The microphony would have been less apparent in a typical TV receiver. Its effects would have been more-or-less cancelled in the intercarrier formation process, given that the vision carrier was equally affected.
Indeed, though I didn't appreciate that at the time.
Do you think that the advent of frame-grid valves reduced microphony? Or did inter-carrier sound solve the problem anyway?
I think that we may reasonably deduce that the frame grid TV valves, when they arrived, had a minimal microphony problem, as they were also used in FM receivers. But the same was true of prior valves developed for TV applications. That the 6J6, developed in WWII more for radar applications, had a microphony problem only became apparent when it was deployed in TV and FM applications. That evidently spurred GE to develop the 12AT7. And it is reasonable to assume that the valves that RCA developed for “high IF” TV receivers, the 6CB6 pentode, 6BQ7 cascode, and 6X8 triode pentode were similarly free from any deleterious level of microphony. All were used in FM as well as TV receivers. Although the 6X8 was primarily a TV frequency changer, the pentode mixer limiting regeneration caused by the proximity if the IF to the lower Band I channels, it was also specified for used as an FM Frequency changer (with the option of triode-strapping the mixer) and an AM frequency changer (triode pentode mode). That was in fact RCA’s answer to the 12AT7, which as well as its TV and FM roles, was also presented for use as an AM frequency changer in FM-AM receivers. In respect of FM performance at least, that was a seen as a better choice for a combined frequency changer than a pentagrid. RCA had backed the latter approach, positioning the 6BE6 as an FM as well as an AM frequency changer, then introducing the 6SBY7-GT as an improved pentagrid, followed by a noval version of it, the 6BA7. Nonetheless, as an interim measure to meet GE head-on, it did issue an application note showing how the 6J6 could be used as a combined FM-AM frequency changer.
When developing the 6X8, RCA probably assumed that “high IF” TV receivers would almost all be of the intercarrier type, so for the TV application alone, it probably did not have to worry too much about the microphony characteristics of the triode oscillator section. But for the FM case, that was a concern that did need to be addressed. In that it was likely that TV front end valves would be used in FM receivers, even if the latter was not a “headline” application, then microphony probably should not have been ignored. But given that the 6X8 was pitched for FM-AM applications, a bigger market than FM-only, then the issue definitely could not be ignored.
Even so, as you say, the widespread adoption of intercarrier sound did solve the problem of oscillator microphony for TV receivers, and thus allowed continued use of the 6J6 as frequency changer for economy applications.
One frame-grid valve whose primary application was FM receivers was the 6JK8, comprising a frame grid RF amplifier and a conventional triode autodyne mixer, developed for early FM stereo receivers. The US industry was late in adopting single-valve front ends (although that was probably a situation where “better never than late” might have applied), but when it did, the concept was developed along three or four different vectors. Microphony in the RF amplifier was probably less of an issue than in the oscillator, but it does show that there was no fear of frame grids when FM was primary.
Cheers,
Steve
I have hardly ever observed objectional microphony with VHF-FM receivers, except under fault condition. Perhaps the equipment manufacturers took care to mechanically isolate the oscillator valve?
Or maybe it's more likely that the higher oscillator frequency at UHF causes a bigger deviation with vibration.
Returning to the QSS and its origins, this 1982 November IEEE paper provides some background:
“The German 2-Carrier System for Terrestrial TV-Sound Transmission Systems and Integrated Circuits for ‘High-Quality’ TV Receivers”, by Ulf Buhse, Valvo Applications Laboratory, Philips Hamburg.
It started with a discussion of the German (IRT) Zweiton stereo/dual sound system, with a second section on high-quality TV receivers.
Following a brief summary of the shortcomings of the conventional intercarrier sound system, the split sound system was considered then dismissed with the following commentary:
“At first glance, an obvious solution appears to be to split the vision and sound channels immediately after the tuner as shown in Fig. 6.
“The vision and sound channels are then entirely separated and each can be optimised for performing its own function. Theoretically this system would be ideal for hi-fi sound reproduction but, in practice, there are several drawbacks. For example, the system is prone to local-oscillator instability. It needs a special tuner, which is up till now not available.”
