Simple isolated mosfet driver

If I understand your circuit, it puts a lot of voltage across the opto output. A shunt circuit wouldn't. But then there *are* optos with 400 volt phototransistors. [1]

A photovoltaic coupler would be nice, as it can drive the gate directly and doesn't need a high-side supply. They tend to be wimpy,

10s of uA output, so they would be pretty slow. An advantage is that there's no sneak current path to the load.

Or do this maybe:

V+-------+---------+ | | | | Q | | | | d +--------g | s c | b | e | | | | | +---------+ | | | load

where Q is a Supertex depletion fet current limiter, cbe is the opto phototransistor, and there should be a zener g-s which I'm too lazy to draw. It does leak a litle current into the load.

John

[1] and I just saw an ad for a 10,000 volt photodiode.

Probably a gassy 1B3. ;)

Cheers

Phil Hobbs

Sno-o-o-ort ;-) ...Jim Thompson

-- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

| 1962 | Help save the environment! Please dispose of socialism properly!

It will work as long as you have some head room to bias the gate.

Also, you should use a limiting zener diode from the Source to the gate to prevent over voltage for Vgs..

For example, if your fet has a full turn on of 10v's above the source, you need to have at least that much head room from the supply that is on the DRAIN side.

In all, it means your drain supply must have 10 or more volts then you'll ever expect on the source. You also need to factor in loss via other circuits like the opto-isolator. etc..

You really should look at a charged pumped isolated high side driver.. It would save you a lot of work..

You could also consider a P channel on the high side :)

It looks like a surface-conduction silicon tube thing maybe, which can be a little leaky. This looks like fun for, well, *something*

With a CTR of 0.5, power gain is several thousand.

I once worked with some guys who made a cone-shaped bulk silicon thing that could be blasted from above with a high-power laser. It would switch megawatts in nanoseconds.

JT is starting to snort again. He should see a doctor.

John

Hmm, I've been doing a few things along that line myself lately...but not at 10 kV!

I vaguely remember seeing those. It was sort of a giant APD iirc.

C'mon, you used to build stuff out of dead TVs and stuff too. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058

email: hobbs at electrooptical dot net
http://electrooptical.net

I was _laughing_ at Hobb's suggestion of a 1B3 as a photo device. Might actually work.

But, in matters photo-multiplication, I'm engaged in designing around a Hamamatsu R8900... discrete circuitry, not integrated.

As for health, I've been two days running up and down a 12' ladder trimming the Christmas tree, then this morning, unfortunately, working

16' up on a skylight in the kitchen that jammed open, madly working to disassemble control arms before the thunderstorm hit... just minutes after I got it closed :-(

And I'm just 2 months from the big 7-0. So I'm probably healthier than you... or Slowman ;-)

(Though, I will admit, all that "laddering" left me with that gnawing ache indicating that the right hip joint will soon need replacement. I'd better hurry and get it done, on the leftist's dime, before Obama declares me "surplus" :-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
                    Help save the environment!
              Please dispose of socialism properly!

Oh, I have great respect for the 1B3, especially as a high-voltage gain element.

John

Simplest soloution if you dont need high speed switching is something like this.

VO1263

Thanks, That might do the trick. A bit expensive though for something that seems relatively simple. The method they use with the fet for discharging the gate might be all I need with any old optocoupler method.

You have any recomendations? Only ones I have seen are for low voltage or for switching only(rather than linear).

My original idea was somewhat similar. The problem is that the initial state is with the load recieving the full voltage. Hammy gave a datasheet where they use a jfet to discharge the gate and keep it at 0V relative to the source when no current. It should work well and also removes any need for protection.

Not quite, it says it's a diode... Vf = 12V or so.

How do they make those, anyway? Are they monolithic? Even with a wide-assed intrinsic region (PIN structure), I don't know of 10kV being held off by a single junction. Besides, such a junction would have to be so thick that diffusion effects within would not be negligible.

I can still imagine them being made monolithically, but laterally, so you basically make a stack of PN's, shorting every other so it makes an always-on SCR.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

Yes but it saves you from having to make an isolated supply.

Page 18 of this pdf shows how to use a HS driver for continuous on time. It also shows some of IR's HS drivers.

Some of Fairchild's HS drivers.

FAN7371 4A, FAN7382 HI/LOW , FAN7361

These people make diode stacks, so maybe that's what it is, a bunch of diodes bonded together, maybe soldered. But it is awfully leaky, so probably they are exposing the edges of the junctions. I'll try to get a sample and see.

These guys are really expensive. They make a nice small C-W stack that I wanted to use, but they wanted like $60 or something for it. I did it with parts for a few dollars.

John

Dead TVs? No, I'm pretty sure that was Dr. Frankenstein, when he was exploring his feminine side...

--
Offworld checks no longer accepted!