Double Triodes for Cascode Amplifiers

[kalee20]

I was looking at data sheets for VHF-TV-tuner-type double-triodes recently, aiming to build a cascode RF amplifier for Band 2 (yes, I know front-end techniques standardised on ECC85 in grounded-grid, but I wanted more performance and the TV folk needed better than the FM radio guys).

Anyway, the heater-cathode insulation rating surprised me. PCC84, for instance, has TWO values, 100V for one section and 250V for the other. Now obviously the traditional series cascode has the upper cathode sitting at a high voltage above ground, so needs a good Vk-h rating whereas the lower section doesn't, but, why would the two sections actually BE different? What did Mullard do differently between them?

To get a high Vk-h rating, extra insulation is necessary. If one section had a bigger cathode tube, it would have different basic characteristics, gm and ra. But that's not the case. And none of the valves I have looked at appear to have different-sized cathodes.

If one section had extra insulation on the heater (maybe an extra few dips in the slurry of insulating oxides) then build-up of insulation would be greater on one heater than the other, so it would be harder to fit in the cathode tube. And I'm also thinking that the cost of running two different processes for the two heaters would outweigh any cost saving of a few milligrams of oxides on the 'lower' heater.

The other thing which surprised me is the PCC84 and 89 have two connections to the 'lower' cathode. Good practice, it allows separation of input and output circuits right up to the cathode itself, and avoids the input-circuit damping due to cathode lead inductance. But - the (later?) PCC189, which has higher gm, doesn't do this and has only a single lead, in fact it has the same pinout as the ECC85 even though designed for cascode, where the bottom triode operates in common-cathode. I would have thought the single cathode lead would give worse performance!

Can anyone shed any light? Synchrodyne is generally good at this sort of valve question!

[Radio Wrangler]

The upper triode will present an input impedance at its cathode of 1/Gm so there will be some signal voltage at this point and any capacitance from the cathode to ground (or to a groundy heater) will rob signal current and reduce the stage gain at higher frequencies.

So a double triode aimed at higher frequency cascode use would benefit from thicker H-K insulation and of a lower dielectric constant ceramic.Being able to rate it at a higher voltage may be a consequence.

Knowing the lower triode will have a decoupled cathode says that C to ground/chassis here will not be a problem, and only low voltage rating is needed. So with less insulation, the heater power could be reduced. This would mean different currents in paralleled heaters, or different voltage drops in series heaters.

David

[kalee20]

Yes I can see that, and thicker h-k insulation on the upper valve is good news both for lower capacitance, and for better voltage stand-off.

But I can't accept that the lower cathode could have a lower-power heater. More insulation in the top one might mean it takes longer to get to temperature, and also that the heater runs rather hotter than the cathode it's stuffed into. But ultimately, every joule of electrical energy that goes into the heater, gets to the cathode. If it were possible to heat two identical cathodes to the same temperature with heaters of different power, we'd have violated Conservation of Energy. (I am assuming that heat loss from the heater wires, and from the tips of the cathode tube, is negligible).

Plan tomorrow is to locate a PCC84. I'll power-up the heater and see if there's a visual imbalance. The sure-fire way to tell would be to open-up and measure each section independently. But I can't bring myself to kill a good valve in the thirst for knowledge.

[Radio Wrangler]

Slightly greater heat loss at the ends. The difference is small, but there. I'm old enough that valves were in the syllabus and exams when I did my first degree, but that also means that it's a long time back to remember. The cathode radiates IR, and the anode reflects some of it back. Each electron emitted represents a further loss of heat, essentially the 'work function' of the cathode material. That was the one thing I never understood - using oxide-coated cathodes to promote emission, but then oxides tend to be poor conductors. The goings-on in the coating of a cathode seem as complex as the goings-on in a transistor.

David

[trobbins]

Are you sure that each cathode's filament has a different insulation voltage rating, or rather that the insulation voltage rating of either filament is the same but just that the voltage rating depends on the polarity to the associated cathode?

