Amplifiers and AVC

[Nick_B333]

After reading up on valve superhets and the related valve data sheets one thing puzzles me - AVC should supposedly be applied to g1 of remote-cutoff pentodes but not sharp-cutoff. I've been looking at IF amplifier circuits, mainly the "All American Five" sets and found that some with 1U4 (sharp) have AVC and conversely others with 1T4 (remote) don't. All of them do have AVC applied to g1 of the frequency converter though.

Is AVC going to work poperly on a sharp cut-off?

[kalee20]

Yes, but not very well.

A sharp cut-off valve will have a much more abrupt reduction of gain as the bias is increased. More subtley, the signal handling capabilities are rather less than a vari-µ stage, and unless the relevant valve is at an early stage in the receiver, it could well get overloaded with strong signals as the AVC voltage will bias it back, exactly towards the region of the grid characteristic where distortion could occur. In an RF or IF stage, this could well result in spurious responses.

[Nick_B333]

I saw that in a design Rowley One: 1R5 1U4 1S5 1S4: A.Rowley from an article in Popular Science. I guess it didn't work very well then unless it was changed to a 1T4!

[JHGibson]

A sharp cutoff valve can be used in a IF amplifier if there is a good sized screen decoupling resistor. This circuit arrangement gives the valve a semiremote cutoff characteristic.
John.

[G8HQP Dave]

You can use AVC on a sharp cutoff valve, but you do need a large screen resistor and it will never be as good as a remote cutoff valve. The large screen resistor means that as the valve current reduces the screen voltage rises and "lengthens the grid base".

It puzzles me that people sometimes use the "wrong" valve. For example, EF93/6BA6 (remote cutoff) as a BFO. For some reason, Heathkits often used sharp cutoff with AVC for an IF stage (e.g. pentode of ECF82) - saving a valve envelope, there are few remote cutoff pentodes sharing with a triode?

Mixer/frequency changer valves are often remote cutoff on the signal grid (e.g. 6BE6, ECH81).

[Nick_B333]

Zenith Transoceanics seems to be another set that do this with 1U4s. Just about everything I've seen applies AVC to the frequency changer though.

Thanks for everyone's comments, I'll continue reading up on this...

[kalee20]

Originally Posted by Nick_G4IRX

Just about everything I've seen applies AVC to the frequency changer though.

That's a subject in its own right!

While applying AVC to the frequency changer is great because it's only handling small signals, so distortion/intermodulation will be much less of an issue, there is the undesirable effect that applying AVC can result in oscillator frequency shifts.

This tends to be worse in 'single' valves (heptodes and octodes), where the oscillator steals a few electrons out of the main stream. However, even in 'twin' valves (triode-hexodes or triode-heptodes), when the mixer section is biased back it can put different loading on the oscillator- resulting in a shift.

As this is most pronounced on short-wave, sometimes, the frequency changer does not have AVC applied for this waveband. I'm now going to check all my radios for this...

[G8HQP Dave]

Yes, frequency changer AVC is best avoided on SW but you might get away with it for broadcast reception on the lower end of the HF band.

[Synchrodyne]

This thread prompted a look through several circuits on hand for examples as to how AGC was arranged.

Firstly, a 1950s Mullard 405-line TV RF/IF system arranged for black level AGC. The full valve line-up is PCC84, PCF80, EF85, EF80, EF80. AGC is applied to the PCC84, with 3.5 V delay, in full to the EF85 1st IF stage, and partially (around 38%) to the 2nd IF EF80, but not to the mixer or to the EF80 3rd IF. So here is an example of AGC being applied to a sharp cut-off valve. The EF80 gain swing was from 25 dB maximum to 5 dB at -5 V AGC. The EF85 gain swing was from 23 dB maximum to -12 dB at -13.5 V AGC. Mullard did not state why it chose to apply AGC to a sharp cutoff valve, but perhaps the EF80 was better overall than an EF85 in the 2nd IF position, and the extra 20 dB of AGC control range with the available control voltage was desirable.

