Calculating instantaneous Current through IGBT or SCR

Peak current can't be more than the LC resonant peak (note, L includes wiring), which is Z = sqrt(L/C). Take peak volts and you get peak amps, or if you're looking at a CW tank, you can do RMS to RMS, doesn't matter. So for 1100V and 1/sqrt(2) ohms, around 778A peak.

I should think a beefy stud type would handle that, as long as the repeat rate isn't too high..? Of course, finding repetitive peak ratings for narrow pulse widths is another thing.

Tim

-- Seven Transistor Labs Electrical Engineering Consultation Website:

I have to select a IGBT or SCR to go into the power supply for a YAG laser. The network that powers the lamp is a LC circuit set up for critical damping with a period of 250 uS. 250 uS is used because it corresponds to the storage limit of the laser material.

A typical circuit might be 40 uH of series inductance, and a 80 uF cap charged to 1100V. The lamp is kept ionized by a current of a few hundred mA of DC, the SCR or IGBT is in series with the cap, and is triggered to dump it into the lamp through the inductor. This gets fun because the flashlamp's impedance changes in a non-linear way during the discharge.

So the question is, what is a good way to approximate the Di/Dt through the switch once closed. From watching a existing system on the scope, the current rises following a very sharp square wave for about 5 microseconds, then follows a quasi sine curve to near the maximum current in about 20 more microseconds.

The limiting factor on the switch data sheet for both types of devices is the instantaneous current. So that is what I need to determine.

In my case this is to replace a Eastern Bloc component that is no longer made, and is pre-internet. It is a SCR in a Puck Package. The lamp power supplies make it to 15 years and then the SCR goes soft or outright fails. The SCRs were made in Estonia, by the semi-conductor division of a big cold war era Steel plant. So tracing the actual part is impossible.

Suggestions?

Steve

Some power components like that are rated for pulsed current. Look for one that is.

The data sheets that I've worked with will give some sort of a graph for pulse duration vs. pulse power. When you unwind everything, you'll find that there's some maximum amount of energy that you can dump into the device instantaneously, and after that there's one or two time-delayed thermal processes that get the heat out of the critical parts.

The pulse shape matters, which confounds things. Some manufacturers use an exponentially decaying pulse, others use a square pulse -- the exponentially decaying pulse makes things sound better than they really are: just off the top of my head I think a rectangular pulse of t0 seconds duration is roughly equivalent to an exponentially decaying pulse with a time constant of 2*t0, but don't bet anyone's life on that, please.

If you assume a rectangular pulse at your maximum measured current and 20 microseconds then you'll be automatically derating things, so you should be fairly OK.