binary voltage multiplier

Well, he is getting more than double the voltage out of the Walton generator, depending on which switches are closed. I've called this a charge pump supply, but perhaps I was using the wrong name. Nice to know the actual name of this style of voltage multiplier.

Someone should ask "Kathy Loves Physics" (youtube) to do an episode on the voltage multiplier as most of her presentations relate to electricity...

John :-#)#

Fun. Of course with the right parts, you can do 2**N multiplication with N stages. (Does require using the relays for commutation, of course.)

Cheers

Phil Hobbs

This circuit can do a 2^N step high-voltage supply with N low voltage, cheap telecom relays. Maybe an incremental linear reg could interpolate between the steps.

I probably won't actually do this, unless a certain customer gets serious, which is unlikely. I was just doodling.

We could go into the multichannel high-voltage supply business with our new power rack, but lots of people are in that biz already.

tirsdag den 26. juli 2022 kl. 15.59.31 UTC+2 skrev snipped-for-privacy@highlandsniptechnology.com:

regulate the input?

Sure. The source could be something like an LTC3803 (geat part) and a standard transformer, one of the DRQ things.

We had one inquiry for a HV supply that was pretty time-agile, which would need a linear reg.

Frankly, I'd rather not be in the HV business. Too much to go wrong.

This stepped C-W was just a goofy idea.

You still only get N-times multiplication at most, though. The 2**N multiplication scheme runs each voltage doubler off the output of the previous one.

Cheers

Phil Hobbs

(We're using different definitions of N here, of course--I'm talking N CW sections and you're talking N relays, with the nth one switching 2^n sections, for 0 <= n < N.)

Cheers

Phil Hobbs

This topology uses isolation transformers with dual secondaries:

I _think_ I got the dots right...

I think that untangles into a 360 volt transformer and a half-wave doubler.

Lately we don't run line AC into boxes, or use line frequency transformers. Too many safety and compliance isues, too much weight and cost.

Ya that sounds right. An old trick in the National Semi handbook from the 70s I think it was, for regulation on the secondary is to wrap an LM317 inside the grid to cathode voltage of a power triode and that works OK for regulating up to a couple kV at low currents without too much thought on how to provide an appropriate reference and error amp for the potentials on that side.

Long lead times on many N fets with Vds > 1kV, not surprising I guess.

Yep. Laptop-style external bricks in 24-ish and 48 volts are pretty readily available up to 100 watts or so and those CUI-type modules can take you most places you need to go initially from there, on the board. The client can use their UL-listed off-line PSU of choice if that's important to them.

I forget exactly how many units UL wants to test to destruction these days just as a minimum requirement for a cert I think it's at least three.

much more than that, e.g.

Dat's a beefy brick.

The downside from a user's perspective is once you start trying to rack up a number of boxes with wall warts and external bricks you start wishing they had internal PSUs again, kettle leads and IECs are much more pleasant from a cable management perspective

Phihomg and MeanWell make warts (big plugs) at 60 watts.

All you need is a power bar with the outlets in the right orientation. The one over my bench is a very nice 20x Tripplite one, with filtering and MOVs, but the prongs are oriented horizontally, so that most warts take up two spaces.

Cheers

Phil Hobbs

torsdag den 28. juli 2022 kl. 16.23.22 UTC+2 skrev snipped-for-privacy@highlandsniptechnology.com:

lots of 65W USB-C chargers like this:

We're shipping rackmount boxes to a big govt facility. They have a group in charge of safety and compliance and all that stuff and would have set us back 6 months and maybe $40K in lab fees if we ran AC into the box. An external 24 volt 2.5-amp laptop-type supply just fixed the problem... no complaints from the users.

The power connector on the back is the usual barrel, but it's threaded for optional mechanical locking.

Yikes. GaN fets are changing things. I wonder what the switching frequency is.

What's the output voltage? Surely not 5.

Yeah, we use warts too. Our standard one is an SL Power ME10A2403B01, which comes with international plug adapters. It's overkill for most of our stuff, except for some of the beefier TE coolers.

It's a drag when those fall out, for sure.

Cheers

Phil Hobbs