(A HIGH-SPEED PRECISION INSTRUMENTATION TAPE RECORDER)
by B. A. COLA and J. M. URITIS - Missile and Surface Radar Division DEP, Moorestown, N. J.

JOSEPH M. URITIS graduated from Newark College of Engineering in 1937 with a BSME. Before joining RCA in 1946, he was a design engineer for Bernard Aviation Equipment Corp., New York City. In 1941, he engaged in the design and test of main propulsion and auxiliary shipboard equipment at the Philadelphia Navy Yard. At RCA, Mr. Uritis has designed disk recorders for broadcast studios and other commercial applications. As a member of the Advanced Development Group, he was responsible for conception of the tape files used in RCA Electronic Data Processing equipment. In video recording, he was active in the design and development of the time division multiplex video recorder, and later, the quadruplex color video recorder. He was responsible for the design of the Broadband Recorder (AN/TLH-I) built for the Signal Corps. When the BMEWS program started, he designed magnetic drums and discs for that project, and later was responsible for design of the TRADEX tape transport. During design of the precision tape recorders for Bell Labs, he was responsible for the mechanical design.
Mr. Uritis holds 8 patents in the recorder field.

BENJAMIN A. COLA graduated from Drexel Institute in June 1950 with a BSEE. At this time he joined RCA and worked on the design of power supplies, video circuits, TV airborne cameras, naval radar tracking equipment and servo mechanisms in the Special Devices Department. A period of one year was spent on systems analysis of shoran bombing and reconnaissance equipment. He received an MSEE degree in June 1955 at the University of Pennsylvania. In August 1955 he was transferred to the newly formed Missile and Surface Radar Section, where he worked on analog data processing assignments for TALOS, ATLAS, BMEWS, DAMP, and TRADEX. His most recent assignment was design project engineer on the development of a precision high speed tape recorder.

This multiple-channel, wideband precision tape recorder was developed as a basic element in a tracking radar system to be used in multiple target environments. All signals received by the radar tracking the target in real time are tape recorded. Thus, all targets viewed by the radar during a "live" exercise can be repeatedly acquired, tracked, and analyzed through playback at a later date. To permit recovery of maximum information, data is recorded in the form of unprocessed radar IF signals; then on playback, the optimum signal-processing system recovers each target parameter of interest.

Radar target parameters such as range, doppler, and angle position must be measured precisely; at the same time, target resolution and relative amplitude must be preserved. Such performance requirements impose challenging demands on a radar tape recorder - particularly signal-to-noise performance, gain and phase tracking between channels, gain linearity, transport velocity stability, and interchannel jitter.

Fig. 1 shows the complete radar tape recorder designed to meet these standards of performance. The radar tape recorder provides 15 channels with a bandwidth capability in excess of 6 Mc; the recorder operates at a tape speed of 1,180 ips and a velocity stability of better than 1 part in 105. The recording time is 6 minutes. The radar system utilizes these capabilities to record 8 channels of 3.5 Mc video radar signals along with 7 wideband channels of timing, digital, and other reference signals for processing by the radar in serial form.

The radar tape recorder (Fig. 2) stores ungated if information from the uhf and L-band receivers so that during subsequent playbacks, targets can be acquired and tracked in angle, range, and doppler.

The receiver-exciter sends repetition rate pulses to the UHF and L-band transmitters where the power is stepped up and sent to the antenna for transmission. The receiver-exciter receives the returns from targets and sends them to the tape recorder and tracking receiver.

The tracking receiver (in conjunction with the range tracker) tracks targets in angle, range, and doppler during real time or from tape playback. The tracking receiver also provides agc, scintillation, doppler, and range information to describe the target to the data recorder. The tracking receiver also provides azimuth and elevation errors to the angle servos that drive the pedestal, keeping the antenna on target during live track.

By recording the radar returns on tapes, it is possible to separate and analyze the targets of a missile shot as many times as desired and, thereby, effect an economy of radar systems and missile firings.

Based on the principal radar requirements of Table I, specifications for the radar tape recorder were determined. Requirements for frequency and polarization diversity and for monopulse angle tracking resulted in a need for eight signal channels. Range and doppler accuracies were improved through the use of a recorded if reference signal and a range clock signal. Additional channels were used for a transport servo-speed reference, pedestal angular position data, and time-of-day codes - resulting in a 15-channel machine (see chart of Table II).

