A. C. LUTHER, Mgr. - Television Tape Recorder Engineering Broadcast and Communications Products Division, Camden, N. J.

ARCH C. LUTHER, JR. graduated from the Massachusetts Institute of Technology in June 1950 with a BSEE degree. He joined RCA in July 1950 in the Broadcast and Television Engineering Department. He has worked on the design and development of sync generators, camera control equipment, monitors, switching equipment, and other items of television studio equipment. Mr. Luther is now Manager of "Television Tape Recorder Engineering" in the "Broadcast Studio Engineering Department", and is responsible for development and design of television tape recorders and accessory products. He holds 25 United States Patents in the field of television. He is affiliated with Eta Kappa Nu, the Society of the Sigma Xi, and IEEE.

The recording and reproduction of color-television picture signals on magnetic tape has posed many challenging problems in addition to those which have been solved in monochrome television tape systems. The most important color problem is that of time-base stability. This paper describes the means developed for color tape playback in the RCA TR-22 Television Tape Recorder,1 which solves the problem of time base stability.

The mechanical scanning process of the quadruplex recorder produces various distortions of the timing of the reproduced picture signal. Many of these effects are residual to the process of segmentation of the picture signal into pieces of 16- or 17-line groups for recording on each transverse track on the tape.

The appearance of some of this type of picture distortion in a monochrome signal is shown by Fig. 1. Each of these effects is readily minimized and made negligible for monochrome transmission by proper adjustment of the transverse scanning mechanism and the recorder circuits, but this does not ensure good color reproduction.

The amplitude- and phase-modulated subcarrier which transmits the color information in the composite color video signal requires far greater timing accuracy.

A picture displacement from line-to-line of 0.03usec is not visible in a monochrome picture, but the same displacement represents a phase shift of 40° to the color subcarrier, which will produce a serious hue shift in the color picture.

The maintenance of such precision on a day-to-day basis with present TV tape recorders requires skilled operators and constant attention, both of which are costly and often just not available in the field. Therefore, it is desirable to develop means to reduce the precision required in the transverse scanning for playback of a tape recording.

A further source of color instability is the time base jitter resulting from non-uniformities of the mechanical motions of the recorder. Even with the most sophisticated servo system available (Pixlock), the residual jitter may be of the order of 0.1 usec (130° of sub-carrier phase) which is intolerable for color. Fortunately, it has been possible to develop devices for correction of both segmentation and jitter errors simultaneously.

Practical experience with quadruplex recorders in the field has shown that it is reasonably easy to maintain the total of all time base errors to less than 1.0 usec peak-to-peak. Therefore, a correction device should have at least 1.0 usec of correction range in order to achieve the goal of non-critical operation of the recorder system. This kind of control range is within the capability of electronically-variable delay lines based upon the properties of silicon diode capacitors. Therefore, an all-electronic corrector is feasible.

Given an electronically-variable delay line, it becomes necessary to develop time base error measuring circuits in order to derive proper control of the delay line. This may be done by the use of phase detector circuits comparing the timing of the TV sync pulses to a suitable reference pulse.

There are two ways to arrange such a system. Fig. 2 shows both open-loop and closed-loop control systems. These two systems differ substantially in their capability for rapid correction of errors.

The closed-loop system, which might at first appear more desirable, is limited in rapidity of correction by the sampling process inherent in the sync pulse phase detector. Since the response of a sampled-data feedback loop must necessarily require a number of sampling intervals for stabilization, and we obtain only one sample per tv line from the sine pulse comparator, the fastest control response of the closed-loop system would be several tv lines in duration.

This would be unsatisfactory for our purpose, since we have seen (Fig. 1) that the segmentation time base errors of the tape process can produce large changes in timing from one line to the next so our corrector nuist be able to change its delay by the full range from one line to the next. In fact, the entire delay change must occur within the blanking interval, so that it is not seen in the picture.

The open-loop corrector is capable of one-line response, and it therefore is the system which has been developed. As applied in the RCA TV Tape Recorders, such a device is called atc (automatic timing corrector). The atc is capable of correcting jitter and segmentation errors for monochrome picture signals, but it does not complete the job for color signals. For complete color operation of the tape recorder, the "color atc" unit is required in addition to the standard "atc". Both of these units are available for the TR-22 recorder as accessories which fit right in the TR-22 console as seen in Fig. 3.

