high voltage supply breadboard

I've been bit by high Qg, so now I usually make sure it's as low as I can find else the driver becomes more robust to drive it.

Cheers

I think a bigger problem is the drain capacitance. At some flyback ratio, all the available energy in the inductor is used to charge the drain capacitance, so you can't get any more voltage output. And that stored energy is lost, wrecking efficiency.

A higher transformer ratio helps there, assuming the capacitance in the transformer secondary doesn't become a problem too. Fortunately, I'm running a few watts out and I don't care too much about efficiency.

Class E, but I can never get the darn things to work. Tesla coil?

Is a bipolar out of the question?

This works well enough:

I originally wanted it to be LLC resonant (hence the oscillator structure), but it turns out the transformer really is a pretty good transformer, so it doesn't really current-limit in a resonant fashion. A more conventional forward converter would behave just fine.

Between the coupling capacitors (the pairs of 5.1nF caps in the doubler) and the foldback current limit in the 'controller', it behaves well enough, above low voltages anyway.

It's big and huge and inefficient, so would probably do a fine job.

Tim

-- Seven Transistor Labs, LLC Electrical Engineering Consultation and Contract Design Website:

Yes, this businees of restricting yourself to an existing part on the BOM can easily be overdone. How about an FQT4N25.

That's interesting, because a bipolar might have lower output capacitance. But it would need a low-voltage high-current base driver. If I stay with the LTC3803 controller, I might need some interposing base driver circuit, running off a +5 supply or something. That's not unreasonable.

GaN would be cool, too, one of those EPC parts maybe.

Doesn't hurt to try!

My secondary restriction is to what's already in stock. We have IRFL210, which is about twice a ZVN4424, close to your suggestion, so I'll try that.

Good lord! How did you manage to drill such *appalling* holes? And why have you gone with SMDs as opposed to THT which would be far less likely to flashover? What's your target Vout? What's the actual application??

I used my Dremel freehand for everything, with a small carbide dental burr. I'd have to hike down to the machine shop to drill round holes.

I'm getting pretty good at Dremeling; SOT23 footprints work fine. 0805 and 0603 parts are easy too. I think that copperclad is 1/2 oz, which carves nicely.

And why

It doesn't flash over at 1500 volts out, and nearly all the parts that we have in stock are SMDs. All the parts on the actual product will be SMDs, so it makes sense to test the real things.

The target is 1400 volts out, and a couple of auxiliaries. The application is driving some optically-active PLZT gadgets, pulsed from

Well, the load IS largely capacitive. Some find it advantageous to add a small output inductor, to ease switch stress.

RL

Inductor where? Which switch stress?

My final load will be about 3 mA average, and will include 0.44 uF of film caps, as energy storage for the output pulser stages.

The holes look worse than they are. Any little burr around a hole means torn out copper, because the copper is very thin. Doesn't matter as long as you're putting a big fat screw/nut/washer over that, anyway.

The cuts are still rough of course, but his technique has certainly improved. :-)

I use a utility knife, myself. This can be used directly with a metal rule to get very straight, smooth lines, or freehand with some care. Gaps of

This is the nicest row of pads I've made:

On the upside, a 50 ohm trace on 1.6 mm stock is a pretty reasonable width (something like 2mm, depending on how close the ground fill lies), easy to carve by hand (or by Dremel).

Tim

I've done prettier stuff. I wan't in an artistic mood on Saturday, just wanted to get it tested. Used copper, not gold.

I've done that, but it's more labor intensive than Dremeling, and wrecks X-acto blades. I drew the pattern with a Sharpie, freehand, and then Dremel'd.

I once had a couple of big sheets of teflon microwave copperclad, which cut and peeled beautifully. Some idiot thought it was trash and threw it out.

CW is a switched cap circuit. Though there is some leakage L in series, all the current will want to peak at the switching transitions. OK at turn-off, not so good at turn-on.

The first multiplier stage and switch will see n x output current, spikey as described. A saturable bead in series with C43 should do something.

RL

Pretty good inventory. BTW, the IRFL214 is rated 250V.

What's an example part? You knew we'd ask.

Spice without leakage inductance isn't showing any radical mosfet current spikes. Leakage inductance softens things up more. It looks OK. None of the parts on the breadboard got very warm.

The breadboard rules. If it's not an issue - thought it might be what you were complaining about - blaming the switch.

RL