Now looking at a range of MF and HF receivers:

Some Eddystone HF receivers, which show some interesting variations:

The 670C has delayed AGC applied in full to RF and IF stages (both UAF42). The frequency changer (UCH42) has full AGC applied on the lower frequency bands up to 5.5 MHz, but not on higher frequency bands, in order to prevent oscillator pulling. The 840C is similar, with the division point at 5.2 MHz.

The more sophisticated 940 has delayed AGC applied in full to both RF stages (ECC189 and EF93/6BA6), the mixer (ECH81) and both IF stages (EF93). In this case it seems that oscillator pulling is avoided by using a separate oscillator valve (EC90/6C4) rather than the ECH81 triode. The 850/4 VLF model has delayed (22 V) AGC applied in full to the RF and both IF stages (all EF93), but not to the frequency changer (ECH81). Although oscillator pulling would be less likely at low frequencies, presumably there was enough AGC control range without including the mixer.

Moving up to the 830/7, delayed AGC is applied in full to the RF (ECC189), 1st mixer (EF95/6AK5), 2nd mixer (ECH81) and both IF (EF93) stages. Both mixers have separate local oscillators, so would seem to avoid the pulling issue that way. In the case of the ECH81 2nd mixer, the triode is used as an oscillator buffer.

And up again to the 880/2, delayed (15 V) AGC is applied in full to both RF amplifiers (ECC189 and EF93), the tuned IF amplifier (EF93), and the first two of the three 500 kHz IF amplifiers (EF93s) but not to the 1st and 2nd mixers (both EF95s) and the 3rd 500 kHz IF stage (EF91/6AM6). Evidently 5 controlled stages were enough, so that Eddystone had the flexibility to avoid applying AGC to either mixer or the final IF stage, perhaps something of a textbook approach. As an aside, it looks as if some features developed for the 880 series (ECC189 cascode 1st RF, EF95 mixer, EK90/6BE6 self-oscillating SSB demodulator) were selectively cascaded down to the 830 series, 950 and 850 series models.)

The above group contains just one example of AGC being applied to a sharp cut-off, namely the EF95 1st mixer of the 830/7. This has 0.05 and 0.001 µF screen decoupling capacitors in parallel, whereas in the 880/2, both EF95s (without AGC) have 0.05 µF capacitors only.

The Bush EU.24 export table radio tunes up to 30 MHz. Delayed AGC is applied in full to the RF (UF42) and IF (UF42) stages on all bands, and to the mixer (UCH42) on all except the highest band (20 to 30 MHz).

The Chapman S6BS and S6 tuners, which tune up to 26 MHz or so, have delayed AGC applied in full to the RF (EF89), FC (ECH81) and both IF (EF89) stages. In this case though there is a separate AGC IF stage, fed from the anode of the 2nd IF, being an EBF80 without AGC applied but with negative feedback. Simultaneously this arrangement (used all of the Chapman valve tuners) would seem to eliminate differential distortion, and modulation rise in the final IF stage that would otherwise indicate not applying AGC or at least reduced AGC voltage. Given that it is a hi-fi tuner, attention to these items would be important; it also had a low-distortion demodulator. On the other hand, oscillator pulling was probably of less consequence in a domestic receiver used for shortwave program listening than in an HF communications receiver. And maybe the ECH81 was better than its predecessors in respect of susceptibility to oscillator pulling? (Early versions of the S6BS and S6 probably used an ECH42, like the S4 and S5 models, but I haven’t been able to confirm this.)

The Armstrong RF103/3 which with a single SW band probably tunes up to around 18 MHz, has delayed (1.2 V) AGC applied in full to the RF (6F15) and 1st IF (6F15) stages, and to the FC (6C9) on LW and MW only, not on SW. No AGC is applied to the 2nd IF (6C15) on any band. The EXP125/2 from the same period (circa 1950) tunes to around 26 MHz; I do not have a schematic, but Armstrong publicity material describes it as having delayed AGC applied to the RF and both IF stages, proportionately arranged to give optimum control, so it was different to the RF103/3. It would be interesting to know what Armstrong used on the late BS125T and RF125T models, but circuit information on these seems to be unobtainable.

