A. D. BURT, S. P. CLURMAN, and T. T. WU
Astro-Electronics Division, DEP, Princeton, N. J.

Discussed herein are some of the improvements in satellite tape-recorder design since the first "TIROS". Elements of the new designs include system requirements, tape transport, sensing heads, motor-drive and speed-changing techniques, and means of compensation for angular momentum. Examples include design applications in the AVCS (Advanced Vidicon Camera System) recorder and the HRIR (High-Resolution Infrared) recorder of the NIMBUS meteorological satellite, as well as in the recorders for other spacecraft.
.

A. D. BURT attended Drexel Institute Evening School and University of Pennsylvania during 1925-1928 and 1930-1932. He was a student engineer at the General Electric Co., from 1928 to 1930, and transferred to RCA Victor Company in 1930, as an engineer, where he did development work on vibrators, magnetic circuits, phonograph pickup, turntable drive systems, and motors. During this period, he proposed the present geometric series of values used for resistors. In 1946, he became Manager of the Record Changer and Tape Systems engineering section of the RCA Victor Home Instrument Division. At AED (Astro-Electronics Division), he is engaged in the evaluation and solution of electromechanical problems related to tape recorders for satellite use. He has seven issued patents and has written and presented several technical papers related to phonographs and tape recorders. Most recently he co-authored a technical article, "The History of the Phonograph," at the request of the "Encyclopedia Britannica."

STANLEY P. CLURMAN received a BSME degree from the City College of New York in 1941, and an MS degree from the Stevens Institute of Technology in 1945. In 1941, he joined the Curtiss Wright Propeller Division, where he advanced to Senior Stress Analyst; from 1946 to 1948, he was a member of the staff of MIT and worked on a Navy materials study program. Later, he became project engineer with Sperry Gyroscope Company. He served as Chief Mechanical Engineer of the Hogan Laboratories, New York City, from 1951 to 1955. Mr. Clurman joined AED (Astro-Electronics Division) in 1958, and is now an Engineering Leader in charge of advanced recorder design. He was part of the original team that developed the first TIROS recorder. During 1960-62, he was responsible for all satellite recorder work at AED, including those for TIROS, NIMBUS, OGO, and several classified programs. He has contributed to work on dielectric tape storage devices. He did early development work on continuous-motion film transports, magnetic couplings, and vacuum-mechanical material considerations. He is now responsible for work on the tape transport for the NASA pre-prototype dielectric camera, of the panoramic scan, continuous-motion filmtype. Mr. Clurman is a senior member of AIAA, SMPTE and IEEE and a member of Sigma Xi. He has published three technical papers and has ten patents issued to him.

T. T. WU received his BSEE from Syracuse University in 1952 and his MSEE degree from Columbia University in 1957. Before joining RCA, Mr. Wu worked on the design of material handling equipment and automatic controls. Mr. Wu joined the Astro-Electronics Division in 1958, where he specialized in logic-system, tape-recorder, and circuit design. He designed and developed circuits for the NIMBUS AVCS and HRIR tape recorders, the digital work generator for Project DAMP. He also worked on the analyst console for Project ACSI-MATIC and the sun-angle computer for the TIROS around station.
(Editor's Note: Mr. Wu recently left RCA, and his photograph was not available.)

The primary function of satellite tape recorders is to store information acquired during phases of the orbit when there is no direct communication link between the satellite and its ground stations. When the satellite is in position to establish a communication link, the recorded information is read out and communicated through the satellite transmitting system. A secondary function may be provided by a recorder which records and reads out at different speeds. This function causes bandwidth compression or expansion of the original data, which may be required for tactical reasons.

An example of this is a recorder which continuously stores low-frequency data during, say 95 minutes of orbit time, and then reads out all data during the relatively short time of ground station contact; say, eight minutes.

The data to be stored may include any of a wide variety of signals, such as video, infrared spectrum, inputs from sensors measuring physical conditions in space, telemetry signals reporting the state of the satellite itself, and communication relay signals. Most of the discussion in this paper will refer to the recording of satellite video signals.