The principal specifications for the radar tape recorder are Table III. Reasons for some of these specifications are described herein. The signal-to-noise (S/N) specification is dictated by the desired system dynamic range. Velocity stability is necessary to assure doppler tracking accuracy as well as to minimize the occurrence of spurious signal spectral lines due to transport velocity modulation. Gain and phase track specifications assure accuracy in the measurement of target angle offset.

Gain and phase linearity maintains good target resolution by avoiding pulse distortion or distortion echoes. Low interchannel jitter is important in measuring accurately the signal polarization vector, range, and angle. Bandwidth is determined by the radar pulse characteristic; at the signal input to the recorder, the pulse is in the form of a cosine squared.

To accomplish the specifications set forth, a basic decision was made to use fm rather than am recording on all tracking video signal channels for the following reasons:

• 1) Gain linearity and gain tracking are very difficult to achieve in an am system because of the effect of the nonlinear magnetization characteristic of the tape and because of variations in tape sensitivity.
• 2) An fm system minimizes the effects of tape drop outs, variations in sensitivity of the magnetics and am variations introduced by changes in the effective head-air-gap with tape pressures. Good gain linearity can be achieved by frequency modulators and demodulators.

.
The choice of FM recording results in the disadvantage of greater circuit complexity because of the above requirements for frequency modulators and demodulators and also the need for more bandwidth than that in an am system. In fm, the bandwidth required to pass the sideband frequencies depends on the desired S/N performance, and on a minimum fm carrier frequency that prevents fold-over of the side frequencies. The final choice resulted in a 4.5-Mc fm carrier frequency with significant sideband frequencies extending from 0.2 Mc to 8Mc.

The 8-Mc bandwidth required that the relative speed between tape and head exceed a minimum value determined by the magnetic head gap-length. Since head gap-length could not conveniently be made less than 40 microinches (0.000040 inch), a tape speed of 1,180 ips was used; at this speed, 6 miles of tape are required to provide 5 minutes of recording time. Because of the requirement for multiple channels, longitudinal recording was chosen.

The basic design criteria for the tape recorder were established, based on a consideration of signal-to-noise, gain tracking, gain linearity, bandwidth and multiple channel capability.

Satisfying the requirements for velocity stability and interchannel jitter depended to a large extent on the mechanical perfection of the tape transport and the performance of the servo speed control. The main elements of the transport are shown in Fig. 3. The recorder components are assembled to a jig plate which is accurately aligned and attached to a rigid frame to assure a precise tape path. The complete assembly is shock-mounted within its cabinet.

The 5-minute recording requirement resulted in 30-inch-diameter reels weighing 50 pounds and containing 6 miles of 1.1-mil magnetic tape plus 2,500 feet of clear leader at each end. The tape is accelerated to full recording speed in 60 seconds and is stopped in the same time. The reels are equipped with quick-disconnect knobs to facilitate fast reloading and permit uninterrupted recording with two transports. Tape guides and tensioners are air lubricated so that frictional contact to the tape occurs only at the capstan edge-guide and at the magnetic heads.

Remote control of recording, subsequent repeated playbacks, and rewinds dictated that the tape remain threaded and taut during and following each run. To accomplish this, programmed acceleration and deceleration cycles were employed at the start and end of the tape runs. A length of clear polyester leader is attached to each end of the tape; transitions from clear leader to tape are sensed photoelectrically to initiate the program.

The capstan-head assembly of Fig. 4 is designed for quick replacement by utilizing an air manifold which automatically seals upon installation of the capstan. Two blocks of magnetic heads, one an 8-track and the other a 7-track are supplied by the RCA Broadcast and Communication Products Division. Heads are prealigned on arms for fast replacement once initial adjustment of the head arm support has been made.

To achieve constant tape speed, the capstan assembly employs an air-turbine drive and hydrostatic air bearings. Capstan speed is sensed by a tone wheel and controlled by a hysteresis brake; both are integral parts of the capstan shaft. To minimize longitudinal vibrations, the tape is stabilized by making simultaneous contact with both the capstan and the heads. Since the tape is not perfectly smooth nor of uniform thickness, it is necessary to relieve the capstan surface behind the area where the heads contact the tape.