The Quad AMII tuner, which tunes up to around 18 MHz, applies delayed AGC in full to the RF (EF89) and FC (ECH81) stages, but only about a quarter of the AGC voltage is applied to the IF stage (EBF89), to minimize modulation rise. This stage is neutralized, and probably operates at relatively high gain. The AGC delay is said to be quite small, in the interests of minimizing the impact of differential distortion. The earlier Quad AM (aka AMI) covered MF and LF only, and lacked an RF stage. Delayed AGC was applied in full to the FC (12AH8) and IF (EF93) stages; the latter was of the negative feedback type which I think would help to minimize modulation rise when operating at significantly reduced gain. With just two stages that could be subject to AGC control, reduced AGC voltage on the IF stage might not have been an option.

The Dynatron T139 tuner/control unit, which tunes up to 30 MHz, does somewhat differently. Delayed AGC is applied in full to the RF stage (W77/EF92) and the mixer (X79), whose triode was inactive, there being a separate L77 oscillator. Neither of the IF stages (both W77) has AGC applied. They both have negative feedback, partial in the case of the 1st IF. There is an RF pre-stage, Z77/EF91 used as a grounded-grid triode, employed above 10 MHz and on all bandspread SW bands; this does not have AGC. The Dynatron view may have been that two controlled stages were enough for a domestic receiver used for program listening, where a communications-type AGC curve was not needed.

Finally, the Marconi Atalanta marine receiver is different again. Delayed AGC (7 volts) is applied partially to the 1st RF stage (EF85) and fully to the 2nd IF stage (EF85 also). Simple AGC is applied to the 1st IF stage (W77). AGC is not applied to the 1st mixer (ECH81 with separate oscillator), the 2nd FC (ECH81) or the 2nd IF stage (Z77).

Thus there is quite a bit of diversity, apparently somewhat dependent upon what weightings were applied to the various parameters (AGC range, oscillator pulling, distortion, etc.) and how many stages were available for control – more meaning greater opportunity to optimize any one without too much compromise on the others.

Cheers,

[Synchrodyne]

The discussions on AGC that have recently come up in the Eddystone 888A thread (https://www.vintage-radio.net/forum/...d.php?t=120853) and the Valve Questions thread (https://www.vintage-radio.net/forum/...146#post806146) reminded me of this old thread, whose title is delimiting and so does not confine it to valve technology or to specific receivers, thus allowing a broader-ranging discussion on AGC generally.

Hence it seemed to be an appropriate place to record some items and issues on the rather big subject of AGC that have been brought to mind by the above-noted recent discussions. What follows essentially observational, and empirical not theoretical, much “what” but little “how” and “why”, and rather “lumpy” in that some items are mentioned only briefly and some treated in more detail, without any correspondence to their real significance. Plus of course there are no doubt major gaps. But of nothing else, it does give some indication as to how big this subject really is.

Firstly looking at the valve era:

A treatment of AGC could be divided into AGC generation and AGC application. Both I think are important, and the details of AGC generation can be quite intricate in the TV case.

AGC Generation: For AM radio receivers, AGC obtained from rectified signal (essentially the carrier) was most common, but amplified AGC was also used. In this case, the amplification could be provided at IF by a sidechain amplifier, followed by rectification, or by a DC amplifier after rectification. E.g. the GEC BRT400 used a grid-leak detector for AGC, which was configured to provide usable DC gain by having the top of the anode resistor at ground and the bottom of the cathode resistor at negative HT.

For FM receivers, AGC, where used, was usually obtained either from a limiter grid, or from across the capacitor of a ratio demodulator.

SSB receivers had their own arrangements, particularly in respect of time constants. For the SSB/ISB pilot carrier case, the AGC was usually derived from the pilot carrier IF amplifier chain, although some such as the Marconi HR24 had the choice of both pilot carrier and sideband AGC. Receivers for suppressed carrier SSB had sideband AGC, usually derived by rectifying IF.

