IEC 61000-4-5 10 kA lightning surge test box ideas

[RAZZITEAK]

Hello!

So, I have a need to test and verify surge emunity for certain products. What the product is or how it should be used is irrelevant. However, what isn't is the cost to have these tested.
That is why i want to do my own testing to see if the product meets the requirments. The test follows the IEC standard  61000-4-5:2014 with a 8/20 (8 us rise time, 20 us duration) 10 kA pulse  !

The test box which i've built consists of a high voltage transformer (6 kV) which charges a bank of very large capacitors (4 uF). The idea is this: When a certain voltage is reached the capacitors should dump all their energy into the DUT (device under test), trough a shaping network, consisting of resistors and inductors. At its current state, the surge process is controlled via a GDT (spark gap). So when a certain voltage is reached, the GDT arces over and the energy gets dumped into the DUT.

However, here inlies the problem. The GDT, altough providing the decired waveform, isn't stable enough to deliver the same peak current into a short curcuit every time. It varies with several kA after every test. Probably since its breakdown voltage varies after every spark.
I want something which can provide a more accurate pulse and do away with the GDT solution completly. And I was thinking about a really beafy IGBT. But they aren't rated for the high current i'm after. And the ones that are, cost as much as a small car.

I'm after a device which can handle a switching current of 10 kA peak with a standard 8/20 us pulse at a 1 minute intervall. And not break down from the 4 - 6 kV that will be across it. Does anybody have experience using IGBT (or similar devices) this far from their working parameters?
Maybe somebody has experience with these kinds of tests and are willing to share some info, that would be fantastic!

[chris_leyson]

Just spent two weeks getting some equipment to pass IEC 61000-4-4 fast transient at 4 to 5kV and I was wondering what the hell they use in the transient generator given that the rise times are 5ns, definately not spark gaps, that's old school, I doubt they use thyratrons, too expensive, best guess is a triggered cascade of avalanche devices.

Anyway, EN 61000-4-5, surge test. Going to be doing surge testing next week so that might be fun, I hope not.
Given that an IGBT rated for the current might cost the same as a small car, how about a big thyristor ? T3sl4co1l, Tim, might have some ideas. Unfortunately there is very little information out there on how to build EFT and surge generators. Watching with interest.

[T3sl4co1l]

You may find this of interest,



This gives an 8/20us surge into a low resistance (as shown), a 1.2/50us surge into a high resistance, and an equivalent 0.5 ohm (or is it 2 ohm, I forget) output impedance (that is, Vpk(oc) = Ipk(sc) * 2Ω).

Obviously, you need some awfully meaty parts to pull this off at 5kV and 10kA, which I guess you've solved adequately.

I suspect EFT machines use hydrogen thyratrons.  Avalanche isn't powerful enough, you only have maybe 100uJ per 2N3904, and yes, the FMMTxxx handle more but they are also slower.  So you need a big stack of them to get there.  Maybe it's worthwhile?

Surge generators probably use SCRs or ignitrons.  Not sure if the latter is reliable enough, but at least being triggerable is better than running a GDT.  SCRs you need to be careful not to violate max dI/dt, otherwise you get the one surge just fine... and nothing more.   Large SCRs have been used on quarter shrinkers just fine, so it can't be too bad.

I've seen pulse generators for particle accelerators, based on IGBTs.  They're kind of special though, not quite the same energy and risetime (faster actually, but even faster at the output because they drive a magnetic compressor).

Tim

[RAZZITEAK]

Very intresting input!

This gives an 8/20us surge into a low resistance (as shown), a 1.2/50us surge into a high resistance, and an equivalent 0.5 ohm (or is it 2 ohm, I forget) output impedance (that is, Vpk(oc) = Ipk(sc) * 2Ω).

Obviously, you need some awfully meaty parts to pull this off at 5kV and 10kA, which I guess you've solved adequately.

Tim

Yes i do have very meaty components. Each capacitor is about 2 l in volume, and i have 4.

I've looked around a bit and it seems as Infineon does have SCR's which might be suitible. However, nothing seems to be in stock! Altough overcommable, they aren't what I would consider cheap.
Nevertheless, do you have any specific SCR/manufacturer in mind or perhaps a place to get them with resonable lead times?

[T3sl4co1l]

You're complaining about the wrong thing (cost) with capacitors that size!

Tim