LTSpice model for a SiC MOSFET

I never liked 807's. Ugly tubes.

Imagine a buck switcher with two HV npn's and one CCFL transformer as both the main inductor and the base driver.

I had a similat thought, posted just before I saw yours. Honest.

Ah yes. For that market, appearance is everything.

A much better-looking bottle would be a PY81 but, although it is specified to withstand a peak anode voltage of 7 kV, the maunfacturers only allow that under cutoff conditions. It also uses a standard B9A base, which is still easily obtainable.

But you don't know enough to realise that the Baxanadall class-D oscillator is a self-oscillating two transistor half-bridge - of a particular sort. The guy with the 1kV 10uA DC source had contacted me because I have the Baxandall paper on my web-site, so he seems to have had much the same inspiration.

His very preliminary example circuit wasn't all that Baxandall-like, which is why I want to put a simulation together that he can play with and get some insight into what it does.

A great audiophool prodcut would be a power amp with a pair of 833's and lots of custom transformers and oil caps and things. Figure about

There is a big difference in the voltage stress seen by the transistors.

The half bridge exposes each to 1kV max, the push-pull to nearly double (I think you quoted 1.6kV).

Baxandall topology was great for tape recorder erase/bias oscillator kinda stuff at low supply voltages but at HV not such a great choice.

If you've ever built a Royer inverter, you should be able to remember the horrible switching spikes you get when you switch from one side of the half-bridge to the other.

The Baxandall configuration does lumber you with a smooth peak at 1.67 times the supply voltage - the peak of the half-sine wave train that has the same average amplitude as the supply voltage - but it does its switching at the lowest voltage points, and doesn't generate high frequency interference.

It's lot better at stepping up low voltages to high voltage than the Royer inverter, and Peter Baxandall seems to have invented it for a photomultiplier power supply, but really high voltage MOSFETs seem to make it a plausible choice for going the other way.

The transformer gets pretty horrible, but it should be practicable.

I have recently been told of one - the output transformer alone cost UKP

It amuses me that there is great prestige value put on using triodes "with no negative feedback", completely ignoring the fact that triodes have built-in non-linear negative feedback because of the influence of the anode potential on the grid-cathode potential gradient. A triode with the negative feedback removed is called a 'Pentode' ..but they are condemned for being non-linear and "unmusical".

My bad. I underestimated the selling price.

That only had to deal with 160-340V, and used bipolar transistors in proportional base drive 2 x transformer type Royer.

But the principle of voltage stress reduction and lower frequency operation is right on the money.

RL

Where does 1.67 come from?

Try asking Grok, Bill. Or else your preferred AI assistant. Works for me.

He could strain his imagination to its limits and google

ccfl transformer

And then he could try designing electronics.

AI does produce quite a bit of nonsense, and we know how enthusiastically you embrace implausible nonsense.

Some of us are more interested in objectively verifiable reality.

Did it find anything for you?

I've been doing that for some fifty years now, and have had some objectively verifiable successes, not that you are temperamentally equipped to appreciate other people's successes.

Try and find some other put-down. You might find one that works, eventually - that does seem to be how you find the "designs" (such as they are) that you post here.

Integrate a series of half-sine peaks that get to 1.67V and the voltage averages to about 1V.

You can do it as purely mathematical exercise, and I did it years ago, and that's roughly the result I got.

There's a voltage drop across the switching FET that's on and the Baxandall circuit doesn't product perfect half-sine waves, so it hasn't got a lot to do with precise reality, but it's good enough for preliminary design.

If senile dementia hasn't set in too far I could probably do it again.

Yes, but thats the centre tap voltage. Due to the autotransformer action the stress across the off transistor will be twice that, or

It took me a while to wake up to that. I think that there's an option where you'd use four switching transistors, two to alternately ground either end of the main inductor and two more to alternately switch the feed inductor into the other end of the main inductor, but I haven't worked it out in any detail. It's very much in the brainstorm state at the moment

I have found a 4.5KV MOSFET, the IXYS IXTT02450HV which could survive in the standard Baxandall configuration.

It's $US45.24 each in small volume (which really is excessively expensive), and I've asked for Spice model, but if Infineon is anything to go by, I'm not going to get it anytime soon.

If either Spice model shows up I'll try and put a simulation together. It has nearly got to the point where I should try and bodge a MOSFET model that I have got access to into something that would fit one or other data sheet, but that's hard work, and my model isn't going to be all that trustworthy.

Well, it does at times produce a lot of nonsense, I grant you. However, it would only cost a few moments of your time to try and you might well be pleasantly surprised. It's getting better and better as time goes by.