What's the best audio output valve?

[GrimJosef]

Originally Posted by G6Tanuki

... "Ultra-linear" wiring of push-pull output valves always seems intuitively a bit strange to me - since it means you're constrained to using an anode HT-voltage no greater than the valve's maximum screen-grid voltage.

Indeed. For some valves of course (the larger members of the KT series, for example http://frank.pocnet.net/sheets/084/k/KT66_GEC.pdf http://frank.pocnet.net/sheets/084/k/KT88.pdf http://frank.pocnet.net/sheets/084/k/KT90.pdf) the maximum anode and screen voltages may be quite similar in any case. But for others they can be very different indeed. The 6550, for example, seems to have caught out a number of audio amp designers who have tried to run it with a screen grid voltage way, way beyond the safe limit. This always ends in tears ... .

The trade-off I guess is between maximum raw power, efficiency and distortion. Ultra-linear operation is a way of optimising the latter two, with an effective compromise between the linearity of a triode and the efficiency of a pentode/beam tetrode. As you say, it can fall down if you care most about power though. For something like a basic PA rig or a guitar amp or a transmitter modulator you'd be better off running in pure pentode/beam tetrode mode and accepting the higher distortion levels. Moving the feedback round into the cathode circuit is a neat solution but it does require an extra entirely separate winding on the output transformer thereby adding to the complexity and cost of what is probably already the most challenging component in the entire amplifier.

Cheers,

GJ

[Synchrodyne]

Originally Posted by G6Tanuki

All the competent designs for *RF* Linear use of the 6-40a specify you need to feed the screen from a regulated [250V] supply. These valves also like high HT+ voltages: 750-850V being typical in SSB service, where according to the specs you can get 87 watts PEP.

That would not necessarily be inimical to partial cathode loading, though. With the latter, the screen grid is at AC ground, so there should be some freedom as to the DC feeding arrangements.

Originally Posted by G6Tanuki

"Ultra-linear" wiring of push-pull output valves always seems intuitively a bit strange to me - since it means you're constrained to using an anode HT-voltage no greater than the valve's maximum screen-grid voltage.

The original Blumlein design allowed for alternative ways of coupling in the screen grid other than direct tapping into the output transformer primary. AC coupling to the transformer tap, with a separate DC feed was one. And I think using a separate primary winding for the screen, tightly coupled to the anode winding, was another. So different screen and anode DC voltages could be accommodated, although it might not have been done much in practice.

Be that as it may, I have the impression that the late high-performance audio output valves, including the EL34, KT88 and KT77, were designed primarily around the needs of UL audio circuits, including the electrode voltages reasonably required to realize their respective power output capabilities in situations where they spent most of their time in Class A, with some excursions into mild Class AB. The higher anode voltage capabilities that came out of these designs were perhaps something of a bonus for other applications. The earlier KT66 had, I think, the same voltage maxima for anode and screen, possibly reflecting the fact that it was intended for use both triode-strapped (in which case it similar characteristics to the PX25) and as a tetrode.

Cheers,

[G6Tanuki]

Originally Posted by Synchrodyne

The original Blumlein design allowed for alternative ways of coupling in the screen grid other than direct tapping into the output transformer primary. AC coupling to the transformer tap, with a separate DC feed was one. And I think using a separate primary winding for the screen, tightly coupled to the anode winding, was another. So different screen and anode DC voltages could be accommodated, although it might not have been done much in practice.

Be that as it may, I have the impression that the late high-performance audio output valves, including the EL34, KT88 and KT77, were designed primarily around the needs of UL audio circuits, including the electrode voltages reasonably required to realize their respective power output capabilities in situations where they spent most of their time in Class A, with some excursions into mild Class AB. The higher anode voltage capabilities that came out of these designs were perhaps something of a bonus for other applications. The earlier KT66 had, I think, the same voltage maxima for anode and screen, possibly reflecting the fact that it was intended for use both triode-strapped (in which case it similar characteristics to the PX25) and as a tetrode

Yes I've seen some designs where there has been an entirely separate "centre-tapped screen supply" feedback winding on the output transformer: I've also seen a design where each half of the output-transformer primary has a series-connected pair of resistors across it, with the 'centre' point feeding the screens - presumably depending on the voltage-drop across the paired resistors (due to screen current) to constrain the screen-voltage.

You're probably right about the later valve-designs being built round the idea of lower HT voltages that are also suitable for direct UL connection to the screens.

A thought: if you had an ultralinear P-P pair of valves with limited maximum screen-voltage but wished to run a significantly higher HT+ on the anodes, what would be the effect of feeding the screens from the usual UL taps but through voltage-regulator tubes or Zener-diodes? They'd 'lose' the excess screen voltage but not compromise the coupling of the audio component needed to provide UL operation. I guess you may need to parallel the 'screen regulator' tubes/Zeners with capacitors to avoid the active devices introducing any non-linearity though.

Hmmmm... where are those 807s and that 800VDC supply?

[GrimJosef]

Originally Posted by G6Tanuki

... A thought: if you had an ultralinear P-P pair of valves with limited maximum screen-voltage but wished to run a significantly higher HT+ on the anodes, what would be the effect of feeding the screens from the usual UL taps but through voltage-regulator tubes or Zener-diodes? They'd 'lose' the excess screen voltage but not compromise the coupling of the audio component needed to provide UL operation. I guess you may need to parallel the 'screen regulator' tubes/Zeners with capacitors to avoid the active devices introducing any non-linearity though

Might well work as long as you have a valve rectifier when the slow rise of HT would allow the capacitor plenty of time to charge. With a solid-state psu though the screen might well break down at or shortly after switch-on.

Cheers,

GJ

[G6Tanuki]

Good point - I hadn't considered that.

Must admit, my favourite "slow start" circuit for when using semiconductor-rectifiers in an amplifier was the trick of using a relay whose coil is wired in the cathode-to-earth path of one of the output valves, and whose contacts switch a series resistor in the power transformer primary. Initially the resistor's in series with the power supply, limiting inrush current and maximum HT-voltage. As the valves begin draw current the relay pulls in and shorts-out the resistor so applying full HT.

I've seen this used in push-pull-6L6 amplifiers where the relay coil's resistance also served as the cathode-bias resistor! There's component-economy for you.

[G8HQP Dave]

Originally Posted by Synchrodyne

That would not necessarily be inimical to partial cathode loading, though.

Remember that dual tetrodes usually have a common cathode pin, as well as a common g2 pin. To do UL or cathode feedback you would need to use two valves.

[mole42uk]

Originally Posted by Radio Wrangler

After you've done it, why not have a go at a somewhat more difficult task? try designing a good transistor amplifier. It's possible, but it's definitely harder than doing a reasonable valve amplifier

Ah, but then I'd have to generate some enthusiasm...... And anyway, I'd probably go with MosFETs, and then I'd have to go Class D because I used to design SMPS, and then I'd want to have microchip control...

Oh, you mean class A with bipolars? Hmm yes, that would get the grey cells active!

[kalee20]

Originally Posted by Radio Wrangler

Perhaps the title of the thread would have been better in the past tense... What WAS the best audio output valve?

Fair point, but it does restrict later developments.

The KT66 was probably the best audio output valve... until MOV developed the KT88. But, it depends what you want.

Originally Posted by G6Tanuki

A thought: if you had an ultralinear P-P pair of valves with limited maximum screen-voltage but wished to run a significantly higher HT+ on the anodes, what would be the effect of feeding the screens from the usual UL taps but through voltage-regulator tubes or Zener-diodes? They'd 'lose' the excess screen voltage but not compromise the coupling of the audio component needed to provide UL operation.

I'd suggest that this would be better than feeding the screens via resistive potential dividers.

One of the benefits of UL operation, in addition to the distortion-reducing partial feedback to the screen grids of the output waveform, is that the power associated with screen-grid current can be captured and used.

I like the pretty ones, just add global NFB and it really doesn't matter.

https://www.vintage-radio.net/forum/...1&postcount=13

[GrimJosef]

Originally Posted by merlinmaxwell

... just add global NFB and it really doesn't matter.

Aaah, if only it were true . The 5881 is a nice valve though.

Cheers,

GJ

[PE9ZZ_JO22KI]

@Radio Wrangler: this thread's title is obviously way too broad. There's no such thing as a "best" valve for any purpose without specifying design constraints. How "best"? Best performance? Best audio fidelity? Best price/performance ratio? Best reuse of old line sweep tubes? Best "WOW" factor? Ad nauseam.

Tjerk, 9ZZ

[Maarten]

The latest and greatest from Philips in audio output includes the PCL805 or EL86 and the EL503 in my opinion. Valves that represented the best the consumer industry could manage before valves went out of fasion.

[Maarten]

Maybe EL503 should read EL522, which might presumably be newer although I don't know of any significant difference. After that, development on consumer valves stopped and most production was moved abroad.