[Synchrodyne]

The first two double triode valves designed specifically for use as cascode RF amplifiers for VHF TV applications were the RCA 6BQ7 followed by the GE 6BK7. The 6BQ7 had two cathode pinouts for the first triode, with the inter-triode shield sharing a pinout with the second triode grid. On the other hand, the 6BK7 had a single pinout for the first stage cathode, and a separate pinout for the inter-triode shield. These appeared to have established two conventions for cascode double triode pinout patterns, and most succeeding designs followed one or the other. One might infer that the higher in-situ input conductance possible with two cathode pinouts was not seen as a significant enough benefit to justify its use by all valve designers.

RCA wrote a fairly long (23 pages) article on the 6BQ7 in RCA Review, 1951 March, available at: https://www.americanradiohistory.com..._Issue_Key.htm. As far as I can see, this made no mention of the fact that the first section cathode had two pinouts. It did mention that both sections were intended to be identical, to facilitate its use in other applications, such as push-pull grounded grid amplifiers. Not said, but implied is that their mechanical assemblies, inclusive of heaters and cathodes, were identical.

When Philips came to issue a cascode double triode, about two years later (1953), its ECC84 evidently followed the pinout pattern of the 6BQ7, and was of generally similar performance. In 1957 it issued frame-grid derivatives, the ECC88 (sharp cutoff) and ECC89 (remote cutoff). The latter followed the pinout pattern of the ECC84 (and so the 6BQ7), perhaps to provide some continuity for the TV front end makers. But the ECC88 followed the pinout pattern of the 6BK7. I understand that the ECC88 was also intended for non-TV uses. Possibly in some of these an internal screen connected to one of the grids was unwelcome.

The ECC189 was described simply as an ECC89 with an altered pinout, switching to the 6BK7 pattern. Possibly some of Philips’ customers preferred this. Somewhere I think I have seen that this pinout was more favourable for TV tuners with printed circuits, but right now I cannot find the reference.

Regarding the heater-cathode voltages, the Brimar datasheet for its version of the ECC84 was written as follows:

Heater-to-cathode 1 90 volts max.
Heater-to-cathode 2 (heater positive) 90 volts max.
Heater-to-cathode 2 (heater negative) 250 volts max.

My guess is that there is no physical difference between the two triode assemblies. Rather the specification has been written around the actualities of cascode operation, in which cathode 2 is going to be much more positive than cathode 1.

The cascode RF amplifier was used quite extensively in American FM tuners, so is a logical choice for a high-performance valve tuner. The attached article shows that it was in FM applications by 1953. The cascode was found in a small number of British FM tuners, the best known of which was probably the Leak Troughline II of 1959 (although it wasn’t the only or even the first example). Given that the US market was quite important for Leak, it is not so surprising the Troughline II was designed with the export market very much in mind. The ECC85 single-valve front end was more of a cost-driven than a performance-driven solution, and this approach did not infect US FM practice until the late 1950s. And once adopted, it resulted in the development of improved valves (along more than one vector) that had no counterparts in Europe.


Cheers,

[kalee20]

Originally Posted by Radio Wrangler

Slightly greater heat loss at the ends. The difference is small, but there. I'm old enough that valves were in the syllabus and exams when I did my first degree, but that also means that it's a long time back to remember. The cathode radiates IR, and the anode reflects some of it back. Each electron emitted represents a further loss of heat, essentially the 'work function' of the cathode material. That was the one thing I never understood - using oxide-coated cathodes to promote emission, but then oxides tend to be poor conductors.

Yes I can see that there would be greater heat loss at the ends, if one heater really was wrapped in an insulating blanket before stuffing in the cathode (the heater itself would be hotter).

Interesting post David, thanks! Although veering off-topic slightly, I had never considered loss of energy from the cathode by emission of 'hot' electrons. I'm going to have to think of a way to calculate that now...