TV receivers varied, but probably the modal approach was to provide both gain and gating, the latter both to access black level in the positive modulation case and to minimize the effects of noise impulses in the negative modulation case. Whilst negative modulation was originally chosen primarily because of the assumed ease of applying AGC, in practice noise protection was found highly desirable on the VHF channels. The gating and gain functions could be and were often combined. Gating could be done either from the signal itself, e.g. with differentiated separated sync pulses, or by using line flyback pulses. Gain might be obtained by amplifying the gated-out agc pulses and then rectifying them, or by rectifying line flyback pulses in a grid-controlled rectifier. The latter allowed the line flyback pulses to both provide the energy input to the AGC system and to do the timing. The Mullard high-gain sync-cancelled system was effectively self-gated; the line flyback pulse provided energy, but not timing. It was also inherently noise-cancelling. Noise-cancelled sync separators using dedicated valves (6BY6, 6CS6) had been introduced in the early 1950s. In 1957 arrived the 6BU8 “Siamese” twin pentode, in which one half did the noise-cancelled sync separation and the other half the noise-cancelled, line-gated AGC.

AGC Application: In a signal chain consisting of a mix of RF, MX and IF stages (from zero upwards of each), the key question was which should be AGC’d and with what proportion and with how much delay.

Delayed AGC, with a single delay, was the norm for domestic AM-only valve receivers. Some FM-AM receivers abandoned delayed AGC on AM because it was convenient to use the EABC80 and like valves for FM and AM demodulation and the 1st AF stage, and that did not easily allow for delayed AGC. One of many examples was the Bush VHF64. It is mentioned because its export counterpart, the AM-only BS64, had a very similar circuit to that of the AM side of the VHF64, but it did have delayed AGC.

Some communications receivers had an additional delay for RF AGC, but that was by no means universal practice. An example was the Marconi Atalanta marine receiver, as noted in post#9 upthread. Some communications receivers had much larger delays for the whole AGC system than was the case with broadcast receivers. Feedforward AGC was found in some professional receivers, such as the Marconi HR24. This had feedforward AGC for the 100 kHz sideband IF amplifiers, but the normal feedback type elsewhere.

With FM receivers, AGC was often used to delay the onset of overload in the RF and/or early IF stages, and so ensure that limiting took place only in the designated stages and as late as possible in the selectivity chain. With limiter grid AGC, there was probably some delay. Where the 1st (of two) limiter grid was used to provide AGC, there was probably a significant effective delay, as such a stage was often an amplifier/limiter that did not begin limiting until higher incoming signal strengths were reached. The Leak Troughline II et seq was an example where AGC was taken from the 2nd IF stage grid back to the RF stage. (The original Troughline took its AGC from the 3rd IF stage (limiter) grid.)

FM receivers with ratio demodulators but without limiters used AGC to keep the audio output level reasonably constant; whether any delay was normally involved I am not sure.

Another FM variation was the use of “fast”, or unfiltered AGC from the limiter grid back to an earlier stage. This AGC thus included any AM, such as noise spikes, that was on the signal, and so acted to cancel it, thus assisting the limiting process. The Quad FM tuner, from the Series B 2nd iteration, did this, with unfiltered AGC from the limiter grid to the 1st IF stage. (The Series A and 1st iteration Series B had filtered AGC from the limiter grid to the RF stage.) The limiter grid was an effective AM demodulator, and for example was used as such in the BBC/Ambassador VHF AM-FM Comparator receiver.

TV receivers, except perhaps very early examples, usually had the RF AGC delayed relative to the IF AGC. One can find detailed writings on this subtopic. The two Mullard papers on its high-gain sync cancelled AGC systems go into quite a bit of detail in respect of both generation and application, and in respect of the latter, RF delay and voltage grading to suit various mixes of RF and IF valves. And RCA dealt with it in detail.

At the valve: With RF pentodes and triodes, it was customary to apply the AGC bias to the signal grid, either shunt fed or series fed as better suited the circuit. Less commonly, AGC bias was applied to pentode suppressor grids.

