|Home | Audio Magazine | Stereo Review magazine | Good Sound | Troubleshooting
Departments | Features | ADs | Equipment | Music/Recordings | History
Reprinted from RCA Engineer, July/August 1982. Copyright 1982, RCA Corporation.
by Charles Repka, Ed Freeman and Gopi Mehrotra--- both engineers at RCA.
RCA's SelectaVision CED (Capacitance Electronic Disc) videodisc system, which originally appeared in a monophonic version, is now available it a two-channel version suitable for both stereo and independent-channel bilingual use. In addition, its sound capacity has been enhanced to levels comparable with those of high-quality audio-cassette, reproducers.
[For an explanation of the basics of RCA's mono SelectaVision system, see Audio, March 1980.] Because of the basic purpose of the system, the design had to satisfy sever al constraints simultaneously. For ex ample, the system had to give good audio reproduction on typical home stereo component systems. The system had to produce acceptable monophonic and o on a typical television system.
Stereo discs and existing monophonic discs had to be forward and reverse compatible. The system had to be able to reproduce two independent audio channels for bilingual or special-purpose programs.
To ensure compatibility with mono players, the stereo signal is fed through an L + R, L-R matrix. This allows mono players to recover the compatible sum signal from stereo discs, while stereo players decode the sum and difference signals back into their original left and right components. When playing a mono disc, a stereo player behaves the same as a standard mono player (it does not apply the decoding matrix).
For bilingual applications, the encoding matrix and the decoding matrix are by passed and the two channels are re corded independently.
For adequate reproduction of high-quality stereo programs through a home system, stereo system goals included a minimum audio bandwidth of 15 kHz, distortion levels of less than 2%, and a dynamic range of at least 60 dB. Satisfying these demands required optimization of the audio-noise performance of the CED system.
The audio channels on the videodisc are recorded on FM carriers. The first channel, at about 716 kHz, is used for the sum or monophonic signal; a second channel, at about 905 kHz, is used for the difference signal (Fig. 1). As in any FM system, the demodulated signal-to-noise ratio is a result of the composite effects of the deviation, the de-emphasis characteristics, the carrier levels, and the magnitude of phase-noise sources.
The requirement for reverse compatibility with monophonic players and the tight frequency-division multiplexing of the audio and video FM signals severely restrict the choice of these parameters.
Reverse compatibility defines the maximum deviation on the sum channel as the same ±50 kHz defined for the monophonic signal. Consequently, changes in deviation large enough to affect signal-to-noise ratio are impractical.
Another direct way to improve audio S/N is to increase the level of the audio carriers on the disc. However, too great an increase produces proportional "sound beats" in the video luminance signal. A correction circuit in all CED players cancels these beats by adding an inverse beat . Extensive measurements of the sound-beat signal and the practical performance of correction circuits showed that audio-carrier levels could be safely raised 4 dB at the outer disc radius and 1 dB at the inner disc radius. Stereo substrates are cut with a linear "amplitude taper" that decreases toward the inner radius, and reflects the available increases.
Additional margin has been gained by improving the performance of the basic disc compound, partly as a con sequence of manufacturing gains made during the initial year of mono phonic production, and partly from formulation changes. The gains made using the above techniques apply to the noise attributed to the carbon in the disc. This noise is conveniently approximated for the audio channel as white noise, band limited by the FM-demodulation systems. It is uncorrelated from channel to channel.
The performance-improvements with respect to this noise source were sufficient to unmask another type of noise in the disc, displacement noise. Displacement noise arises when variations in the position of the disc surface with respect to the stylus electrode displace the estimates of zero crossings on the FM carrier. Such displacements can arise from surface roughness of the disc or vibration of the stylus. The underlying mechanisms cause this noise to be correlated in both carriers.
When matrixed by the stereo de coder, they reinforce one another in the left channel, and nearly cancel in the right channel.
Control of these noise sources proved extremely difficult. Cancellation of these noise components was potentially possible (the behavior of the right channel providing tangible evidence of this possibility), but fundamentally required adding random noise back into the system along with the cancelling information. Because of this trade-off, it was desirable to reduce the main contributors of such noise below detect able levels. A resonance in the cartridge was damped by a design change, reducing its contribution by about 10 dB at 10 kHz, and disc surfaces were improved by process changes. These improvements successfully lowered the "left-channel" noise back into balance with the "right-channel" noise.
To further improve stereo performance, noise-reduction systems were investigated. After evaluation of sever al systems, the newly developed CBS CX system was selected over competing alternatives. Although some other systems promised more noise reduction, CX encoding was attractive for several reasons: It was possible to foresee that the masking of noise by high-level signals could be made adequate on the disc, thus assuring acceptable performance of the CX system.
The CX system had been designed with a primary emphasis on compatible play of unexpanded material. This not only assured reverse compatibility of stereo discs on monaural players, but also solved the problem of presenting adequate dynamic range to stereo inputs, while restricting dynamic range on television r.f. inputs. Only the com pressed material is fed out the TV r.f. modulator, while the expanded material drives only the stereo outputs.