That ignored the fact that the Japanese manufacturers were already producing split-sound TV receivers, so had presumably overcome whatever perceived difficulties they presented. Also, in Europe, TV receivers for the French systems E and L were of the split-sound type. One could argue that the oscillator stability requirements for an AM sound channel might be less stringent than for an FM sound channel in that normal diode demodulators were essentially untuned and were insensitive to spurious PM. But that was not the case where the AM sound IF channel included a quasi-synchronous demodulator with a tuned tank circuit, as was the case with the TDA2543 IC.
Be that as it may, the paper moved on to describe the quasi-split sound (QSS) system, including the various requirements to ensure that it worked properly.
One point made was that in the QSS channel, the phase steepness needed to be the same at the vision and sound carrier frequencies, in order that any oscillator jitter produced equal phase jitter for each. If not, any oscillator jitter would result in phase jittering that was different for the sound and vision carriers, this showing up as spurious phase modulation of the sound carriers.
The benefits of quadrature demodulation in the QSS channel were also noted. That is, the limited vision carrier used for the creation of the intercarriers is phase-shifted by ninety degrees, which means that it would not demodulate its own double-sideband AM.
In summary it was said that the quasi split sound system with quadrature demodulation bore the possibility for hi-fi sound reproduction, and that it was a variant of the intercarrier method using the advantages but avoiding the disadvantages.
Following that there was a discussion of actual TV receiver circuitry, using the TDA2545A and TDA2546A QSS ICs, which included the previously discussed features, including quadrature demodulation. The TDA2545A was the basic QSS IC, whilst the TDA2546A added one FM intercarrier channel, nominally at 5.5 MHz.
For a more detailed treatment of QSS theory, the reader was referred to this paper:
H. Achterterg, U. Buhse, H. Schwarz:
Aufbereitung des Fernsehtonsignals mit den integrierten Schaltungen TDA 2545 und TDA 2546 nach dem Quasiparallelton-Verfahren
(Quasi-Split Sound Processing with the Integrated Circuits TDA 2545 and TDA 2546)
Valvo Entwicklungsmitteilung Nr. 79, Nov. 1980
I have not found that paper. But I think that it is reasonably inferred that QSS was “invented” by Valvo, and that the TDA2545 and TDA2546 were the first QSS ICs.
(As an aside, I have some Valvo papers (in English) from about the same period, and they are very comprehensive, multipage documents Without trying to find them, my recollection is that they covered the TDA1072 (AM), TDA1074 (AF) and TDA1028/TDA1029 (AF) ICs. Thus I imagine that the TDA2545/TDA2546 paper would be similarly comprehensive.)
I have had a subsequent thought about the apparent German preference for QSS over split sound. And that is that QSS might be simpler where two sound carriers were involved, as with the Zweiton system. With appropriate care over group delay, etc., both can pass through the same QSS IC without adverse interaction, and both may be demodulated by the same limited and phase-shifted vision carrier to produce two easily separated intercarriers (at 5.5 and 5.742 MHz for system B.)
The Japanese approach to split sound was to convert the (single) sound IF down to 10.7 MHz and then process in standard FM IF ICs, usually with AFC feedback to the 10.7 MHz conversion oscillator. Perhaps it could have been done with two sound carriers (33.4 and 33.158 MHz), converting to 10.7 and 10.942 or 10.458 MHz, but those numbers look to be a little close for separation with standard FM filters in the 10.7 MHz range. A noticeably lower second sound IF would have helped there, but then it would have been non-standard. And if one postulated that to make separation of the two sound carriers sufficiently easy, it would be preferable to move to a second IF around half of 10.7 MHz, that is in intercarrier territory anyway.
Incidentally, in the valve TV receiver era, second sound IFs, where used, had to be chosen carefully in order to avoid interference effects, and evidently 10.7 MHz was a non-starter. With solid state equipment, probably more so in the IC era, economic inter-module isolation was less of a problem, and given adequate screening, 10.7 MHz could be used without problem.
