Xenon strobe

Is that 12% variation in total light output integrated over the duration, or peak?

I did not look at light output when testing my flash, but I did find that both peak current and delay between trigger and flash varied quite a bit when I had a just adequate trigger voltage. With just enough trigger to reliably flash I had random delays from 3 to 20 usec, and about 10% variation in peak current. With the maximum trigger voltage I had available (20 kV) the timing jitter went down to around 1 usec and the peak current variability reduced to a few percent, less after the tube warmed up a bit, which only took about 10 shots at 5/sec and 18 J. So you could try a higher voltage trigger transformer and see if that helps, or try your existing trigger transformer from a separate supply at its highest allowable primary voltage.

The peak current of my flash also increased around 5% as the tube warmed up. So the best way to get uniform total light output despite thermal (and other) effects might be to integrate light output as measured by your PIN diode and shut the flash tube current off when you have enough.

Glen

The 12% change that I am seeing is in the integrated light output, using a photodiode and integrator circuit. In practice, the strobe is used with a camera and the exposure of each shot is also visibly different. I have built a "rough" circuit with a IGBT to turn off the lamp when the integrated potodiode signal reaches a threshold. The circuit works and I am able to control the overall brightness of the flash now but still get up to

12% flash to flash difference. I believe that this is due to the delays in the photodiode / integrator etc, such that the feedback signal are out of phase with the actual light output from the lamp, and the variation is not being removed. I think that this does suggest that the variation is in the brightness of the initial burn of the lamp. I need to make a faster feedbak circuit to get better results.

I will also take your advice (Glen) and try increasinging the trigger voltage.

The faster feedback sounds like a good idea to me. The IGBT gate drive might be the hardest part to make fast, you will need a driver that can sink a lot of current to get the gate charge out fast.

Also make sure your PIN diode is not saturating and missing variations in peak brightness, if it is monitoring the flashtube directly instead of reflected scene light this could be a problem.

Thanks for the advice

Mike