[Radio Wrangler]

The pretty ones are nice, and a lot of cult audio seems to be designed primarily to impress the eye.

I've wondered about trying to promote the 3-500Z as audiophile output stage. Class A single ended triode capable of a hundred or so watts. Just give 'em a google and look at the images of glowing anodes and bright heaters. They LOOK the proverbial million dollars.

There is no (well not too much) truth in the rumour that I only want to promote them because I think 2.5kV on top caps might thin the ranks of phools a little.

Will anodes with fins which go up-and-down sound better or worse than those with fins which go round-and-round?

THe high anode impedance and high voltage will really show up transformer quality. Putting them in grounded grid with a transformer coupling from the driver stage allows yet another important transformer to be added to the design..


David

[mole42uk]

Originally Posted by Radio Wrangler

Will anodes with fins which go up-and-down sound better or worse than those with fins which go round-and-round?

No, no - you need top caps with integral cooling fans. Preferably ones with blue LEDs around the circumference to make more pretty.

[turretslug]

Don't forget the "all in the best possible taste"- and utterly spurious- gold plating.

On the original theme, perhaps the 6L6 is high on the list. It spawned the 807, inspired the KT66, and is still appreciated in 5881 etc. derivatives now, 80 years on from its original inception. A sensible size envelope for the power and a base type that stood the test of time. A few more types with B9d base would have been good, but that's the way things go.

[Radio Wrangler]

Aye, the 6L6 was a landmark design. They are available and right now the prices don't seem too silly. Besides, with the original metal 6L6, no-one can see what shape and how smooth the anode is... and as for getter shape

I think we're on the brink of a new era when supplies of the pundit-preferred audio valves are all used up and the demand switches around anything available with gain, so all power valves become rare and expensive. enjoy what you can while you still can.

Don't forget the 5B/254M !

David

[Marconi_MPT4]

For me, the low price and inherently lower distortion of the 6L6 still makes it one of the best all round performers. I am currently using a couple of old Brimar 6L6G valves in the Williamson amplifier with bias set to reduce combined anode and screen dissipation to 21.5 Watts instead of 25W for the KT66. Apart from output power being reduced a little it gives comparable results at much lower cost and importantly meets the original spec for frequency response and distortion! In this application, no need to blaze a trail to obtain specific GEC KT66.

Somewhere I still have a pair of STC 4304CB triodes designed for transmitting applications up to 500MHz. Now, these light up brightly with a 24W filament each and suspect being power triodes would fetch 'phool' money. Hmm, just looked up price on internet....

Still relatively cheap, the PL/EL509 triode connected seems to produce some nice characteristic curves and I calculated the anode impedance (ra) is somewhere around 200 to 300 ohms together with a high slope over 11mA/V. Definitely would need stoppers to eliminate parasitic oscillation!

Rich

[GrimJosef]

Originally Posted by Radio Wrangler

... Don't forget the 5B/254M !

You certainly won't if you should happen to touch it while it's running .

Cheers,

GJ

[bikerhifinut]

I have followed this thread with some interest.

Whilst I understand some of the digs at "Audiophiles" I don't get this inverted snobbery sometimes.
Most of the "Audiophile" amplifiers on sale at more or less affordable prices seem to use current production Valves, even the "radio transmitter" valve jobs are made from valves in production somewhere in the world.
I'm thinking 845 and then down to the likes of KT150/120/88. (not forgetting 6550 etc etc)
EL34's are ten a penny thanks to the guitar amp market as are EL84's.

The people building amps from strange and rare directly heated 1920's and 30's triodes are probably as rare as the valves themselves in these days of few "hands on" hobbyists.

However to address the original question, I still think the "Best" audio valve ever, certainly in terms of "Bang for buck" was and is the humble EL84 and variants. Works extremely well in very simple push-pull circuits (Mullard 5-10 anyone?), and also versatile in simple SE circuits. And as long as you keep it within its working parameters it's a very robust little device.

My personal "Runner up" would be the KT88. My everyday power amp is a kit based one using KT88 in a circuit very similar to the Leak designs biased entirely into class A and with a couple of compensating networks for stability.
I sometimes swap the output valves around and mostly find the differences between KT88 and 6550 to be more subtle than some critics would have you think, and horror of horrors I have found a preference in this particular application for a quad of used KT90 that a good friend donated for me to have a play around with.

I'm not as clever as most of the people here in terms of electronic knowledge and experience, but I do trust my ears.
No offence intended to anyone.
Andy.