In addition to its data record-reproduce function, the recorder must be integrated into the satellite system. This usually means that a stringent budget of weight, power, and space has been allotted to the recorder. Finally, extreme ruggedness and reliability must be designed and demonstrated in order that it survive the launch and orbital environments, and to ensure continued operation over increasingly long missions.

The object of this paper is to discuss some of the more recent recorder developments and considerations at the RCA Astro-Electronics Division since the launching of the first Tiros satellite.1

Satellite projects which require recorders usually invoke specifications which include type of signals, signal bandwidth, storage capacity, weight, power, size, etc.

An example of such system specification occurred in the program for the Nimbus weather satellite. The Nimbus project required the development of two high-performance recorders: an Advanced Vidicon Camera System (avcs) recorder and a High Resolution Infrared (hrir) recorder.

The avcs recorder was specified as a four track, single-speed machine which records and reproduces three channels of video signals and one timing channel. The three video signals are received from three independent vidicon cameras, each of which scans 800 TV lines per picture in 6.5 seconds. Each video signal has a bandwidth of 60 kc and is frequency-modulated by the recorder's circuitry before being recorded on tape.

Each video tape track is capable of storing 64 individual picture frames, including a tape wastage allowance for stopping and starting the transport between each frame. The timing channel carries a 50 kc subcarrier signal which is amplitude-modulated 50% in accordance with a standard Minitrack pulsed timing signal.

The hrir recorder was specified as a four track, two-speed machine which records one channel of infrared analog data and one timing channel for an interval of 128 minutes, and reads out all data in 8 minutes. The infrared input signal has a bandwidth of 5 kc, and is also frequency-modulated by the recorder's circuitry before being recorded on tape.

The timing data is a 10-kc sub-carrier signal which is amplitude-modulated 50% for Minitrack timing pulses, similar to the avcs unit.

Recording at low speed, the recorder runs for 64 minutes from start to end-of-tape and records on two tracks. At the end-of-tape, the motor is automatically reversed, and the input signals are switched to the alternate two tracks. This extends the recording time to 128 minutes with a loss of approximately 1 second of data during the reversing interval.

During reproduction, the tape speed is increased to eight times the recording speed, and all four tracks are read out in parallel, in 8 minutes. Two of the tracks are read out, of course, in the inverse sequence to that in which they were recorded.

For both recorders, it was specified that the flutter be kept to a minimum, consistent with the resolution of the rest of the system. Both recorders were required to have a maximum uncompensated angular momentum of 0.015 lbin-sec.

The use of fm subcarrier recording for video signals involves both advantages and disadvantages for the recorder designer.

The advantages are: There need be less concern with flatness of response or partial loss of signal level than would be the case for "direct" recording, since fm demodulation, when preceded by full limiting circuitry, is virtually AM-insensitive. There is less concern with harmonic distortion; in fact, limiter circuits reduce all signals to square waves at one point in the playback process.

A disadvantage is that the signal-packing density of the tape is reduced below that possible in "direct" recording.

In the Nimbus recorders, we have used full metal-faced magnetic heads with 90 micro-inch (0.000090-inch) gaps. When used in conjunction with tape having thin oxide coating, we have been able to reliably use maximum subcarrier packing densities of 4,000 cycles/in.

Since the Nimbus avcs video signal is transformed into a subcarrier frequency which deviates between 120 and 73 kc, the selected tape speed of 30 ips causes a maximum packing density of 4,000 cycles/in and a minimum packing density of 2,433 cycles/in. At these high packing densities, no bias is required for good sub-carrier reproduction.

Saturation recording is used, and the record-head current is varied with frequency to give the optimum playback signal level. Fig. 1 shows the playback characteristic of the avcs magnetic head at a tape speed of 30 ips. The optimized recording current is also shown here. In typical four-channel heads, the four individual characteristics are matched within 2.5 db within the usable range of 73 to 120 kc.

Signal erasure is most commonly accomplished by fixed permanent magnets, in order to reduce circuit complexity. This is only feasible where information may be destroyed during the first playback and, where playback is permissible during rewind, after recording. Mechanically moveable magnets are regarded as undesirable devices for satellite application.

