PWM Class D amp using tubes?

[PeanutNore]

I'm helping a friend with a design for a solid state class D amp he has created from scratch using some basic parts - a 555 timer, some standard op amps, and a MOSFET H-Bridge motor driver IC. I didn't know anything about class D beforehand but have a pretty good handle on it now, and it seems to me that there is nothing that would make it impossible to use tubes as the output driver(s) in a class D PWM amp without requiring an output transformer (just a choke for an LC lowpass filter).

A single power tube biased into cutoff to act as a switch with a DC blocking capacitor after it would be the simplest method and I don't see how this would not work. But, can something like a 6V6 or 6L6 actually switch on and off at 200khz or more?

For more power and to avoid the large DC blocking output capacitor, 2 tubes in a half bridge would be even better. These are usually done with N-channel mosfets and a bipolar supply. This will mean the "bottom" tube in the half bridge has "B-" connected to the cathode rather than it being at ground, so it will need a bias supply that is even more negative. I drew a simplified diagram of how it seems to me that this would work. Imagine B+ is +220vdc, Bias 1 is -60vdc, B- is -220vdc, and Bias 2 is -280vdc. The PWM input voltage swing will need to be large enough to keep the tubes out of the linear range.

[loudthud]

Tubes have no problem switching at 1MHz and above, but you need many in parallel to be able to switch the required current. Next you have to deal with the voltage drop that the tubes require to be able to switch the high current. Then you have to deal with what it takes to drive many tubes at a high frequency. You would be better off operating the tubes into a higher impedance and using an output transformer to couple to a speaker.

[J M Fahey]

Tubes are VERY VERY inefficient so any advantage of using class D is irrelevant.

Besides, let's put some numbers into it.

A 6L6/EL34 (forget 6V6/EL84 , etc. which are even worse) can put some 200 mA peak into a load.

For 100W into 8 ohms you need Ipk = 5A which means 25 tubes per side, or 50 total.

I'm not even delving on how to drive them.

100W/8r means 40V peak.
Add to that the 60V drop across each tube so you need +/- 100V rails.

And filament power for all those tubes (and the rest of the amp).

Just for comparison, a +/- 100V rails, class D amp using MosFets can give you some 500W RMS into 8 ohms, work cold (ok, warm) , weigh and cost 10% of the other one ... or less.

Now, as a theoretical concept, even a Lab experiment, yes, you can make one.

Say, a 2 6L6 amp driving a 400 ohms load for 8W RMS?

As you see, the higher the impedance, the better it is for tubes.

When you reach kiloohms they work very well ... but that means Output Transformers ... as we all know.

Yes, it will work as drawn.
It's not Physically impossible, just impractical.

[headsmess]

watch everyone scream... "man, you cant post on a thread from 2014!"

yes i can. do a google. only result. i have to chime in. someone else might be avoiding work like i am... i just dug up the front path. gimme a break!

been thinking of it with russian rod tubes, (1zh29b etc?) extra complicated... theyre directly heated pentodes.
the "power tubes" have an 800W pulse rating iirc... i think its 5W in normal operation. whatever. a battery powered 800W tube amp would be hilarious...

nevertheless... tubes can take overloads, pulses... so i think you can calculate everything for 50% duty...?

my line of thinking is keep the transformer. you still need the high voltage.

single ended or push pull, i see no reason to change the standard topology... both anodes positive if PP. grids are no issue.

its going to be designed for 200khz, rather than your lowest frequency, thats all.

meaning the core is a lot smaller. number of turns much lower. same ratios, say... iirc, 22:1 is ~8k@8R? been a while... but you only want say, 88:4 turns on the OPT itself.

(edit>) issue... thinking that the speaker is anywhere near its rated impedance when hit with 200khz... ummm. may have to think a bit more on that output bit. has to convert 200khz to...audio. hmm.

bass is no longer based on the size of the transformer itself... potential for lightweight bass amps?

seems like the transformer will be really easy... no interleaving.

negative feedback on the grids ensures a hard switch.

a project for this evening maybe... might try it with some ax7s... probably in for some awesome crossover distortion

[uneumann]

In 1997 US Patent 5,612,646 was issued to David Berning for a tube-based class D amp design.
Check out his latest at https://www.lineartubeaudio.com/prod...near-amplifier
The technology is described here... https://www.lineartubeaudio.com/the-technology
I recall seeing guitar amps based on this idea a few yrs ago.

[teemuk]

The Berning amplifier actually uses MOSFET switching devices but I guess that with sufficiently obscure logic it could be labelled as a "tube amp". Whatever sells best. ;-)

[Mick Bailey]

Perhaps the Russian 6C33C-B triode could be used to drive the speaker directly. It's been used in OTL designs driving loads as low as 4 Ohms with a fair degree of success.

[teemuk]

Yes, but why?

I understand that there's certain appeal in tubes, which is attributed to specific circuits employing them distorting somewhat nicely and reacting dynamically. They make nice analog effect processors.

But if we use tubes to build a class-d amp circuit then we can't really expect that it behaves similarly. How it behaves will really fall down to that circuit design - and the simple class-D setups are often horrid regarding operation at "overmodulation" state. Horrible overdrive performance, often "horrible" in the sense that they die quickly. Practical class-D is so much more than just devices switching. We want them to switch reliably, at proper sequence at super high frequency, and we want to prevent overmodulation from creating very nasty hard clipping distortion or cross conduction. It's not a Bassman amp.

Even OTL tube amps are far from being comparable to one's generic guitar amp. Futterman, the OTL pioneer, pretty much gave up on tubes because MOSFETs effortlessly did everything he arduously tried to achieve with tubes. And they did it so much better. Their characteristic curves are nearly identical to pentodes anyway and the OTL amps aren't exactly comparable to a generic Bassman. Most of them are more in the realm of sterile "solid state-ish" Hifi and blow up when overdriven. In practice, something like a MOSValve amp has much more of that guitar amp -style "tubeness" than an OTL tube HiFi amp.

[uneumann]

I recall seeing guitar amps based on this idea a few yrs ago.

Not sure if this is based on the Berning patent, but the GAGA is the guitar amp I was thinking of...

https://milbert.com/news

Has anyone actually played one? I see one listed on Reverb 3 yrs ago.

[Mick Bailey]

Yes, but why?

I can understand the novelty in doing something different, even if the results are not as good as doing the same thing another way. Power amps built with dozens of NE5532 opamps in parallel, choke-coupled amps, etc. Or my SE transmitting tube HiFi running 1kv plate voltage and producing as much clean power as a pair of EL84s in PP.

[teemuk]

Perhaps the Russian 6C33C-B triode could be used to drive the speaker directly. It's been used in OTL designs driving loads as low as 4 Ohms with a fair degree of success.

Yes, it can be done. The 6C33 traditionally gives rather low-ish output impedance in OTL ...when few are run in parallel as cathode followers within a loop of high negative voltage feedback. It's a perfect setup to make a tube emulation of a generic sterile solid state Hifi amp that hard clips, except that there are also arduous difficulties in providing sufficient drive for unipolar devices like tubes, they still don't work properly with four or eight -ohm loads, and overdriving may end up being catastrophical for the amp. But novelty factor is duly admitted.

And yes, tubes can be used for class-d. The very first patent for that class of operation was from times when solid state practically didn't even exist.

The tube can switch on and off though in a traditional class D audio amp it is a worse device to select for that than a MOSFET. But let's skip that and go to the real deal; the driver circuitry. Switching sequencing, dead time sequencing, modulation, overdrive performance, under and overvoltage protections, etc. Or do we just close this discussion at theoretical level that establishes what is known already, that a tube can work as a switch?