silly power supply

Yeah, we have a little Cockroft-Walton for biasing APDs that works up to

450 V or so. Like yours, it uses a 1:1:1 toroidal transformer, but ours is wired up as a 3:1 step-up autotransformer, driven by a gate driver and rectified by a bunch of BAT46es.

Cheers

Phil Hobbs

The zener stack provides 0/5/10/15 volts to bias an EUV (13 nm) photodiode. The customer insisted on those voltages, with the provision that a 32-bit password be written into an FPGA to enable the

15 volt option, and that we not reveal the password to them.

My previous circuit used an ISDN transformer, but they are getting hard to buy.

Have you looked at LTC3803? It's great for making regulated autotransformer flyback HV supplies with a DRQ-series transformer.

Mine are nearly all negative, and need a few mA at most, so so I just use an unregulated C-W based on a PA2004 transformer, BAT46 dual diodes, an op amp, and a 2N6520ish PNP cap multiplier/emitter follower output stage driven by a level shifter. Running the C-W around 2 MHz makes all the caps small and cheap, and the extra loss is a nit.

The small caps also make it intrinsically safe, which is comforting.

Cheers

Phil Hobbs

Is your intrinsic safety about shocking people, or is it about igniting explosive atmospheres? I'm doing some capacitive tank gauging where the latter matters. There have been some cases where level sensors have blown planes out of the air.

I recall that LLNL considers 9 joules to be the fatal shock threshold. The limit for detonating explosive gasses is much less.

fredag den 23. december 2022 kl. 02.04.59 UTC+1 skrev John Larkin:

why not used the third winding and save two diodes in the rectifier?

It's actually being used to drive another bridge rectifier to make -10 volts.

I should totally redesign the power supply. The -10 turns out to be just 39 mA so something else might be done.

Over the next couple of years, we might save $50K on parts, so it's worth a day of redesign.

One trick that not everyone may know: The recurring question is, where is the current going? And specifically, how much current is this supply delivering? The trick is to diode OR into the rail from an external supply and tease that voltage up until it takes over from the regular supply, and note the current. Gotta be careful to not blow up the DUT, of course. Synchronous buck switchers might fight being over-driven on their outputs.

LTM8023 has a catch diode so shuts down fine when I do this. It's one of the expensive parts that has to go.

We should always design low-ohm shunt resistors into power supplies. Input side would generally be OK. A ferrite bead would work.

Just shocks for us.

It's very hard for us to get liability insurance, because none of the underwriters can be brought to understand what we do.

Our HV supplies use 10 nF caps at 2 MHz, and have a pair of LND150s in parallel at the output, so there's no way to get hurt. (Although, blowing up your airplane is not one of the ways I considered.) ;)

Cheers

Phil Hobbs

It's called minimum ignition energy. There is a large literature. Think fractions of a millijoule. Here is one industrial example:

.

Joe Gwinn

On a sunny day (Fri, 23 Dec 2022 11:02:03 -0800) it happened John Larkin <jlarkin@highland_atwork_technology.com> wrote in snipped-for-privacy@4ax.com:

What happened to winding to few turns on a pot-core or some E | core? Any voltage you like and any current you like if a flash (s to speak) Voltage Multipliers work fine too.

Too complicated for John Larkin, He's much happier bodging up a circuit that he can wrapped around a transformer that he can buy off the shelf.

Not exactly. But there are cheap Chinese coil-winding machines around, and you can buy spools of enamelled wire pretty quickly.

<snipped Jan's examples, which don't say much about the transformers involved.>

This is what seems to be evolving. That DRQ127 transformer costs about a dollar and is multi-sourced. Great parts.

(I think I got that right.)

Home-made transformers are too labor intensive, in engineering and production. We want to buy inductors on reels for automated pick-and-place. Only exotica like transmission-line transformers, or thermal oddities, are worth inventing.

The real challenge is to *not* design transformers.

Joe Gwinn snipped-for-privacy@comcast.net wrote in news: snipped-for-privacy@4ax.com:

Here is another. Seems to get missed a lot.

What's the coupling coefficient? Like all manufacturers, they quote series inductance as exactly 4x parallel, so you can't estimate k from the datasheet.

Cheers

Phil

The DRQs vary with size and value but are typically around 0.99.

They arc around 2 KV!

On a sunny day (Sat, 24 Dec 2022 05:21:50 -0800) it happened John Larkin snipped-for-privacy@highlandSNIPMEtechnology.com wrote in snipped-for-privacy@4ax.com:

That is the circuit diagram The first things I am missing is some peak current sense in Q7 Q4

Nice!

The soldering sucks, you need through-hole to stans vibration or a ceramic coil former screwed to the board, or btter the chassis.

Well, that is better :-)

I really do not know, most coils in ferrite core transformers can be done with one volt per turn calulate primary inductance.. use 15 kHz up as in TV,,,, Flyback gives more volts per tuns of course.

RF is different calculate L use 220 pF if < 50 MHz etc

Everything always works here... not much time spend 'designing' am just a neural net,,,

Must have wound hundreds of LF an RF coils and transformers,

Hire some guy to do the simple winding ... There are plenty companies that will make them too..

Especially for higher power stuff it is all custom..

>

<snip>

The rubber band's an awesome hack, Jan! It's good enough to remove my reluctance to use the E-I cores on hand for day-to-day experimentation. Because it eliminates the single use constraint applicable when the E-I core is bonded together. Now E-I cores are available for use in any project. One of my experiments involves a ring modulator. An E-I core ought to make it possible to lower its carrier frequency from 200KHz to 10Khz. Previously, an E-I core would not be frivolously wasted on such an experiment. Danke,

LT Spice will rotate parts, at least in 90 degree steps.

I like the diodes that way.

On a sunny day (Sat, 24 Dec 2022 18:17:20 -0000 (UTC)) it happened "Don" snipped-for-privacy@crcomp.net wrote in snipped-for-privacy@crcomp.net:

What do you use a 10 kHz ring modulator for if I may ask?

It's simply an experiment to see how a low frequency affects conversion loss and IMD. Danke,