A situation in which permanent-magnet erasure was not permissible existed in the hrir recorder, since two pairs of tracks were recorded in series by tape reversal and head switching. A fixed permanent magnet would erase the first pair of tracks during recording of the second pair. To meet this problem with a minimum of components, two identical four-channel head blocks are used.

During recording, two channels in the "downstream" block carry the recording current while two channels in the "upstream" block perform erasure. After tape reversal, the two previously unused pairs of channels are used similarly, but the "upstream" and "downstream" roles are reversed. The result is that, in each block, two channels are used for erasure and two channels are used for recording. During playback, only two heads in each block are used.

In the design of our transports, we have followed the concept of minimizing the number of elements in contact with the tape, including the elimination of edge-guiding components, since these tend to generate new disturbances in the tape movement.

We have also used transport configurations in which the magnetic coating does not make moving contact with any surface, except for the necessary case of the magnetic heads. This approach reduces tape wear and maximizes the tape life.

Such a configuration is used in the tape transport shown in Fig. 2. In this picture, the magnetic heads have been removed to show the capstan area clearly. The tape is stored on and exchanged between two coaxial reels which are in parallel planes, approximately 3/4 inch apart.

The tape leaves one reel, passes around a series of four rollers, and enters the second reel. Two of the roller axes are inclined at slight angles to the reels' axis in order to lead the tape out of the plane of one reel and into the plane of the second reel. These angles are computed so that, if all components were perfect, the tape would 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.

The tape passes around one of the rollers twice-once upon leaving one reel, and again upon entering 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°. 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.

An additional factor which is necessary to permit the above mechanism to function is a technique for providing positive tape tension. This is done by constant-torque "Negator" springs, which torque the two reels in opposite directions. The resultant tape tension is nearly constant, and is independent of the direction of tape motion, the tape speed, and the presence of motor torque. Since the tension is always present, and the tape has a large angle of wrap, a very satisfactory frictional grip between the tape and capstan is developed, and the capstan can drive the tape without the need of pressure rollers. This is a desirable situation, since pressure rollers must be regarded as noise generators.

A brief discussion of the Negator mechanism will be of interest. If each reel were torqued by separate constant-torque springs, they could be regarded as being torqued by weights hanging on strings. This system, shown schematically in Fig. 3a, would work well, but it has a drawback.

The Negator springs would have to rotate through the total number of reel revolutions. Referring to Fig. 4, it will be seen that for the case of a 7-inch diameter reel, 1,200 feet of tape would require 740 turns of the reel. It is not feasible to get this large a number of turns directly into Negator springs. Some form of gearing could be used, of course, to couple the springs to the reels and reduce the number of turns. In any high-quality recorder, however, it is usually desirable to eliminate all toothed gearing, no matter how indirectly it is coupled to the tape-head realm.

If the spring system is mounted on one reel and coupled to the other reel the two reels are torqued one against another (Fig. 3b). The springs now "see"
only the relative turns between the two reels. Again in Fig. 4, for the same example of a 7-inch diameter reel, it will be seen that the maximum number of relative turns between reels is 50 for 1,200 feet of tape. This is a reduction in required spring rotation to nearly 1/15 of the total reel turns, and permits a feasible spring design. It also reduces the spring energy, and, therefore, the spring weight, for the case cited, to 1/30 that for direct torquing!
The Negator spring mechanism for the Nimbus tape transport is visible in Fig. 5 facing the base casting on the side opposite that of the reels.

The hysteresis-synchronous type motor has been widely used in RCA satellite recorders. It is, admittedly, not the most efficient motor type available. However, it has a number of other important advantages.

It provides an exactly constant speed, when driven by a precision oscillator, without any of the complications or "dither" of a servo system.

It has good starting-torque characteristics, and, since its rotor has no specific orientation, it has none of the difficulties (characteristic of polarized-rotor synchronous motors) of pulling into synchronism with high-inertia loads.

Since no brushes are involved, the reliability is high and is limited only by the failure rates of ball bearings and stationary windings - an irreducible minimum for conventional motors.

