A very silly circuit

PV optocouplers are cool for driving isolated gates, but maybe not available in your junk box.

I'm designing a kilovolt-range linear voltage regulator, and a PV is a really easy way to drive the pass transistor gate.

Hey, a packaged SIP dc/dc converter would make a nice SSR gate driver. $4 or something from Mouser.

That is the point of this circuit. A PV photocoupler needs a power source. This circuit *is* the power source.

Since speed is not an issue, a 555 and a few capacitors in a doubler would work to drive the Mosfet.

No, a PV optocoupler makes it own voltage. Just add it to the 12 volts to drive the gate.

Since when? They use a phototransistor, not a PV solar cell.

Exactly. I was also wanting reasonable switching times to avoid SOA problems.

I considered that. One of the surplus houses has a steal on DIP dc-dc converters, but I'd have to send off for 'em.

I also considered direct-coupling and charge-pumps, but those were partsier.

This was just for fun, to see what it would take. That's why it's a very silly circuit--I spent more time posting it than designing it.

Dissipation is surprisingly high--just 10A through a 10 milliohn FET makes 1W, which is about as much as I'd want in the original tight, insulated box under the hood.

IRL I swapped in a spare from a now-unused accessory. It's hard to beat relays for toughness and low contact resistance.

The radiator cooling fan is the one that failed. The swapped-relay works, but I won't know until summer if the temp sensor driving its coil is good. :-)

Cheers, James

Or even just a discrete oscillator in a charge-pump arrangement.

I think the parts-counts work out to be about the same, and I just plain trust transistors to be tougher than ICs under the hood.

Cheers, James Arthur

The PV couplers John and Tom are talking about actually do exactly that--they couple an LED to a solar cell array, which drives a FET.

They're slow, and expensive, but neat. I'm not sure if they come sufficiently hefty to run a radiator fan, and small enough to fit in a 2 x 1 x .5" metal case though.

Cheers, James Arthur

Rickman is obnoxious and generally wrong, so I ignore him. Sloman clone.

I mean the little ones, like this:

Really cool. They make about 8 volts, plenty to drive a mosfet gate. Yes, they will be slow when driving a lot of capacitance. We pay about $1.60 each.

This is essentially what's inside small SSRs. Which can cost half that price!

Vishay makes a dual, 13 volts typ output. In series, that's 26! Four make 100 volts! Must be good for something.

How much current in the load? Might have some SOA problems with the soft turn-on/turn-off??

Might make a Schmitt arrangement, charge up cap _then_ connect to gate and vice versa for turn-off. ...Jim Thompson

Oh, that's pretty sweet. 24uA is enough to turn on a pretty big FET in

I have some Panasonic PhotoMOS SSRs by the partial-reel but they're unsuitable here--several ohms, with no way to access the PV array.

Yup, gotta be good for something. Driving varactors without fuss or EMI, for one.

Cheers, James

Yes, I see now that he used *exactly* the term PV. Sorry for that. I've never seen one before.

I don't actually know. The hypothetical load was an Acura radiator cooling fan motor, whatever that draws.

I don't think so, but I didn't check. The switching times are well under 1mS, purposely, with the gate capacitor there to intentionally slow things down. Lowering C1's value speeds the switching, if needed.

Okay, I just checked the SOA on a 1.2 milliohm FET, and it looks like 100uS switching is just enough to avoid SOA up to 10A (for this random monster FET). It might need a tougher FET or slightly faster switching to be solid. Unknown, since the load is a mystery.

Right now it switches Q3 'on' in 30uS (fan-start, the worst case load) and 'off' in 120uS. Not bad, but might need beefing.

Yup. Hate to use extra parts though, if avoidable. In your world parts are free. In mine they have to be bought, inventoried, and soldered.

Cheers, James Arthur

If you are actually going to build this, why not use any of the many SSR devices in a package? What will this circuit do that a commercial unit won't? This looks like a much larger circuit than you could do with an SSR.

The one that failed was made by who?

I guess that's why they call them PV optocouplers.

On Saturday, October 3, 2015 at 2:09:48 PM UTC-4, snipped-for-privacy@yahoo.com wrot e:

There's an oscillator (I think) in the diff pair transformer part but I'm not seeing it.

(At home, I don't have ltspice here.. well other machine.)

George H.

I'm not building it, that's why it's a very silly circuit. I already swapped in a spare relay from an unused accessory. This was just for fun, to see what it might take.

Will it fit in a ~2 x 1 x .5" package, to fit in the original space?

It wasn't marked. It lasted almost thirty years bolted onto an automobile's radiator bracket, so I have absolutely no complaints. Failed with coil open.

Cheers, James Arthur

In many places cars are fine with a permanently dead rad fan.

NT

The biasing biases both transistors linear, and the base drive from T1c guarantees positive feedback. Oscillation ensues.

The quasi-diff pair makes a modified blocking oscillator, but with a defined current-limited drive into the transformer, instead of relying on Q1-Q2 characteristics to set that.

It's set to use 20mA, mostly to power the indicator LED.

Cheers, James Arthur