Mixers: The conventional wisdom was that AGC was better not applied to mixers. But it was close to being unavoidable on broadcast receivers of the simple (MX + IF) type. With the (RF + MX + IF) line-up, it might be avoided, particularly on the HF bands (as in the Eddystone 670C, noted upthread) but then in a high quality unit with restricted AGC on the IF stage, full AGC on the MX stage was then probably necessary. In communications receivers with more stages, say up to 2 x RF, MX1, IF1, MX2, 3 x IF2, there was more flexibility, but even so, AGC’d mixers were still found. See post #9 upthread for some examples of various configurations.

TV receiver practice was for a long time no AGC on the MX, but then in 1961 Mazda introduced its 30C17 triode-pentode that was designed for AGC. Clearly, someone thought that the benefits of an AGC’d mixer outweighed the disbenefits, so some circumspection is needed around the conventional wisdom.

AGC and Distortion: In radio receivers, differential distortion and modulation rise distortion were potential issues, although probably more so for broadcast receivers than for communications receivers. Differential distortion arose with conventional delayed AGC systems that had a dedicated rectifier with standing bias. Keeping the delay small minimized the problem; it could be avoided with a separate delay diode. It could also be avoided by using amplified AGC via a separate IF sidechain. Modulation rise could occur in the final IF stage, with big signals and high AGC bias. Having zero or reduced AGC on this stage helped avoid it. But this was a luxury not readily available to broadcast receivers of the simple (MX + IF) type, as typically full or near-full AGC was required on both stages in order to obtain an adequate AGC range. Degeneration in the IF stage evidently helped, and was sometimes found in high-quality units. With the (RF + MX + IF) line-up, reduced AGC voltage, perhaps one quarter, could be used on the IF stage. With amplified AGC via a sidechain, full AGC on the final IF stage seemed not to be a problem.

Remote vs. sharp cutoff: For AM radio receivers, the conventional wisdom pointed to the use of remote cutoff valves, but sharp cutoff and semi-remote cutoff valves were also found in AGC’d positions, and even connected to the same AGC line, as remote cutoff valves in the same receiver. The Eddystone 830/7 was an example, see post #9.

In TV receivers, in the early days, AGC was applied to sharp cutoff valves, such as the 6AG5, Z77 and EF80. Possibly this was because they were what was readily available from the valve inventory. But arguments were made in favour of using AGC’d sharp cutoff valves; for example they required a lesser AGC voltage magnitude for a given control range, and so less AGC system gain and they remained until the end of the valve era. Remote cutoff TV valves appeared in the early 1950s, and some sources suggested that they were better for positive/AM systems. Sometimes both types were mixed in the same receiver. Later, semi-remote cutoff valves appeared as if a halfway house. From the RCA and Mullard writings I have seen, the solution was not so much on one side or the other, but rather once having made the valve line-up choice, it was essential to get the AGC levels and delays for each stage exactly right. Valve slope (in)consistency played a part, and resulted in the use of selected subsets, as in the case of the 6CF6 relative to the 6CB6, and the 6JH6 relative to the 6BZ6.

For FM receivers, both sharp and remote cutoff types were used in AGC’d positions, without any obvious reasons as to why one or the other would be chosen. Leak had used the semi-remote cutoff (in situ) ECC84 as RF amplifier in the Troughline II and 3, but switched to the sharp cutoff ECC88 in the Troughline Stereo, with appropriate biasing changes. (It would also have had the option of using the remote cutoff ECC89/ECC189.)

With RF pentodes and triodes, it was customary to apply the AGC bias to the signal grid, either shunt fed or series fed as better suited the circuit. Less commonly, AGC bias was applied to pentode suppressor grids.

Now to the solid-state era:

This brought with it not only benefits and solutions but also and additional problems.

Amplified AGC: Insofar as DC gain was more easily obtained than with valves, amplified AGC was easier and probably became more common, moving to being the norm once AGC generation was done within ICs.