It permits the use of power-saving circuitry. By starting the motor at high voltage and switching to a lower value after synchronous speed has been reached, the rotor is magnetized to a higher level than if operated at the lower voltage.

Low levels of flutter are usually required in a precision satellite recorder. It is, therefore, important that the motor not contribute significant disturbance to the mechanical system.

A hysteresis motor, when powered by regulated AC voltage, has no torque variation except for a slight ripple due to the cyclic variation of magnetic-reluctance path during the course of one rotor revolution.

When the motor is a high-speed type - powered, for example, at 400 cycles - this torque ripple will be 400 cycles and 800 cycles, and will generally be well-attenuated at the tape-drive capstan. Moreover, a high-speed motor (e.g., 6,000 to 12,000 rpm), when coupled to the capstan by a low-compliance transmission like a Mylar belt, will contribute a significant flywheel effect to the capstan motion.

In applications where more than one tape speed is required, we have used one of two techniques. If the speed-change ratio is less than 10:1, we have used combinations of pole switching and electrical-frequency change.

A typical case is the motor for the "Nimbus hrir" recorder, which has two independent windings, permitting operation as a 4-pole or 8-pole motor. By switching from 400-cycle energization of the 8-pole motor, the rotor speed is dropped from 12,000 rpm to 1,500 rpm.

When the required speed-change ratio is much larger than 10:1, this technique becomes undesirable, since the lower-speed mode of operation will become very inefficient, and will transmit increased torque ripple.

For higher-speed change ratios, we have developed a two-motor, belt-and-pulley planetary transmission. This system is illustrated best by the breadboard-demonstration model in Fig. 6.

Here, two motors are coupled to the same output shaft by a planetary belt mechanism. The output-shaft speed will be the sum of the input contributions from each motor. Each motor is coupled through an appropriate reduction ratio to the planetary-system input. By driving either motor singly, and immobilizing the second motor, the output shaft speed may be changed over any ratio required.

An interesting option of this system is that it can provide a four-speed transmission device if the two motors are operated both singly, as described above, and also, simultaneously with like and opposite directions of rotation. This scheme, however, will only allow independent selection of two of the speeds.

An extremely valuable second attribute of this device for two-speed applications is that it will permit using, for the low-speed mode, a low-power motor which does not have to fill the needs of the high-speed mode.

When power budgets are extremely low, and also different for high- and low-speed modes, a designer may be extremely grateful for this situation. A prototype of the planetary drive for another tape recorder designed by RCA is shown in Fig. 7.

In the design of speed-change mechanisms we have, in general, avoided clutches and other time-honored mechanical speed switching devices in favor of purely electrical switching. This has been done for reasons of reliability.

The attitude of an earth-oriented satellite is controlled by a stabilization system which has a limited corrective capacity. When a tape recorder starts or stops, a reactive torque is developed. This will have an affect on the satellite attitude. Also, while the recorder is running at constant speed, there is a gyroscopic effect which increases the effective inertia which the stabilization system must control.

Both of these effects will degrade the precision of attitude control, and will waste energy in the stabilization system. Both of these effects are proportional to the net angular momentum of the recorder, and both effects will be eliminated if the net angular momentum can be made equal to zero.

This condition requires that IΏ = 0 at all positions of the tape on the reels and at all positions of the Negator springs between the reels.
(Anmerkung : was ist "IΏ" ? - Question: waht is IΏ)  

In this design, the change in momentum resulting from these two variables is sufficiently complex to preclude meeting the condition IΏ = 0 with any simple means of compensation. Increasing the diameter of the reel hubs will reduce the variation resulting from each of these variables for a given length of tape. The price paid for such a reduction, however, is an additional fixed value of momentum, additional weight, and an increase in size.

Consideration of all of the several factors resulted in the selection of a hub diameter of 6.00 inches for a tape length of 1,250 feet. With 1.0-mil-base Mylar tape, this results in a build-up to 7.56-inch diameter (hub plus tape) when all of the tape is on one reel, and to 6.82-inch diameter when half of the tape is on each reel.

