensitive heads have therefore not been promoted for
space-probe service as yet.
CHAPTER 14
Complete Recording Systems
It probably would be fairly realistic to say that the magnetic
recorder is currently a necessary evil in any complex electronic data
gathering system. The recorder has been gathering technical data
for us for 15 years or more, and the technical personnel involved in
the gathering process are now completely accustomed to the peculiarities of the recorder. Although ItIl occasional complaint expresses the
discontent which the typical user feels from time to time about the
shortcomings of his recorder (Ratner [1965]), and one occasionally
hears of proposals to replace magnetic recording with a photographic
or similar technique, by and large the technical community suffers in
silence with the recorder problems of the current state of the art.
Like a benevolent tryant, the magnetic recorder impressed men originally by being so much better than anything else available that it was
accepted gladly. Disenchantment with the new regime soon set in,
however, with the realization that the recorder was not quite as good
as it was originally expected to be. The gradual step-by-step improvement from year to year of recorder capabilities has, however, kept the
using public from extensive overt expressions of unhappiness with
recorder limitations. Nevertheless, the tape recorder does have many
shortcomings, and great ingenuity is employed in its application today
to minimize the effects of those shortcomings.
Typically, a tape recorder is used in a complex system. A great
deal of the system design must now be based on satisfying the recorder
limitations, that is, on so designing the balance of the system as to
minimize the effect of the imperfections in the recorder. Until something better comes along, this necessarily must be the way in which
data acquisition operates. One of the purposes of this chapter is to
discuss the way in which this compensating process takes place and
the overall results so achieved.
271
272
MAGNETIC TAPE RECORDING
The approach here will be simply to describe several typical datagathering systems in which tape recorders are used. These wiII
include two rather simple and one rather complex satellite recorder
applications. In another application, a ground-based recorder is used
directly with a satellite, where the experimenters responsible for the
satellite have specifically rejected the use of an on-board recorder.
The final example will be one of the use of magnetic recorders in a
ground installation for research and development data gathering and
analysis in some of its more complex forms. This latter system is
representative of many of the data reduction systems which are
associated with the complex gathering systems.
The Interplanetary Monitoring Platform (IMP) is an example of a
research satellite employing no on-board recorder and making the
necessary compromises inherent in the decision not to carry a recorder.
The information presented is intended to show the complexity of data
and coding involved in the basic satellite decisions and some aspects
of the data reduction problem inherent in the lack of the on-board
recorder.
The neit Operational Plan presented is that for the Orbiting Geophysical Observatory, OGO-A also known as EGO for its extremely
eccentric orbit. The recorder developed by RCA for use in this program is described in chapter 13 which includes a discussion of the
provision in the recorder design for its use both in EGO and POGO.
The data reduction and ground recording requirements of this satellite
are covered here.
The third Operations Plan discussed is that of Nimbus. Nimbus
is the most complex of the satellites involved and requires overall
system performance of the recording channel greater than any previous
satellite. It involves two separate on-board recording functions, the
recorders for which are discussed in chapter 13. It also employs
the particular flutter compensation technique discussed in the section
immediately preceding the description of the Nimbus operational plan
below.
THE INTERPLANETARY' MONITORING PLATFORM (IMP)
The mission of the Interplanetary Monitoring Platform project
(IMP) is described as follows in NASA-GSFC Operations Plan 1064, paragraphs 1.0, 1.1 and 1.1.1 through 1.1.5:
Mission
The Interphmetary Monitoring Platform (IMP) Project
consists of a series of spacecraft which wiII contain scientific
experiments to provide comprehensive data from the intensity
and charge spectra of cosmic rays, information on the solar
wind and the interplanetary magnetic field.
COMPLETE
RECORDING
SYSTEMS
273
Objective8
There are fiye objectives of the IMP project. The objectives
may be detailed as indicated below:
1. To make a detailed study of the radiation and environment of cislunar space and to monitor this region over a
significant portion of a solar cycle. (This infonnation
necessary for support of ProJect Apollo requires that
an operational IMP be in orbit at all times during this
project.)
2. To study the quiescent properties of the interplanetary
magnetic field and its dynamical relationship with particle fluxes from the sun.
3. To develop a solar-flare prediction capability for Apollo.
4. To extend our knowledge of solar-terrestrial relationships.
5. To further the development of relatively inexpensive
spin-stabilized spacecraft for interplanetary investigatIOns.
Figure 14-1 is a photograph of IMP.
(from a NASA photograph)
FIGURE
14-1.-Tbe Interplanetary Monitoring Platform (IMP) tlight unit.
274
MAGNETIC TAPE RECORDING
The orbit of IMP is extremely eccentric with an apogee of 110,000
nautical miles and a perigee of 105 nautical miles. The period is 98.5
hours. In this respect IMP is similar to the Orbiting Geophysical
Observatory (OGO) but. almost every other aspect of its plan differs.
Despite the extreme distance over which data must often be transmitted in such an eccentric orbit, the information from IMP will be
transmitted direct to the ground recording installation without onboard tape recorder delay. Because the total amount of information
to be transmitted in any unit time is relatively low, a certain amount of
digital on-board storage has been provided for part of the experiment.
Otherwise the data is transmitted directly. That is, the output of a
part.icular sensor modulates the telemetry transmitter directly through
an encoder and the data is recorded in real time on the ground.
Some nine experiments carried on this spacecraft are divided according to the type of output obtained from each. The output signals
from four of these experiments are in analog form and those from
the other five are in digital form. The output of these multiple signal
sources must, of course, be sampled. This sampling is not in the
simple iterative form used for the conventional PCM commutatordecommutator system, ~nce the input signals are not in identical
format. For example, to quote from the Operations Plan:
"Rb-vapor magnetometer: the Rb-vapor magnetometer output will
directly modulate the transmitter for 81.92 seconds once every 5.46
minutes."
Similar statements are made concerning each of the other three
analog experiments. A certain period of time during which the analog
signal is telemetered directly is assigned to each experiment. The
digital experiments are sampled on a more conventional basis in which
instantaneous sampling of the output is done at regular intervals and
the samples are stored in a small on-board digital memory. The digital data in the memory is then extracted on an irregular time schedule
compatible with the analog sampling. The basic digital encoding
technique is via PFM with which a modified burst FM analog coding
has been integrated. The coding technique and experiment array
may be understood from the following quotation from the Operations
Plan (paragraphs 4.3.6.1, 4.3.6.1.1, 4.3.6.1.2) :
Experiment Outputs
Following is the list of experiments grouped according to
the output of each (analog or digital) :
Analog
Low-energy proton analyzer (Ames)
COMPLETE RECORDING SYSTEMS
275
Plasma probe (MIT)
Thermal-ion and electron spectrometer (GSFC)
Fluxgate magnetometers (GSFC)
Digital
Range vs. energy loss (University of Chicago)
Total energy vs. energy loss (GSFC)
Neher-type ion chamber (University of California)
Othogonal Geiger counter telescope array (GSFC)
Optical aspect sensor (GSFC)
Analog
The outputs (between 0 and plus 5 volts dc) of the analog
experiments are converted into frequency outputs from 15 to
5 kc by analog oscillators. Zero volts corresponds to approximately 15 kc while plus 5 volts corresponds to approximately
5 kc. Several inputs may be gated into a single oscillator.
The oscillator outputs are later divided by 16, so that the modulating frequency is in the range of 5/16 to 15/16 kc.
Digital
The outputs of the digital experiments (except the optical
aspect sensor) are fed into the DDP (Digital Data Processor)
which does much of the accumulation and digital subcommutation. The output of the DDP goes to a series of pulsed digital
oscillators which accept three bits of information and encode
this as one of eight discrete frequencies in the band from 5 to
15 kc. These oscillator outputs are also later divided by 16.
The eight possible frequency levels correspond to the eight possible combinations of 3 bits. Thus digital information, which
is already in a digital form, may be easily encoded.
Figure 14-2 shows the IMP telemetry format for "normal"
sequences.
The ground receiving station instructions for acquiring the
telemetry data from IMP are somewhat unusual in that, for example,
the satellite has no command channel, cannot. be commanded to
operate in any particular way and therefore transmits all the time.
This means that no command data must be transmitted or recorded
and several modes of recording operation are ne<>essary to guarantee
that all the data is dbtained from the compromise transmission design
required by continuous operation of the satellite. There are two different recording modes provided for. In one, the tape recorder
operates at 1% inches per second and all data is recorded on one set of
duplicated recorders. In the second mode, the data is recorded simultaneously at 1% and 30 inches per second, again with individual
recorders duplicated. The following note' about additional safety
276
MAGNETIC TAPE RECORDING
I ' I .. I I .. I I • I I I
~
- - 'TO'''\'
o
--@
II.
1I
"",,-.c.'If
~
110
•
..
.0
v, 1....1....1
."MGol.
Of C.MIC.A.ttO)
'''UX ....~" ...
@
(UMIY 0' (.MltAr.o)
®
~l?12'n??lilanllal??al??illaZ2?2??1
... ••
"
.@.
tOM (.""Wac.&.,,,"W 0' (,1tt.U')
@
.~
1\
~O•• (""\'I
9.
@.
!.MC..G.'" \.0••
.. ---@
'l
1......"( \.0"
®
- - 1'O~A\. 'MI.• •'( "
t
T
(@ ...." ..... ) - - - -
1???2l1llllllilll'??IZ2 lillllll>
C,""".I.a. lU\lW 0' c..\.Ir)
DIG.IT .. L. t.LOt ..
® @ ( ! ) I - -_ _ __
G.I.'.I. .. '~\.c..C.O.t.
101
to"
c. ... c.
•
.
100
,;
..,
S
~
:!
..
•
•
10
II
.~
Itt
I
I
I
7
I
2 2 2
- - TOT"'\. U4E...f.'t' "1\ 1."'1..6" ...OU
010
..
<~ I
I
I
?
Z 2
?
2
2
Ciaot.lGot.R "1.\.I.~Pt.
I 2 2 2 2 2 , 2
t.A.~\~)
I
I
I
2 I~
(UM,.,. 0" C.MIC. .. ctO)
IS "NALO~
Z ?
?
I
2
c..rc.
?
?
to'"
?
2
,
7
6..M P \ . I . ' » ) - - - - - - - - - - - -
'LU .. Go.."..
c.loU.VI.l.a(UNlY OT
'"
Z
c. ... \-".}
2 1 7 7 7 2 7 7=::4>
t)IGt,'U.\.
c.\.OCk.
®)-----+-------~~--~--------<~)------
001
"0
to,,", c;.M.....l.a(u ...., zt>
=:::zt>
G. ...
,.l. •
2
?
2
2
Z
?
I
I
Z
?
2
2
?
I
2
it>
• '" aUG OUT 0,. DDP ...o~ ..
cz:z:tzz;z:Ol ..M..\.o. tt."D Ou1' C.ON1'U.OU~ .....0 .I.to.".o aU.'T
"",. ,,1...'O ..... NC.r. ,,1trt.....wl.1'I." " ,,(0 ,. ~t.tO"D """D OUT)
4. " '.. "c.
1'1.~1.""'I.T"'t t.1.Q\,l'E.Nc..1. 1 • • \1. . ,• • •
,1.c..O.. M
a\,l"'T\ (~&"") ,. .... !. 0.'. 'I.e.. 'OL.\..OW'NGo "0.'.'1.(. aL..MK
6'tH(. (.MA.MNt.L. eUa,,., ""-«.1. 0.1..4 "Cot. P'OL.LOWINf'." 0.0. ,t.e.. .~"M"
"~"'MP,",'M" .....:,t, ~ t: .CV,ft
' ••• "'NC. O.c..
'.n.
~,,~t.1'~
OI"1'''~
'''fl.
(from
FIGURE
II
N ABA report)
14---2.-The IMP telemetry format for "normal" sequences.
The terminology is typically somewhat obSCure. The format designates what
information is transmitted in what channel for each frame. The entire process
is sequential here, that is, a frame in which samples from each channel are transmitted in time order is followed also in time order by another frame in which a
new set of samples is transmitted in each at the various channel times. For some
experiments (note "flux gate A" in frame 3), the entire frame is occupied with a
data transmission. A frame identification symbol is transmitted in the zero
ehannel and there are other circumstances under which all or part of this
sequence are preempted for steady transmission of certain data.
factors in obtaining a complete record quoted from the Operations
Plan is of interest (paragraph 6.1.3.2) :
"There will be at least 30 minutes of recorded telemetry overlap
between stations, 15 minutes of recorded overlap between tapes of 1%
COMPLETE RECORDING SYSTEMS
277
inches per second and 10 minutes of recorded overlap between tapes
of 15 ips and 30 ips."
The reference to speeds of 15 to 30 ips undoubtedly means that, since
either Ampex FR-lOO or FR-600 recorders may be used, the FR-600
with its somewhat more modern heads and electronics is capable of
adequate band width at the lower speed.
