Sweet folly: a tube SMPS

[Elvee]

Let us indulge in a little fantasy.

Many people are against the use of SMPS's in audio.

Other say that tubes are outdated and are worthless compared to solid state.

Here is a project to reconcile them all: a tube SMPS.

This example delivers ~300V @ 60mA, and uses a single pentode.

It operates at 120KHz, and the transformer could even be air-cored.

If a P-series tube is used, the filament can be powered by a capacitive supply.

Just for fun, of course, at present it has only been simulated, not built yet.

[iko]

Can you please post the sim file? I'd like to have a little fun with this.

[Elvee]

Here it is, with the pentode model:

[the_manta]

Amazing :-)
This really looks interesting. Good way to spend the tons of PL504/PL36 :-)
I'm not really good familiar with SMPS. What kind of circuit is this ? Looks like self resonating (500µ & 4.7n). Where is the green signal measured ?
IMHO it wouldn't be a problem to use Valve rectifiers for the secondary. Fast switching shouldn't be a problem.

[Andrew Eckhardt]

If we should have all this heat and sunusoidal stuff couldn't we put some output voltage feedback on it too?

[iko]

Doesn't look like the schematic you posted first.

[ryuji]

interesting idea, but at that frequency it doesnt matter if its valve or SS, your not going to hear anything that its producing for carrier, the 'bad' smps's that people talk about from time to time run on 16~30 kHz which on the lower end is is possible to hear diode switching impulses

[Elvee]

Originally Posted by the_manta

Amazing :-)
This really looks interesting. Good way to spend the tons of PL504/PL36 :-)
I'm not really good familiar with SMPS. What kind of circuit is this ? Looks like self resonating (500µ & 4.7n).

Yeah, it is basically a class C power oscillator.

Where is the green signal measured ?

It is the anode signal of the pentode.

t If we should have all this heat and sunusoidal stuff couldn't we put some output voltage feedback on it too?

Mind you, it is not that inefficient; I pushed the circuit to its limits to get 300V on the load, which wasn't very good for efficiency, but at more modest power level, I had ~60% efficiency (excluding the heater), and tube experts could do much better.
Inductive coupling could improve the situation, and the oscillator could then be made a blocking one, but this would degrade the relative cleanliness of the signals.
I have also tried Royer versions, which are even cleaner and offer better efficiency (but at the expense of a second valvle).

Some mode of feedback could be included, via the screen grid f.e.

Doesn't look like the schematic you posted first.

Ooops!
Sorry, wrong file, this should be the right one:

interesting idea, but at that frequency it doesnt matter if its valve or SS, your not going to hear anything that its producing for carrier, the 'bad' smps's that people talk about from time to time run on 16~30 kHz which on the lower end is is possible to hear diode switching impulses

I think the switching frequency is not the only objection, the internal impedance is another one.

[trobbins]

You could even simulate the conducted emi into a mains LISN for fun.

Do you get the same mains frequency output ripple as what is on Vht?

Ciao, Tim

[DigitalJunkie]

I recall some of the older Tek scopes using a little switcher in an AL box,for the HV voltages for the CRT.

I might go see if I can hunt down the circuit.

http://i199.photobucket.com/albums/a...s/Tek545HV.png

It uses a 6AU5 (V800) and a 12AU7 (V814)