Delayed AGC: Delaying the RF AGC relative to the IF AGC, not widely used in valve radio receivers, became more common, even in broadcast receivers.

Mixers: AGC was seldom applied to solid-state mixers.

Devices: Some bipolar transistors were especially designed for AGC’d amplifier positions, an example being the Mullard BF167 for TV early IF stages.

TV AGC Line Gating : Gating using line flyback pulses still remained common practice. But typically the flyback pulses were used for timing only, not for energy input as in the valve case, given that as said, DC gain was readily obtained with transistors. Sometimes, AGC was applied to bipolar UHF RF amplifiers, whereas it was seldom or perhaps never used with valve UHF RF amplifiers. In the solid state era there was probably no circuit complexity excuse for not using black level AGC with positive/AM systems (e.g. Decca did a quite neat dual-standard black-level solid state AGC system for its Professional 23 model), although one should never underestimate the creativeness of the setmakers along this vector.

Difficulties: Small-signal bipolar RF devices were generally viewed as being “disadvantaged” in the CM, IM and spurious response department, and to add to their woes, they did not always behave well under AGC, suffering from non-negligible input admittance changes that in the TV case particularly, resulted in undesirably skewed bandpass characteristics. To some extent this happened with valves, but evidently less so and more easily ameliorated. Forward or reverse AGC was chosen as best suited the situation, the former being common in TV vision IF amplifiers. One remedy was the cascode with AGC bias applied to the upper rather than the lower transistor, something that could not be done with valves. This kind of cascoding could be done with jfets and mosfets as well as with bipolars, and in the latter case, in a single device, namely the dual-gate mosfet, which then became a popular device for AGC’d RF and IF stages where ICs were not a good fit. A further development was the triple transistor circuit, essentially a long-tailed pair with the signal applied to the bottom transistor and AGC bias to one of the upper transistors, not being the one from which the output was taken. Motorola integrated this as the MC1550, which was offered as the basis for a TV vision IF system that avoided bandpass skewing, then quickly moved to a balanced version of the same circuit incorporated into its MC1350 IC.

Integration: TV AGC generation moved into ICs with the Motorola MC1352 of 1968 and the early “jungle” ICs of about the same time, such as the Philips TAA700 and Motorola MC1344. Typically these ICs derived IF and RF AGC voltages from the demodulated video signal, with programmable delay for the RF AGC, and were equipped to allow line gating by a modestly-sized line flyback pulse. At least some of the jungle ICs also provided noise-gating or noise inversion. In the case of the MC1352, the IF AGC was used within the IC, and that established a precedent for subsequent vision IF ICs. FM RF AGC generation was integrated with the RCA CA3089 IC of 1971. The RCA CA3088 AM radio IC, also of 1971, provided delayed AGC for an external RF amplifier if uses. In the communications field, the initial release of the Plessey SL600 series (late 1968 I think; announced in WW 1969 March) included an SSB AGC generator, the SL621. Later there was the SL623C which combined AM demodulation, SSB demodulation and AM AGC generation, and the Sl1625 which did AM demodulation and AGC generation..

PIN diodes: PIN diode attenuators were another solid state innovation, and had the advantage that they did not reduce the signal-handling capability of the following active devices. Conventional AGC with active devices had the fundamental problem that reduced gain was achieved by reducing the mean current pushed into the load for a given signal input level. And reduced mean current meant reduced maximum current swing. Any early use of PIN diodes for RF attenuation ahead of the 1st RF amplifier was in the Fisher TFM-1000 FM tuner of 1966. The PIN diodes were controlled by delayed AGC bias developed from a discrete sidechain with IF amplifier (relatively narrow band), rectifier and DC amplifier. This tuner also had conventional RF and IF AGC. By the late 1970s, PIN diode attenuators may have been more common in high-performance FM front ends. The Ambit EF5804, for example, had a PIN diode attenuator at its input fed from a discrete rectifier and DC amplifier whose input was taken from the mixer output, thus making it wideband. The EF5804 also had conventional RF AGC on both of its mosfet RF stages, this being wideband and developed by a rectifier and DC amplifier within the tuner head.