The track assignments for the two recorder speeds describe pretty
clearly the recording technique. Track 1 for the low speed (1% ips)
mode direct (analog) records the AGC levels of the receivers in the
form of IRIG Channel 1 and 3 VCO outputs. Track 2 carries the
control track in the form of 60 cps modulated on 1 kcps in order to
permit synchronization of the recorder on playback. Tra,ck 3 records
the phase-detected data signal derived from the diversity combiner
in direct mode. Track 4 carries the Minitrack time standard in the
form of 1 kc modulated with binary-coded-decimal time, recorded
in direct mode. Track 5 carries the phase-detected data signal in
an FM recording mode. Track 6 carries the Minitrack t.ime standard in the form of serial decimal time code recorded in FM, and Track
7 carries a voice commentary and WWV t.ime, recorded direct.
By using both FM and direct mode recording, data signal response
down to direct current with a relatively wide overall channel bandwidth can be obtained.
The track assignments for the high speed (15/30 ips) record scheme
are similar. Track 1 again carries the receiver AGC signals, direct
recorded in the form of VCO outputs. Track 2 carries a direct recording of Minitrack time in the form of 1 kcps modulated with binarycoded-decimal time, mixed with the standard speed control track
(60 cps) modulated, in this case, on 18.24 kcps (for playback synchronization) . Track 3 records directly the phase-detected signal
from the diversity combiner and Track 4, also direct recorded, carries
a 10 kcps reference frequency derived from the Minitrack format generator. Track 5 carries a direct recording of WWV time from t.he
WWV receiver. Track 6 carries an FM recording of t.he Minit.rack
t.ime standard in t.he form of a serial decimal t.ime code and Track 7
carries a direct-recorded voice commentary.
Not only in the eccentricity of its orbit but also in much of its data
handling is IMP an unusual satellite. Because data must be recorded
when the satellite is quite far from the earth, the data rates have been
held down as low as possible. Pulse frequency modulat.ion is chosen
as the first step in the encoding scheme as it appears to maximize
the signal-to-noise ratio with relatively simple encoding and decoding
equipment. The use of comb filters permits the decoding of PFM
data at an optimum signal-to-noise ratio by picking up noise only in
278
MAGNETIC TAPE
RECORDING
the total bandwidth of all the comb filters combined rather than in
the bandwidth from the bottom of the first filter passband to the top
of the last filter p3$band (chapter 4). With these very low recording rates, the tapes are reproduced for data reduction at 16 times the
recorded rate. (Data recorded at 1% ips is reproduced at 30 ips.) The
mixture of digital and analog data on one channel is somewhat unusual, as is the method of transmitting an 8-bitdigital sample. In
this technique, mentioned above, 8 different pulse repetition frequencies appear at the output of the PFM encoder. A burst of the selected
frequency is transmitted, chosen on the basis of the number between
o and 7 that is to be transmitted at a particular time. In other words,
3 bits of digital data are converted to octal form and transmitted as
the pulse frequency corresponding to that number. This technique
was undoubtedly chosen because the system has the relatively good
signal-to-noise ratio required to transmit the analog information effectively. This signal-to-noise ratio makes octal coding possible with the
corresponding improvement in transmission efficiency over straight
digital coding.
Because the reduction of the data from IMP is, although not standard, relatively simple, it provides a good sample of the basic techniques
involved in any satellite data reduction. The actual reduction process
is described in the following (edited) quotation from the Operations
Plan (paragraphs 9.2 through 9.6).
A block diagram of the functions involved is given in figure 14-3.
9.2 Analog-Telemetry Tape to Digital-Computer Tape
Telemetry data is designed to be recorded at 1% ips, on tape
compatible with Ampex FR600 playback equipment operating
at 30 ips, allowing a theoretical reduction of processmg time
approaching 16/1. This feature of the IMP telemetry system
YIelds a substantial improvement in the rate of processing data.
Information recorded on tape at the acquisition stations must
be of uniformly high quality; a major problem in formulating
a processing philosophy can develop if station operations are
other than satisfactory. Systems operation will be set up on
the assumption that the data-acquiSItion stations will perform
adequately. The problem of correctly recording coded time on
tape simultaneously with telemetry information will be checked
by an on-board satellite clock.
Necessary to the PFM encoding technique is the utilization
in the data-processing operation of comb filters, the function
of which is to improve the SIN ratio by reducing the noise bandwidth. The conditioned signal is sampled at a particular rate
and in a particular mode determined by the output of the frame,
channel, and sequence sync detector under the control of the
programmer. The time code is recovered in a separate operation which depends upon an initial automatic or manual readin of time, and subsequent updating using the reference fre-
COMPLETE RECORDING SYSTEMS
279
-ANALOG TAPE
SIMULATED
MASTER
DATA TAPE
(TEMPORARY) RAW WORK
TAPE (LINE CORRECTEDJ--
(TEMPORARY) RAW WORK TAPE _ _ _
EXPERIMENTER'S AND OA AND PP TAPES/
"
EXPERIMENTERS PRINTOUT
(from a NASA report)
FIGURE 14--3.-Block diagram IMP data reduction system (see text).
quency recorded on the telemetry tape. The sampled signal
is converted to a digital form and stored sequentIally in the
magnetic-core memory of the output buffer which produces
the digital tape. Each time a frame sync pulse is detected, the
updated time, frameA and sequence identification are also stored
in the memory. A nag bit indicates the correspondence of updated time and recorded time. At the appropriate time, the
entire sequence of stored characters is dumped on an output
digital magnetic tape.
Since the analog-to-digital tape-conversion process operates
at 16/1 ratio relative to acquisition time, there is no need for
preprocess editing of tape. During the initial processing of
step I (fig. 14-3), both visual and recorded displays will
monitor the quality of tape recording and playback of satellite
information. For those samples of data during which there
is insufficient SIN ratio to accurately recover the data, a flag
will be recorded, indicating a signal-dropout condition. Special
purpose processing equipment will be employed for the Rbma~etometer signal when the spacecraft is within 5 earth
radIi (2% of orbItal period). In summary, the equipment consists of:
(1) FR600 playback tape deck.
(2) Comb filters.
(3) Synchronizer detector and identifier.
1f88-008 ()........65..-19
280
MAGNETIC TAPE
RECORDING
(4) Analog-digital conversion (discriminator providing a
precision of 1/1000, and filter-tooth detector providing a
precision of 1/100).
(5) Time decoder.
(6) Programmer-control.
(7) Core memory buffer and tape drive.
(8) Digital tape deck.
9.3 Master Data-Tape Production
The output tape from step I consists of a string of BCD
characters with tape identification, time, and data recorded in
a single block corresponding to an individual sequence. A
prime function of step II is to check the time-code data for
accuracy with reference to the satellite clock and to perform the
necessary corrections. In addition, the informatIOn content
will be checked and corrected, for character-count consistency.
The information is put into format logically with correct time
and identification, and a master data tape is generated and
printed simultaneously. In addition, a punched-card log of
the telemetry tapes recorded on the digital master data tapes
will be maintained for summary and searching functions in step
III. The medium-scale computer of the Data Processing
Branch, Data Systems Division, will also be used for this phase
of IMP information processing. These master data tapes will
form the library from which the experimental data tapes will
be prepared.
9.4 Experimental Data-Tape Production
The experimenter's data tapes will be prepared using the
master data tapes produced in step II as mput. The accompanying telemetry-tape log on punched cards will provide a
means for automatically reading a master data tare at random,
so that a chronological data sequence can be readIly generated.
In addition, it will provide documentation for both steps II
and III. The prime function of step III is to decommutate
the individual experimenter's data to separate tapes in different
formats and provide an accompanying printout.
9.5 Master Trajectory-Data Production
This operation is principally a merging of data and an extensive computational effort which produces the final master
trajectory information. It is best accomplished on a large-scale
computer system such as the IBM 7090, which is planned to be
employed for this purpose. Approximately 2 hours of 7090
time will be required per week of trajectory. Duplication of
the trajectory tapes and the printouts for each of the experimenters will be accomplished on the 7094 computer in the Data
Processing Branch.
9.6 Data Evaluation
The station telemetry tapes will be received by the Data
Processing Branch for evaluation and reduction. A representative cross sample from each station will 'be reviewed for quality
and quantity of usable data and checked against expected sta-
COMPLETE
RECORDING
SYSTEMS
281
tion performance. A summary of tapes by station will be
maintained as a control input for statIon operation and as a
guide for processing. Tapes will then be arranged in chronological order to enable processing in a time-history sequence.
It is anticipated that there will be three categories of tapes.
They are (1) unusable tapes resulting from inadequate signalto-noise ratio, interference, or operator error; (2) questionable
tape requiring extra handling to recover the data; and (3) good
tapes of sufficient quality to warrant immediate machine processing. The first class of tapes will be retained for archival
purposes and possible exploitation as the state of the art is
advanced. The remaining two categories will be processed
through the system described in the preceding paragraphs of
this section.
OGO-A
As described in the Operations Plan (Paragraphs 1.1 and 1.2, Plan
11-64) :
The primary objective of the Orbiting Geophysical Observatory
program is to conduct large numbers of significant, diversified
geophysical experiments for obtaining a better understanding
of the earth-sun relationships and the earth as a planet. The
secondary objective of the program is the development and
operation of a standardized observatory-type oriented spacecraft consisting of a basic structure and a subsystems design
which can be used repeatedly to carry large numbers of
easily integrated scientific experiments in a wide variety of
orbits.
OGO-A is the first mission in the OGO program and is also referred to
as EGO for Eccentric Orbiting Geophysical Observatory. For data
handling, eccentric is the key word because it refers to the very wide
range of distances this satellite will have from the earth. Actually the
apogee is 149,000 kilometers (80,400 nautical miles) and the perigee is
275 kilometers (148 nautical miles). The rotational period of the
satellite is 63.85 hours, a figure many times that usually considered to
be a reasonable satellite period.
In the OGO-A mission, magnetic tape recorders have the following
tasks to perform:
1. Ground-station recording of FM/FM telemetry data on con-
ditions aboard the Atlas and Agena launch phase.
2. On-board recording of the data from a large number of experinwnts in the form of wideband PCM. This on-board data
is stored during the orbit period and "dumped" at a high rate
of speed while the satellite is in view of a ground station.
3. Recording at ground stations of PCM data telemetered via
playback from the on-hoard recorder.
4. Recording at ground stations on a semi-continuous basis of
282
MAGNETIC TAPE RECORDING
special purpose telemetry which produces data on the spacecraft incompatible with the PCM telemetry and which hence
is transmitted FM/PM. This data consists primarily of noise
and electrical phenomenon measurement related to a study of
of very low frequency (VLF) radio transmission.
Another function is required of magnetic recording in the launching
of the vehicle and its injection into orbit. This function has now
become so routine as hardly to require mentioning. The Minitrack
function is performed in that a transponder aboard the spacecraft is
interrogated from various of the ground stations and returns a signal
to the ground station which is received by a complex extended antenna
array on the ground. This antenna array is so instrumented as to give
precision phase and amplitude information from the received signal,
from which information the spacecraft's relative slant range and track
can be determined. Typically the raw Minitrack phase and amplitude data are recorded, as well as computed phase data reduced to digital numerical form and slant range calculations in digital form. This
recorded information is used later to determine the spacecraft's track
and orbit with precision.
There is some overlap between the functions of the various recording
subsystems. For example, data from the sensors on board the spacecraft which determine operating conditions during launch and orbit
injection are telemetered to the ground by a system having quite different ground station recording requirements from that used with spacecraft sensors for determining orbital conditions. The orbital satellite
sensors are carefully integrated with the PCM system used for transmitting the data from the orbital experiments, and use the same telemetry transmitters and receivers. The Atlas and Agena launch vehicle
sensors are independently instrumented for the individual parts of the
vehicle and are telemetered with still another system with its own
ground recording requirements.
The extremely eccentric orbit and correspondingly long orbit time
of OGO is unusual. During much of the orbit, little information can
be obtained for the satellite whether it is in "view" of the ground
station or not. Data is collected relatively slowly in distant parts of
the orbit and then "dumped" relatively fast while the satellite is near
the earth. For satellites with a more nearly circular orbit the data is
collected slowly and dumped fast, not because of change in distance
between satellite and earth but because only in a limited part of the
orbit is the satellite in "view" of the relatively few tracking stations
available. (Compare these two modes of operation with that of IMP,
discussed earlier.)
However, OGO has three different basic bit rates in its PCM system.
COMPLETE RECORDING SYSTEMS
283
These are 1,000, 8,000, and 64,000 bits per second. All data which is
recorded is taken at 1,000 bits per second but data which can be transmitted in real time direct from the spacecraft to the ground station
can be obtained at any of the three bit rates. Depending on receiving
conditions and the signal-to-noise ratio of the telemetry link, the
ground station can command real time data at a rate appropriate to
the transmission conditions obtaining. Recorded data is always
played back at a 64 to 1 increase in speed in order to take full advantage of the short time during which the satellite is in good "viewing"
distance.