These diameters result in a reel turn-differential of some 36 turns. Two Negator springs are employed to maintain the tape tension over this differential. The action is such that the springs are wound on a "sun" hub fixed to one reel when all the tape is on either reel. When half of the tape is on each reel the Negator springs are wound on two "planet" hubs, mounted on the other reel.

The tape velocity, as it is unwound from one reel and wound on the other, is a constant and is determined by the surface velocity of the capstan. This results in not only a change in inertia, but a change in angular velocity of the reel-tape system as well. This, together with the cyclic transfer of Negator-spring material, results in the rather complex variation in angular momentum of the reel-tape-Negator system.

The reel-tape-Negator system rotates in a direction opposite to that of the motor-pulley-roller system. The former varies in angular momentum, while the latter has a constant angular momentum. Early in the design, an analysis showed that the angular momentum of the reel-tape-Negator system exceeded that of the motor-pulley-roller system. The logical point to add the compensating momentum is at the capstan shaft, since it serves there a dual function of momentum compensation and flywheel action which reduces wow and flutter.

Experimental tests using a ballistic torsional pendulum, and without any compensating flywheel on the capstan shaft, gave an "average" value of 0.27 Ib-in-sec of uncompensated angular momentum. A flywheel was designed to compensate this value of angular momentum using a dense alloy -  Heavimet - to keep its weight at a minimum.

Fig. 5 shows the flywheel mounted on the capstan shaft. Fig. 8 shows the experimental results obtained with this flywheel incorporated into the tape transport.

Both avcs and hrir systems require "fm" modulators to modulate the video signal before its recording on the tape. These systems also have "am" recording electronics for recording reference or clock signals.

The "fm" modulators used in these systems are voltage-controlled oscillators (vco) which are DC coupled to provide low frequency response.

These vco's incorporate specially developed circuits, and their performance surpasses that of the corresponding modulators in the Tiros weather satellite.

The new circuits have
.

• extremely high linearity over a very wide range of modulation;
• high frequency stability - as little as 0.004 %/°C variation;
• improved reliability;
• symmetrical square wave output; and
• high-precision frequency and deviation setting.

.

The improvements are due to a basic change in design. On Tiros, the vco utilizes a free-running multivibrator as an oscillator. One of the problems inherent in multivibrator circuit design is a small probability that the multivibrator might not oscillate when the DC power is applied.

In the Nimbus system, a relaxation oscillator is used which will not lock-up under any condition. The relaxation oscillator generates a linear sawtooth waveform, by charging a capacitor from a current source and by discharging it through a high-speed switch. The sawtooth waveform is used to trigger a flip-flop which produces a symmetrical square wave with constant amplitude. In the avcs system, a hybrid tunnel diode and transistors are used in the high-speed switch. The fall time of the sawtooth is about 0.04 usec, while the sawtooth period is variable from 5 to 7 usec.

The avcs circuit is designed to produce a square-wave output of 73 kc when the input is -6.5 volts, and 120 kc when input is -11.5 volts. In the hrir system, an all-transistor circuit is used, because of lower frequency requirements. The circuit produces a 10-kc square-wave output when the video signal is 0 volts, and 8.25-kc output when the video signal is -6 volts.

The linearity of these circuits is from 0.1 to 0.2%. The temperature stability is 0.02 %/°C for avcs, and 0.004%/°C for hrir, over the temperature range between 0 and 50°C. To record the reference frequency which is AM-modulated by the Minitrack signals, a simple, direct, record amplifier is used for recording. The high-frequency bias normally used in direct recording is eliminated, since it is not required to reproduce with high linearity in this case.

In the hrir system, electronic switches are used to switch record, erase, and playback signals between two pairs of head.

During the playback, the video signal is amplified, limited, and filtered in both the avcs and hrir systems. The preamplifier is designed for low noise, high gain, and low power drain. It is followed by a limiter circuit to remove any amplitude modulation due to the head-response characteristics or tape dropout. The harmonics generated in the limiting process are then removed by a low-pass filter.

The reference frequency, with Mini-track modulation, is reproduced by a high-gain preamplifier. Since it reproduces only the reference frequency, a compensation network such as used in a conventional direct reproduce system is not necessary. The elimination of the compensation network reduces the size and weight of the circuitry.

