Really interesting, Adam! I've built a few can crushers myself, using very similar techniques to yours. My largest has a 15mF (0.015 Farad) 1200V capacitor bank and needs a minimum inductance of about 3 uH to protect the SCR. Typical peak current is 40kA. The capacitor bank is constructed as three 400V 45mF stages in series. Each on of those has three paralleled ~1kA stud-mounted diodes to prevent reverse-charging the caps. I started with a single one per stage but they failed after a few shots. As far as I understand, "non-repetitive" means that sufficient time must elapse between events that there is no build-up of heat between pulses. In other words, the thermal initial conditions of each pulse are the same. SCRs are fairly slow to turn on and they turn on by propagation of a "wave" of charge carriers which travels much slower than the speed of light (roughly about 2 mm per microsecond or 2 km per second - about the same speed as a high explosive detonation wave). The need to limit dI/dT during turn-on comes from the fact that the current handling ability of the SCR is proportional to the fraction of the device which is conducting. If the current rises too fast, you end up carrying almost full current before the entire SCR junction area is conducting. Some large SCRs can only handle about 100 A/microsecond dI/dT I experimented with using a saturable reactor to limit dI/dT until the SCR turned on fully. I found that a very large core was needed to hold off saturation until the SCR was fully on. The reason I did this was because high dI/dT is very beneficial in a can crusher, so ideally I wanted the extra inductance to "go away" once the SCR was fully on. To David's original question, though: 100nF at 1kV is not much energy. I wouldn't expect too much difficulty in switching that, except that most means of switching it are likely to be rather lossy since the cap will be mostly discharged by the time that the switch is fully ON. This is probably true even for a mechanical switch since the arc will start and complete before the contacts actually touch. Sean On Thu, Apr 5, 2018 at 2:39 PM, Adam Field wrote: > On Thu, Apr 5, 2018 at 1:47 PM, Van Horn, David < > david.vanhorn@backcountryaccess.com> wrote: > > > > > High energy capacitive discharge is fairly well trodden ground. See if > any > > of these give you some food for thought. > > http://www.electricstuff.co.uk/destructotron.html > > http://www.capturedlightning.com/frames/shrinkergallery.html > > https://hackaday.com/2012/09/28/how-a-quarter-shrinker-works/ > > > > > > These are primarily "one shot" applications, I need tens of pulses per > > second. > > > > > I designed and built a few of these for my day job (larger capacitor make= r > you would recognize). They are "hockey puck" SCRs with multiple diodes in > parallel to catch the ring back. I used single SCRs as it made the design > much easier. You need to know how much di/dt you want to handle. I starte= d > with these (there are probably better parts on the market now): > > https://www.vishay.com/thyristors/list/product-94414/ > > These have good voltage handling and are capable of 1000 A/us of current > rise. To get the max di/dt out of an SCR you need to drive the gate as ha= rd > as possible. For the Vishay SCR I used a microchip mosfet driver at aroun= d > 20V if I recall correctly. To manage di/dt I use inductance in the circui= t, > which is usually just extra wire. Using v=3DL(di/dt) you can work out how > much inductance you'll need to protect the SCR. To measure wire inductanc= e > I just back calculated the ringing frequency with my known capacitance > after firing the circuit with a low voltage. > > The device sheets always specify "non repetitive" and so far I'm not sure > what that means exactly. They are happy with large currents at 5 to 15Hz > with some factor of safety in the di/dt limit. The largest di/dt happens = at > the beginning of the sinusoidal discharge. > > This is an LT spice circuit that track really well with "real life" > > https://i.imgur.com/bWBy5N6.png > > For timing the discharges I used a PIC that's controlled by UART. > > Interesting SCR stuff: > > http://educypedia.karadimov.info/library/an1008.pdf > https://www.dynexsemi.com/assets/downloads/dnx_an4999.pdf > > > Also, SBE (not my employer) is using a physical discharge contact for the= ir > large capacitor work: > > http://www.sbelectronics.com/wp-content/uploads/2009/11/ > PMC2006-presentation.pdf > -- > http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive > View/change your membership options at > http://mailman.mit.edu/mailman/listinfo/piclist > --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .