KT88 Class A push-pull

[Colossal]

All:

The Komet Vimana (a lovely sounding amp) is claiming 54Wrms from a pair of KT88s running in Class A push-pull. The Vimana 25 is EL34 based also running in Class A push-pull and it puts out 25W. I have a true class A EL34 push-pull amp and it is putting out about 25W cathode biased. The Vimana and Vimana 25 are also cathode biased and the specifications are saying specifically "true class A". I am not disputing that, but 54Wrms from a pair of KT88s? With a quite flattish loadline (7-8k), centered idle point (centered between saturation and cutoff), and idle current right up against the max diss curve, I am not seeing how you can get much more than 26W out of a pair of KT88s. Am I missing something?

Any thoughts?

[Stevem]

Well whatever it can produce in single ended class A is exactly half of what it can produce in push pull class A.

Since a EL34 typically only does 11 watts in SE class A, I too can’t see how two 88s in PP class A outputs that stated wattage.

[Colossal]

Thanks, Steve,

I am scratching my head at this one. The =C= Svetlana datasheet shows a KT88 application for SE operation and they list 19W output (with the plate at 400V and screens at 225V). They do not list the Raa however, which is of no help. Other applications list 14W or SE KT88.

[jabguit]

KT88 SE.png

[Colossal]

KT88 SE.png

Yeah, I have spent a while fiddling with various tools and the most I've been able to theoretically squeeze out of a pair of KT88s running class A push-pull is 31W. 27W from g1=0 to saturation is typical.

[didit]

Hello Dave —

Any chance of actual measurements? They explicitly imply it has some distortion. With high plate voltage and a matching transformer impedance its almost believable that they’d achieve power with wretched looking output pushing a 1kHz square wave through.

Best ..

[Colossal]

Hello Dave —

Any chance of actual measurements? They explicitly imply it has some distortion. With high plate voltage and a matching transformer impedance its almost believable that they’d achieve power with wretched looking output pushing a 1kHz square wave through.

Best ..

Hi Ian,

I have not been inside a Silver Cloud (the earlier, limited release) or the production version, theVimana, so have no other information to go on, other than what is stated in the specifications on Komet's website. The Vimana 25 is listed as a 2xEL34 Class A push-pull amp and it is specified as 25W, and I believe this. I have my own 2xEL34 Class A push-pull build and the tubes are dissipating 24.5W each at idle, so I believe Komet is doing something similar with their Vimana 25. I don't believe Komet is playing the marketing card of saying "it's true Class A because it's cathode biased" but I am trying to understand how they are claiming 54Wrms for the 2xKT88 amp. You could be right about that being based on smashing the amp with a square wave, but you'd have to push it into Class AB to get that power out, unless I am not understanding the conditions. For true class A p/p, I'd expect a high idle current (~105-110mA for a KT88), lowish plate voltage so as to be centered on the load line, and a relatively flattish load line (6k6 to 8k or so), but that only yields 27W or so. The specifications also do not say it is an ultralinear amp either. It just says true Class A and individually cathode biased KT88s.

[R.G.]

If it was a car amplifier, I'd think that they estimated peak instantaneous power for a 1uS period.

[Stevem]

And on the cars charging system you would need a 5” diameter by 1.5 ft long 1 farad cap to sustain that !

[didit]

I can, using some dubious theory to help lift a few parameters, get to roughly 33W --- It'll be a tube eater. I'm skeptical there's benefit derived beyond spec'manship. Ultralinear doesn't help juice total measurable power but will smooth off a bit of roughness.

..

[Colossal]

I can, using some dubious theory to help lift a few parameters, get to roughly 33W --- It'll be a tube eater. I'm skeptical there's benefit derived beyond spec'manship. Ultralinear doesn't help juice total measurable power but will smooth off a bit of roughness.

..

Thanks. Your results match my own. I was able to get to 31W in one scenario, but the rest yielded 26-27W. I am interested in this because of my own build which I use as a clean pedal platform. I tested it with an old pair of 6550s, but the EL34s were better sounding. I am wondering about KT88s though, hence my interest in the Komet amp. But it is likely that they are not achieving 54Wrms then.

[Fischerman]

This will probably be of no help but I tried building a mini-Major. A 2x KT88 Marshall Major type of thing. I got custom transformers from Chris Merren and he said with the ~500v power supply that his PT would provide that 100W output was no problem.

The amp never sounded that good so I sold the OT, bought a MM Hiwatt OT and built my 'David Gilmour Hiwatt' into it. It was a Hiwatt chassis with Hiwatt-ish layout/boards so it all turned out well.

[Stevem]

This from Kevin O’conners book does 50 rms from four 6550s in SE, so to me this proves that your not getting 50 rms from two 88s even if you where able to use up to there 800 volt rating of there plate.

Might be able to use just 3 and use them up faster, also note it’s not cathode biased.

[Colossal]

Thanks for posting that, Steve. Interesting design. Yeah, the amp in question can't be class A if it's putting out 54Wrms.

[Helmholtz]

The theoretical maximum efficiency of a class A amp is 50% at full clean output.
So if PDmax is 40W, you could get 20W in SE and 40W in PP.
But that would require ideal power devices (tubes) that swing to 0V at the peak current.
As no such tube exists, class A efficiency is always lower than 50%.
Additionally the OT typically "steals" around 5% of the output power.