No more high voltage PNP transistors?

Does it gotta be PNP bipolar? Does it gotta be cheap?

There are dpak n-channel mosfets up to 1500 volts. I cascode them with an optocoupler to avoid nasty gate voltage drivers.

Depending on your application a depletion-mode MOSFET might do for the pull-up, as in...

...Jim Thompson

The lowest voltage they have is 2500V. PN OC025 Very simple via 1.7 Vf led at 50mA. PD is only 125mW but for most apps of this nature, I think that would do just fine.

They go up to 25KV, and it looks like they are working on a 50Kv type.

jamie

It's only for bursts that it needs to do this much. SOT223 would be nice.

For at least one of them, yes. But I'll see if I go N-channel or do the cascade.

Not within reason. Meaning $1 would be ok, $2 painful, $5 not ok.

The drive in my case would be nasty. I could do an RF+rectifier drive but it ain't pretty.

Unfortunately they only go to 500V. The STN9360 at least goes to 600V but I'd have to find out if there are any pathologies when using it as a linear device. It's from ST and getting answers from their support has proven to be like kicking an oak tree unless it's a key account project and this project isn't.

We just bought 50 pieces of 2SK4177, 1500 volt n-channel, for $3.25 each. 1KV, in quantity, should be a lot cheaper.

Optocouple it! If you don't need speed, a PV coupler would be spiffy. It becomes a kilovolt linear SSR.

Two depletion fets and a regular optocoupler would work, too. Cheap.

On a sunny day (Tue, 08 Dec 2015 13:33:04 -0800) it happened Joerg wrote in :

Use a NPN upside down?

Everything is relative ye noo

high voltage optocoupler?

Upside down wouldn't work. You have to use it inside-out. ;)

Cheers

Phil Hobbs

Do tubes come in P channel? Can you make holes in vacuum? Do you need to use antimatter? Interfacing could be a problem...

Well, you could use an ion gun. Might be a bit hard on the anode, as well as being thousands of times slower.

Cheers

Phil Hobbs

Very good, and even US-sourced.

I could do that. But then I'd be about at the complexity of a beanstalk which would also work here and be even cheaper.

A 6KD6 in P-channel, that would be the solution :-)

Sure, but only the isolation path is high voltage, not the output device. So then I'd also need depletion mode FETs which are a bit boutiquish and don't come >500V either.

This isn't bad:

That is nice. I'd probably add some angst diodes.

But I'll use the FMMT560 in a stack. At about 15c cents a pop it is hard to pass up.

On a sunny day (Wed, 9 Dec 2015 10:41:46 -0500) it happened Phil Hobbs wrote in :

Unless Joe post a circuit it is hard to tell, I think he often refers to inside out modified transistors by accident so to speak.

If it is a single stage output I really see no problem making it NPN. And dual NPN totempoles were in many audio amp output stages, even video out drivers.

That will need a couple of resistors to bias the halfway-up base. That's OK if you don't mind that current.

The FMMT560 has a guaranteed hfe of 80 until way past 10mA so the resistor losses are reasonable. I needed a resistor there anyways and it had to be split up because of voltage rating limits so it doesn't even cost me extra. Nice thing is it also comes in a SOT223 flavor for when more muscle is required.

Of course, one must not forget the transistor capacitances in all that when things have to act fast. That's where the beanstalk has a real disadvantage. But the simulator says it's got plenty of margin.