And yes oxides are insulating generally, though get them to red-heat and I could accept that they become sufficiently conductive. After all, glass does!

Originally Posted by trobbins

Are you sure that each cathode's filament has a different insulation voltage rating, or rather that the insulation voltage rating of either filament is the same but just that the voltage rating depends on the polarity to the associated cathode?

As per Synchrodyne's post below yours, the ratings really are specified differently in the data sheet!

Originally Posted by Synchrodyne

My guess is that there is no physical difference between the two triode assemblies. Rather the specification has been written around the actualities of cascode operation, in which cathode 2 is going to be much more positive than cathode 1.

Thanks for the informative post (as usual), Steve. The two sections look identical to me, too.

Writing the spec around the intended use makes sense, the only thing is, if the h-k insulation really can withstand 250V (cathode positive to heater), I'd have thought they would emphasise this throughout, rather than just for one section.

And the links: looks like my evening reading material beside the fire, has been thoughtfully provided!

[trobbins]

Due to this valve being associated with a particular application, it would seem plausible that they both could have the same max Vhk rating (by design and part preparation), but only one triode is screened for that better rating, and hence it is on the datasheet so designers can confidently use it for that application.

I have some ECC84's so will see if I can grid leak bias each triode for fun.

[woodchips]

You have raised an important point here, that changing the size of the cathode will affect gm and ra, possibly other things too.

Yet you can buy all sorts of double triodes with long plates, short plates, rivetted or welded anodes etc etc and they all sell as ECC8x valves. I have read all sorts of books to try and find what, exactly, defines the characteristics of valves, gm, ra, etc. And failed. Searching online just produces thousands of pages of rubbish. A valve is a mechanical construction, copy exactly the physical shape and size of a valve and you will have a 1960's ECC83, but you don't, for some reason. There are other variables, mainly metal purity and, most important, the cathode coating, but what influence do these have? Gas adsorption in metal could be something?

The people who did all this work are long dead, all we have left is the reports they wrote, and they are not very useful. Possibly because the real difference was actually the cathode material, and that was a commercial secret.

[Bazz4CQJ]

Originally Posted by woodchips

The people who did all this work are long dead, all we have left is the reports they wrote, and they are not very useful. Possibly because the real difference was actually the cathode material, and that was a commercial secret.

I think that for most of the time that valves were being researched and produced, the people involved lacked really good tools for investigating the finer and perhaps more critical aspects of the chemistry and physics that were involved.

In particular, the huge advances in chemical analysis instrumentation that came only after valve production ended include;

• the refinement of electron microscopes and the EDX elemental analysers that could be built in to them
• True "surface sensitive" analytical techniques which enable you analyse the outer few atoms of a surface
• Ion probe techniques that allow you to dig holes in a surface only atomic layers in depth and analyse the material as you burrow in.

All those techniques, which were so useful in semiconductor development, really came to fruition from about the late 1960's onwards and the valve developers must very often have been working in the dark.

It's only when you spend time using an electron microscope that you realise just how incredible dirty the world is and frequently you can find particles on a surface of all kinds of extraordinary substances, the source of which are impossible to account for. It may be that as much the composition of the cathode materials was secret, it was actually quite ill-defined for those who were using it.

I really take my hat off to all those who helped developed valves during the period in which that work was done .

B

[Synchrodyne]

An interesting point is that the first (and subsequent) VHF TV double triode cascode valves were designed for the purpose. Evidently the valve makers did not see that the then-existing double triodes were optimal for cascode VHF RF amplifiers, whilst the burgeoning TV receiver market justified the development of special-purpose valves.

In 1947 GE had introduced the 12AT7 double triode for use as an FM and VHF TV mixer-oscillator applications, as well as others (up to 300 MHz) for which it was suitable. For the FM and TV applications, it was an improvement over the (rather microphonic) 6J6, and at least a match for the (Loctal) 7F8. With its split heater, the one valve was suitable for both the 6.3-volt parallel and 150 mA series-string heater cases.

