2002-12-08
Second Try
I've solved a few problems, and created a few more with this prototype. The flashing over is a hard problem, to run at more than 7kV I need to strip down and rebuild the entire ladder network, which I don't think I'll bother doing for this unit. Rather I'll learn from experimenting on this fairly harmless device for when I build the next.
The gap extinguishing problem has gone away since I started using my coronatron supply to charge the unit. It offers much lower current drive than the laser power supply, and the improved current limiting at the base of the network is allowing reliable extinguishment. The pulse repetition rate has dropped, but it is still quite acceptable.
I now have fairly reliable 30kV supplies after a troll of the very addictive eBay scored a pair of TV flybacks for $5 each and a range of other exotic devices that are sure to make appearances in projects on this site really soon! Searching for high voltage capacitors on eBay is fun... The biggest range is available from Europe, with Russian surplus devices the best all-round. While they appear to be non-pulse rated, and will likely have an ESR too high for ideal Marx generator use, I've won a few auctions on different types and will see how they go. They are likely to beat these 220p 15kV EHT multiplier supply caps at any rate. No luck on high voltage resistors, but I figure with a bit of careful study I can make chains of normal 1W devices behave. I've bought a tape of 500 of them for future experiments.
While I wait for the new caps to ship from Europe, I went ahead and added the rest of the local caps I had to the prototype, here it is part way through the construction, at 16 stages, or about 112kV on good shots with its currently crippled 7kV per stage:
Check out the corona on the second last stage capacitor lead. I have obvious loss problems all over the place. The construction is pretty poor, I have not clipped leads to length, or taken any HV soldering precautions to reduce corona. The thing is laying flat on the desk, and the gaps are not tuned at all. The spark length is not all that long, about 65mm on the best shots, and it gets quite thin and pale if stretched much beyond 40mm.
It is very loud! I would not like be near it operating for more than a few minutes without hearing protection. Its erected capacitance is only 13p, which is tiny. Each stage stores only 5.4mJ at the current voltage, or a total of 86mJ for the entire erected unit. That's quite a wimpy stored energy. It is inherently safe though, a full discharge of this unit into your body hurts (as above), but is not too likely to cause harm.
With stored energy scaling as the square of voltage I can improve the device a lot if I only could realize the full voltage rating of the capacitors.
Even with the current limited power levels I have noticed some strange effects. One which I have seen mentioned in the past, usually about Van de Graaff generators, is the observation of dark patches in the sparks, near the anode end of the spark. It does not photograph well, all my shots are long exposure, but it is definitely there. It is not always seen though, about a quarter of the sparks exhibit it, and it seems to correspond to the weaker sparks. Could it be gap jitter starving the discharge?
Finally for now, here are some pictures of the test setup, and the thread like streamers from near the maximum discharge length, sorry it is so noisy, there was a severe lack of light. I think I've hit a scalability limit with the number of stages and the charging resistance, perhaps inductive charging is worth the (large) effort? Maybe a I can score a quantity of good quality chokes on eBay and try it out...
The spark gaps need some attention too. I think I'll have to go the whole hog and carefully construct them using dome nuts. The result is probably worth it, particularly on the higher power device I am designing. A triggered 1st gap seems like a good feature too.
Leave a comment on this article.
Parent article: Marx Generator.