To make the playback of data completely independent of the timing
within the orbit, two tape recorders are provided. One of them
records at any particular time and the other is ready to take over when
for any reason the operation of the first is interrupted. A simple
operation interrupt would occur when the first recorder came to the
end of its tape. Under these circumstances, the second recorder would
then proceed to record. Another interruption might be the demand
from the ground that the tape recorder play back its recorded data.
The following quotation from the operations plan document will indicate clearly the format used for the recorders (part of paragraph
4.3.1-4.2) :
The OGO-A Observatory will carry two on-board magnetic
tape recorders for storage of PCM data. The data will be
stored at the rate of 1,000 bits per second and played back on
command at 64,000 bits per second. The data is played back in
reverse order from which it was recorded. Each of the two
spacecraft recorders is capable of recording for a maximum of
12 hours. When one recorder has been filled, recording is
automatically switched to the second recorder. When the
spacecraft accepts a command to play back stored data, one of
the following two sequences takes place depending on which
recorder is in operation at that time:
1. If data is being recorded by the first recorder when the
command is given to play back, recording will be
switched to the second recorder while the first recorder
is playing back. Upon completion of first recorder
playback, recording will revert back to the first recorder
and that data stored by the second recorder (during
playback of the first recorder) will then be played back.
~. If data is being recorded by the second recorder when
the command is given to play back, data will continue
to be recorded by the second recorder while the entire
first recorder is played back. Upon completion of firSt-
284
MAGNETIC
TAPE RECORDING
recorder playback recording will be switched to the first
recorder and all data stored by the second recorder will
be played back.
The time required to complete a playback sequence depends
on the amount of data stored. Playback of one completely
filled recorder requires 11.25 minutes.
Because the total period of this satellite is 63.85 hours, the 24-hour
recording capability does not, of course, cover the entire orbit. It
does not even cover the entire relatively distant part of the orbit.
For example, 20 hours into the first orbit, which is about 12 hours
before apogee, the spacecraft is 132,000 kilometers from the earth.
At apogee it is 150,000 kilometers away at some time after 32 hours,
and at 54 hours it appears from extrapolation on the subsatellite plot
that it would be about 100,000 kilometers from the earth. It appears,
therefore, that the recording scheme is not used necessarily to assure
that the data will be dumped when the satellite is near the earth but
rather to assure that the data will be received when the satellite, even
though distant from the earth, is in clear view of any specific receiving
station.
The ground recorders for this satellite are used in a fairly typical
manner during each of the several phases of the flight. The Goddard
Space Flight Center Space Tracking and Data Acquisition Network
(STADAN) records the data from the normal tracking part of the
operation, that is, during the stable established orbit, on seven-channel
Ampex FR-600 recorders. The tape recorder speeds will be 30 inches
per second for a 64,000 bits per second bit rate, 3>3,4 inches per second
for 8,000 bits per second and 33,4 inches per second for 1,000 bits per
second. All seven channels of the recorder and its monitor voice track
are used. The track assignments are quite typical of such an
application.
For example, in the ca.se of the wideband POM telemetry, track 1
and track 7 are considered to be relatively poor tracks because of the
possibility of edge damage to the tape and the most important data are
placed on tracks 3, 4, and 5. Track 1 records the receiver AGC's and
a 10-kcps reference frequency. These signals are mUltiplexed by using
the 10 kcps reference frequency direct and mixing it with FM telemetry
subcarrier signals placed in IRIG channels 1, 3, 5, and 7, to cover the
AGC's of the four receivers used in the double-diversity system.
Track 2 takes the raw unconditional PCM data from the receiver outputs after it has been passed through one of the receiver-pair diversity
combiners. It is recorded as a direct signal. Channel 3 is also recorded direct and consists primarily of conditioned PCM data which
has passed through a signal conditioner. Track 4 carries the direct
recording of the time standard system which involves decimal time of
COMPLETE RECORDING SYSTEMS
285
day derived from the time standard generating system. Track 5 is
also recorded direct and records the bit-rate clock which has been
derived from the PCM synchronizer and signal conditioner. The
purpose of recording the bit-rate clock is to permit quick reduction of
the PCM data in the acquisition station if the local PCM synchronization is operating correctly. Since the raw, unconditioned PCM data
is available in the record, a better synchronizing job may be done later
on playback if the ground-station job is inadequate. Track 6 is an
FM-recorded time-standard system which gives a serial decimal time
code, a key input to any such tracking operation. Track 7, which is
also recorded direct carnes the command signals that have been sent
from the command transmitter to the satellite. Track 8, which is
really the voice monitor track, carries a direct audio recording of the
voice commentary and also contains the WWV reference time.
The special-purpose telemetry transmitted in FM/PM form is recorded on an identical recorder somewhat differently implemented.
On track 1, as before the receiver AGe's are modulated onto IRIG subcarriers and recorded directly. On track 2 there is a mixture of the
binary time code from the time-standard system and the control track
signal which is used to synchronize the playback of the recorder to the
power system of the playback point. The control track signal is a
60 cps signal amplitude-modulated on an 18.24-kcps carrier. On
track 3, the FM data output, direct from the diversity combiner, is
direct recorded. On tracks 4 and 5 there are two separate 10 kcps
reference signals recorded from the time-standard system. On track
6 the serial decimal time code is recorded via FM from the timestandard system. On track 7 the satellite commands are recorded
direct, and on track 8, the monitor track again, the voice commentary
and WWV time are recorded.
A significant difference between the implementation of the two recorders is the absence of the control track signal from the PCM
recorder. Since any time error can always be removed from a PCM
signal by storing it in a buffer at an irregular rate containing time
errors and clocking it out under an accurate time reference, the absolute
speed of the playback recorder is not important. For the specialpurpose telemetry, where the signal is in FM/FM form it is important
that the playback be synchronous with the original recording if the
data is to be transmitted accurately and hence the control track is
needed. It is also significant that it appears that the accuracy of the
playback from a conventional recorder is good enough here that no
flutter or wow compensation is considered to be necessary. It is further interesting to note that the tape speed, 'as of the time of producing
the document from which this data is taken, for the special purpose
286
MAGNETIC TAPE RECORDING
telemetry had not been decided, although it was assumed at that time
that it would be 30 inches per second near perigee and 15 inches per
gecond at all other times.
An interesting compatibility problem is shown by the handling of
the vehicle sensors from the Atlas and Agena vehicles. This data is to
be recorded primarily by the Manned Space Flight network stations
and is therefore recorded on their standard tape units which use not
half-inch but one-inch tape. The one-inch machines are not used intensively in this application since many of the tracks are empty. The
compatibility problem arises in that the data for both these vehicles
will be reduced by the Lockheed Missiles and Space Company which
has responsibility for the Agena data. 'Since Lockheed uses a halfinch standard data format, all data taken by the Manned Space Flight
network on one-inch machines must be transferred to half-inch tape
before Lockheed can deal with it.
The track assignments for the one-inch machines are as follows: On
track 1 the diroot video output from the 244.3 megacycle telemetry
receiver is recorded. (This seems somewhat strange because an edge
track is easily damaged.) On track 2 a voice monitor is recorded and
on track 3 a 100 kcps reference signal which will be used later for FM
flutter compensation. Tracks 4, 6, 8, 10, 12, and 14 are not used.
Track 5 contains WWVH timing, and track 7 records the 244-mcps
signal strength frequency modulated on a 10.5 kcps VCO, and the 244mcps video signal strength modulated on a 5.4 kcps VCO. Track 9
carries the direct video, described as a backup, but probably the primary track, since it is more protected from edge damage than track 1.
Track 11 carries serial decimal time and track 13 carries a 50 kcps reference. When these data are transferred to a half-inch machine the
track assignments are: Track 1 244.3-mcps video, track 2 voice, track
3 WWVH timing, track 4 the signal strength (AGC), track 5 the
video backup, track 6 serial decimal time, and track 7, the 50 kcps
reference.
In specifications for the on-board recorder forming part of a complete data transmission system the problems, discussed in chapter 13,
of providing a reliable enough unit within the space, weight and power
limitations of satellite use must be considered. The performance
of the on-board recorder must, in the interest of system efficiency, be
limited as far as possible. Therefore a study of the tradeoff between
limitation of the performance level required of the on-board recorder
and the possible complexity this limitation adds to ground recording
and data reduction must be performed.
COMPLETE
RECORDING
SYSTEMS
287
The state of the flight recorder art can be expected to be inadequate
to provide completely the performance required from the on-board
recorder if the precision of that unit were to determine the overall
system precision. It is naturally not possible to abandon an entire
line of research because at first glance the on-board recorder appears to
require techniques beyond the state of the art. It is, however, possible so to design the complete data transmission system that the limitations of the on-board recorder are formally integrated into the system
and the data redudion technique provides automatically for correction
of initial recorder deficiencies. A complete recording and data recovery system can then be devised which permits an overall accuracy
higher than that delivered by the original recorder itself.
Before deciding that the state of the art is inadequate to provide
satisfactory on-board recorder performance it is necessary to examine
the state of that art very carefully. Many ingenious recording modes
and permutations of standard coding systems have been used to avoid
the on-board recorder limitations directly. The simple compensation
techniques used, for example, in FM flutter compensation in the
ground-based recorder can be extended to the flight unit. The reference
tone can be transmitted to the flight unit at the same time that the data
is transmitted and reproduced from the flight unit at the same time
that the data. is reproduced. Without the system being concerned
with where the flutter originates, a conventional flutter-compensation operation performed on the data reproduced from the ground
recorder will remove most of the flight-recorder flutter at the same
time.
The decision in the case cited to use FM recording might, of course,
involve initially a careful consideration of whether the increased
bandwidth of the FM system was worth the overall data improvement
that the complex double-FM transmission system involved. In the
same way, the decision to use PCM recording for on-board data always
involves an acceptance of the greatly increased bandwidth that PCM
requires over direct recording. As accuracies required have increased,
the decision to accept the bandwidth problems of PCM is apparently
being accepted more freely by system designers.
After selecting the flight recorder recording mode with care, the
quality of the data played back using routine compensation techniques
or using redundant coding to increase accuracy may still be too low to
provide adequate overall system performance. Complex compensation can then be considered. By complex compensation is meant an
approach in which the entire data transmission system is designed
288
MAGNETIC TAPE RECORDING
specifically to provide compensation for a fully expected and
engineered-in limitation of the on-board recorder.
A classic example of this sort of compensation is used in recording
the data from the Nimbus AVCS (Advanced Video Camera System).
Whereas for the earlier Tiros weather picture satellite with its relatively low 300-line-per-picture video resolution, adequate pictures
could be reproduced from the on-board recorder without compensation
for its flutter, this is no longer possible with the 800-line-per-picture
images produced by Nimbus. In order to deal ~ith such highresolution pictures when the satellite recorder has typical on-board
recorder flutter, a compensating system peculiar to the signal involved
has been devised.
When the satellite is interrogated and the signal is telemetered to
the ground station, a reference tone is transmitted along with the video
signal. This reference tone is recorded at the same time as the video
signal and is played back by the satellite recorder simultaneously with
it. It therefore has the same time irreglarities as the video signal.
Either at the time of reception by the ground station or after reproduction from a high-performance ground recorder system, the reference tone is passed through a discriminator. The instantaneous amplitude of the signal recovered from the discriminator is proportional (in
a conventional discriminator) to the instantaneous flutter in the satellite recorder. That is, its instantaneous value is proportional to the
instantaneous fractional speed error in the recorder. The signal is
then integrated, producing now a signal proportional to the instantaneous displacement of the point on the tape being reproduced from
the position it would have if the tape motion were perfectly regular.
Since this is a scanning system which produces television pictures
by "painting" them onto a monitor screen, the disturbance produced
by the existence of flutter will be the displacement of picture elements
from their proper positions when they are laid down at a uniform
rate by the scanning beam. The reference signal, which is now proportional to time displacement along the tape, is also proportional to time
displacement along the scanned picture line. The derived reference
signal is therefore used to add a corrective displacement to the uniform sweep of the cathode ray tube beam which is "painting" the television picture in such a way as to restore the picture elements to their
proper positions. The brightness of the scanning beam is dependent
both on its current/voltage product and its velocity; the disturbance
of the uniform movement of the scanning beam should normally produce irregularities in brightness. However, in this case, the bright-
COMPLETE RECORDING
SYSTEMS
289
ness variation, which is proportional to the rate of change in the
displacement and hence to the flutter, appears to be unimportant.
Displacement of the picture element is apparently a severe enough
fault that, when it is corrected by the means described, the residual
brightness errors appear invisible.
NIMBUS
The Nimbus weather satellite data handling system is one of the
most complex applications of magnetic recording to satellite data.
Essentially all data from the satellite is returned to the ground from
on-board recorders and the precision required of the data is such that
compensation for on-board recorder irregularities must be provided.