The 12AT7 could be and was used in cascode amplifiers, but apparently it was better suited to fixed-frequency rather than variable frequency applications. Philips included it (as the ECC81) in its initial “World Series” of TV valves, and included a cascode RF amplifier circuit in its Book IIIC. I understand that this kind of RF amplifier was used in some French early 819-line TV receivers of the single-channel type. As a VHF TV mixer-oscillator, the 12AT7 was largely superseded by the arrival of ad hoc triode-pentodes in 1951, this change driven by the movement of the standard American TV IF from 20+ to 40+ MHz. It remained a staple in FM mixer-oscillator service for quite some time, though. The early GE application information also showed that it was suitable for use as an AM mixer-oscillator, particularly in combined AM-FM receivers. Zenith used it that way for many years, although GE itself did somewhat differently. Perhaps as a rearguard and interim action, RCA subsequently also proposed that the 6J6 double triode be used the same way in AM-FM receivers. Hitherto RCA had backed – wrongly, as it turned out - the pentagrid for FM mixer-oscillator use, from the 6BE6 through the 6SB7Y to the 6BA7, so turning to the 6J6 was perhaps a “hospital pass”. It got itself up to speed, as it were, with its 6X8 triode-pentode in 1951. Primarily this was a VHF TV mixer-oscillator intended for use with 40+ MHz IF strips, but it was also specified as an AM and FM mixer-oscillator. In the FM case, the pentode mixer could be operated as was, or triode-strapped where lower noise was required (and one imagines to ensure that it had a match for the 12AT7 in this role.)

In respect of the VHF TV cascode requirement, evidently GE saw that the 12AT7 would not do, and so despite its established versatility developed the 6BK7, which was broadly similar to RCA’s 6BQ7, which appears to have been developed mostly on a “clean sheet” basis.

In Europe one supposes that the need for valve count economy drove the development of the one-valve FM front end. I think that there had been one or two cases in the USA where the 12AT7 had been used as a single-valve front end, but the use of two valves was the majority choice at the time. One may deduce that neither the 12AT7, nor the new VHF cascode TV valves were ideal for the application, and that development of an ad hoc double triode was justified. This appeared as the ECC85. I am not sure who did the work, but it might have been Telefunken rather than Philips. It was not mentioned in early 1950s Philips literature, such as Book IIIB, which covered the “new” range of FM-AM valves.

The cascode double triodes were also used in video and audio circuits (although dedicated audio double triodes such as the 12AX7 (ECC83) and 12AY7 were also used in this mode), and towards the end of the 1950s, as RF amplifiers in HF receivers. For the latter, the remote cutoff ECC189 seems to have been the valve of choice. It could be set up with a range of curves with different cutoff vs. bias points, presumably to allow proper matching into gain structures and AGC systems. For FM applications, the sharp cutoff types seem to have been preferred. Depending upon upper stage biasing arrangements, these could be set up as strictly sharp cutoff or as semi-remote cutoff (with a square-law curve), and this duality was noted in respect of the 6BQ7 and ECC84, for example. But the ECC88 seems to have been slated for strict sharp cutoff applications only; perhaps because the ECC89/ECC189 was available for and better suited to the range of remote cutoff curves. Leak used the ECC84 for the Troughline II and 3, and the ECC88 for the Troughline Stereo.


Cheers,

[turretslug]

I wonder if the move away from the dedicated cascode-intent double-triodes, such as the 6BQ7 and ECC84, with one triode having two cathode pin-outs and the other triode grid commoned with the inter-valve screen was driven as much by marketing considerations as anything. A double-triode is potentially a very useful and flexible device with application at AF, RF, control, gating and logic circuitry (they could be found throughout the then burgeoning TV, hi-fi and computing scenes, for example)- but one of the cascode-intent devices with inter-valve screen commoned to a grid becomes a lurking problem if that grid is made "live" to signal. Later devices like the ECC85 and ECC88 showed how a less-dedicated pin-out would sell like hot-cakes to all aspects of the electronics industry, yet still made a credible cascode stage for the VHF front-end case ('scopes often being full of ECC88s in differential amplifiers and timebases as just one case), whereas the ECC84 remained as a "what if..." one-trick pony with a fairly small production potential.

