[Synchrodyne]

Originally Posted by G6Tanuki

OK it's not FET but one of the late-1960s RSGB books I've got [Amateur Radio Techniques?] mentions a US "Fisher" FM-tuner where the antenna is fed through a single tuned-circuit then into a 4-hot-carrier-diode ring-mixer - the article mentions that this was done to give better strong-signal handling than could easily be achieved with current-generation bipolar devices.

Thanks. I was unaware of that one. I suspect that it was quite rare in the FM tuner world.

Originally Posted by Radio Wrangler

I'm not sure whether the MC1496 or the RCA CA3027/3028 family came first.

One thing is definite, they aren't 'Gilbert Cells'. The transistor tree mixer predates the Gilbert cell, and the Gilbert cell is a logarithmic current-based 2 or 4 quadrant linear mixer.

What Barrie Gilbert invented was the idea of adding diode connected transistors on the substrate so that currents could be converted bach to voltages with a logarithmic law which compensated for the antilog law of current partitioning in a long-tailed pair.

His cell is in 4-quadrant linear multipliers and a lot of interesting stuff from ADI, but it seems that everyone sticks his name on the thing that he improved, and don't know of the final thing. It's a bit like everyone attaching Baird's name to every theatre and not knowing about television.

Ah, I wasn’t aware of that distinction, but is does explain why Gilbert’s papers on the topic seem to have arrived at about the same time as ICs incorporating the transistor-tree mixer, and not before.

Many sources seem to miss this. This site is probably typical: http://www.radio-electronics.com/inf...multiplier.php. The diode addition is mentioned, but it is not shown as the key difference that makes a Gilbert cell out of a transistor tree. Traps for the unwary (such as me).

I think that the RCA CA3028 predated the MC1496, and was aimed mostly at military and commercial applications. I am not sure what was the target market for the MC1496. Not consumer, as then it would have had a MC13xx number. But MC15xx was used for industrial/military components, and there was an MC1596 that I think was a MIL-spec MC1496.

The CA3028 was quite versatile, and could be used as a cascode or differential amplifier with various agc options, as a differential limiter and as a singly-balanced mixer. As far as I know the Fairchild µA703 was differential only, and the Motorola MC1550 cascode only. RCA also had the CA3026 that contained a pair of “half-trees” that could be used together as a full-tree mixer, and was so deployed by Leak in the stereo decoder part of its Stereofetic tuner. I don’t know if RCA or others proposed using dual-gate mosfets for the mixing/switching part of stereo decoding, but Motorola released its first IC-base decoder (non-PLL type) around 1968, and that was the direction in which the industry went, moving to PLL in 1971.

Given that FETs and ICs both competed with each other and complemented each other in the transitional era, some more than passing discussion of ICs in this thread seems unavoidable. Leak well summed up the situation in the name that it chose for its first solid-state FM tuner.

Originally Posted by G8HQP Dave

I think putting two dual-gate MOSFETs in front of a mixer on Band 2 is asking for trouble, unless you can be sure that stations will be sufficiently widely spaced in frequency that the excessive gain is more than compensated for by the improvement in RF selectivity.

One gets that impression from the Ambit commentary on the EF5804, and its inclusion of PIN agc as well as regular agc. Yet two RF stages became the norm for “supertuners” by the later 1970s. To some extent that might have been marketing-driven, perceived as necessary to justify its positioning even if of dubious or even negative technical merit. That seemed to happen in the audio field, although as observed earlier in this thread, the upper end consumer HF receivers were not immune, and so they had features that were not found on professional equipment.

Whereas the use of two FET RF stages in FM tuners had no significant precedent in the valve era, conversely, the use of two RF stages and four gangs was common practice in valved HF receivers, more often in single conversion types with IFs around 450 kHz, but also in some double conversion types (e.g. Marconi) with 1st IFs around 1.6 MHz or even higher. On the other hand it would appear that the use of two RF stages in those HF receivers that used FETs in the early signal stages was very rare. Admittedly, the early FET era corresponded with major changes in HF receiver topology including upconversion to high 1st IFs, which eliminated the image problem, and the use of very linear mixers that could handle high signal levels and needed less “protection” from out-of-band signals, these doing away with the need for exceptional front end selectivity. But even where FET-based HF receiver signal pathways followed valve practice, single RF amplifiers were the order of the day, and the fourth gang was obtained by using bandpass inputs. This suggests that the FET mixers used in such receivers were less tolerant of large signals than their valve predecessors, and so would not be happy with too much RF gain. One example was the Eddystone EC958, which had a jfet/mosfet cascode single-stage RF amplifier feeding into a dual-gate mosfet mixer, with a 1st IF of 1335 kHz. On the HF bands, the input was double-tuned bandpass, with a single-tuned interstage between the RF and mixer stages. Given that Eddystone had used two RF stages (and 4 gangs) on its top valve receiver, the 880, which had a higher 1st IF, one might reasonably infer that the single RF stage choice for the EC958 had something to do with not overwhelming the mixer.

Marconi, notwithstanding its frequent use of two RF stages in the valve era, also used a single RF stage, dual-gate mosfet, in its Apollo marine HF receiver. It had a bandpass input and single-tuned interstage. In this case the mixer was an SL640 bipolar unit which might have needed a bit more protection than say a dual-gate mosfet. The Marconi Hydrus HF ISB receiver might be a useful case to study here. It seems to have been the solid-state successor to valve receivers such as the HR22, and according to the advertisements, it was FET-based. However, circuit information for the Hydrus seems to be unobtainium.

Anyway, perhaps the professional HF receiver examples with one mosfet RF stage ahead of the mixer are more indicative of the best trade-off for overall performance (within the boundaries of what was possible with mosfets), whilst the FM examples with two mosfet RF stages reflect the influence of perceived marketing needs overriding sound technical judgment.

Cheers,