Surge testing pulse generators

I think older ones back in 1980s just used contactors.

piglet

Hmm, interesting. At 4kV I suppose that would be more like a triggered spark gap - mechanically triggered.

Hydrogen thyratron does sound good, even today. Doing it with any kind of semiconductors is an intimidating prospect, and will probably wind up costing much more!

I've played a little with one before -- a different brand, but pretty beefy all the same. Capable of EFT, 1.5/50 / 8/20 surges and a couple of other things (RF burst, etc.?).

Physically speaking, I might just have to rate the power input filter as the most impressive part of the unit. It likely accounts for a majority of the weight, and handles huge pulses on three phases (two lines plus ground, so you can surge any combinations of all three).

Didn't get a look inside it, of course, as we needed it for calibrated testing, :) and the rental place probably wouldn't have been too amused if their precious cal and "warranty void if removed" stickers got bumped. I could see evidence of, I think, unusually large reed (or possibly mercury, or both?) relays, and stud diodes or SCRs, inside some vent slots.

We only needed it for EFT, but FWIW, it operated silently in that mode, for any intensity up until my hacked test fixture started arcing over.

Rental BTW is a few $k/month. If you have a pile of things to do precompliance on (or that are consistently failing and you can't figure the damned things out from just one day at the test lab), it's worthwhile.

Tim

-- Seven Transistor Labs Electrical Engineering Consultation Website:

If you read Hesterman's article, you come to the final statement that the waveforms defined by the combined surge standard are not physically possible using the circuit recommended in the standard.......all you can do is produce waveforms that meet a specific level of current or voltage in time (not both) within ~5% of the definitions. It's unfortunate that the original sources for this standard were not more forthcoming, or comprehensive.

As the larger components employed are of a necessarily fixed value, the best you can do is hard-switch component values for range and tune the initial voltage for amplitude, thought fine-trimming an electrically small choke might still be practical, if pulsed linearity is maintained.

As with any higher frequency circuitry, you should not expect any single component to be ideal, but to combine their qualities, in order to approach the intended function repeatably and over time.

As testing involves application of surges to working units, the isolation and coupling methods required to prevent the line source of supply from affecting or being affected by the surge is not elementary.

Much can be learned by applying preliminary surges to unpowered circuitry, considering the relatively low amplitudes of source - however the source remains a critical supply of follow-on currents in working circuitry - one that the surge source alone will not provide. This is important in demonstrating survivability or repairability.

As was stated elsewhere, in the simplest of embodiments, a contactor with low bounce (and renewable contacts) is probably the most economical switch for basic precompliance characterization lash-ups. As long as the voltage and current waveforms are reliably produced, the kind of switch used is not relevant to a test sequence.

Both surge hardware and circuitry being tested should be physically enclosed to avoid operator hazard.

RL

ive.

ge_Surge_Immunity.pdf

I was thinking perhaps a slug of lead could be shot between 2 springy conta cts, not with a gun but something lower energy, and that a little deformati on of the lead could ensure good contact for a usable amount of time. But I have zero experience with kit like this.

NT

This IGBT is just fine for the job

Please take a look at this design i've built two years ago for a client here in France. It's a 6KV/4KA surge tester.

It's far from a DIY design. Habib.

That is the working realm for thyratrons and ignitrons.

to OP

are you going to do this yourself as a "for credit" test or an unoffical test?

it is a lot easier to rig up a tester that is not exactly compliant but good enough to give yourself good confidence that your device will pass the "offical test" in an offical test lab.

Mark

This IGBT is just fine for the job

Please take a look at this design i've built two years ago for a client here in France. It's a 6KV/4KA surge tester.

It's far from a DIY design. Habib.

JB

The best device for this, and it will never have a high slew rate pulse dump, is a good DC supply, and a good heavy wound isolation transformer.

A high slew rate dump will have to be a huge cap bank.

Just like the railgun guys set up. And these guys...

Like this...

Kilovac ? But I don't see any higher than 3.5kv onthe TE website.

Cheers

Mercury-wetted reed relays are useful in this fashion (it takes a while to engage and to disengage, but first contact conductance has very good risetime).

It is not official testing. I have a design that passed the official testing but I wish to see whether I can achieve robustness beyond what is required by the standard, to reduce warranty costs etc..

To begin with I would like to do some un-official, comparative testing to see what works best, then send it off to be tested officially again (which should be easy since the old design passes that). So for the in-house testing, the surges need to be adjustable and repeatable or measurable, and the waveforms only need to be very approximately right (though it is more important that the test represents real life than the standard).

Chris

Thank you for that datasheet!

I guess if one has to ask where to buy it, and how much it costs, then one probably wouldn't like the answer.

I agree that it looks like a fairly involved project. I think if I tried that, I could end up spending more on components than the hire of a tester would cost, even if I didn't manage to blow up any IGBTs, which I am not at all confident of.

Chris

Your welcome.

This double IGBT is just extraordinary, believe me even those made by mitsu bishi electronics which are not bad at all ; the ABB reference here is outs tanding by comparison the competitors. 0-4000A in less than 150ns !

You can to buy it directly from ABB. Price 650 EURO each.

I had built two surge generators, one with a 470nF (6KV/4KA) the other with 2.2uF (2KV/1,2KA). Did not remember exactly but overall price for specific components approx. 10k EURO for the two genes, better to hire services fro m an official lab to drive IEC 1000-4-5 tests for you.

H.

I did wonder about a huge mercury relay, but thought things would go very badly when a mercury arc struck, which seems likely sooner or later. I'd definitely give that option a miss.

NT

I made my own IEC surge-tester many moons ago. Wound up making the capacitor from mylar layout film and aluminum foil, and the inductor from copper tubing. The switch was a spark gap made of tungsten welding rod, gap adjusted to control output power.

I don't remember what we used for power--a neon-sign-looking sort of transformer, I think.

The enclosure was made of plywood, with two widely separated push-buttons in series as a safety to fire the pulses. It was in NO way safe, of course. The output was wicked dangerous.

The overall result was a relaxation oscillator of sorts, made a hell of a racket, and pretty serviceable surge pulses.

Cheers, James Arthur

Plus the contacts don't burn. If you really overdo it, though, they can explode and spew mercury vapour everyplace.

Cheers

Phil Hobbs