The CX system generally reduced the audibility of "ticks" and "pops"; some other systems exaggerated their effect.
A variety of economical circuit executions with interchangeable component sources are possible with CX.
Only inaudible waveform distortions arose from forcing the CX-encoded signal through an FM system which has a limiter.
A bonus arose from the decision to use the pre-emphasized signal in the CX control loop. Dynamic range in any basic FM system diminishes at a rate of 6 dB per octave beginning at the pre-emphasis break frequency. How ever, the action of the CX system reduced this to 3 dB per octave for steady-state signals. This characteristic suggested that additional pre-emphasis could be added in mastering, while still maintaining greater than benchmark dynamic range. Empirical work on a variety of program tapes confirmed this effect with complex audio waveforms, and a second break point was added at 6.1 kHz (25 µS).
Subsequent de-emphasis in the player contributes significant noise margin to the system (approximately 3 dB, A-weighted S/N).
CX Noise-Reduction System
The CX process provides up to 20 dB of noise reduction by means of a complementary compansion process.
Audio signals are compressed by a 2:1 factor (on a dB scale) during the disc-mastering process. During playback, the demodulated audio signals are expanded by a 1:2 factor (Fig. 2).
A simplified block diagram of the CX compression system is shown in Fig. 3.
Voltage-controlled amplifiers in the feedback loops of the left and right channels are adjusted by a control signal derived from the left- and right-channel outputs. Unlike other compansion systems (for example, Dolby or dbx), compression takes place with no spectral alteration. This characteristic permits playback of the compressed signal without an expander for compatible reproduction.
Compression occurs at a 2:1 characteristic only for signals greater than -40 dB with respect to a reference. All signals below-40 dB retain a 1:1 characteristic. For the CED system, the reference operating level is defined as ±25 kHz deviation of the FM sum-channel sound carrier, with a 1-kHz sinusoidal modulating signal applied to both left and right inputs in phase.
Audio signals below 100 Hz are attenuated by 6 dB per octave before being applied to the control-signal generator. This prevents gain pumping that can be caused by strong subsonic signals. System pre-emphasis is added to the audio signal before it is applied to the control-signal generator.
This improves system headroom (mar gin from standard signal level to maxi mum signal level before significant nonlinearity occurs), and noise-reduction characteristics, at high frequencies. The resultant compression-frequency response for single-channel steady-state sinusoidal-input signals is shown in Fig. 4. (Note that this response curve applies only to the effect of the compressor and pre-emphasis on individual sine-wave signals. Response with mixed-frequency signals, such as music, would essentially re main flat.) The transient response of the control-signal generator is nonlinear and has been designed to minimize audible side effects. A block diagram of the control-signal generator is shown in Fig. 5.
Filtered left and right signals are fed to the input cell where they are passed through ideal full-wave rectifiers to pro duce absolute-value-level signals. In addition, a minimum signal-level reference is generated. The largest of the level signals or the minimum level reference is rectified and passed through a complex network that generates control responses for rapidly rising, rapidly falling, or slowly changing audio-signal levels. The compressor and expander are the inverse of one another.
The Stereo-Mastering System
The audio portions of videodisc mastering systems have been redesigned to accommodate stereo quality levels. Figure 6 shows the audio block diagram of a videodisc mastering sys tem. Stereo audio programs can be supplied either on tracks 1 and 2 of the one-inch C-format tape or on a separate half-inch audio tape. Because of the limited audio performance of C-format VTRs (wow and flutter, especially at half speed, stereo signal phase tracking, and distortion), half-inch audio tape is the preferred format. Video and audio are synchronized by using the SMPTE time code, recorded on the videotape and on track 4 of the half-inch audio tape.
The mastering process takes place at half speed. The appropriate scale-factor modifications have been made to all tape machines, equalization net works, and time constants within the audio-signal path. From the tape machine, the audio signal is fed through a Dolby A decoder. (To ensure the highest possible quality, all stereo-program master tapes are Dolby A-encoded.) Following the Dolby A is a filter set.
The filter can be used to optimize the bandwidth of the audio chain to match the bandwidth of the program material.
Many film sound tracks, for example, have a bandwidth of only 8 kHz. By optimizing the audio chain to fit the program, system S/N can be improved by 2 to 3 dB. For high-quality programs with low source noise and wide band width, the filters are set for a 16-kHz bandwidth. The filters can also be used as notch filters to remove single-frequency spurious tones from pro gram sources.
After bandpass filtering, the signal is CX-encoded. The CX encoder used in the signal chain incorporates equalization circuits within the control loop that cause the encoder to compress the audio signal as if it were pre-emphasized. Actual pre-emphasis takes place later in the signal chain.
After CX encoding, the left and right stereo signals are added and subtracted in the stereo matrix to form sum and difference signals. The sum and difference signals are then pre-emphasized and limited to peaks equivalent to 100% (±50 kHz) deviation. Audio program material can have a peak-to-average ratio greater than 10 to 12 dB, which means that some form of signal limiting must be applied to prevent overmodulation. Limiting is applied to the sum and difference signals. This permits a higher average signal level before limiting as compared with conventional left/right limiting.
