Tube DIY Asylum: Re: Thinking out loud about HF filament supplies by Kurt

The link below is what Thorsten had to say about this subject.

I did finish building a box that delivers a 32 KHz open loop sinewave (not phaselocked to anything) to 71A or 45 heaters. My evaluation after using this heating system is that it "cleans up" the sound, making it less lush. This turned out to make the 71A a bit sterile and the 45 to be just remarkable in performance. I did find problems with using this method on anything other than the output tubes. It made bad sound when applied to my 01A drivers also. So I have battery filaments for the 01A DHT drivers and ultrasonic for the outputs.

I found out that a high distortion sinewave was inaudible compared to low distortion, but if it AMs or FMs then there will be noise. I did not use square waves but I found it's tolerant to high distortion sinewave so long as it's very stable.

Here's how to experiment with this without building something terribly complicated: Get a good function generator or audio signal generator, a synthesizer would be even better. Then get an SS stereo or multichannel amp to drive individual small high frequency toroidal transformers (for this I used small 1" diameter ferrite toroidal cores with about 20 turns for primary and 10-20 turns on the secondary, depending on the resistance of the output tube). The secondary is the heater voltage. Hopefully the SS amp has enough BW for the frequencies you want to try, and hopefully it has built-in protection against excessive loading/shorting in case of mistakes.

I made a big mistake on winding my toroids at first. I used magnet wire which got cut by the edges of the toroid and the ferrite is conductive. So then I used 24ga teflon-insulated wire and checked it out with an ohmmeter.

My personal opinion is that if you use steady sinewave you won't have a problem with digital sources. And anyway, the real benefit shows up with the higher resolution sources, analog. Digital has enough problems that a sinewave is not going to further degrade much I think.

Also, I think phaselocking is not really important from a technical standpoint. If you use a crystal-controlled oscillator to get the steady sine or square wave and set the frequency in some place close to a desired innocuous place, the beat frequencies that come out by not being phaselocked ought to be very low drift frequencies, which would be infrasonic. A "free-running" synthesized function generator could show if this works out okay.

Ultrasonic heating (Open in New Window)