Pulse Discharge Guys,
Had a close call today (26-Jan-2002)...
I was in the driveway shrinking some quarters for fun and really had everything dialed in. They were coming out exactly like I wanted and were very consistent. I was getting a little complacent and bored. Then I got the bright idea to see what would happen if I turned it up to 11 (on a scale of one to ten).
I started to charge the cap for the shot then had a moment of clear thinking and grabbed my full face shield and tossed my flak jacket over my winter coat (just in case the containment failed). The plan was to run it up to 13kv (26kj). Looking back I realize the largest shot I'd ever done was 15kj. My power supply has an upper limit adjustment and instead of setting it at 13kv I just turned it up near max (20kv) planning to let off the charge button when the meter read 13kv. I was standing 10-15 feet away with the bulk of the machine between me and the containment.
Things started out fine and the cap was charging like usual but it was taking a really long time once it got to 12kv (my supply charges at 1kj a second so I should have realized something wasn't right). The last thing I remember was seeing the needle move past 12.5kv and the power supply upper limit light come on, then everything went quiet, and thud the shock wave hit me like being hit by a truck. I was disoriented and realized that something had gone really wrong and I wasn't thinking too clearly. Luckily I only have a sore chest and neck and my hearing came back after a few minutes (whew) although the ringing hasn't completely dissipated .
Post mortem - The shot never went off and it was immediately apparent that there had been a flash over between some HV wiring and that the wires had EXPLODED. Looking at the shadows cast by the explosion I was able to determine that two wires had exploded but I'm still not sure of initial point of failure. One wire was directly connected to the terminal on the cap with the other end going to the charge and discharge contactors (It's completely missing along with the tubing it was in). The other wire was a ground wire for the body of the contactors and is missing a couple inches. What I think happened is the HV charge lead flashed over to the body of the contactor then the ground wire completed the circuit.
How much energy - The bang was big enough it brought out everyone in the neighborhood and some people jumped in their cars and drove down to see what happened (and I live in a relatively rural neighborhood). Initially I thought the charge was around 22kj but my crappy meter could be way low (as Bill and Kevin previously pointed out). The power supply was set at 20kv which equates to 56kj and it hit its upper limit. The charge time of 45 seconds or more supports a charge of at least 45kj. Anyway you slice it, it was a big bang and I'm really lucky!!!
I've attached some post failure pictures.
Be careful of those unintentional exploding wire experiments !!!
After quite a delay I finally performed an autopsy on my PDM. The good news is that most of the subsystems are still working properly and I found the weak point where the flash-over occurred.
The small Jennings contactor (RJ2B-26S) took the brunt of the force with serious damage to some of the HV wiring. Additionally, most of the vinyl tubing which was used to provide extra insulation on the HV wires was burnt. The small contactor is dead which is no surprise due to the damage on it's Normally-Closed HV post (the one which was tied directly to the energy storage cap).
The larger contactor took a hard hit with some of the ceramic looking a bit worse for wear but it still functions fine. Additional collateral damage was done to the back-board where the components were mounted, the insulator on the energy storage cap, and soot was deposited on or impregnated in the rest of the components.
The obvious cause of the problem was operator error but the weak spot where the flash-over occurred was due to the way I mounted the small contactor. It was mounted into a hole in a square aluminum plate and the plate mounted to 1/4" hardboard. Unfortunately I mounted the plate on the wrong side of the hardboard which placed the grounded mounting bolts .25" closer to the HV stud (the stud that WAS directly connected to the main capacitor). To make the stresses worse there was a ground wire attached to the closest bolt. This reduced the clearance to less than .5" (Yikes). If the contactor was installed correctly and the ground wire re-routed the clearance would have been .8-.9".
During the design of this machine I tried to predict the various failure modes. Unfortunately I never considered the possibility of the contactor failing via an external flashover. In fact I placed this contactor in the safety loop thinking an over-voltage would cause an internal flash-over thus connecting the energy storage cap to the bleeder resistor network. I never considered an external flash-over and the ramifications it posed.
Testing individual components and the re-built machine
My friend Kevin Ottalini suggested I margin test the re-built machine before re-connecting the energy discharge cap. Funny thing is this was exactly what I was doing at the component level and will now perform this test on the entire machine.
Kevin's suggestions are:
Questions and comments Copyright © 1997,2006 Brian D. Basura This site was last updated 04/02/06