Cheers,
Steve
Some further insights into QSS performance were provided in another 1985 IEEE paper:
System And Realization Aspects of Multistandard TV Receivers
Wolfgang Weltersbach
Philips Group of Companies, Hamburg, West Germany
The paper addressed high-performance multistandard receivers, including multichannel sound facilities, with a particular focus on IF strip requirements.
In respect of the sound IF, it was said that split sound would be best from a quality viewpoint, but that it made high demands on the tuner and tuning system, which could not be addressed in an economic way for consumer TV sets of the time. (That assertion was debatable, but it did seem to be the European position at the time.)
That being so, QSS was seen as the optimum system, and the basic multistandard approach was shown in block schematic form:
Early implementation of QSS was done using quasi-synchronous (passive carrier regeneration) quadrature demodulation, e.g. as with the TDA2545 IC. But improved results were obtained by using active carrier regeneration, using a PLL, as shown in this block schematic:
The performance comparison was shown in this chart:
The active carrier regeneration case performance was shown with and without the tuner, indicating that the latter was responsible for some degeneration. The upper curve, without the tuner, was said to be close to what was achievable with fully split sound. That does illustrate that the use of split sound required a high performance tuner if the latter was not to be the limiting factor. It may also be seen that in the QSS active carrier regeneration case, the tuner was limiting, whereas in the QSS passive carrier regeneration case, it was not. Regular intercarrier performance was not included, probably on the basis that it had no relevance in a discussion of high quality circuitry, but perhaps also because it would not have looked very good.
The multistandard QSS IC with passive carrier regeneration was realized as the TDA3845, although it might have been subsequent to the above-mentioned paper. The earliest mention of that IC that I have found is 1989, but that was from a casual rather than an exhaustive search.
Whether a multistandard QSS with active carrier regeneration was ever developed I do not know. For the paper, the laboratory model was realized using available mixers, VCOs, and opamps.
The question of passive vs. active carrier regeneration for the vision IF channel demodulation function was also considered, with passive chosen basis the following analysis.
“Initially the PLL looks very attractive, it improves the performance of the video signal. However, depending on the application field, some drawbacks have to be taken into account:
- lock-in problems for special picture contents (side band locking)
- threshold for weak IF signals, pull-in problem
- interferences or beat frequencies due to the oscillator.
“With a sophisticated circuit design and special tuning systems these problems may be solved in TV sets. Specially a European multistandard receiver for positive. modulation system (L) adds one severe problem more. The PLL has to handle a zero carrier signal during synch pulses.
“For the time being the European setmakers are sticking to the conventional quasi synchronous detector as far as is known to the author.
“In the US market (standard M) the PLL aspects look more attractive, because the linearity of the demodulator offers advantages for NTSC.”
The final comment connects with an upthread commentary about the Zenith/National Semiconductor LM1822/3 vision IF with fully synchronous demodulation.
Interestingly, the quasi-synchronous case is shown with a ±0.75 MHz narrowband filter in the reference channel ahead of the limiter, so that the latter captured only the DSB part of the vision signal. In practice, this was usually (mostly?) omitted, with some reference bandwidth trimming done by the demodulator tank circuit. But this post facto operation would not undo the phase modulation damage done by limiting a sideband asymmetric signal. It was noted that the delay caused by the narrowband filter would need to be matched by a suitable delay in the wider bandwidth signal path, and I imagine that this might have been a stumbling block at the consumer product level. Not mentioned was that if the reference signal had been through the main vision IF filter and so subjected to the Nyquist slope, then there would also need to have been an “anti-Nyquist” filter, perhaps obtained by suitably sloping the top of the ±0.75 MHz filter.
Back to the main theme, we can say that the QSS technique was fairly well defined in terms of key points. Ahead of the main vision IF bandpass filter, the feed from the tuner was split, with a second branch carrying both vision and sound carriers passed through an appropriately shaped double-peaked filter, amplified (with gain control) together, the sound carrier then being synchronously demodulated by the limited (and preferably quadrature phase-shifted) vision carrier, to produce the intercarrier(s).