One of the Nimbus recorders is also unusual in that the record and
playback speed are the same, although the recorder is actually used
to "dump" the data in a fraction of the orbit time.
The television cameras in Nimbus have a large angular coverage
toward the earth. If Nimbus has taken a picture of one part of the
earth, about two minutes elapse before it has reached such a position
that the desired overlap between the current picture and one just taken
occurs. The angle of coverage and overlap are such that Nimbus need
only take about 33 pictures per orbit. Eight seconds are required to
scan a single Nimbus picture off the sensitive surface of the vidicon.
Actual recording during an entire orbit therefore lasts only a little
over 4 minutes. This 4 minutes of recording is then played back for a
ground station at exactly the same rate at which it was recorded, in
effect a dumping operation, although carried out at the recording
speed.
There are many similarities between the video systems of the Nimbus and the earlier Tiros. Tiros produced pictures with an equivalent
horizontal and vertical resolution of about 300 television lines whereas
Nimbus has a resolution capability of 800 lines in both directions. It
was permissible to playback the Tiros television signal from the onboard recorder and place it directly on the cathode ray tube monitor
on the ground for photographic recording without compensating for
deficiencies in the on-board recorder. However, the higher resolution
of Nimbus showed up the flutter and wow of the on-board recorder
as unacceptable errors in geometry of the reproduced picture. Compensation for the on-board spacecraft irregularities was therefore
provided, as discussed in the preceding section.
The Nimbus system includes not only the so-called AVCS or Advanced Vidicon Camera System for daytime pictures of the earth but
290
MAGNETIC TAPE RECORDING
also the so-called HRIR or High Resolution Infrared monitoring
system. The HRIR system is in effect a crude mechanically scanned
television system operating in the infrared and thus permitting pictures to be obtained at night when starlight would not be adequate to
produce pictures in visible light. Along with this double imaging
system the satellite transmits data from several other on-board sensors
besides the image forming ones and must send extremely accurate
attitude and position determining data.
The attitude and position data is necessary so that the precise relationship between the pictures taken and the position of the satellite
over the earth can be established so that properly-integrated continuous picture coverage can be obtained. In order to use the pictures most.
effectively, a grid system indicating latitude and longitude lines, is
generated in a computer on the ground and applied to the final images
produced so that the weather pictures are accurately referenced to the
earth. The satellite therefore has a recorder for the video system, a
recorder for the HRIR system and a recorder for the PCM, attitude
and other satellite on-board data. The operation of these systems is
best understood by the following quotation from the Nimbus Data
Handling System (NDHS) Manual (X650-64--189, paragraph 3.2) :
Typical Data Handling Sequence
A typical sequence of data handling operations at the NDHS
during and immediately following a normal 8-minute station
pass is described below.
Since a high degree of flexibility' is inherent in the NDHS
data processing system, the sequence of operations indicated
are representative rather than absolute.
After spacecraft acquisition, ground and spacecraft time correlation is performed by the Command Console Operator.
Then, the spacecraft is commanded to transmit "A" stored data,
a direct A VCS picture, and stored HRIR data. The following
PCM data sequence occurs:
"A" stored raw data are received and recorded at the
station over approximately a 4-minute interval; during
this interval the Computer Subsystem calibrates and records attitude data and selected parameters required for
meteorological data processing (metro parameters).
Following the 4-minute playback, the Computer Subsystem provides a synopsis listing of computer-recommended command instructions and mode status, based on
an analysis of "A" stored data.
During the time remaining to the end of the station pass,
selected. "A" real time parameters are received, calibrated,
COMPLETE
RECORDING SYSTEMS
291
and printed out. This provides data for a near real time
evaluation of the effect on the spacecraft of commands
transmitted during the current pass.
The "B" telemeter is considered a backup data source
and "B" real time data would be received when it is not
possible to receive" A" data.
Concurrent with PCM data sequence described above, the following
meteorological data sequence occurs:
The direct AVCS picture is received during the first 3 minutes of a station pass which occurs during earth day, thereby
providing picture data for assessment of A VCS performance
in real time.
Following the direct picture period, the spacecraft is commanded to transmit stored A VCS data, which are then received
and recorded over approximately a 4-minute interval. During
this period, time points are stripped off and stored in the Computer Subsystem approximately every 8 seconds. Meanwhile,
the computer uses part of each 8-second interval to calculate
time/position points.
The remaining free time of the Computer Subsystem is used
to smooth the attitude data and write a tape containing calibrated metro data and smoothed attitude information.
The smoothed attitude and metro data tape is rewound. At
the GILMOR NDHS the attitude and metro data tape is then
transmitted to the Goddard NDHS via the wideband data link.
The smoothed attitude data are also retained in the computer
core storage for calculation of time/attitude points at the end
of the A VCS reception period.
Continuing with a consideration of the meteorological data sequence,
reception of HRIR data also starts at the time of spacecraft acquisition
and continues for about 7 minutes. When reception :i3 complete, the
following events occur:
The A VCS-HRIR Mincom recorders A and B are rewound
and HRIR gridding begins using Mincom A for data playback.
At the G ILMOR NDHS the tp"idded HRIR data are transmitted to the Goddard NDHS VIa the wideband data link. The
transmission of the data is simultaneous with the gridding process and is completed approximately 17 minutes after the start
of spacecraft interrogatIon.
During HRIR gridding, Mincom B is played back for the
computation of A VCS horizontal and vertIcal sync Qffsets.
As soon as the HRIR gridding and offset synchronization
activities are completed, AVeS gridding is started. It requires approximately 32 minutes. Transmission of the gridded
triplets from the GILMOR NDHS to the Goddard NDHS is
automatic and simultaneous with gridding. At the GILMOR
NDHS, the transmission of the gridded A VCS data begins
immediately' after transmission of the PCM Engineering Units
Tape descrIbed below.
292
MAGNETIC TAPE RECORDING
The off-line processing sequence is less fixed and will depend upon
the requirements established by the NTCC for the particular station
pass. Typical off-line processing would be performed as follows:
Off-line processing of PCM "A" stored data can begin at. the
end of the acquisition period, starting with the reduction of the
entire "A" stored record to engineering units. Simultaneously,
a sync summary is printed out to prOVIde a complete hiStory of
interrupt words, mode changes and bad frames, sub frames or
words.
The Engineering Units tape produced at the GILMOR
NDHS is transmitted to the Goddard NDHS for further
analysis.
The spacecraft status (electrical energy balance and nitrogen
status) is completed and the results printed out. These data
are also punched on paper tape for transmission to the NTCC
when this processing is performed at the GILMOR NDHS.
Selected items are limit-checked (Limit Summary Module),
function listed (Data Listing Module) and subjected to the
averages and extremes routine (Average and Extreme Module).
These results are printed out and made available to the NTCC.
The seqhence of operations described for this typical case requires
a station pass of 8 or more minutes.
It should be noted, however, that a pass time of little more than 6
minutes would suffice for all but the 3 minutes of the HRIR data. A
pass time of little more than 4 minutes would allow the reception of
the "A" stored record and most of the stored A VCS data (assuming
no direct picture). Pass time in excess of 8 minutes, of course, would
allow reception of additional real time data.
A common variation of the above sequence will be the interrogation
of "A" real time data prior to the receipt of "A" stored data. The
alteration of the sequence in which PCM data are acquired from the
spacecraft poses no data handling problems since the telemetry computer program can automatically recognize the type of data ("A"
stored, "A" real time or "B" real time) being input and will automatically initiate the appropriate data processing subroutines.
CENTRAL DATA PROCESSING FACILITY
NASA Technical Note D-1320 describes in detail "A Central Facility for Recording and Processing Transient Type Data." This
facility was constructed at Lewis Research Center to accomplish the
objectives which are set forth in this edited quotation from the introduction to the technical note referred to:
In 1954 a central automatic data processing system was placed
in operation at the Lewis Research Center. This system met
COMPLETE
RECORDING
SYSTEMS
293
all design objectives and is still in operation; however, it is only
useful for processing steady-state-type data. Since that time,
advances in the state of the electronic art have made it possible
to build a data system capable of recording and automatically
processing transient-type data. Increasing quantities of such
data are recorded at thIS center as a result of experiments such
as vibration testing, heat-transfer experiments, and rocket testing. The data system described in this report came as a natural outgrowth of the need for an improved means of processing
transient-type data and the proven usefulness of a central automatic data processing system. This system makes possible the
analysis of data by mathematical techniques that would be too
time-oonsuming if done manually, and gives results of much
greater accuracy than could be attained by older methods of
processi~.
Recordmg is done by two types of equipment, which may be
classified as analog and digital. These two recording systems
are independent and complement each other. The analog system has the advantage of high-frequency response, while the
digital system gives greater accuracy but with lower frequency
response characteristics.
This facility was installed, of course, to fulfill the data processing
requirements of the Lewis Center as a research operation. The particular choice of equipment and data reduction techniques was related
directly to this function of the Lewis Center. However, almost every
function that is performed in this facility is directly related to a function which is now performed in data reduction from essentially every
NASA or military flight or research program. Certain aspects of the
techniques involved are specialized to the Lewis Center needs of transmitting data over long distances within the Center and concern the
problems of long telephone cables. Other parts of the operation are
essentially identical with those that would be performed for flight
data originating from a boost vehicle, a satellite, or some reentry test
vehicle.
In a typical flight program there is a certain amount of local use at
the data acquisition station of raw, unreduced data. This function is
very similar to the quick-look facility which is described in the Lewis
operation. Whereas at Lewis the data is recorded at the time of the
test and the tape is immediately available, for most flight operations
the data is recorded fairly remotely from the central management of
the flight. Either tapes must be transmitted to a data reduction facility, or the data itself may be transmitted by land lines. In some cases,
immediately after acquisition, crucial data is played back at reduced
speed to fit the transmission bandwidth of the land lines available and
thus transmitted to interested parties hundreds of miles away.
294
MAGNETIC TAPE RECORDING
The following description of the Lewis installation is derived directly from the material of the technical note, and is not intended to
be an exhaustive review since the best source of that is the technical
note itself.
Digital System
The digital recording system has 128 input channels. The
voltage resolution is 0.1 percent of full scale which results in an
overall system accuracy of 0.25 percent of full scale. It operates at a sampling rate of 4,000 samples per second. This
sampling rate is limited by the analog-to-digital converter available and may be distributed among channels as desired. For
example, if there are 12 channels of data being sampled, the
sampling rate for the individual channel would be 333 times
per second.
The operation of this digital system is carefully designed
around minimizing the amount of complex and expensive equipment needed. This system, therefore, has as few analog-to-digital converters as possible, in this case, only one. By multiplexing, it is possible to use this single converter with great efficiency. In more complex systems several converters may be
used with even more complex multiplexing systems.
The multiplexing performed for this data recording system
consists in switching in a systematic way between the selected
ones of the 128 input channels to connect these to from 1 to 8
output channels. For example, the multiplexing scheme might
be for channels 1 through 16 to be sampled in rotation and the
samples connected to output channell, at the same time input
channels 17 through 32 would be likewise treated and the output
sent to output channel 2, and so forth. In this way 128 input
signals are made available 8 at a time to the recording system.
In this particular unit, the switching is accomplished with
oil-damped sealed reed switches which are switched by magnets
which rotate past the array of switches arranged in drum form.
The eight output channels of this multiplexer are fed to
"analog storage units." These storage units, in effect, open a
switch for a short period of time and arrange for an internal
capacitor to store charge to produce a voltage equal to the instantaneous value of the input signal at the time of switch opening and then to close the switch and hold the charge on the
capacitor. This is in effect a "sample and hold" circuit. In
order, then, these eight samples are interrogated by the analogto-digital converter. The converter used in this particular system employs successive approximation to provide a complete
conversion in 22 microseconds with an ll-binary-bit output.
The unit is operated at two different speeds, either 1,000 or 4,000
conversions per second depending on the recording mode.
OOMPLETE REOORDING SYSTEMS
295
The output of the analog-to-digital converter is recorded on
tape and is sent at the same time to various "quick look" facilities. In the quick-look facility provision is made for examining
the data after it has passed through all stages of the reduction
process except recording and reproduction. The digital information is presented on numerical displays and is also reconverted to analog form for this purpose by plotting on an x-y
recorder or put into other visual presentation form (storage
oscilloscope, etc.) for examination by the operator.
The recording operation is performed on 8 parallel tracks in
a digital "tape handler." The data word consists of the 10 data
digits plus the polarity digit produced by the ll-bit converter
plus 7 other bits which identify the individual data word. In
other words, each successive data number as converted by the
analog-to-digital converter is identified serially in recording.
The 8th track serves to keep a complete running record of the
data on a larger scale. In addition to the individual data reading identification provided by the 7 word bits, the so-called
"ancillary coding" on track 8 further indexes the data as to
block number, test number and facility (of origin) number.
For the complete index identification, the ancillary code requires
some 96 frames.