The double-width module of the atc unit contains the electronically-variable delay line (evdl). This line consists of 84 sections with fixed iron-core coils and silicon diode capacitors. The "horseshoe" arrangement of this line shown in Fig. 4 effectively bends the line in half to fit it in the module length without introducing any discontinuity. The rest of the double module contains the video driver and output amplifier circuits which complete the video path through the ATC unit.

As seen in Fig. 5, the video passes through the line in single-ended fashion, while the control of the line is push-pull. This is necessary to balance out transients which are coupled from the control busses to the video path by the capacitance of the diodes every time the control voltage is changed.

Since the evdl is to be used in an open-loop control system, it is essential that the delay vs. input voltage characteristic be highly linear for proper tracking of control voltage to be maintained over the entire delay range. The capacitance-vs.-voltage characteristic of the silicon diode is highly nonlinear and of course delay varies as the square root of capacitance which is a further non-linearity.

Therefore, a nonlinear amplifier (nla) is introduced in the control path to correct for these effects. The delay-vs.-voltage characteristic at the input of the nla is then accurately linearized. The nla also contains the phase splitters and error driver amplifiers for feeding the delay line.

Driving the control voltage to the delay line is a difficult task in itself, since the diodes represent a capacitance load, and a fairly fast response must be achieved - during which time the positive and negative busses must remain accurately balanced so that transients coupled to the video will cancel.

Furthermore, the drivers must at all times present a negligibly low impedance to the line over the entire video passband because the driver represents the ground return for the capacitors of the delay line. This is accomplished by the combination of complementary-symmetry emitter followers on each error bus and additional fixed capacitance to ground to insure a low impedance at the higher video frequencies.

As the diode capacitance is modulated, both the delay and the characteristic impedance of the line will change. The line is terminated at both ends with fixed resistances which are accurately matched to the line impedance at the center of the delay range. As modulation away from this point takes place, an increasing mistermination results. The total amount of delay modulation which can be achieved by a line of given length must therefore be limited so that the reflections produced by mistermination never become great enough to be visible. This is a reasonable compromise.

Additional considerations in the delay line are concerned with the location and duration of the delay-change transient which occurs when the delay is set for each tv line in the picture. Since a large timing shift may occur when the recorder switches from one head to another during the playback process, it is essential that the head switching take place before the timing error is measured to set the delay for the next tv line. Head switching in RCA recorders is located during the horizontal sync pulse; therefore, the trailing edge of sync is used for delay error measurement by the atc. This arrangement allows the greatest tolerance for switching pulse location.

The delay-change transient must be contained within the width of horizontal sync, because it must not interfere with the timing of the leading edge of sync (which is used by receivers) and it must not interfere with the color burst, which appears just after the trailing edge of sync (Fig. 6). It is inherent that the delay-change transient in such a variable delay line cannot be shorter than the delay of the line itself, and in practice it is longer than this because of rise time limitations in changing the control voltage to the delay line.

Therefore, we split the delay line in two and provide separate control to the two parts to allow a sufficiently short delay change transient to fit within the sync pulse. Fig. 7 shows the arrangement for doing this. You will note that the delay-change transient in the resultant signal occurs ahead of the sync edge which is measured to produce the delay change. This is done by adding fixed delay (dotted in Fig. 7) in the video path before the signal enters the variable delay line.

The atc system as described is capable of reduction of timing errors by at least 25 times (Fig. 8).

So far we have described the atc system operating only for monochrome signals. Of course the basic action of reducing time base errors will also be in the right direction to stabilize a color signal, but there are still important limitations to color operation based on an atc system which works only from the edge of horizontal sync.

The first limitation is noise. Using the sync pulse for timing measurement of the signal provides only one sample per tv line, and a relatively wide bandwidth is needed to handle the signal.