Against that, any intercarrier-producing technique that used the vision and sound carriers after they had passed through the main vision IF bandpass filter (usually including some relative suppression of the sound carrier) would fall into the regular intercarrier class. That group covered a multitude of sins. Way down in the basement, as it were, would have been the case where the intercarrier was not separated from the video baseband signal until after the video amplifier, as was found in some valve-era receivers. At the top end, and perhaps performance-wise matching QSS with PLL demodulation, was the technique found in some late vision IF processing ICs, such as the Motorola MC44302A multistandard unit, mentioned upthread. This used PLL synchronous demodulation following a common IF filter and gain strip, but separate demodulators, in-phase for vision and quadrature for intercarrier, in the latter case for AM as well as FM sound.
Separate vision and intercarrier demodulators at the end of a common IF strip was not new, and it was found in the valve era where separate diodes were sometimes used. Early in the IC era, the Motorola MC1331, derived from the MC1330 quasi-synchronous vision demodulator, had separate intercarrier demodulator, albeit that it was of the in-phase type.
QSS may be seen as an inspired sidestep at a time when the conventional thinking was that the only way to avoid the limitations of conventional intercarrier sound was to go to split sound. Something it also brought along was a “proper” implementation of quasi-synchronous demodulation, with a carefully conditioned reference carrier. This had certainly been mooted in respect of quasi-synchronous vision demodulation, but as far as I know seldom done in consumer ICs. Rather, improvement was achieved by going to PLL fully synchronous demodulation, which technique was also segued into QSS to take the latter close to the split-sound performance level.
Cheers,
Steve
As always, very interesting. I suppose we in the UK should count ourselves lucky that the Beeb developed NICAM.
Thanks, Sundog.
My impression is that NICAM was somewhat better than either Zweiton or BTSC.
In the past I had “daily” exposure over several years to each of the three, sequentially BTSC in the USA, Zweiton in Australia and then NICAM here in NZ.
That said, the comparisons, being well-spaced sequentially, were hardly rigorous. Although the audio electronics and the speakers were the same throughout, the listening rooms were much different, as also were the “front end” electronics. In the USA I used a Luxman T407 outboard TV tuner, which had split sound. In Australia and NZ I used available reasonably high quality VCRs or DVD recorders as the source. Presumably these had QSS sound channels, but probably not aimed at as a high a target as would be dedicated outboard tuners. Even so, I’d rank NICAM as the best, followed by Zweiton and BTSC.
There was an oddity that cropped up in one Sydney location where we lived for a few years. We were in (or rather on the edge of) a ravine that was just down the hill from the Gore Hill ABC channel 2 transmitter, so nominally in a high signal strength location, although because of the topography, it was not line-of-sight. Stereo TV sound reception could be flakey at times, jumping in and out, but on the other hand, neither mono sound nor vision were at all noisy. Anyway, I cranked up the ICOM R7000 and connected it to the TV aerial, to have a look at signal strengths. Voila! Vision and the 1st sound carrier were good, but 2nd sound was quite low and varying. I guess that the combination of direct (or not quite direct) and reflected waves down in the ravine conspired to produce a relatively high Q dip right over the 2nd sound carrier. As I recall, making a slight change in the aerial pointing direction brought the 2nd sound carrier to a level that was satisfactory most of the time.
In the USA, I was also able to make a comparison between BTSC and FM Multiplex. The local (DFW) PBS TV and FM stations continued the practice of occasional simulcasting for a while after the TV station had started with BTSC. FM Multiplex was definitely better (I was using a Carver TX-11a tuner, fairly well thought of at the time). For example, some caption buzz was apparent on the TV sound, even via the T407 in split-sound mode, and the noise floor was slightly higher. I suspect that there was more compression on the TV sound, as well, whereas the PBS radio station prided itself on using minimal compression.