The terms "word," "frame," and "block" will be identified
as they are used here to avoid confusion. A data "frame" is
the series of 8 bits or pulses which are laid across the 8 tracks in
the digital recorder and a word is made up of 3 such frames.
On tracks 1 through 7 there are word number identification bits
1 through 6 and a parity bit. The 8th track for this frame carries an ancillary coding bit. Frame 2 contains the 7th word
bit and the polarity bit and bits 1 through 4 of the binary number which constitutes the date, followed by an additional parity
bit and ancillary coding on track 8. In this ca...-.e bit 1 is the
most significant and bit 10 the least significant. of the data bits.
Data frame 3 contains data bits 5 through 10 and an additional
parity bit as well as the ancillary coding data. The parity bits
here are, as usual, placed in the 7th position to guarantee that
the 7 bits in a row always contain an odd number of ones. If
t.he data information has an even number of ones, t.he parity
bit is made unity and if the data bits have an odd number of
ones the parity bit is made zero. The ancillary coding actually
does not. fit frame by frame but. is transmitted at every other
frame and really does not enter into the frame proper.
A "block" is a group of 32, 64, or 128 data words. The blocks
of data are numbered in sequence, starting with 1 from t.he
beginning of the reel of magnetic tape; t.he block number essentially indicates reel data time and is useful for editing the data
later. A "reading" which is a special term used in this particu781H)28~20
296
MAGNETIC
TAPE
RECORDING
lar data reduction system, consists of any number of blocks of
data needed to record a data event from one facility. Thus a
reading is bounded at the beginning by the start of the recording from a facility and at the end by the stopping of the record
equipment after the data event.
The ancillary coding track contains the following information in order: (Every other frame is used, that is, there is no
recording in odd-numbered but only in even-numbered frames
and up to 96 frames are needed to contain the complete unit of
ancillary data.) First, a block start indication, the one piece of
data that is recorded in both even and odd frames (1 and 2).
It is followed by the block number in binary number form, the
reading number in the same form, an identification of the facility
number in the same form, and finally, an identification of the
computer program which is to be used in later reduction of the
data. The ancillary code concludes with a single index bit in
frame 94 (always a one) which notes the end of the code.
This kind of ancillary coding is universal in data acquisition
because it is always essential to identify each piece of information with its exact time of occurrence and the corresponding
physical event. Reference is made in describing the satellite
data acquisition techniques to such items as "'VVYV reference
time" and "binary-coded-decimal real time"; these data perform essentially the same task as does the ancillary coding
described here.
Data editing takes place, once the recordings have been made,
in order to identify important or erroneous parts of the data.
Various display systems are designed for displaying in numerical or graphical form the various pieces of data and their
location for an operator who scans through the recorded data.
The ancillary coding is essential to locating this data within
the mass recorded on a single tape. In the editing process,
data of interest is selected and all errors are identified eitJler by
the operator or automatically by the equipment. Such errors
might be incorrect frame identification, a loss of consecutive
order in blocks or the like. When the data is to be transferred
to a computer for reduction, the same error identification is
automatically performed, and information on the errors is
transmitted simultaneously to the eomputer along with the data
from the tape.
The individual interested in the results of this recording and
data reduction can, in the editing process, select the items of
particular interest to him and can decide exactly what is to be
done with the data and so indicate by establishing a specialized
computer program. ·When he has completed his selection, the
data reduction process is then automatic and the tape may
either be played over telephone lines to the computer or may be
physically transported to a digital tape playback unit at the
computer.
COMPLETE
RECORDING
SYSTEMS
297
Analog System
In the analog data reduction and acquisition process in this
facility the methods of data identificat.ion are quite similar to
those of the digital facility. A 14-track analog tape recorder
is used, 12 tracks being assIgned to parallel recording of 12 sets
of data and 2 tracks being used for identification and playback
system control. Every 50 milliseconds a marker signal is :placed
on one of the control tracks; a binary block number, identIfying
serially each individual 50-millisecond interYaI from the start
of the recording, is placed on the same track between these
markers (a "block" has now become a 50-millisecond interval).
FM recording with a bandwidth of 0 to 10 kcps is used for the
analog system. With this wide frequency response, analog data
recording is used for high-frequency analysis; it may be accompanied by simultaneous digital data-taking if several aspects
of a single experiment are to be analyzed. Because the analog
system is subject to drift of gain and zero, careful calibration
procedures are followed.
Seven "zones" occur in sequence on the tape after the start of
the record. Zone one carries the record number on one of the
control channels (this is a 4-digit binary-coded-decimal number, the first two digits of which identify the facility of origin
and the last two the reading number) ; this zone lasts for 2 seconds. Zone two is the zero-calibrate zone; for 3 seconds the
transducers from which data are to be recorded are set to zero
and the corresponding FM frequency recorded on the tape.
Zone three carries a 3-second full-scale calibrating signal corresponding to the zero-calibration of zone two. Zone four is
called the "pre-data" or "non-usable-data" zone and lasts from
150 milliseconds to 2 seconds, during which, automatically or
manually, the normal operation of the transducers is reestablished. After t.his pre-data zone has passed, zone five or the
"usable data" zone then occurs to the end of the test.
The end of t.he test data zone is established by the sending of
a "stop" command from the test facility. A l%-second zerocalibrate signal followed by another 1%-second full-scale-calibrate signal is automatically recorded on the tape. The zones
are identified by signals on the control channels as well as by
block and reading numbers.
The output of the analog tracks can be analyzed either by an
analog or a digital computer. For the analog computer, the
functions of compensating for drift of gain and zero during the
experiment as well as for wow and flutter may be required. The
same search functions as in the digital system are necessury for
editing and for informing the computers as to their data reduction duties. Facilities are provided for looking at the results
of the records either in graphical or numerical form, and automatic search is provided.
298
MAGNETIC
TAPE
RECORDING
The analog data may be played back at the speed at which it
was recorded or at one-half, one-quarter, one-sixteenth, or onethirty-second of normal recording speed. These slower playback speeds are required because the analog computer is seldom
able to handle the full bandwidth of the actual experimental
data. Slow speeds are also required because for precision data
analysis, the data may be digitized and analyzed by the digital
computer. The digitizing process is, of course, limited in speed
to the capability of the analog-to-digital converter.
This brief description of the particular facility installed at the
Lewis Research Center parallels closely what might be said for the
somewhat more elaborate data reduction facility at Langley Research Center and in many other government and industrial installat.ions. Likewise the data reduction processes involved in handling
data from satellit.e and space probe experiments is parallel to, although obviously not identical with, the techniques used at Lewis.
Multiplexing, for example, is performed onboard the vehicle for most
flight experiments. Elaborate identification is not possible on the
vehicle; identification is usually supplied in the ground data acquisition system. Timing data and signals derived on the ground for
synchronizing the later demultiplexing operation are usually provided
at the time of ground acquisition. Quick-look facilities are also provided at the data acquisition point to indicate whether, in fact, the
experiment from the flight vehicle is proceeding properly and to decide
if it is not, what if anything can be done about this.
For example, in the special-purpose telemetry on board the IMP
satellite, certain data about low-frequency radio transmission are
transmitted continuously to the ground. The experimenter in this
case happens to have access to a large (150-foot diameter) receiving
antenna and installation and can listen directly to his dat.a as it is
received from the satellite. Without requiring transmission of data
to him from an acquisition facility within the NASA complex, the
experimenter can determine that certain commands should be sent to
the satellite to modify its operation to extend the utility of the experiment. This is a rather special kind of "quick-look" ability!
Typically, the data output from almost all data acquisition plants
in the NASA complex except the self-sufficient ones such as those at
Lewis and Langley is eventually copied and transmitted to experimenters in various parts of the United States who will use the data
in different ways. The functions indicated from this brief description
of the Lewis Center are, however, essentially those which will be used
by the various experimenters. For example, vibration data from the
COMPLETE RECORDING
SYSTEMS
299
launch vehicle of the Agena is transmitted to Lockheed, the contractor
for the vehicle. (This is referred to above in connection with the
OGO-A satellite.) Lockheed engineers will look at the actual vibration data on oscilloscopes or on oscillograph charts and may then
perform either analog or digital spectrum analysis of this vibration.
Whatever the nature of the data from a satellite or space experiment, it receives the same general sort of handling for each user. The
Lewis data handling facility is therefore a representative "microcosm"
of the world of data reduction.
Page intentionally left blank
Page intentionally left blank
CHAPTER 15
Utilization Factors in Magnetic Recording
Ten chapters of this survey, following the introduction, were concerned with the nature and performance of elements of magnetic recorders. Chapter 13 discussed methods of testing and evaluating
recorders. Chapter 13 surveyed the' special requirements of miniature high-environment recorders and some of the ways in which these
requirements are met. In chapter 14 attention was shifted to the
use of the recorder in a complete system and to an almost philosophical
discussion of what the task of the recorder is within such a system.
What now are the utilization factors with which we should be
concerned~
The external performance specifications of the magnetic recorder
have been continuously improved since its first use for technical data
recording. The bandwidth of the recorder has been widened, the
signal-to-noise ratio has been improved and the flutter and distortion
have been reduced. At the same time, the effectiveness with which the
magnetic recording medium is utilized has increased. The longit.udinal density of recording, for example, has increased from 500
cycles per inch for the original hi-fidelity broadcast audio recorders
operating at 30 inches per second to 8 kilocycles per inch for home
audio use, and 12,500 cycles per inch routinely for data recorders----now 17,000 cycles per inch is in the offing. Some increase in lateral
packing density has also been made but not all in the same order of
magnitude. These individual increases have resulted in an overall
increase in the amount of data recorded per square inch of medium.
For this recording density to increase, both the medium and the
methods of using it had to be improved. Controlled very narrow
gaps, improved head finishes, and, most important, increases in uniformity and surface smoothness of magnetic tape, have all contributed
to the improved density.
It is tempting to try to establish a criterion for individual recorder
301
302
MAGNETIC TAPE RECORDING
performance on the basis of achieved recording density. With care,
this can be done, but the value of the criterion is limited. It has been
suggested that one might set a criterion for recorder packing densities
on the basis of the classic Shannon information theoretical measurement of channel capacities (Shannon [1948]). The Shannon channel
capacity statement is so beautifully simple that it is often misinterpreted; the misinterpretation is particularly easy in the case of
magnetic recording. Recording density will be briefly analyzed here
using a "conservative" approach to the Shannon criterion:
0= Wlog2
(S+N)
N
This classic Shannon formulas states only that, for a communication channel with a bandwidth in cycles per second of W, with an
average signal power of S, and a "white" noise power of N, the capacity in bits per second is W times the logarithm to the base 2 of
signal plus noise power over noise power (and this is the important
part) fo1' an optimum coding method. The Shannon channel capacity is the capacity for optimum coding but not necessary for any
less than optimum coding. The optimum method of coding is
definitely not implied or included in the formula.
If we take the full width of a strip of magnetic tape and consider
that this is a communication channel, the coding process for this
channel will include the decision as to how many tracks the tape shall
be divided into, what the spacing between tracks shall, be and the particular mode of recording to be used on the individual track. One
might try to evaluate the performance of a particular code choice by
comparing the total information per linear inch of tape with that for
optimum coding or for ultimate channel capacity. Unfortunately,
this cannot be done because, although the calculation can be performed
for the individual tracks, no such calculation can be performed for the
tape as a whole. There is no meaning to the term signal-to-noise
ratio or bandwidth for a one inch strip of tape until the decision has
been made, for example, to use a one-inch wide record/reproduce head
combination with it or to divide it into 20 separate tracks of a given
width. The Shannon criterion, therefore, cannot directly be used with
magnetic tape to compare any recording mode choice with the ultimate
capacity of the tape.
The Shannon criterion is useful, however, in another sense.
Slightly restated, the Shannon channel capacity formula gives the capacity in digital terms, i.e., in binary bits per second, of an analog
channel of a given bandwidth and signal-to-noise ration. The Shannon criterion, therefore, can be used, not as a statement of the ultimate
UTILIZATION
FACTORS
303
but as a statement of analog capacity in digital terms. Having stated
capacity in this way for an individual track as a specific channel, one
can then obtain overall performance figures for a given width of tape
containing several tracks. This overall performance figure is a measure of the effective storage density on the tape. This conversion to
digital form permits comparing analog and digital coding choices on
the basis of total information capacity.
An indirect method is also available for deriving a figure of merit
for the effectiveness with which a given coding choice, particularly
an analog one, is used in a given recorder. If, following Eldridge
(Eldridge [1963a]), one determines for a given tape and head combination the width of the track which will give unity signal-to-noise
ration, this track width can be used as a criterion of recorder performance. One can, for example, measure with care the signal-tonoise ratio and bandwidth capabilities of a practical narrow track on a
piece of tape. In performing this measurement, the tape and the rest
of the parameters of the record/reproduce unit can be specified. Then,
using the theoretical relationship that for a given longitudinal recording density, the power signal-to-noise ratio is proportional to
track width, the figure which Eldridge experimentally verified, one
can derive the track width for unity signal-to-noise ratio. This Wo as
Eldridge calls it, is the smaller the more effectively the tape is used.