Wideband RF AGC: Wideband RF AGC loop for radio receivers, separate from the main AGC loop, were not unknown in the valve era, but it seemed to be fairly scarce. But it was much easier to do with solid state. Another example was the Eddystone EB35 Mk III HF receiver of the mid-1970s. Here the 40841 dual-gate mosfet mixer was the input stage. It had AGC, as well a signal on gate 1, the (wideband) AGC bias coming from a jfet amplifier and rectifier fed from the mixer output. This was a rare example of an AGC’d solid-state mixer, but given that Eddystone chose not to include an RF amplifier, it may have been unavoidable.

Integration of radio receiver wideband RF AGC had happened by the early 1980s, if not before. National’s LM1863 AM radio IC included a wideband AGC generator for an external RF amplifier, typically a jfet/bipolar cascode. Its LM1865 FM IF IC used a mix of wideband and narrowband AGC to provide what it called an automatic local/distant switch. RF AGC was applied earlier when the wideband detector indicated the presence of strong nearby signals.

TV AGC Noise Inversion: The Philips TDA2540/1 TV vision IF ICs of 1974 departed from established practice abandoning line-gating in their AGC generator section. Rather internal noise inversion was used with what was otherwise sync tip AGC. The idea was not new, having been used in previous discrete solid state circuits (e.g. the Beovision 3000, I think). But I can only guess that Philips chose to do it this way to facilitate its use in VCRs, which did not have readily accessible line flyback pulses. Otherwise, to retain line gating, it was necessary to generate gating pulses from the video itself, and sometimes that was done. The Sony TUM-100 TV tuner of the late 1960s had a discrete transistor line sync flywheel circuit to generate the pulses. The BBC UN1/642 TV receiver used a TAA700 jungle IC to separate the sync, which then fed a discrete multivibrator circuit that generated the gating pulse for the TAA700 AGC system. Some later TV vision ICs included self-generation of line gating pulses. For example the National LM1823 of the early 1980s used separated line sync pulses for gating. The accompany application note, #391, provided a detailed commentary about gain distribution between RF and IF and the proportioning of AGC. The Philips approach of abandoning line gating by no means became universal, and other vision IF ICs, such as the TDA440 et seq, continued to make provision for it.

Returning to the Philips TDA2540/1, the two variants differed in their RF AGC outputs, catering respectively for NPN (TDA2540) and PNP (TDA2541) RF tuner transistors. This duality was found elsewhere, for example with the TDA440N and TDA440P. There was also a version with RF AGC output to suit a mosfet RF amplifier, namely the TDA2544, but that seems to have been obscure in Europe. I think that the Toshiba TA7607/7611 may have been a clone of the Philips IC series, and here precedence seems to have been given to the mosfet RF AGC version (TA7607) over the NPN transistor RF AGC version (TA7611). With its sync tip AGC, the TDA2540/1 was suitable only for negative modulation systems. Its counterpart for positive modulation systems was the TDA2542, which had mean level AGC. One assumes that this rather poor choice derived from the decision not to use line gating.

Two Loop TV AGC: Introduction of two-loop AGC, with separate RF and IF loops, to the IC world seems to have been done by Plessey in the late 1970s, in connection with its SAWF program. Initially it had introduced the SL439 and then the SL1430 ICs to act as IF preamplifiers and to match a following SAWF. This was followed by the SL1431 (NPN RF) and SL1431 (PNP RF), which added a wideband RF AGC generator to the basic SL1430. Wideband TV RF AGC did not seem to become general practice, though. For example, a late vision IF IC, the Motorola MC44302A, had delayed RF AGC derived from the IF AGC. This was line gated, the gating pulse being internally driven by a line PLL system timed from separated sync pulses. It handled both negative and positive modulation systems, and with positive, the AGC, whilst basically black level, was also adjusted for peak white level.

One conclusion is that in the solid-state era, the device makers certainly paid a lot of attention to AGC systems.

Cheers,