Therefore, the timing measurement will be affected by the random noise which is added by the tape recorder process, producing an effect known as positional noise. For monochrome purposes, the positional noise is at or below the threshold of visibility, so it is not serious; but in color it represents a substantial phase jitter which results in hue instability. For this reason, it is desirable to use the color atc, so that the greater number of samples per tv line (due to the 8 cycles of sub-carrier in the burst) and possible use of narrower bandwidth will reduce the positional noise effect.

A second problem with the use of sync pulse comparison for color is the fact that there is no specified phase relationship between the edge of sync and the color subcarrier. Therefore, the sync pulse can not be used to control the absolute phase of subcarrier.

It is even possible that there can be time modulation introduced between the sync pulse and the color subcarrier so that it is essential that the final color timing correction be done by comparing the phase of burst with a fixed reference.

A further problem arises because the correcting action of monochrome atc is not perfect due to residual errors in calibration of the open loop. As stated previously, a reduction factor of 25:1, minimum, is considered practical, but this can still leave enough residual to disrupt color performance if large errors are being corrected by monochrome atc. The cascading of the color atc process means that any residual can be theoretically reduced by another 25:1 factor, which then makes it negligible even for color.

Color atc employs the same open-loop control system which we have just described for monochrome signals. The only difference is that the error detector uses burst instead of sync (Fig. 8). The color atc delay line is shorter, because a delay range of 360° at subcarrier frequency (0.28 usec) is all that is needed. Because of the shorter line, it is not necessary to split it to contain the delay-change transient within the sync pulse. The nla and driver considerations are the same as monochrome, in fact, identical circuits are used throughout.

The error detector for color atc is substantially different and a lot more complicated. This is because it is necessary to make a phase detector with a linear range of 360° at subcarrier frequency. This is accomplished by operating the detector at half-frequency, so that the desired 360° is only 180° to the error detector (Fig. 10).

Both the burst and reference subcarrier are divided by multivibrators. The divided burst is used to form narrow sampling pulses and the reference signal forms a sawtooth waveform. A special inhibit function on the burst path guarantees that sampling always occurs on the center half of the sawtooth waveform.

By measurement of the phase error of burst with the color error detector circuit, a control signal for the color atc variable delay line is formed. This is able to correct the phase errors in the composite color signal and thereby stabilize the timing of the signal to provide excellent color performance. However, the job is not done, because the signal must still be passed through the processing amplifier of the tape recorder in order to clean up the blanking and sync pulses and insert a new color burst that will be free from noise. This requires special handling in order to not distort the color components of the signal.

The system for processing the color signal consists of splitting the composite color signal into high-pass and low-pass components (Fig. 9). The low-pass component does not contain any color information and therefore can be passed through the standard monochrome tape recorder processing amplifier to clean up blanking and sync. The high-pass signal contains all color information and is processed by special circuits in the color atc modules to clean up the blanking interval and insert a new burst. Then the high-pass signal is added back to the low-pass signal just ahead of the output amplifier of the processing amplifier.

Color "atc" thereby provides time base correction on the tape playback signal to a residual error level of a few nanoseconds with respect to the subcarrier reference signal. This allows color tape recording with essentially no bandwidth or color response limitations. With careful operation, the TR-22 with color atc can record and playback color program material which is indistinguishable from the original live signal.

1. A. H. Lind, "The TR-22 - RCA's Transistorized TV Tape Recorder," RCA

pictures

Fig. 1 - Three photos from the picture monitor showing scene material with a) quadrature, b) skewing, and c) scalloping distortions.
Fig. 2 - Open-loop and closed-loop timing-error correctors.
Fig. 3 - TR-22 tape recorder module area highlighting the monochrome ATC and color modules.
Fig. 4 - ATC delay module showing 84-section variable delay line.
Fig. 5 - Electronically variable delay line.
Fig. 6 - Waveform of color horizontal-blanking interval, showing location of head switching and ATC delay-change transient.
Fig. 7 - The ATC, showing splitting of delay line and error driver amplifiers to shorten delay-change transient and locate it properly.
Fig. 8 - Picture monitor photos showing typical scene material a) before, and b) after ATC.
Fig. 9 - Color ATC.
Fig. 10 - Waveform of the color error-detector.