[Synchrodyne]

Returning to the issue of FM tuners and front end performance in difficult reception conditions, this comment made in the Radford entry in the 1967 London Audio Fair catalogue is I think illustrative:

“The standard FMT.2 tuner uses low noise silicon planar transistors in the signal frequency unit. This tuner will now be available to special order fitted with field effect transistors. The F.E.T. model is being made available primarily for foreign markets where conditions demand the highest possible performance in respect of cross modulation, image rejection and adjacent channel selectivity”.

What variety of FET was used by Radford was not stated, but given the early 1967 timing, jfets seem more likely. The FMT.2 had been shown at the 1966 Audio Fair, and in the catalogue it was described as having a four-gang front end with a double tuned circuit ahead of the input transistor.

One may surmise that Radford found FMT.2 front end performance not good enough in some export markets, hence the introduction of the FET option. Radford was somewhat unusual in that it introduced a solid state FM tuner before it introduced solid state amplifiers, the opposite of what several others, such as Leak, Quad and Rogers did, their hesitation in some part due to the knowledge that valve performance could not be matched with bipolar devices.

Radford’s use of a double tuned circuit ahead of the RF amplifier was interesting in that it went against the conventional wisdom that at VHF, it was better – for noise performance - to have a double tuned interstage and single tuning at the input. Presumably Radford was concerned to get as much selectivity as reasonably possible ahead of the first bipolar transistor. Whether the FET version was similarly configured is unknown. But Rogers chose the same configuration for its Ravensbourne FM tuner in 1968, that is a bandpass input circuit ahead of the dual-gate mosfet RF amplifier followed by a single-tuned interstage to the dual-gate mosfet mixer. But placing the bandpass between the RF amplifier and the mixer appears to have been more usual with following designs such as the Revox A76, although some makers stayed with three-hang front ends, as in the Leak Stereofetic and the Quad FM3. The next step, adopted for “supertuners”, was the use of two dual-gate mosfet RF stages, usually with bandpass interstages and so six gangs in all. The EF5804 FM tuner head shown and described in Ambit catalogue #3 was an example. Actually, the Ambit catalogues #1 and #3 between them illustrate a hierarchy of FM front end designs and their performance characteristics, with commentary (I think it was done by Bill Poel) that is quite accessible from the layperson’s viewpoint. It is evident that the use of two RF stages and associated additional tuned circuits was primarily driven by a desire to reduce spurious responses, including but not limited to half-IF. On the other hand, the extra gain ahead of the mixer could compromise strong signal handling, and the EF5804 was fitted with a pin diode attenuator ahead of the 1st RF amplifier and operated by a wideband agc loop driven from the immediate post-mixer signal. This was additional to the “normal” agc on gate 2 of both RF mosfets and derived from the IF strip. (As an aside, somewhat later – early 1980s I think – the National LM1865 FM IF IC – was designed to provide both narrow and wideband RF agc outputs in addition to the other functions that had previously been delivered by the CA3089 and its ilk.) The EF5804 gain was quoted as 46 dB, whereas the Larsholt 8319, a representative four-gang/one RF amplifier design, was quoted at 32 dB. So the second RF amplifier was adding around 12 dB gain, and commensurately reducing mixer overload headroom, the later being compensated for by the input attenuator. But apparently such was required to address the challenge of “sorting out” multiple strong stations in the typical US urban environment whilst retaining good fringe reception capability.

The use of two RF stages probably had no precedent in valve FM front end practice. Even the use of four gangs, with say a bandpass interstage, was probably rare in valve days. Nevertheless, it could be said that in the solid state era, domestic FM tuner front ends followed the pattern set in the valve era and then improved upon it with additional pre-mixer selectivity and gain. The dual-gate mosfet was the active device of choice for the RF amplifier and mixer stages in order to achieve maximum spurious response suppression. There was no general move to an altogether different approach, such as the use of a bullet-proof mixer with high level oscillator injection preceded by a minimum amount of RF gain and selectivity. Nor I think was there any significant move to the use IC bipolar mixers in place of dual-gate mosfets, even though such devices were developed for TV tuner use. Single bipolar ICs that did the whole front end job, such as the TDA1062, seem not to have made much impression at the higher performance end. There was also the SD6000, which was an IC containing two dual-gate mosfets, one for the RF amplifier and one for the mixer. It needed an external oscillator though.

ICs suitable for use in consumer equipment became available at about the same time as FETs, and these dominated in FM tuner IF strips from that time, initially with one simple “building block” IC, such as a µA703 or CA3028 per stage, followed by increasing aggregation of functions. The arrival of the multifunctional CA3089 in 1970 was a major nodal point in the transition to ICs. Nevertheless, sometimes additional IF gain ahead of the CA3089 was desired, as well as buffering between cascaded ceramic block selectivity filters. This was sometimes done with FETs rather than bipolar transistors or ICs. The Meridian 104 was an example, with a single agc’d BF961 dual-gate mosfet between the front end and the block filter (which fed into an HA12411 IC, one of the CA3089 derivatives). As the gain stage was ahead of the block selectivity, one assumes that the designers used a mosfet in part for best linearity and overload performance. Another, perhaps unusual example was the Eddystone 1002 FM-HF combination receiver. Here two dual-gate mosfet stages, without agc, preceded the single ceramic filter and a CA3089. But the unusual part was that those mosfet IF stages followed a Mullard module bipolar front end. I imagine that the Mullard module was quite low gain, perhaps lower than that provided by the kinds of front ends that might normally be used with a CA3089-based IF strip. So the use of two mosfets might have been to make up the gain deficit to enable the CA3089 to limit on suitably low signal levels. The previously mentioned National LM1865 FM IC included a pre-IF gain/buffer stage that obviated the need for discrete pre-IF stages.

Cheers,