Quote:
Originally Posted by russell_w_b
Ever considered a thyristor crowbar circuit? (a thyristor alone would do your app) which, upon receipt of a sudden current pulse caused by a valve flashover, would open the a.c. supply vacuum switch and simultaneously (meant to be simultaneously...) dump the d.c. HT to deck via an ignitron.
|
Hey Russell, it's a good idea and indeed, I think it is now essential to protect the power valve as well as the power supply downstream. Thank you
I will also put a gas discharge tube (ex Radar modulator) which strikes at 28kV, across the tank circuit. This will hopefully be a lower impedance pathway than the adjacent winding, a few centimetres away. 20KV is the theoretical tank circuit voltage, so this may strike too easily - I'll have to experiment. I could also put a current sense component on the earthy side of the tube, to trigger a crowbar.
I am wondering about the current sense component - you say a current transformer is ideal. I think I should full-wave (bridge rectify) the RF component and then sample it through a (Op amp) comparator that is preset to an appropriate level.
Then I can latch up a 555 and send a pulse to a thyristor as you suggest...
Does that sound fast enough? It all needs to be over in less than 5mS, to protect the rectifier chain in the power supply. They're good for a huge overload for around 10mS, but perhaps not repetitively. I know 5mS is actually quite a long time, but it's important for me to think this through.
-------------
Back to David Knight's transmission line model, I now have a related question on corona inception voltage.
Until I have the transformer that is being wound for me, I am building for test purposes a temporary EHT power supply, which is an SMPS effectively.
The pulse transformer came out of a Panasonic microwave oven inverter.
I was driving it resonantly last night at 28KHz with my MOSFET power modulator and
the open circuit secondary of the pulse transformer produced corona at both ends.
This was from a drive voltage of anything above just 18V. The original circuit was had a power IGBT as the switch, and was also powered from a bridge-rectified mains voltage, not a MOSFET at 18-30V! And so I have two questions:
1) Does the corona at both ends of the open circuit secondary relate to David Kinght's experiments? Neither of the ends is connected to the ferrite core. There seems to be no other plausible explanation.
2) How am I getting corona, in any case, at such a low drive voltage? Does it mean that corona inception voltage is lower at high frequencies than at lower frequencies? AND/OR, is it partly explained by the fact that my MOSFET is hard switching on and off at around 25nS, whereas the original IGBT was switching on and off at several hundred nS?
Hence my dI/dT is much faster, way more than an order of magnitude faster... Could that be it?
Thanks everyone