All anecdotal, of course. The only encounters I had with the Korean version of Zweiton and the Japanese FM/FM TV stereo sound systems were with hotel TV receivers, from which little could be gleaned.
Regarding NICAM reception techniques, as far as I know, some form of intercarrier technique was almost always used, probably quite often with QSS. I imagine that NICAM may have been more robust against the receiver RF/IF path degradations, but perhaps not completely so. That might need a little more light research.
Cheers,
Steve
Whether a multistandard QSS with active carrier regeneration was ever developed I do not know. For the paper, the laboratory model was realized using available mixers, VCOs, and opamps.
In fact Thomson Consumer Electronics R&D Laboratories described a QSS IC with PLL vision carrier regeneration in a 1991 IEEE paper “Advanced Multistandard TV-Sound IF Integrated Circuit”, by M.A. Reiger and R.K. Koblitz.
Here is the block schematic for the IC (whose commercial designation was not provided):
This IC also provided for AM sound (for systems L and L’), which was routed through the QSS IF amplifier stage, but demodulated separately. The nature of the AM demodulator was not disclosed, other than that it appeared to be untuned.
The FM intercarriers were available for external further processing, but internal processing was also provided. Another PLL provided a second conversion from 4.5, 5.5, 6.0 or 6.5 MHz to 500 kHz, at which frequency the intercarrier was subjected to integrated bandpass filtering and demodulation, the nature of the latter not being given. The Zweiton 2nd FM carrier at 5.74 MHz was converted to 260 kHz, and similarly processed.
It looks as if the IC was intended for use in multistandard receivers that used a 38.9 MHz vision IF (VIF) for all systems, or at least all systems except L’. Different sound IFs for systems L and L’ would not appear to have been a problem; both could be run through the QSS IF amplifier to the AM demodulator, which as said, appears to have been untuned.
It was also said to be suitable for NICAM sound, with the intercarrier output used to provide the NICAM intercarrier for external processing. A NICAM decoder IC suitable for both the system B,G,H and system I variants was described in another 1991 IEEE paper from SGS-Thomson Microelectronics, “NICAM Decoder for Digital Multichannel TV Sound Broadcast”, by Gary Shipton and Godfrey Onyiagha. However, nothing was said in relation to the “quality” of the NICAM intercarriers delivered to the IC.
In the case of the QSS IC, it is reasonable to assume that the choice of PLL carrier regeneration was made in the interests of obtaining demodulated FM sound quality that was close to that of a split sound system.
Something not mentioned, but perhaps a valid question, is would there have been any problems with using a PLL-type QSS IC and a PLL-type synchronous demodulation VIF IC in the same receiver. That would imply the presence of two same-frequency (say 38.9 MHz) PLLs in close proximity. I suppose that it might have been simply a question of providing adequate screening and perhaps physical separation of the respective ICs. With quasi-synchronous ICs, there would be similar-frequency external tank circuits, quadrature for the QSS IC and one in-phase and one quadrature (AFC) for the VIF IC, and it appears that avoiding any unwanted mutual coupling was quite doable.
Still, one wonders if Motorola’s choice of a combined VIF/SIF in its MC44302A, and yet another approach by Siemens (to be described in following posting) were to some extent an effort to avoid same-frequency PLL duality.
Cheers,
Steve
The previously mentioned Siemens approach to maximizing sound quality with the QSS system was described in another 1991 IEEE paper, “A World Standard Video and Sound IF IC”, by R, Heymann and H. Kriedt.
In this, the VIF and SIF had separate amplifiers within the IC, so that before demodulation, it was akin to a split sound receiver. The VIF amplifier was followed by a PLL demodulator. The SIF amplifier output was split into two pathways. One went to an AM demodulator of the untuned quasi-synchronous type, for systems L and L’. The other went to a mixer whose other input was taken from the PLL, essentially a quadrature reconstructed unmodulated vision carrier. The mixer output was thus the intercarrier.
