Charge pump for bias/switching supply?

[Ptron]

I was wondering about the possibilites of using a charge pump/s to derive a bias/switching supply from a 5v tap. Anyone ever experimented with this sort of thing? Dumb idea? Are there even any practical circuits that could do this?

It would have to be a voltage octupler at least to get to 40v and I'm guessing 10mA rating minimum although I guess I really don't know how much current your typical bias supply draws.

I suppose there is the diode/cap ladder multiplier but then you're talking about adding 8 more e-lytics to the circuit.

Just kinda thinking out loud here.

[loudthud]

I've tried just using a filament transformer in reverse. I needed about 60Vac so I tried using a 12.6V 4W transformer supplied from the 6.3V filament supply. I don't recall the numbers but I didn't get as much voltage out of the thing as I thought I would. I attribute my problems to the following:

Small transformers have poor voltage regulation. Designers take this into account when they design the parts. So, a 12.6V transformer might actually have a 14 or 15V winding that loads down to 12.6V at the spec'd max current. This eats your lunch and sticks you with the tab when you try to run them in reverse.

I suggest you use a 6.3V 1amp transformer and a full wave bridge on the output side to make your DC. There is a temptation to just raise the impedance in the bias adjust circuit, like using 100K pots, but this could lead to other problems depending on the tubes used.

[Ptron]

Thanks Loudthud, I'm already using a transformer as such. I'm really just curious about the prospects of using a charge pump here. A high voltage octupler+ doesn't seem practical to me but maybe someone knows otherwise.

[Steve Conner]

Sure there are. I just built a boost converter that generates 50V from a 9V battery. When I tested it, it could go up to 130V, but I only needed 50. I used a chip from ON Semi called the MC34063, with an external HV switching transistor, but there are dozens of switched-mode controller chips like this.

Whether you actually want a switched-mode converter pumping out EMI inside a tube amp is another question altogether. The MC34063 in particular is a cheap and nasty controller that can have audible frequencies in its output under some conditions.

[Gregg]

The MC34063 in particular is a cheap and nasty controller that can have audible frequencies in its output under some conditions.

I have some experience with that one but I could never make it inaudible if I follow the schematics. For some reason the inductor starts whistling and it goes into the audio. According to datasheet the chip needs 1.25V on one of its pins. After some experiments I noticed that when that voltage is less than 1.25V the thing stops whistling but then the trimpot is useless.

[Don Moose]

The thing with switchers is that you gotta shield the fark out of the nodes where the inductor connects to the switch and where the switch connects to GND ... then you gotta run the switch frequencies as absolutely high as they'll go - 600kHz to 1MHz.

That super-high-frequency mode then makes selection of inductor, switch and driver components ad lot pickier.

Stick with transformers.

[Steve Conner]

Gregg, I guess I should have said "practically all conditions" :P My MC34063 circuits have always whistled or sizzled too, except the one I made the other day. But I think if it were more heavily loaded it would squeal too. I believe the problem is that the MC34063 operates in an all-or-nothing mode, it doesn't have proper pulse-width modulation.

When that pin drops below 1.25V it means that you've lost regulation. The regulation process is what makes it whistle, as the oscillator is pulsed on and off at an audible rate

If you want a quieter switcher, I'd recommend trying a TL494 or LM257x instead.

In a previous life I designed switched-mode drivers for telecoms lasers, that had to meet a SNR spec. I managed to meet it, but no matter what I did, I could always detect some trace of the switching frequency in the laser output, using a photodiode and FFT analyser. The customer didn't seem to mind though

[Gregg]

I've made several tube pedals with HV switching supplies but I used a simpler one where increasing the clock frequency /above 25kHz/ affected the voltage. I was able to get 330V feeding two 12AX7s. In both cases even without regulation /I assume because of the low current load/ the voltage fluctuations were in the range of a couple of volts.
Although it's claimed that such power supplies make the sound harsh and/or hard all people who tried those pedals liked them and didn't make any remarks concerning this issue /they didn't know about it anyway - hello blind testing!/.

[Ptron]

Well, how about this? I copied it from Forrest Mim's Radio Shack book. No switcher involved. Caps could be pretty small, like 10V.

[loudthud]

I just built the octupler using 220uF 35V caps and 1N4152 diodes. I added a 56K resistor as a load. With 5.00VAC input I got 46.1V out. With 6.31VAC input I got 59.9V out.