I realise that the above is straying rather far from the OP, but I'd also wondered if the then relatively new dedicated efficiency diodes with highly insulating h-k construction had taught manufacturers about making reliably insulating series cascode devices and the cascode seemed sufficiently new and promising to justify a dedicated valve? Maybe they felt that the extra expense of a highly isolating h-k construction for a recognised new "performance" type could be covered by charging a premium that would compensate for a more limited market. As outlined above there may not have been significant difference in this aspect, anyway- more a case of what they wanted to tell folk in the data sheet.

[kalee20]

Originally Posted by woodchips

You have raised an important point here, that changing the size of the cathode will affect gm and ra,... I have read all sorts of books to try and find what, exactly, defines the characteristics of valves, gm, ra, etc. And failed.

I haven't studied detailed valve theory, but I believe μ is largely determined by anode-cathode spacing relative to grid-cathode spacing. gm is determined by length of assembly (double the length of everything, and pretty obviously you double the mA thus also the mA/V), grid-cathode spacing and grid pitch. ra is of course μ/gm. Metal types make almost no difference to basic parameters; even the cathode, as long as it emits enough electrons, likewise. Specific materials are chosen for reasons of life, production-ease, outgassing, cost, etc.

Originally Posted by Synchrodyne

One may deduce that neither the 12AT7, nor the new VHF cascode TV valves were ideal for the application, and that development of an ad hoc double triode was justified. This appeared as the ECC85.

Yes, this one does seem strange to me, why it was found necessary as it's operating in E/PCC84 territory - and the screen commoned with g1 would not have been a problem. What exactly was wrong with the xCC84's characteristics that the '85 did better at?

Originally Posted by turretslug

I realise that the above is straying rather far from the OP, but I'd also wondered if the then relatively new dedicated efficiency diodes with highly insulating h-k construction had taught manufacturers about making reliably insulating series cascode devices and the cascode seemed sufficiently new and promising to justify a dedicated valve?

Certainly a high h-k rating is important for the cascode. Some efficiency diodes have a spiral wrap of insulation around the heater. Others have the heater in its own tube, sitting centrally within the (larger) tubular cathode, which gets heated purely by radiation. Either way, capacitance would be lowered, which helps RF operation. Down side is they take ages to warm up, and are power-hungry, which is bad news for tuning drift in front-ends. Of course, h-k rating is even more important for efficiency diodes, we're talking of kilovolts peak.

My question really is, given that the 'upper' section needs 150V or so of h-k insulation, did manufacturers actually make it differently to the lower section? I find it difficult to believe the lower section was made inferior...

Pinouts: The PCC84 / PCC89 pinout, which provides for two cathode connections for the lower triode, looks ideal for cascode applications, but also seems under-utilised. I have just been looking at Mullard's 'Valves, Tubes, and Circuits' No. 15, this shows the two leads connected in parallel. And they recommend it! It will result in an improvement, but only a halving in cathode lead inductance. Why not promote complete separation of the input and output circuits, right up to the cathode itself? As per the figures below:

[Synchrodyne]

Originally Posted by kalee20

Yes, this one does seem strange to me, why it was found necessary as it's operating in E/PCC84 territory - and the screen commoned with g1 would not have been a problem. What exactly was wrong with the xCC84's characteristics that the '85 did better at?:

I suspect that one would need to go back to the developmental write-up for the ECC85 to ascertain what was the motivation for a separate FM front end valve. As the cascode valves were optimized for “half” HT voltage on each section; perhaps it was thought desirable to optimize the FM valve for full HT. And perhaps there was less concern about performance much above 100 MHz, whereas the cascodes were designed to work at 200+ MHz. Maybe the differences made the ECC85 slightly cheaper to produce.