If, for example Wo is smaller in one case than in another, the track
width necessary for a given higher t/w,n unity 8ignal-to-noi~e ratio
will also be less for the first than for the second case. It, therefore,
will be possible to pack more information on a given width of the tape
in the first case than in the second.
One could postulate a situation requiring no mechanical guardbands
between tracks and placing as many tracks of width W 0 across the
tape as possible. One could then derive the individual track channel
capacity for zero signal-to-noise ratio in this case and multiply by the
number of tracks, thereby obtaining a figure for total information recorded on the width of the tape. Although not a true statement of
Shannon's criterion for an ideal channel, this is a useful figure for
comparing recorder performance. This ultimate capacity is not a
practical number, since guardbands are necessary and a zero signal-tonoise ratio system is not useful for most applications. However, the
calculation provides a means for comparing recorder performance
where two recorders may be coded and operated in completely different
ways. It would otherwise be difficult to calculate, under these circumstances, whether one was really doing a better job than the other.
Such a determination is useful in deciding which way to direct the
development of a high performance recorder system when several
means of coding are available and a choice between them is difficult.
304
MAGNETIC TAPE RECORDING
The preceding discussion referred to the problem of specifying
analog system capability in digital terms. The discussion implied
that so specifying analog capability is useful because it gives a coherent
basis for a performance comparison between digital and analog systems for accomplishing the same thing. This comparison can be made,
however, only for the recorder itself and not for the overall system.
Given an analog signal which must be recorded, the user's criterion is
"How much tape must be used to record the desired signals with the
desired characteristics?" If the signal is to be recorded directly in
analog form, the calculation of tape utilization is very straightforward. To record the same signal in digital form, all the criteria
and design parameters involved in analog-to-digital conversion as well
as the accompanying limitations of sampled data systems must be
considered. The Nyquist criterion that slightly more than two samples
per cycle can transmit. completely the information contained in an
analog signal is sometimes used in calculating the conversion efficiency.
Users of sampled data systems, most of them admittedly working at
very slow data rates (in the order of a few cycles per second) , find this
too optimistic a calculation by far (Edwards [1964]). Ten, twenty,
and even thirty samples per cycle are often used when attempting to
collect data in this form (chapter 4). The primary difficulty arises
here because the bandwidth of the system is seldom under complete
control.
In other words, in much data collection, certain useful slowly varying data must be extracted from a very noisy band containing disturbances of frequencies much higher than the highest in the data of interest. If one samples at just slightly more than two samples per cycle
of the useful data, one also samples the unwanted data at the same time
and the phenomenon of "aliasing" or "frequency folding" results in
the desired data being distorted by the presence of undesired data of
higher frequency (Susskind [1957]). The data can be filtered before
sampling to eliminate these higher disturbing frequencies. In so
doing, one must be careful not to change the amplitude of the desired
data, i.e., the filter must have extremely uniform amplitude and phase
response in the useful pass band. A filter designed to have this conservative inband characteristic will not attenuate out-of-band disturbances very rapidly.
To obtain a particular accuracy of the sampled data in the most
conservative way, the filter must attenuate signals at any frequency
where aliasing or frequency folding can take place. The potentially
disturbing signal at this frequency must be attenuated to below the
accuracy percentage desired. In practice, this means that the two
samples per second criterion must be applied, not to the upper useful
UTILIZATION
FACTORS
305
frequency, but to the highest frequency· 01 possible disturbing signal
which UJ not attenuated below the desired accuracy percentage. A
simple example may illustrata this: To obtain data samples with an
accuracy of one percent, undesired signals must be attenuated by 100
to 1 or 40 db. With an 18 db per octave presampling filter of essentially flat response up to the upper useful frequency,40 db attenuation
would be reached at a frequency 6 to 8 times that of the upper useful
one. This means that the two samples per cycle criterion must be
based on, say, 8 times the upper useful frequency or 16 samples per
cycle of the useful frequency. For higher accuracies the situation is
correspondingly more severe.
The problem in the tape recorder may actually not be this bad
because external physical influences may reduce the amplitude of undesired signals of frequency higher than the upper useful band. How
much this amplitude is reduced, however, is not usually known, and it
is therefore almost impossible to make a general specification of what
the sampling characteristics should be for an analog-to-digital conversion system. Since these conditions cannot be specified, it is essentially impossible to compare an analog-to-digital conversion followed
by digital recording with direct analog recording as a method of coding
for magnetic recording.
Although, as Eldridge shows, narrow track digital recording is by
far the most effective way to use magnetic tape, this advantage may
have little significance in the practical case. There are, however, circumstances under which this process of conversion to digital and
recording on narrow tracks may be extremely useful. The use of
digital recording as the ultimate standard of efficiency is, however, an
oversimplification. The factors derived from the discussion above
should be brought into any analysis of tape recorder utilization.
Page intentionally left blank
Page intentionally left blank
APPENDIX
Review of General References
Since this survey was not designed to be a definitive text on magnetic
recording, it is appropriate to include here capsule reviews of the
major textbooks to which reference has been made. These capsule
reviews are intended to direct the inquiring worker toward the source
most useful to him for solution of particular problems. By characterizing the emphasis and point of view of each of the works as a. whole,
these reviews may help a newcomer to the field decide which ones
might well become part of his technical library. The reviews represent the personal opinions of the author.
The most recent work is "Magnetic Tape Recording" by H. G. M.
Spratt (Spratt [1964]) now in its second edition. Mr. Spratt is an
Englishman and almost all the illustrations and equipment references
in his book are to English or Continental equipment. The Spratt
book stresses both the characteristics and the manufacture of magnetic
tape and concentrates on magnetic recorders used for sound. The
emphasis on sound and audio phenomena is somewhat disconcerting,
but does not disguise the excellence of the fundamental approach to
magnetic recording problems. For example, Spratt paraphrases four
different explanations of the action of high-frequency bias in linearizing analog recording. Spratt has extracted from the applicable
current literature most of the recent descriptions and analyses of the
fundamental processes of magnetic recording.
The sections on tape manufacturing, tape materials, and tape testing are rel~tively complete and contain information not found elsewhere. The sections on the principles of sound reproduction and
electroacoustics fall beyond the scope of this survey as do the descriptions of audio tape recorders. Spratt does not refer dirootly to pulse
recording or to the use of recording in connection with computers, but
he does include excellent references to dropout problems.
The Spratt book, although specialized, can be recommended to a
newcomer on the basis that its author is very careful to include as
307
308
MAGNETIC
TAPE
RECORDING
many aspects of magnetic recording as possible and to take no controversial position on issues which have not been resolved. Although its
emphasis misses the area of interest in this survey, it is actually as
broad as any text in English which attempts to deal in detail with the
major parts of magnetic recording.
For a detailed discussion of the magnetics and the relationship
between head and tape in pulse recording, "Digital Magnetic Recording," a recent book by A. S. Hoagland of IBM, can be recommended
(Hoagland [1963]). This book summarizes the current understanding of the relationships most important in magnetic pulse recording
and is remarkably free of any commercial biases. It, however, does
not consider any of the mechanism of drum, disk or tape recording
and has little to say about dropouts. It is particularly useful in giving numerical approximations to field distributions around the magnetic heads in terms that can be used to analyze head-tape system
performance.
The title, "Physics of Magnetic Recording," of a new text by C. D.
Mee (Mee [1964]), also now of IBM, although accurate, is somewhat
misleading. Mee considers every aspect of the head, the tape and the
interaction between the two in considerable detail. There probably
are more accurate references here to the important work currently
going on in increasing the understanding of the fundamentals of magnetic recording than in any other text. The bibliography is particularly useful, being relatively brief but clearly critical, in that the author appears to have read and analyzed each reference and to have
decided on specific criteria to include it. I will confess to a prejudice
against the "bubble" mechanism which is promulgated by Mr. Mee as
an aid to the understanding of the action of ac bias, but other workers
in the field may not suffer from this prejudice. The bubble theory, it
should be pointed out, is not advanced as an explanation but simply
as a study or calculation aid by Mr. Mee.
This text is the second in a series of monographs on selected topics
in solid state physics edited by E. P. Wohlfarth. It is easily missed
in a technical library or book store because it seems to be classified
under Solid State Physics and to be alphabetized under W for Wohlfarth. This isolation is not appropriate to the really basic utility of
this text.
An earlier book which covers most of the areas of magnetic recording of interest to NASA is "Magnetic Tape Instrumentation" by
Gomer L. Davies (Davies [1961]). This text covers the state of the
entire field of magnetic tape instrumentation in approximately 1960.
It includes more discussion of magnetic recording mechanisms of the
instrumentation type than any other reference in English. However,
APPENDIX
309
because it covers a very wide range and is a relatively short book, the
coverage is not very deep for anyone subject. It is to be recommended
for a rather complete study of data irregularities resulting from flutter
and of criteria for flutter performance in instrumentation recorders.
"Magnetic Recording Techniques" by \V. Earl Stewart (Stewart
[1958]) appeared in 1958 and is another relatively short book on magnetic recording with an emphasis on audio recording. Three-eighths
of the book is devoted to appendices on magnetic recording standards
of audio, some definitions of magnetic quantities, a magnetic and
sound recording glossary, and the text of four papers written by
others. These appendices are of mixed utility at the present time
because the standards are obsolete as is some of the technical work
quoted in the light of current understanding of the magnetic recording process. Most of the material dates from 1953 but it does represent the best work that was available at that time.
In addition to these five English texts, each of which covers all of
magnetic recording to some extent, there are a number of more specific
works directed to the user of specific devices such as, for example, the
video tape recorder for broadcast use. I do not believe these texts are
useful for an introduction to any subject except that for which each
is specifically written. The five general works are therefore recommended to the newcomer.
There is a sixth work on magnetic recording on which some comment should be made. It is "Technik Der Magnetspeicher" edited by
Fritz Winckel (Winckel [1960]), a collection of the work of 14 German engineers and scientists in a rather massive (612 pages) German
review of the field of magnetic recording as of 1960. I am familiar
in detail only with certain portions of this book. It appears to be
extremely complete in its coverage of some of the important parts
of magnetic recording technology but it is also padded with rather
specific material on equipment and on special problem solution, most
of which is now obsolete. I also believe, on the basis of rather solid
evidence, that some of the theoretical analyses contained in this text
are simply incorrect. This is necessarily a controversial subject, but
the book can be used with profit by an experienced worker in magnetic
recording since many of the analyses, controver~ial or not, are extremely complete. The newcomer to the field, even if well equipped
with a reading knowledge of German, should approach this book with
care.
Page intentionally left blank
Page intentionally left blank
References
Ampex Corporation, 1964, "Ampex FR-900 Instrumentation Recorder," (brochure) 2109S, August 1.
C. E. Anderson, 1957, "The ~lodulation System of the Ampex Video Tape Recorder," Jour. Soc. Mot. Pic. & TV Engr., vol. 66, pp.I82-184; April.
E. Bahm, 1963, "L'ltra-Reliable ~Iotors for Spacecraft Tape Recorders," J.P.L.
Space Programs Summary. Xo. 37-23. Yol, IV. pp. 2~270; August I-September 30.
-ibid, "Development of a Hysteresis-Induction Motor," J.P.L. ·Space Programs
Summary, No. 37-24, vol. IV, pp. 229-233; October I-November 30.
E. Bahm, 1964, "Developments of a Hysteresis Motor for Spacecraft Tape Recorder Applications," J.P.L. Space Programs Summary, No. 37-26, vol. IV,
pp.13&-138; February I-March 31.
G. P. Bakos, 1964, "Design Aspects of Ferrite Magnetic Heads," Proe. Int.
Conf. on Mag. Rec. (MAGREC), London, pp. 8&-87; July 6--10.
M. F. Barkouki and I. Stein, 1962, ''Theoretical and Experimental Evaluation
of RZ and ~RZ Recording Characteristics," 1962 West. Elect. Show and
Con. (WESOO~), Paper No. 8.1; August.
S. J. Begun, 1937, "Recent Developments in Magnetic Sound Recording," Jour.
Soc. Mot. Pic. Engr., vol. 28, pp. 464-472; May.
E. E. Bisson, 1964. "Friction and Bearing Problems in the Vacuum and Radiation Environments of Space," "AdYllDCed Bearing Technology," NASA SP-38,
Washington.
E. E. Bisson and W. J. Anderson, ed., 1964, "Advanced Bearing Technology,"
NASA SP-38, Washington.
B. M. Brenner and G. ~Ieyer, 1964, "Time-Base Errors in Analogue Recording,"
Proc. Int. Conf. on Mag. Rec. (MAGREC), London, pp. 26--29; July 6--10.