Here is a block schematic:
The sound section is shown in more detail here:
As in the Thomson case, the IC functions include processing of two FM channels through to demodulation, again using another frequency conversion. In this case though the second conversion VCO had a range of 10 to 14 MHz. The final FM IFs were not given, but it was said that the second conversion was not necessarily used on one of the four incoming frequencies (4.5, 5.5, 6.0 and 6.5 MHz). From that one may deduce that any one of the four could be chosen as the final IF, which would thus be in the range 4.5 to 6.5 MHz. If 4.5 MHz were chosen, then the VCO frequencies for the three others would be respectively 10.0, 10.5 and 11.0 MHz. In the 6.5 MHz case, they would be 11.0, 12.0 and 12.5 MHz. The Zweiton second FM carrier would be 242 kHz below the final IF. The two final FM IF filters were external to the IC, as were the tank circuits for the quadrature demodulators. One may suppose that the modal choice was a 5.5 MHz final IF, second carrier at 5.74 MHz, so that standard filters could be used. That would have required VCO frequencies of 10.0, 11.5 and 12.0 MHz for the 4.5, 6.0 and 6.5 MHz cases.
There does not appear to have been a pinout for the raw intercarrier, which suggests that this IC was not intended for use with NICAM. However, that is something that may well have been remedied in the production version.
The vision PLL was actually described as being an FPLL, or frequency phase-locked loop. It had an AFC circuit operating in parallel with the PLL itself, this to allow a wide (5 MHz) capture range but a suitably narrow loop bandwidth. The AFC circuit was switched out when the loop locked. The FPLL was in fact an earlier Siemens development, in part aimed at obtaining some improvement with conventional intercarrier sound. I’ll elaborate in a future posting.
Best intercarrier buzz performance was said to have been achieved with a PLL bandwidth of 15 kHz. But because of the use of inexpensive standard tuners it was necessary to set the bandwidth to 100 kHz to avoid the effects of vision carrier phase modulations within the tuner such as oscillator pulling or phase noise of the varactor diodes. That resulted in a reduced FM sound signal-to-noise ratio. If nothing else, this confirms that the tuner could be a limiting factor.
The quadrature FM demodulators were a departure from normal, in that they interposed an amplification stage between the tank circuit and the multiplier. This was said to allow the use of a lower quality (meaning, I think, lower Q) tank circuit, with the net result of lower distortion.
That was an interesting development, perhaps attending to what might have been a previously neglected area. I suspect that the intercarrier quadrature demodulation ICs used in many TV receivers were similar to the early types, such as the TBA120 series, and were not necessarily optimized for low distortion. In the FM radio field, the progression had been somewhat different after the initial flush of ICs. The RCA CA3089 of 1971 was a step change, and it was intended for use with a double-tuned quadrature coil where very low distortion was required. It set the pattern for subsequent ICs from multiple makers that offered improvements along various vectors. Another important marker was the National LM1965 of c.1984, which inter alia included a feedback circuit that allowed the achievement of two-coil linearity with a single quadrature coil. The Siemens circuit appears to have been a different way of achieving a similar result. Some of the Japanese split-sound TV tuners of the 1980s, in which there was a second conversion to 10.7 MHz in the sound channel, used standard FM IF subsystem ICs.
Cheers,
Steve
The simultaneous use of PLL demodulation in both the QSS and VIF pathways, with separate PLLs operating at the same frequency, might have been seen as potentially problematical. Probably it would have required separate QSS and VIF ICs, and careful component layout to avoid mutual interference. I am not sure if in fact it had been done before Sanyo developed SSP.
I have now found and read the 1947 Parker article mentioned in my initial posting, namely: Parker, L.W., TV Intercarrier Sound System, Tele-Tech., 6:26 (October, 1947).
The article was billed as being a relatively simple explanation of the intercarrier technique, and its overall tone was on the optimistic side.
Apparently the technique had been proposed by Parker a few years previously. That would explain why Parker’s work seems to have taken precedence over that of Dome. Also, the “intercarrier” name was applied by the FCC. Previously it was referred to a “difference frequency” system.
Cheers,
Steve