An article in Electronics World 1959 August (p.52ff) on the then-new American single-valve FM front-end practice mentioned that cascode TV valves such as 6BQ7 and 6BZ7 could be or had been used, as well as the 6AQ8 (American designation for the ECC85) and 6DT8 (an American development that was generally similar to the ECC85.) A Radio-Electronics 1960 March article (p.38ff) describing the Standard Coil one-valve FM front end unit gave the suitable valve types as 6AQ8 and 6DT6. This might indicate that these were perceived as being directionally better than any of the cascode valves. A subsequent American development was the 6JK8, with a frame-grid triode RF amplifier and a conventional triode autodyne mixer. I suspect that this may have been developed to provided improved front-end performance for stereo receivers. Another development of the single-valve FM front end was the 6C9 double tetrode on a decar base. Both of those are evidence that directionally tweaking performance along the FM (100 MHz) vector, rather than simply accepting what the cascode valves delivered, was a desideratum.


Cheers,

[Synchrodyne]

Originally Posted by kalee20

Pinouts: The PCC84 / PCC89 pinout, which provides for two cathode connections for the lower triode, looks ideal for cascode applications, but also seems under-utilised. I have just been looking at Mullard's 'Valves, Tubes, and Circuits' No. 15, this shows the two leads connected in parallel. And they recommend it! It will result in an improvement, but only a halving in cathode lead inductance. Why not promote complete separation of the input and output circuits, right up to the cathode itself? As per the figures below:

Perhaps one needs to go back to the original Philips literature for the PCC84 to find a decent commentary on the two cathode pinouts for the lower triode. Mullard’s apparent lukewarm attitude may have been local rather than corporate. The appropriate Philips source document appears to be: "The PCC 84 double triode," Electronic Applications Bulletin (Philips), 14, No. 8/9, 1953, although I have not succeeded in finding it.

I do need to correct what I said earlier about the early sequence of cascode valves:

Originally Posted by Synchrodyne

The first two double triode valves designed specifically for use as cascode RF amplifiers for VHF TV applications were the RCA 6BQ7 followed by the GE 6BK7. The 6BQ7 had two cathode pinouts for the first triode, with the inter-triode shield sharing a pinout with the second triode grid. On the other hand, the 6BK7 had a single pinout for the first stage cathode, and a separate pinout for the inter-triode shield. These appeared to have established two conventions for cascode double triode pinout patterns, and most succeeding designs followed one or the other.

That was wrong in respect of the 6BQ7, which did not have two cathode pinouts. Rather it had the same basing as the 6BK7, with one cathode pinout and a separate pinout for the internal screen.

The apparent first American cascode valve with two lower section cathode pinouts was the GE 6CB7 of late 1955. Of it was said:

“It is designed to give considerably more gain and a better noise factor on the higher vhf channels than previous cascode tubes. The increased gain is achieved by a new basing connection featuring two cathode leads for the driven-grid section, with the shield between sections internally connected to the grid of the grounded-grid section.”

In fact though it was a rebased 6BZ7, the latter being the third of the American cascode valves, I think developed by Lansdale.

The timing of the 6CH7 puts it behind the Philips ECC84/PCC84, making the latter the first cascode valve to have had two lower section cathode pinouts. In that case one might have expected Philips to offer a reasonable commentary supporting its choice. But I guess the prosaic possibility cannot be outruled that Philips simply followed the precedent it had established with its EF80 pentode, where the second cathode pinout was said to be beneficial when it was used as a VHF TV RF amplifier.

How much effect the 6CH7 had on American practice I am not sure. Soon after it became available, the triode re-entered the scene as a lower-cost VHF TV RF amplifier in “neutrode”, “guided grid”, etc., guise, thus diminishing the cascode market.


Cheers,

[Nymrod121]

A fascinating thread, thank you all
Best regards
Guy