A. D. Burt, S. P. Clurman and T. T. Wu, 1963. "Design of Satellite Tape Recorders after Tiros I," .Jour Soc. of Mot. Pic. & TV Engr., ,-o\. 72, pp. 788791; October.
W. G. Clement, 1964, "A Xew Head for Recording Twenty-Four Tracks on
Quarter-Inch Magnetic. Tape," J.P.L. Space Programs Summary, No. 37-27,
vol. IV, pages unknown; April I-May 31.
P. T. Cole, 1963, "A Recording and Reproduction Technique for Pulse Code
Modulation," Internal Memo, NASA Goddard Space l<'light Center.
P. T. Cole, H. J. Peake and C. F. Uice, 1962, "Application of the Modularization
Concept to Satellite Tape Recorders," NASA Technical Note TN D-1451;
November .
.J. H. Conn, 1964, "Vibration Isolation of Satellite Tape Hecorders," NASA
Technical Memorandum TM X-942; February.
L. G. Cox, 1962, "A Wide-Range Wow and Flutter Indicator," Jour. -Soc. Mot.
Pic. & TV Engr., vol. 71, pp. 9-12; January.
E. D. Daniel, 1955, "Flux-Sensitive Reproducing Head for Magnetic Recording
Systems," Proc. Instn. Elec. Eng. (Brit.), vol. 102, part B, 442--446; July.
311
788-028
~65---21
312
MAGNETIC TAPE
RECORDING
E. D. Daniel, 1963, "A Preliminary Analysis at Surface-Induced Tape Noise,"
Proc. at the 1963 Intermag. Conf., Sect. 7: 7-1-1-7-1-5; April.
E. D. Daniel and P. E. Axon, 1963, "The Reproduction of Signals Recorded on
Magnetic Tape," Proc. Instn. Elec. Eng. (Brit.), vol. 100, part ITI, pp. 157167; May.
E. D. Daniel and E. P. Wohlfarth, 1962, "Fine Particle Magnetic Recording
Media," Jour. Phys. Soc. Japan, vol. 17, p. 670.
G. L. Davies, 1954, "Magnetic Recorders for Data Recording under Adverse Environments," IRE Trans. on Audio, vol. AU-2, pp. 133-137; SeptemberOctober.
G. L. Davies, 1961, "Magnetic Tape Instrumentation," McGraw Hill Book Co.,
Inc., New York, N.Y.
J. A. Develet, Jr., 1964, "Fundamental Accuracy Limitations for Pilot-Tone
Time-Base Correction," IEEE Trans. on Audio, vol. AU-12, pp. 53-55; MayJune.
S. Duinker, 1960, "Durable High-Resolution Ferrite Transducer Heads Employing Bonding Glass Spacers," Philips Res. Rep., vol. 15, pp. 342-367; Augnst.
R. A. Edwards, 1964, "Some Systems Engineering Aspects of Data Acquisition,"
Data Syst. Des., vol. 1, pp. 20--29; February.
D. F. Eldridge, 1960, "Magnetic Recording and Reproduction of Pulses," IRE
Trans. on Audio, vol. AU-8, pp. 45-47; March-April.
D. F. Eldridge, 1961, "Quantitative Determination of the Interaction Fields in
Aggregates of Single-Domain Particles," Jour. App. Phys. suppl. to vol. 32, no.
3, pp. 247S-249S; March. (Proceedings of the Sixth Conference on Magnetism
and Magnetic Materials.)
D. F. Eldridge, 1963, "DC and Modulation Noise in Magnetic Tape," Proc. of the
1963 Intermag. Cout., Sect. 4: 4-4-1-4-4-7 ;. April.
D. F. Eldridge, 1963a, "A Special Application of Information Theory to Recording Systems," IEEE Trans. on Audio, vol. AU-H, pp. 3-6; January-February.
D. F. Eldridge and E. D. Daniel, 1962, "New Approaches to AC-Biased Magnetic
Recording," IRE Trans. on Audio, vol. AU-10, pp. 72-78; May-June.
R. C. Falwell, K. W. Stark and A. F. White, 1963, "A Precision Endless-Loop
Recorder for Space Applications," NASA Technical Note TN D-1542.
G. J. Fan, 1961, "A -Study of the Playback Process of A Magnetic Ring Head,"
IBM Jour. Res. & Dev., vol. 5, pp. 321-325; October.
A. Gabor, 1960, "Digital Magnetic Recording With High Density Using Double
Transition Method," IRE 1960 Int. Conv. Rec., pt. 9, pp.179-185.
General Kinetics, Inc., 1963, "A Theoretical and Practical Evaluation of the
Dynamics of an Endless-Loop Tape cartridge," Final Report Contract NAS
5-2435.
C. E. Gilchriest, 1957, "The Application of Phase-Locked-Loop-Discriminators for
Threshold Improvement and Error Reduction in FM/FM T{'l{'metry," Proc.
1957 Nat. Telem. Cont.; May 27-29.
C. P. Ginsburg, 1957, "Comprehensive Description of the Ampex Video Tape
Recorder," Jour. Soc. of Mot. Pic. & TV Engr., vol. 66, pp. 177-182; April.
P. C. Goldmark, C. D. Mee, J. D. Goodell and W. P. Guckenburg, 1960 "A 1 % ips
Magnetic Recording System for Sterophonic Music," IRE Trans. on Audio, vol.
AU-8, pp. 161-167; ,September-October.
C. W. Hansell, 1945, "Report on the Magnetophone," Office of Publication Board
Report, PB 1346.
D. G. C. Hare and W. D. Fling, 1950, "Pic.ture-Synchronous Magnetic Tape
Recording," Jour. Soc. Mot. Pic. & TV Engr., vol. 154, pp. 554--566; May.
REFERENCES
313
A. S. Hoagland, 1963, "Digital Recording Techniques," John Wiley & Sons, Inc.,
New York, N.Y.
IRIG, 1960, "Telemetry Standards," IRIG Document No. 106-60, Telemetry Working Group, Inter Range Instrumentation Group, Secretariat, White Sands, New
Mexico; December.
F. Jorgensen, 19(11, "Phase Equalization Is Important," Elec. Ind_, vol. 20, pp.
98-101; October.
F. Jorgensen and I. Moskovitz, 1962, "Reduction ot Tape Skew in Magnetic
Instrumentation Recorders," IRE 1962 Int. Conv. Rec., pt. 9, pp. 157-160.
A. S. Katz, 1964, "Space-borne recorder triples packing density," Electronics, vol.
37, pp. 84-88; August 24.
E. Kietz, 1963, "Transient-Free and Time-Stable Signal Reproduction From
Rotating Head Recorders," 1963 Nat. Space Elec. Symp., Paper 4.3.
E.S. Kinney, 1953, "Magnetic Tape Drive Designed for Minimum Speed Variation," Mach. Des., vol. 25, pp. 219-221 ; October.
R. E. Klokow and C. M. Kortman, 1960, "Predetection Storage of Telemetry
Data Using Wideband Magnetic Tape Recorders," Proc.l960 Nat. Telem. Conf.,
pp. 501-520.
J. H. Licht and A. White, 1961, "Polyester Film Belts," NASA Technical Note
TN D-008, May.
H. Lindsay and M. Stolaroff, 1948, "Magnetic Tape Recorder of Broadcast Quality," Audio Eng., vol. 32, pp. 13-16; October.
L. D. Lipschultz, 1964, "Dynamic Measurement of Small Separations by a Light
Interference Methoo," Proc. Int. Cont. on Mag. Rec. (MAGREC), W. 87-90;
July 6-10.
Lockheed, 1962, "A Spacecraft Recorder Modularization 'Study," Item A under
Contract NAS 5--1853 between Lockheed Electronics Co. and Goddard Space
Flight Center.
D. W. Martin, 1963, "A New FM Multiplex System for Precision Data Recording,"
1963 West. Elect. Show & Conv. (WESOON), Part 6, Paper 7.2.
J. G. McKnight, 1959, "Signal-to-Noise Problems and A New Equalization for
Magnetic Recording of Music," Jour. Audio Eng. Soc., vol. 7, pp. 5--12;
January.
J. G. McKnight, 1960, "The Frequency Response of Magnetic Recorders for
Audio," Jour. Audio Eng. Soc., vol. 8, pp. 46-53; July.
J. G. McKnight, 1961, "The Effect ot Bias Amplitude on Output at Very Short
Wavelengths," Jour. Audio Eng. Soc., vol. 9, pp. 98-102; April.
J. G. McKnight, 1962, "Wow and Flutter/Time Displacement Error," (Letter to
the Editor) Jour. Soc. Mot. Pic. & TV Engr., vol. 71, pp. 428; June.
D. D. McRae and H. Scharla-Nielsen, 1958, "FM/FM Demodulation," Proc. 1958
Nat. Telem. Conf., pp. 273-277.
C. D. Mee, 1964, "The Physics of Magnetic Recording," North-Holland Publishing
Co., Amsterdam.
W. C. Miller, 1947, "Magnetic Recording for Motion Picture StUdios," Jour. Soc.
Mot. Pic. Engr., vol. 48, pp. 57-62; January.
R. V. Monopoli, 1962, "An Inertia Compensated Magnetic Tape Recorder," Data
Syst. Engrg., vol. 16, ,pp. 47-48; January.
B. M. Oliver, J. R. Pierce and C. E. Shannon, 1948, "The Philosophy ot PCM,"
Proc. IRE, vol. 36, pp.1324-1331; November.
O. J. Ott, 1962, "Factors Affecting the Design and Performance of Predetectioll
Recording Systems," Proc. 1962 Nat. Telem. Conf., Paper 4-4.
314
MAGNETIC TAPE RECORDING
G. Parkinson, 1965, "Whittaker Magnetic Tape Fabricated to Stand 600° F,"
Elec. News, vol. 10, pp. 37; January 18.
C. B. Pear, Jr., 1961, "Flutter in Magnetic Recording of Data," IRE Trans. on
Audio, vol. AU-9, pp. 159-166 ; September-October.
H. P. Peloschek and M. H. M. Vrolijks, 1964, "Dense Ferrites and the Technique
of Glass Bonding for Magnetic Transducer Heads," Proc. Int. Conf. on Mag.
Rec. (MAGREC), pp. 82-84; July ~10.
L. R. Peshel, 1957, "The Application of Wow and Flutter Compensation Techniques to FM Magnetic Recording Systems," IRE 1957 Nat. Conv. Rec., vol. 5,
pt. 7, pp. 95-110; March.
R. L. Price, 1963, "Magnetic Feedback Modulator Improves Accuracy in FM
Recording," 1963 West. Elec. Show and Conv. (WESCON), part 6, August
20-23.
R. H. Ranger, 1947, "Design of Magnetic Tape Recorders," Tele-Tech, vol. 6, pp.
5~57, 99-100; August.
V. A. Ratner, 1965, "Wideband tape-recorder users claim obsolete standards
impede technology," Electronics, vol. 38, no. 2, pp. 90-94; January 25.
Raymond, 1962, "Final Engineering Report No. 607, 10' Bit Engineering
Recorder/Reproducer," under Contract No. 950105 between Raymond
Engineering Laboratory, Inc. and Jet Propulsion Laboratory, pp. 10-11;
August 24.
Raymond, 1964, "Progress Report No.5, Spacecraft Magnetic Tape Recorder
Heat Sterilization Study," under Contract No. 950617, between Raymond Engineering Laboratory, Inc. and Jet Propulsion Laboratory, pp. 8-9; January
23.
A. H. Reeves, 1942, U.S. Patent 2,272,070, assigned to International Standard
Electric Company; February 3.
L. Riley, 1962, "Predetection Data Collection System," Proc. 1962 Nat. Telem.
Conf., Paper 9-1
R. W. Rochelle, 1963, "Pulse Frequency Modulation Telemetry," Proc. 1963 Int.
Telem. Conf., vol. 1, pp. 438-445.
O. H. Schade, 1948, "Electro-Optical Characteristics of Television Systems, Part
III-Electro-Optical Characteristics of Camera Systems," RCA Rev., vol. 9, pp.
490-530; September.
K. W. Schoebel, 1957, "The Design of Instrumentation Magnetic Tape Transport
Mechanisms," IRE 1957 Nat. Conv. Rec., vol. 5, pt. 7, pp. 111-123.
G. H. Schulze, 1962, "Applications of a Light Mass Capstan Tape Recorder,"
Proc. 1962 Int. Telem. Conf., Paper 9-5.
G. Schwantke, 1961, "Magnetic Tape Recording Process in Terms of the Preisach Representation," Jour. Aud. Eng. Soc., vol. 9, pp. 37-47; January.
'V. T. Selsted, 1950, "Synchronous Recording on 14-inch Magnetic Tape," Jour.
Soc. Mot. Pic. & TV Engr., vol 55, pp. 279-284; September.
'V. T.Selsted, 1965, "A New Instrumentation-Class Tape Recorder of Simplified
Design," Hewlett-Packard Jour., vol. 16, no. 5, pp. 1-7; January.
