silly power supply

If it were domestic kit I can imagine some just putting a pair of AAs on the board.

NT

What's an AA?

The planet's most popular electrochemical cell. Once known as the penlight.

NT

I am NOT going to put an AA battery into a $5000 aerospace VME module to furnish an amp of opamp power!

Lol. I said domestic kit.

NT

You are talking English, apparently. I speak American.

Domestic kit means household feline here.

:)

I presume a simple capacitor diode rail booster is unsuited.

NT

My manufacturing and test people hate multiple versions and options. I can avoid problems, and maybe redesign, by using that screaming fast THS amp and boosting the supplies. And boosting the supplies is a fun little circuit thing.

That might work. The cap sizes and peak currents would be pretty high, at 1 amp out, but that's probably managable.

If I made a 5 volt p-p square wave (same idea as my current transformer thing) and used that to charge pump the 12 volt rails, I'd get maybe 16 volts at light load, and I could size down the caps so it droops to maybe 13 volts at full load. +-16 is a bit high, but something could be done there.

It's not much different from my transformer-coupled full-wave boost, maybe a little softer and noisier. It would work.

For how long might the output stay at the rail?

Use a Schottky diode to stop the cap being charged more than 0.4V negative, and it only has to provide the peaks, not a continuous 1A.

Clifford Heath.

This circuit looked brilliant, but I decided to breadboard it anyhow, just for fun. Good thing I did, because it's actually terrible. The opamp can't really drive the gates of the available giant mosfets.

With a more serious gate driver, it seems to work fine.

It makes a bit more voltage than I really need, so I can use a cheap schottky bridge rectifier per channel, and tolerate a little more drop than four discrete diodes.

I do need to shift the 400 KHz uP PWM output down, since the gate driver rides on -5, but that's OK.

Den tirsdag den 17. januar 2017 kl. 03.02.19 UTC+1 skrev John Larkin:

just use an ir2153/ir2085 ?

The VME has lots of power available, right? So a DC/DC converter can make your power supplies, and a few minor tweaks (diode-clamp supples to +/-12) will handle power sequencing. It's ugly, of course, but those auxiliary power patches are what keeps some semiconductor suppliers in business.

So a DC/DC converter can make your

Getting, say, +-14 at 1 amp step loading is interesting. I think I have it worked out, posted elsewhere.

Huge dead times on that first one. Besides, we don't have those in stock, and it's a PITA to create a new PADS part, create a new stock bin, buy parts, all that.

I really like the overdriven complementary follower circuit. There's no shoot-through, no dead time, no snubbing, and I can tweak the gate resistor to make the edges as soft as I like, to avoid making nasty spikes. There are really no DC-oops destruct paths anywhere.

If I had step-down transformers available, that IXYS driver could do the whole thing.

Would connecting the T2 primary winding between +5 and the FETs provide you any value (like less ripple on your 5V supply)?

That's interesting. It's too hard to think about, so I Spice'd it. The

Thanks for the idea.

How about one primary to +5 and other to 0V?

piglet