The limiter is a two-channel unit with the highest input to either channel con trolling the limiting action on both channels. This prevents changes in stereo separation when the limit circuits are active. Limiting is by means of gain reduction (rather than clipping) as much as possible, thus avoiding obvious distortion of extreme peaks.
The output of the limiter drives the audio modulators whose output is then summed with the FM-modulated video signal. The composite signal is amplified and used to drive a piezo-electric cutting head that scribes the signal onto a copper substrate [2 to 4].
The CED Stereo Player
The RCA SGT 200 stereo player is similar to the SFT 100 and SGT 100 monaural players in many respects.
The video circuitry, control circuitry and playback mechanism are identical on the three players. The only visible differences in the stereo player are the stereo-indicator light on the front, and the bilingual switch and audio/video output jacks on the back.
The main blocks of the complete stereo signal section consist of two de modulators, mute/track-and-hold circuits, and matrix and CX decoder circuits (Fig. 7). The 716-kHz L + R carrier is demodulated by circuitry on the main circuit board. The 905-kHz de modulator and other circuits are located on a stereo-board subassembly.
The stereo player uses the same custom-designed phase-locked loop IC used in the SFT 100 player to demodulate the audio and video signals .
However, the voltage-controlled oscillator gain has been reduced, and out put amplifier gain reduced by a similar amount, to improve the audio S/N. The loop filter has been modified to in crease the audio bandwidth.
The 905-kHz demodulator functions the same way as the 716-kHz demodulator with the exception of squelching capability when the carrier is missing.
With mono discs (no 905-kHz carrier), the 905-kHz demodulator IC puts a "low" on the squelch line and this function is used to turn off the stereo light and disable the decoder and CX ex pander. The above three functions are also disabled whenever the bilingual switch is moved from normal position to select either channel 1 or channel 2.
Defect gates within each demodulator are triggered by disruptions in the FM carrier signal which could cause undesired disturbances in the audio output. Such disruptions may some times include phase disturbances, amplitude dips or dropouts, and excessive noise.
When triggered, these gates signal the switching control circuit, which in turn triggers a track-and-hold circuit that follows each demodulator's audio output. This T/H circuit then maintains the audio signal level until the defect passes. There is some signal discontinuity when the hold is released, but far less than if the audio signal had dropped to zero or suddenly peaked at the defect.
This technique is very effective in reducing ticks and pops. However, if high frequencies are present in the software, a track-and-hold circuit can degrade the tick-and-pop performance. To avoid this, the track-and-hold circuit has a charging time constant of approximately 98 µS which limits the hold function to frequencies below 1.6 kHz while allowing the circuit to track the signal at frequencies above 1.6 kHz. A C-MOS quad-switch (CD4016) is used for this function.
These switches are also used for muting the player or selecting any language by muting one channel at a time.
The left and right signals are recovered by adding (L + R) to (L- R) and (L + R) to- (L- R) respectively. An inverter amplifier is used to invert the gain of the (L- R) signal. When independent audio channel 2 is selected, the gain of this amplifier is changed to + 1 to get in-phase signals at each output. Left and right signals from matrix output are also added back to get an (L + R) signal for the TV modulator.
The CX decoder is designed using low-cost, commonly available ICs. Two CA324-type operational amplifiers and one LM13700-type transconductance amplifier are used. The function of the CX decoder is to provide complementary expansion of the previously com pressed signal.
The decoder consists of a full-wave rectifier for each channel and a filter/ peak-detector that gives a d.c. output corresponding to the peak signal from both channels. This d.c. is amplified to the right level for proper dead-band operation. This d.c. voltage goes to a time-constant circuit and finally controls the current of the transconductance amplifier, thereby changing its gain. A precision limiter is used to ad just the knee of the -40 dB break point. A single-pole RC network is used at the output of the transconductance amplifier to provide necessary de-emphasis at 25 µS. The L and R expanded signals are then connected to the output jacks through the buffer stage and given an additional de-emphasis of 75 µS.
We thank our colleagues at the David Sarnoff Research Center in Princeton, and RCA SelectaVision VideoDisc Operations in Indianapolis, for the opportunity to be their spokesmen.
1. Gibson, J. J., F. B. Lang and G. D. Pyles, "Nonlinear Aperture Correction in the RCA VideoDisc Player," RCA Engineer, Vol. 26, No. 9, Nov./Dec. 1981.
2. Weisberg, H., "Manufacturing the VideoDiscs: An Overview," RCA Engineer, Vol. 27, No. 1, Jan./Feb. 1982.
3. Kell, F. D., G. John and J. Stevens, "VideoDisc Mastering: The Software to Hardware Conversion," RCA Engineer, Vol. 27, No. 1, Jan./ Feb. 1982.
4. Brandinger, J. J., "The RCA CED VideoDisc System," RCA Review, Vol. 42, No. 3.
5. Pyles, G. D. and B. J. Yorkanis, "VideoDisc's Video and Audio Demodulation, Defect Detection, and Squelch Control," RCA Engineer, Vol. 26, No. 9, Nov./Dec. 1981.
(Source: Audio magazine, Feb. 1983)