[Synchrodyne]

Originally Posted by GrimJosef

It's not a new idea. Williamson used triode-connected KT66s in the output stage of his amp in 1947.

Triode-connected KT66s (and other tetrode/pentode) output valves might have been reasonably common in the post-WWII era until the ultralinear configuration was well-enough established by 1955 give or take. At the time the perception was that triodes were better for “high quality” work than tetrodes/pentodes. It appears that Osram data for the KT66 envisaged triode as well as tetrode operation from the beginning. Larger output triodes, such as the PX25, were directly heated. Possibly that was either necessary or highly desirable – I have seen commentary to that effect, but do not have it to hand right now. So using an indirectly heated tetrode/pentode in triode configuration might have been easier for amplifier designers. From the valve manufacturer perspective, it allowed one series of output valves to fulfil both triode and tetrode/pentode functions.

Some of the early Leak “Point One” amplifiers used triode-connected KT66s, including the TL12. On the other hand, the TL25, aimed more at industrial than domestic applications, used tetrode-connected KT66s. Leak started its swing to the ultralinear output configuration with the KT61-based TL10 in 1954.

Originally Posted by DangerMan

You could say the EF86 is a special case.

Mullard's given reason for using a pentode in the first place was that they wanted the extra gain it gave as well as its inherent microphony and hum reduction. This gave them chance to use heavy overall negative feedback in order to reduce distortion but still end up with designs that gave full output from about 200mV input.

It is noticeable that Leak went another way with the Stereo20 and its sisters, which used triodes in the voltage amplifier and yet had much the same sensitivity and distortion, and no doubt if all EF86's other attributes had been available in a triode, or twin triode, it would probably have found use just for its screening.

It's pretty doubtful you could get enough gain from a triode of the era for use in a preamp with feedback-applied RIAA or tone controls set anywhere other than "flat". Not with decent distortion and hum/noise figures from AC heaters, anyway.

The Leak valve stereo power amplifiers all used a ½(ECC83) input stage followed by an ECC83 phase splitter, whereas their mono counterparts of the same era stayed with the EF86 + ECC81 combination. Clearly the gain distributions were different between the mono and stereo models. Whether the stereo configuration of ½(ECC83) + ECC83 was better (quieter) in practice as well as in theory than the EF86 + ECC81 mono combination I don’t know, but whatever the difference, Leak did not see fit to adopt it for the mono models. The fact that there was no such animal as a ½(ECC83) would have made it difficult, although there was the possibility of using a triode-connected EF86 plus an ECC83. On the other hand the use of a ½(ECC83) input stage in the stereo case allowed a single ECC83 to serve both channels, as compared with, for example, using an EF86 in each channel. Perhaps this was a consideration as much as using a triode input stage for its own sake. Then the higher gain ECC83 phase splitter became a necessity.

The Leak control units also provide an interesting illustration. Leak moved from a single-stage pentode (EF40) to a two-stage (ECC40, then ECC81, then 2 x EF86) configuration, the (2 x EF86) being definitive. In the two-stage models, both EF86s had shunt feedback, the input EF86 providing the initial gain and (disc) equalization, and the second EF86 providing tone controls and filter functions. But the second EF86 was either pentode or triode configured, according to the overall gain required for specific models.

On equalized input stages generally, it seems that the EF86 with shunt feedback was adequate (for the times) in a signal-to-noise ratio sense down to input sensitivities of around 4 mV. For greater sensitivity (say down to around 2 mV) the next step seemed to be an ECC83 configured as a series feedback pair and with DC heating. This was used by some of the tape recorder makers, Brenell and Revox amongst them as far as I know. In any event DC heating would have been very desirable because of the unbypassed cathode resistor of the first triode of the pair, but whether it was otherwise needed because the ECC83 had higher relative hum levels than the EF86 is not clear. Less commonly, the ECC83 was also used as a low level input amplifier in a shunt cascode configuration, a (perhaps surprising) example being the Murphy T1 tape recorder.

Mention of the cascode brings us back to Leak, its Troughline II FM tuner having a cascode (ECC84) RF amplifier, considered beneficial at Band II frequencies, but unusual in UK FM practice. Evidently the VHF cascode was developed as the optimum way to use two triode RF input stages ahead of a chain or pentode amplifiers when a single stage would not provide enough gain to more-or-less nullify the noise contribution of the first pentode. Whether it was developed with TV receivers in mind I don’t know, but its availability surely helped TV receiver design. At Band III frequencies the cascode could provide enough low-noise gain to allow the use of (very noisy) pentode mixers. Such may seem counter-intuitive at Band III frequencies, and for this band alone probably would not have been used. But the otherwise desirable upward migration of TV receiver IFs to the point where they were nudging the bottom end of Band I made the pentode mixer very attractive for that Band, avoiding the need for channel frequency-dependent neutralization of triode mixers, something not particularly wanted for mass-produced equipment.

Whilst the cascode RF amplifier was near-universal for TV receivers (until frame-grid beam triodes became available at least), there was evident ambivalence at FM Band II frequencies, where many makers chose pentode RF stages as being “good enough” from the noise viewpoint. Apparently there was a “rule of thumb” that if a single-stage and therefore low-gain triode RF amplifier were used, then a (relatively low noise) triode mixer was necessary, this being encapsulated in the ubiquitous ECC85. Pentode mixers required pentode (or cascode) RF stages to provide enough gain to nullify mixer noise contribution. High slope pentode RF stages seem to have been preferred over the vari-mu type because of their directionally lesser noise contribution.

Maybe the RF triode noise benefits extended downwards to 20 MHz or lower. Towards the end of the 1950s, cascode 1st RF stages made their appearance in some HF receivers. Previously, Dynatron had included a triode broadband “pre-RF stage” operative only on some of the SW bands in its T139 tuner/control unit. In fact this was a Z77 high-slope pentode configured as a grounded-grid triode. Use of pentodes strapped as triodes was not that rare in HF receivers anyway, such as for stand-alone oscillators, cathode followers, etc. Possibly the use of readily available standard valves was a factor here. In its early 1950s TA160 bandspread export receiver, Murphy used a high-slope pentode (6F1) tuned RF stage because of its lower noise contribution as compared to the vari-mu type.

So the empirical evidence is that the whole triode vs pentode issue (inclusive of pentodes strapped as triodes) is in parts rather fuzzy.

Cheers,