Microwave Products (MPD) officially blessed my design, in writing. Then it didn't work. And like clockwork, MPD officially agreed it wouldn't. My contact, Leroy Karthman ran some tests: He said GE gaps met that spec. MPD's gaps don't. But they were the same, I remanded. In retrospect, I probably just don't have the "inside scoop" on conditioning ignitors. When you first get a 7512, the input appears shorted. Ignitor voltage-pulse viewing confirms it's shorted. But somehow, firing commences anyway, born of the current. This stage of gap-life doesn't respond well to negative on the opposite electrode. Injected hydrogen-plasmoid is like a cathode, it's an ionized electron source. Cathodes obviously complement anodes. Device polarity is thus dictated. After succeeding I saw ignitor breakdown-voltage creep up with use. The rise eventually stopped, and then even dropped somewhat, before stabilizing at about 1,300 V. Perhaps with equilibrium, hard starting also subsides.
Meanwhile, my task had grown in difficulty: my next rendition would be with the gap-baseplate referenced to negative 50 KVDC. The 7512 is rumored ample for 70 KVDC holdoff. The raw-supply does better than 95 KVDC open circuit. The cap-banks are rated 100 KV. But I was dead in the water. And swapping the main leads is so obvious. But how? I needed a 100 KV-isolation 12 KV 40 A trigger-pulse transformer; or to float the actual driver. Even Chuck's ignitron option led to the same dead-end. ETM pulse-transformers isolate by virtue of the 40 KV-wire used for the secondary. Going that route also meant I'd be needing 3 ea (floating) holding-circuit boxes, a (floating) driver-box, loaded with three driver-boards; high-voltage resistors for voltage-balancing across the series-ignitron-string; pulse transformers; anode heating, case-cooling; driver-box and holding-circuit 120 VAC isolation-transformers, 3 ea, 50 KV-rated: I added it up mentally; suddenly I was looking at adding a bay--to the middle of five machines. I lowered an ignitron down the hole and hooked it up using a stock ETM pulse-transformer. Anticlimactic, the ignitron did work. It foil-tested fine, within limits. But all the reasons for escaping ignitrons remained. Once again, I was kept in check. And I'd seen the GE gaps do so well.
It turned out swapping the main terminals was the hot setup. I scaled up ETM's traditional pulse-transformer design to achieve this. I used 100 KV wire for a secondary. My chosen core resembles the one used in the Cober repetitive-firing crowbar circuit I saw at Hughes. The Cober thyratron-drive and single-turn HV-wire primary gave way to a two-turn ribbon-type primary, driven by Tom's SCR. Foil-film caps do 1 KHz while delivering 133-Amps peak, per cap. 600 V-rated caps see 400 V-amplitude, hopefully safely beneath the internal corona threshold. A FET interupts capacitor-charging-current for 100 microseconds' settling-time with each firing pulse. A 50 Watt resistor limits recharge rate. Electrolytic charge at the raw supply droops from 400 V-initial to 350 V-final (due to the ~500 outgoing charge-transfers; over the half-second pulse-burst string).
A fault-relay latches based upon successful application of 400 V to the pulse capacitor. Dropout of this relay interrupts the recharge-source. It gets an initial-reset upon turn-on, and another with each push of the reset button. Integrator-desensitized, the fault-relay ignores normal firings. Acting as both a circuit-breaker and a recloser, it flags the failed or misfired SCR (or/and anything else that might short the +400 VDC drive train). This is an offshoot of my work on the original Crane-50 K repetitive crowbar. It had Triac's that went bad (which smoked corresponding-resistors, prior to the advent of relay protection). Now, a bad Triac is flagged by a warm resistor, when blown. |