The recorders use AC motors. It is necessary to have a power inverter to convert primary satellite DC power to AC power to drive the motor. The Tiros recorder system required a DC power input of about 18 watts to the power inverter, compared with 2 or 3 watts for the electronics. Thus, any increase in power-inverter efficiency will markedly reduce the total recorder power-system requirement.

In the avcs and hrir systems, particular attention has been paid to the circuit and transformer design to increase the overall efficiency well above 90%. In addition, a high-voltage start and reduced voltage-run method is used to reduce further the dc power requirements. In the present design, the dc power requirement is approximately 8 watts to the power inverter to drive the recorder motor after it has been switched to its running voltage.

A latching relay is used to provide the means of switching output-transformer taps under load. To avoid damage to the relay contacts from arcing, and to avoid switching transients to the power transistors, zener diodes are used for suppression. The original circuit was still operating without miss or degradation after 1.8 million operations.

In the avcs motor circuit, a novel braking system is used to provide a quick stop. The braking is done electrically, in the motor, whenever DC power is removed. A latching relay is used to do the switching, and no power is need for this circuit. A life test was conducted on this relay with no failure in 1 million cycles of operation. The electrical braking reduces the stopping time from more than 4 seconds to about 0.5 second.

Further, it provides a small residual magnetic locking torque to prevent the coasting due to Negator spring torque. Since no mechanical linkage is used in this system, no wear-out problem is present, and high reliability is assured.

The transport and all circuitry are enclosed with a hermetically sealed enclosure, shown in Fig. 9. Some idea of the interior packaging arrangement may be obtained from Fig. 10, in which the upper half of the enclosure has been removed.

Each vertical array of circuitry consists of two epoxy-fiberglass circuit-board assemblies which have been cemented together for mutual stiffening. Each ensemble is then given a conformal coating of epoxy resin to immobilize all components and leads.

The enclosure is hermetically sealed by a Viton 0-ring joint and pressurized with 16 psi of a gas consisting of 90% nitrogen and 10% helium, the latter being included to permit measurements of leakage rates.

The maximum allowable initial leakage rate for an enclosure while in a high vacuum chamber is 1 x 10 -4 cm/sec. Using this initial leak rate, and assuming an exponential rate of decay of pressure, it has been calculated that the pressure will reach 0.1 psi, absolute, in 23 years.

This is regarded as still pressurized, as far as outgassing effects of lubricants and other materials are concerned.

During steady-state operation, the avcs recorder draws 10 watts at 24.5 volts-DC. The hrir recorder draws 9 watts in its high-speed mode and 7 watts in the low-speed mode, at 24.5 volts dc.

These units will operate without damage or degradation of performance under the highest laboratory vacuums, and within temperature ranges of -15 to +60 °C. They will survive, without damage, vibration levels of 25 g-RMS, and random noise between 20 and 2,000 cps.

Typical flutter values for the avcs recorder are 0.02%-rms between 0.5 and 30 cps, and 0.10%-rms between dc and 5,000 cps.

A more meaningful demonstration of the recorder's capability may be seen in the 800 tv line video samples (shown in Figs. 11 and 12), which were recorded and played back by the avcs recorder functioning as a link in the complete video system.

1. J. A. Zenel, "Narrow-Bandwidth Video-Tape Recording Used in the Tiros satellite", Journal of SMPTE, November, 1960.
2. The Nimbus project on the AVCS and HRIR recorders is conducted by the DEP Astro-Electronics Division under the sponsorship of the National Aeronautics and Space Administration on Contract No. NAS5-877.

pictures :
Fig. 1-Response curve of magnetic heads.
Fig. 2-Closeup of transport.
Fig. 3-Reel torquing techniques.
Fig. 4-Reel turns vs. tape length.
Fig. 5-Underside view of transport.
Fig. 6-Breadboard of the planetary drive.
Fig. 7-Prototype planetary drive.
Fig. 8-Angular momentum variation curve.
Fig. 9-Hermetically sealed recorder.
Fig. 10-Recorder packaging.
Fig. 11-Video test pattern.
Fig. 12-Sample video copy.