The velocity stability specification of 1 part in 105 imposes a severe runout requirement on the capstan assembly; the angular velocity generates flutter components of 87cps and higher multiples which are above the bandwidth of the capstan-servo correction capability.

Therefore, the dynamic runout of the capstan must be held to less than 20 microinches; it is done by using the ultimate in precision grinding, hand lapping, and dynamic balancing. An additional benefit in reducing the effects of capstan runout is derived from depressing the heads into the tape a*bove the capstan grooves; thus, the heads may be firmly positioned at a fixed radius from the rotational axis of the capstan. Head contact pressure is controlled by a very precise adjusting screw which advances the head only 5 mils per revolution. Pressure applied is measured as an increment of the capstan hysteresis brake current.

Mechanical resonances throughout the tape path produce variations in tape tension, resulting in minute (but not inconsequential) variations in tape velocity at the capstan. Such variations can be caused by improper tape guide design, or by insufficient damping where resonant systems exist in the tape path, as at the tape tensioners. Since these effects are not necessarily uniform, or in-phase across the width of the tape, they produce the undesirable effect defined in the specifications as "phase jitter." The allowable phase jitter between signals (recorded half the width of the tape apart) is ±20 nsec; this corresponds to roughly a differential displacement of 24 microinches.

Various refinements effected to reduce phase jitter include the following:
.

• 1) very smooth, solid flange reels which are two-plane dynamically balanced,
• 2) windage guards which completely encircle both reels,
• 3) an adjustable edge guide adjacent to the supply reel to control lateral weaving of the tape as it enters and leaves the reel and
• 4) precise alignment of all tape guides.

.

Compressed air, supplied by a remote unit at 100-psi nominal pressure, operates the transport. Air is oil free and dried to -40°F dew point. Outlet air-supply filters and filters at the air inlets to the cabinet and to the capstan air-bearing line protect the air-pressure regulating valves and the close-fitting capstan bearings. In the event of a compressor failure, a reserve air supply brings the transport to a safe halt.

Pressure-sensitive switches are set a few pounds below the lower limit of the normal pressure variation range of the air supply. When the air-supply pressure falls below a predetermined value, turbine air is cut off and the transport goes into its programmed stopping cycle. Mechanical brakes on the reel spindles stop the reels quickly to minimize tape spillage in the event of a power failure or loss of tape tension.
.

Each tape reel is belt driven by a two-phase 400-cps induction motor and controlled by an independent servo system. The control signal for the reel servo loop is obtained from the tape sensor device nearest the particular reel involved (see Fig. 3). The stator of the induction potentiometer is excited by an AC reference supply and produces an error signal proportional to the velocity error of the tape at the reel. The Ac-induced voltage in the rotor, which is the error signal, is amplified and applied to a carrier-referenced demodulator; then, the dc output of this circuit is used to drive the operational amplifier shown in Fig. 5.

An additional feedback loop is provided by an AC tachometer on the servo motor shaft. The 400-cps signal is demodulated to dc to make it compatible with the tape error signal; it is easier to realize the proper stability networks in the dc domain, and AC quadrature problems are completely eliminated - providing a more stable design with less effort.

The tachometer feedback signal is combined with the tape speed-error signal in a summing operational amplifier to compensate for the non-linearity of the magnetic amplifiers which are as the servo power amplifiers. The summing operational amplifier output is the servo control signal used to drive the magnetic amplifier; the magnetic amplifier controls the 400-cps power supplied to the reel servo motor.

The capstan servo drives the tape with highly precise speed control by two servo loops: one a frequency lock, and the other a phase comparison control to obtain high sensitivity (Fig. 6). Pulses derived from the capstan tone wheel and proportional to tape speed are passed through a 400-usec delay line; capstan speed is varied by the servo until the pulse frequency corresponds with the delay of the line. The 400-usec delay line is equivalent to a tone-wheel frequency of 2,500 pps (this servoing of the capstan speed to near synchronous value is called a frequency lock). When a frequency lock occurs, the phase of the reference 2,500 pps begins to zero beat with the tone wheel; thus, a phase error is produced which adjusts the capstan speed to a full phase lock.

The phase detector producing the speed-control signal is extremely sensitive; the slightest change of capstan tone-wheel frequency results in a considerable phase error. Once the capstan servo is locked in, complete control of the speed is maintained by the phase loop. The basic speed control of the capstan driving the tape is obtained by applying air pressure to the capstan turbine. The turbine air pressure would drive the capstan at a speed much higher than needed; however, the servo controlled hysteresis brake reduces it to the desired value.