W. T. Selsted and J. A. Dinsmore, 1961, "A Mechanical Braking System With
Feedback and Servo Characteristics," Prod. Eng., vol. 32, pp. 41; July 3.
W. T. Selstedand R. H. Snyder, 1954, "Magnetic Recording-A Report on the
State of the Art," IRE Trans. on Audio, vol. AU-2, pp. 137-144, SeptemberOctober.
C. E. Shannon, 1948, "A Mathematical Theory of Communication," Bell Syst.
Tech. Jour., vol. 27, pp. 379-423; July, pp. 623-656; October.
REFERENCES
315
L. F. Shew, 1962, "High-Density Magnetic Head Design for Noncontact Recording," IRE 1962 Int. Conv. Rec., pt. 4, pp. 53-62.
E. P. Skov, 1964, "Noise Limitations in Tape Reproducers." Jour. Audio Eng. Soc.•
vol. 12. pp. 280-293; OctOber.
H. G. M. Spratt, 1964. "Magnetic Tape Recording," Temple Press Books. Ltd.,
London. England.
K. 'V. Stark, 1963, "Positive Drive, Non-Drag Overriding Clutch," Patent disclosure, dated May 24, 1963.
K. W. Stark, 1964, "Development of a 1,200 Foot Endless-Loop Tape Transport
for Satellite Applications." NASA Technical Note ~ D-2316.
I. Stein. 1961a. "Pulse Resolution from Magnetic and Hall Reproduce Heads,"
1961 West. Elec. Show & Conv. (WESCON). Paper 13-3.
I. Stein. 1961b, "Analysis of the Recording of Sine Waves," IRE Trans. on Audio.
vol. AU-9. pp. 146-155; September-October.
W. E. Stewart. 1958. "Magnetic Recording Techniques." McGraw-Hill Book
Co., Inc., New York.
'V. F. Storer. 1963. "An Intermittent McXion Digital Tape Recorder-Reproducer,"
JPL Space Programs Summary No. 37-24. vol. IV, pp. 222-229; October INovember 30.
P. A, Studer, 1964, "Development of a Brushless DC Motor for Satellite Application." NASA Technical Note TN D--2108; February.
A. K. Susskind, ed., 1957, "Notes on Analog-Digital Conversion Techniques," The
Technology Press of Massachusetts Institute of Technology and John Wiley &
Sons. Inc .• New York. N.Y.
J. F. Sweeney, 191)2. "A Method for Measuring the Changes Introduced In Recorded Time Intervals By A Recorder-Reproducer." Trans. IRE. PGA-7, pp.
24-29; May.
M. R. Townsend, P. Feinberg and J. G. Lesko, Jr., 1963, "A Medium Data-Rate
Digital Telemetry System," Document X-65()....63....174. Goddard Space Flight
Center; September.
P. W. Uber, 1964. "Bearing Load Life Calculations." Talk given to Intercenter
Conference of Tape Recorder Working Groups. Goddard Space Flight Center;
January 29-31.
R. A. von Behren, 1963, "New Oxide Reduces Tape Noise," Electr. Prod. vol. 5. no.
11, pp. 34-35 ; April.
R. L. Wallace, Jr., 1951. "The Reproduction of Magnetically Recorded Signals."
Bell Syst. Tech Jour .• vol. 30. pp. 1145--1173; October (Part II).
M. B. Weinreb, 1961, "Resu.lts of Tiros II Ball Bearing Operation in Space,"
internal note, Meteorology Branch. Goddard Space Flight Center, March.
N. Weintraub, A. W. D'Amanda and R. B. Resek, 1964. "An Advanced Tape Recorder for Spacecraft Applications." paper submitted to Amer. Inst. of Aero.
& Astro (AIAA) Jour. during 1964, publication date unknown.
W. K. Westmijze. 1953, "Studies on Magnetic Recording," Phil. Res. Rept.. vol.
8, pp. 148-157, 161-183,245--269,343-366.
F. Winckel, ed.. 1960, "Technik der Magnet speicher. " Springer-Yerlag. Berlin/
GottingenjHeidelberg.
'V. Wolf. 1960. "An Investigation of Speed Variatiolls ill a ~Iagnetic Tape Recorder With the Aid of Electro-Mechanical Analogies." Jour. Aud. Eng. Soc.,
vol. 10. pp. 119-129; April.
Page intentionally left blank
Page intentionally left blank
Index
Acceleration, reel, 114
AGe level recording, 277
Amplification effect, vibration, 201
Amplifier, reproduce input, 167
Analysis, reliability; 181
Anhysteretic magnetization, 63
Assignments, track, 277
AVCS (Advanced Vidicon Camera
System)-Nimbus,287
AVCS recorder, 235
Bandwidth, fiutter, 208, 209
in phase shift coding, 55
Bearings, duplex pair, 265, 266
effect of shock on, 195, 199
torque variation in, 191
Belt, butt joint in, 91
capstan drive, 90, 91
pla'stic, 91, 199, 225, 258, 259
plastic, fabricating, 200
plastic, failure problems, 261
Bias, AC, 5
AC, effect on overload in record
amplifier, 171
Bias intensity, effect of, 68
Biased recording, AC-, 66
AC-, understanding, 93
DC-, 59, 60
Bit-parallel PCM, 51
Bit-serial PCM, 51
Blackout, fiame, recorders for, 253
Block,295,297
Blocking (of tape), 158
Brake, string, servo hold,back, 114
Brake holdback, 105
servo, 83, 112
Braking, dynamic, 82
Brakes, constant-torque, 198
Capacity, channel, 302
Capstan, 80
differential, 78, 79, 98, 221, 230, 240
single two-diameter, 98, 99, 101
dual, 98, 99
pusher, 92
Capstan drive, 80, 81,90
Capstan surface finish, 93, 94
Carrier erase, 55, 217, 220
Cartridge, reel, 105
Choice, recording method, 41
Chromium dioxide, 157
Class I modular fiight recorder, 184186
Classification, fiight recorder, 184-186
Cleaning, tape, 119
Closed-loop configuration, 78, 79
Olutch,overrunning,l97
in two-speed operation, 247
non-drag, 197
spring, 196, 197
Coating, gravure, 160
knife, 160
Coaxial reels, 104
Cobelt drive, 257
Coding, ancillary, 295,296
optimum, 302
Cogging,81,90,187
Commutation, 188
Compatability, tape, 165
tape Width, 286
Compensation, fiutter, 8, 221, 237
electronics of, 73
limits of, 28, 137
picture element displacement, 288,
289
for on-board recorder limitations,
286,287
Compliance, vacuum, 84, 119
Conversion, analog-to-digital, 294
CondUctivity, surface (of tape), 165
ConfigUrations, recorder/reproducer,
78
Constant-current recording, 172
Constant-flux recording, 58, 172
Construction, modular (mechanical) ,
244
typical electronic unit, 176, 177
ContinUous-loop analysis, 121, 122
Coupling, tape-capstan, 100
317
318
MAGNETIC
TAPE RECORDING
Creep,tape,atcapstan,97
Crosstalk in heads,146
d~/dt reproduce heads, 142
daDlage, tape,l14
daDlping, Dlotor, 90
vibration, 201, 202
Data reduction plan, 279
DC-biased recording, 60
DC response in recording, 8
Delay, bin, recorder/reproducer, 123
precision for correlation, 121, 123
DeDlagnetization loss, 62
Density, inforDlation-packing on tape,
301
lateralrecording,l50
Detection, reel diaDleter, 111
Dill'erential, plastic belt, for two-speed
operation, 261
Differential for two-speed drive,
231,232
Differentiated output (of heads), 59,
60
Dill'erentiation in reproduce heads,
142
Digitizing of analog data, 138, 298
Diphase coding, 55
Diphase recording, 54
DiscriDlinator, FM recording, 38
phase-locked-loop,39,137
pulse counter, 38, 137
Display noise data, 206
Dispersion of tape coating, 21
Distortion, FM recording, 40
Disturbance, tiDle, in transverse-scan
recorder, 125
Drive, Bight recorder, two-speed, oneDlotor, 227, 248, 268
two-speed, one Dlotor versus two,
266-269
two-speed, two Dlotor, 231, 232, 266,
267,269
tape, by capstan, 92, 100
by pressure roller, 92, 100
longitudinal for transverse-scan recorder, 124
Dropout depth, 212
Dropouts, 160
dependence on recording density,
212, 213
DUDlping, data, with identical record
and reproduce speed, 289
DynaDlic range, 169
Dynamics of endless-loop tape pack,
242
EGO (OGO-A), 237
Editing, data, 296
Efficiency, motor, 188, 191, 265, 266
ElastoDlers, probleDls with, 92, 93, 131,
226
Electronics, Bight, 177, 218, 222
FM record, 37
ground-based, 176, 177
Equalization, ell'ect on distortion, 203,
2M
effect on noise, 60, 205, 206
effect on signal-to-noise ratio, 170
eleD1entary recorder, 61
phase, 172
pulse, 175
Equalization of phase-shift-Dlodulation pulse signals, 175
Equalizer, reproduce, 60, 61
Equivalent Dlechanical circuit of openloop recorder, SO, 89
Era-se, perDlanent magnet, 153, 237
Errors, channel-to-channel time, 138
head..switching, 174
tiDle displaceDlent, 127-129
correction with variable delay
line, transverse-scan recorder,
139
Unlit to correction of, 28
Expansion, tiDle-base, 263
"55 d/~" loss, 73
Failure, bearing, 192, 193
Feedback, tachoDleter, 246
Fields, recording, 69, 70
Filters, comb, 278
high-D18ss Dlechanical, 132, 133, 13r.
Finish, head surface, 148
tape surface, 161
FlaDle-blackout recorders, 18, 253---255
]<'lexibility, recorder, 15
]<'lutter, 115
broadband, 132
cUDlulative, 208
INDEX
Flutter, 6-43------<:Jontinued
data, computer removal of, 137, 138
signal spectrum broadening by, 7
}<'lutter analysis, 132
Flutter and time dLsplacement error,
relation between, 14, 129
}<'lutter compensation, 37, 39, 287
}<'lutter effects, 7
Flutter in data, 137, 138
}<'lutter meter, 207
}<'M recording, 8, 36
wideband, 37
}<'ll recording standards, 35, 36
}<'lux, constant-, recording, 172
Flux distribution in tape, pulse, 74, 75
}<'lywheel, viscous-damped, 133
Force, holdback, in endleEl8-loop recorders, 240
takeup, in endless-loop recorders,
240
Forces, tape motion disturbing, broadband, 94
low frequency, 94
}<'ormat, telemetry, 276
}<'rame,295
Frequency, supply for AC motors,
265
}<'requency limitations, recording, 62
Frequency response, 206
Friction, interlayer in endleEl8-l00p recorders, 239,242,243
}<'riction between capstan and tape,
93,94
}<'ringing effect, 141
Gap, record, 68
Gap depth, 65, 67
etIect of head wear on, 150
Gap length, reproduce, 65
Gap loss, reproduce, 66
Gap spacers, 144
Gemini, 221, 234
Geometry, head-tape, transverse-scan
recorder, 125
Gridding (for Nimbus program),
291
Guides, tape, 198
air-lubricated, 118
319
Guides, tape, 11-28--0mtinued
coupling of moving to tape, 198
crowned-roUer, 235
fixed,118
moving, 118
non-twist, 199
Guiding, tape, 104,118
Guiding, trough, 261
H-film, 200, 158
HRIR (High Resolution Infrared)
monitoring system, 289
HRIR recorder, 235
Hall effect, 143
Head, C-core, 142, 144, 14;)
ferrite, 144
flux-sensitive reproduce, 142, 143
Hall-etIect, 143,147, 148
metal-taced, 148, 149
modulator, 142,143
multitrack, 145, 149, 151
narrow-track, 150, 151
plug-in, 86
retractable, 86
video recorder, 145, 146
Head assembly, transverse recorder,
125
Head bUmps, 67
Holdback, brake, 217
servo, 111
string, 113
torque-motor, 107
Hold-down, reel, 116,117
IMP (Interplanetary Monitoring Platform),273
Impedance, reproduce amplifier input,
168
Induction m Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.6 Linearized : No Create Date : 2018:05:16 13:23:51-08:00 Modify Date : 2018:05:16 14:12:01-07:00 Has XFA : No XMP Toolkit : Adobe XMP Core 4.2.1-c041 52.342996, 2008/05/07-21:37:19 Metadata Date : 2018:05:16 14:12:01-07:00 Producer : Adobe Acrobat 9.0 Paper Capture Plug-in Format : application/pdf Document ID : uuid:ab70afe6-8184-c240-84a0-ee8b6e45c6b3 Instance ID : uuid:9ffad986-b9f1-a741-8e67-f3c90f94cdc5 Page Layout : SinglePage Page Mode : UseNone Page Count : 337EXIF Metadata provided by EXIF.tools