semiconductor mag

For those of us peasants still using 2N3055s, what would such parts enable one to do?

John hasn't bothered to tell us that, but GaN fets are fast, and you can use them to run your inverters at several MHz, if you can find ferrites that are resistive enough to tolerate these frequencies.

He's not fond of wound components that he can't buy of the shelf, perhaps because he'd have to learn about that kind of detail if he wanted to get special purpose parts wound.

I'm doing a physically small high voltage pulse generator just now. The goal is at least 750 volt pulses with < 2 ns rise/fall into a 50 ohm load. [1]

Avtech did it with SiC fets with drift step-recovery diodes to snap up the edges. That's a lot of work. SiC is magic.

I rarely use bipolar transistors any more. Mosfets are faster and easier to drive.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Presumably he means an Eaton DRQ inductor.

If you want to get to put together an inverter to get up 750V, off the shelf transformers rarely offer the kind of step-up that you want to do it right. Photomultiplier power supplies did seem to rely on custom wound transformer and custom made core formers too.

I quite like transmission line transformers.

I used one back in 1979. It was a bit of cheat, but it worked. They are fast.

Ghiggino, K.P., Phillips, D., and Sloman, A.W. "Nanosecond pulse stretcher",Journal of Physics E: Scientific Instruments, 12, 686-687 (1979).

We use several of the DRQ series. I'll use a DRQ127 this time, probably 33 uH, as an autotransformer flyback followed by a few stages of C-W multiplier. That's cheap and easy.

Just apply dollars.

Actually, a PMT supply should be a C-W multiplier. It doesn't make sense to generate a high voltage and then waste a bunch of power to make a multi-tap voltage divider.

But I don't think people use PMTs as much as in the past. Photodiodes and APDs and SPADs are more sensible.

This is easy:

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Not as cheap as a properly designed inverter, but if you have a phobia about designing transformers, it's going to look a lot easier.

Back in the day, the photomultiplier power supply market was small but competitive. Cambridge Instruments shipped one with every scanning electron microscope, and never bothered to design or make their own.

It's an attractive idea, but the photomulitplier supply voltage is often the gain control mechanism for the whole imaging system. Having a system where the gain control works faster on the last dynodes than it does on the dynodes at the top of the chain makes the idea less attractive.

The last dynode chain that I specified had a high voltage zener between the last dynode and ground (which improved the linearity of the photomulitplier response), and three zener (for a 270V drop) between the photocathode and the first dynode (which minimised the multiplication noise).

Tell me how you woulkd get that with a string of Cockroft-Walton multipliers.

Depends on the application. Photodiodes are insensitive. APDs are noisey. So are are SPADs, and SPADs can only detect single photons, which makes them unattractive when your source pushes out too many photons.

The coax connectors aren't all that cheap. Cheaper than the ones that I thought about using back in 1978, but still not cheap.

I do dream of a transmission line transformer wound with RG405 but it would be horribly expensive.

The surface-mount U.FL connectors cost us 6 cents each.

The "winding" is a stock connectorized cable assembly, costs us $9.

The pot core halves are 20 cents each. So the whole tranny is under $10. Asssembly takes a couple of minutes.

The turns would short. And it would be interesting to wind.

Our company policy is to avoid horribly expensive.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Not if you wrapped them with transformer tape.

Not if you used the conformable version - which uses solder-soaked braid as the outer.

As is everybody's. If you wanted something that was good (as opposed to barely adequate) to a couple of GHz, you'd have to violate that policy.

Sometimes gold-plating does make sense.