The capstan speed is locked to a crystal oscillator by servoing the tone wheel during the record mode. During the playback mode, the control track, a recorded 2,500-pps reference frequency standard, is servoed to the same crystal oscillator. The slight difference between the capstan speed and the tape speed is removed by this playback servo technique. Electronically variable delay lines are used in those channels requiring additional velocity refinement corrections to the radar data (Fig. 7).

The technique of comparing the playback of a recorded 2,500-pps crystal clock with the same crystal clock and using the resultant error voltage to drive the capstan cancels out all low-frequency speed variations of the tape. However, this method does not remove high-frequency components of speed variation resulting from such sources as capstan asymmetry and high-frequency tape flutter. The capstan turns at 87 cps and introduces Fourier series tape-speed components 174 and 261 cps in both record and playback modes; such components cannot be servoed out because of the limited response of the capstan. To remove the effects of the capstan asymmetry and other speed deviations, electronically variable delay lines are inserted into all channels that need precision speed stabilization; a control voltage advances and delays each channel of data in order to make the tape information appear to be coming from a perfectly constant speed tape.

The control of these electronically variable delay lines is accomplished with the closed-reference automatic-time-correction (atc) loop of Fig. 7. When the capstan servo has locked in, the reference playback signal averages 2,500-pps - but there will be expansions and contractions of the pulse periods for the reasons previously described. This reference playback signal is inserted into an electronically variable delay line that removes the high-frequency speed variations. The tape-playback 2,500-pps reference is compared to the clock-reference 2,500 pps in a phase detector; the resultant error signal changes on a pulse-to-pulse basis. This error is amplified and the atc loop stabilized by passing the signal through an operational amplifier with the proper transfer function.

The resultant amplified error output is used to drive the reference delay line, completing a closed loop. It is apparent that the error signal created will advance or delay the playback pulses to slave them to the clock-reference 2,500 pps by removing the high-frequency components of flutter. At this point, the 2,500-pps reference passing through the delay has had the stabilization of the capstan servo control and the atc servo electronic correction. The signal that drives the reference delay line also drives the six signal delay-lines because the speed correction required is approximately the same for all channels.

The interchannel jitter must be small in order to make this type of delay line correction effective; moreover, this condition is assured by meticulous design of the tape transport. There must be very little phase jitter between channels, since the correction signal of the atc servo loop is not only applied to the reference line, but also (as an open-loop correction) to all of the signal delay lines.

To measure intertrack jitter during the mechanical development of the tape transport, the precise measuring technique of Fig. 8 was used. It consists of recording a 1.5-Mc sine-wave signal upon the two tracks in question, playing these channels back, and then mixing them down to 200 kc. The exchange of phase angle between the two signals remains the same at 200 kc as it was at 1.5 Mc, and it can be handled by a commercially available phase meter. The output of the phase meter was observed on a scope, spectrum analyzer, and pen recorder. A typical pen recording of the intertrack phase jitter is shown in Fig. 9.

The most interesting aspects of the signal electronics for the radar tape recorder relate to the problems of linearity and bandwidth in the precision fm channels. Fig. 10 shows the elements of a typical "fm" channel. In the record mode, the signal in the form of radar video pulses is converted to an "fm" signal in a modulator. It is then amplified in a record amplifier which drives the record-playback head as a constant-current generator. On playback, the signal is amplified in a preamplifier and then in a playback amplifier which includes aperture equalization to compensate for the magnetic-head frequency response characteristic. The signal is then reconverted to a radar video signal in the frequency demodulator.

The bandwidth required in an fm system is dictated by the rise-time requirement; the system S/N ratio is a function of this bandwidth.

(This paragraph is skipped - it is too special)

From this expression, it can be shown that a peak frequency deviation of 1.5 Mc achieves S/N performance greater than 40db assuming that the ratio of peak carrier to rms tape noise per megacycle is 48db.

The radar tape recorder was designed to record radar pulses with a video bandwidth of 3.5 Mc; for the frequency modulation of these signals, a 4.5-Mc carrier was used to minimize the effect of "fold over" (side frequency terms) in the fm signal.

The primary problem of equalization is related to the need for compensating the amplitude-versus-frequency response of the magnetic playback process. The flat amplitude and linear phase responses required to meet the performance specifications for the radar tape recorder were achieved. The following equation, however, shows that the amplitude response of the playback process is not flat and further that the phase response is linear. Therefore, the equalization must compensate for the amplitude response of the playback process without introducing a nonlinear phase characteristic.

Analysis of the magnetics problem results in the following equation for the output voltage from a playback head, assuming a sinusoidal recorded flux on the tape.

Two tape recorders have been in continuous operation on the pacific missile range for almost a year and have recorded and preserved much valuable data for further analysis and evaluation. The advanced techniques employed have proven invaluable in the study of missile re-entry.

Credit for work on the original Tradex tape recording system, which was a forerunner of the high-speed precision tape system described in this paper, is due to H. R. Warren and his associates in the DEP Communications Systems Division.

Also, credit for magnetic head design is extended to the Magnetic Head Development and Design Group of the Broadcast and Communications Products Division, under the direction of B. F. Melchionni. This project was completed successfully because of the tireless efforts of many people with varied skills. It would be unjust not to mention two of the technical giants, Thomas Bolger and Jackson, who did so much to bring this project to fruition.

pictures - Tables

TABLE III - Performance Specifications
Tape speed, 1,180 ips, max.
Speed stability, 1 part in 105
Number of tracks, 15
Tape, mylar 1" wide, 1.1 mil thick
Tape length, 6 miles
Record-playback, 5 min @ 1,180 ips
Start-stop time, 1 minute
Rewind time, 10 minutes
Reel size, 30" diameter
Capstan drive, air turbine
Capstan bearings, air bearings
Tape guides, air guides
HEADS:
Gap length, 50 u inch
Track width, 20 mils
Gap scatter, 40 u inch max.
Track spacing, 63 mils

Mounting: precision magnetic record-playback heads on precision machined rocker arms which are completely replaceable with only one adjustment, head pressure (by two self-locking coarse and fine adjust screws).
Environment: 70° air conditioned

FM RECORDING (8 channels):
Modulation frequency response, 6 cps - 3.0 Mc
within ± 1.5 db Signal to noise, 40 db min., (0 to P)/RMS All distortion, - 40 db max. Gain tracking: differential gain between any two
channels over 40-db dynamic range within 0.5 db. Phase tracking: differential phase between any twc
channels over a 40-db dynamic range within 10°
of 1.5-Mc signals. Phase linearity: Within 3° of 1.5 Mc signal for 1
cycle variation; within 1° of 1.5 Mc signal for
greater than 1 cycle variation. Phase jitter: 0.04 usee maximum between adjacent
channels; 0.1 usee maximum across entire tape. Frequency stability, 30 cps deviation of 1.5-Mc
signal Crosstalk, 50 db max.

DIRECT RECORDING (7 channels, 6 equipped for digital signals):
Signal to noise, 30 db min.
All distortion, - 30 db min.
Crosstalk, - 50 db max.
Amplitude modulation, 10% max.
Phase jitter, 0.04 usee maximum between adjacent
channels; 0.1 usee maximum across 15 channels. Frequency stability, 30 cps deviation of 1.5-Mc signal

DIGITAL ELECTRONICS (4 channel capability):
Pulse repetition rate, maximum 1.2 Mc
Pulse width, minimum, 0.4 usee; maximum, 50%
duty cycle Rise time: 0.1 usee. max.

Fig. 1 - The high-precision instrumentation tape recorder.
Fig. 2 - Acquisition of data by tracking radar plus recording and storage of such data by precision tape recorder.
Fig. 3 - Tape transport. Fig. 4a - Capstan-head subassembly.
Fig. 4a -
Fig. 4b - Detail of capstan and heads.
Fig. 5 - Reel servo system.
wo sind 5 und 6 ???
Fig. 7 - Automatic time correction (ATC) servo
Fig. 8 - The intertrack jitter measuring system,
Fig. 9 - Typical pen recording of intertrack phase jitter (shown are tracks 4 and 10).
Fig. 10 - Elements of a typical FM channel.
Fig. 11 - Unequalized head output versus amplitude. (0 db = 1 millivolt.)
Fig. 12 - Aperture correction circuit and phase-gain curves obtained.