|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
OK, I've been thinking a bit more about ultrasonic DHT heater supplies. I've looked at possible designs and topologies and run some simulations. I'm almost ready to do a real design and fab some PCB's just to try it out (for my own amusement).For sake of economy, I'd prefer to start out with a fairly small (power-wise) circuit, suitable for a 300B or 6B4 or the like - maybe 12 watts / 2 amps maximum. If it's worthwhile the design could always be scaled up for 10V/5A filaments (211, 845, 813, etc.)
I've looked at different schemes to generate a ~100kHz sine wave or square wave, using PWM, PFM, and analog methods. The analog methods are big, inefficient, and expensive, so I've given up on them. I also looked at ways to make a sine wave, but found difficulties - to generate a ~100kHz sine wave digitally (like a "class-D" amplifier) takes really high clock rates (like 1MHz), and gets into more exotic magnetics and topologies; to do a resonant converter, it looks to me like it takes a pretty big inductor to get the "Q" high enough into a 4 ohm load (300b filament) (though I could be wrong here).
So, what I've wound up with is a square-wave converter. The downside is that, being a square wave, it will produce odd harmonics (300kHz, 500kHz, etc.) and it will have more of a tendency to couple into things. I'm guessing that this will not have too much of an audible effect...
What I envision is a small PCB, maybe 2.5" square. The input would be 5V or 6.3V RMS AC (so you can retrofit it into an amp with AC filaments) which is rectified, buck regulated, and turned into a square wave running at the input line frequency x 2048; 102kHz (50Hz line) or 123kHz (60Hz), phase locked to the AC so it doesn't cause a beat between multiple units. The output would be regulated via a third winding on the output transformer, using a very slow (like <10Hz) regulation loop to keep any noise below the audio - this should work since a filament is a pretty static load. The output winding (connected to the filament) would be an isolated, center-tapped winding with no other connections, so common-mode there's only a small capacitance (maybe ~100pF) to the primary.
This can be done with small off-the-shelf switching magnetics and standard power supply stuff. I'm actually looking at a fluorescent lamp ballast controller, the UC3872 from TI, to control the switching.
So... first of all, does this sound like a worthwhile thing?
And secondly, I don't have a 300b amp, only 813 SE and 6B4 push-pull amps. I'd need a victim... uh, volunteer or two... with a decent SE amp (and decent ears) to evaluate it.
If it was worthy I'd probably turn it over to somebody who wanted to sell PCB's or kits.
Follow Ups:
Pete,I'm sourcing parts for a pair of 01A-71A monoblocks, and intend to use lead-acid batteries on the filaments. Initially, that is. I'd much prefer to go to ultrasonic AC, as described by Kurt Strain, but that'd be a whole new project. If you end up making a PC board to implement such a supply, I'd be very interested in 4 of them, even though I would be installing them a few months down the line. In any event, thanks a lot for mentioning the topic and describing your ideas.
Sure Pete, I would evaluate it. If I had nothing to do with time, I could pull out my analog 160kHz DHT filament supply, modify it a bit, add a step down transformer and compare it. My sinewave is synthesized by a micro with a slow start.TonyB
Pete,I am putting my hand up to be an evaluator.
Kevin Carter
K&K Audio
www.kandkaudio.com
Hi Pete,Want someone in the UK with 50Hz mains?
I have a SE DC 6B4G Mono Blocks and soon to add 211's on the end of it.
This post is made possible by the generous support of people like you and our sponsors: