[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Terry's DRSSTC - thoughts...



Original poster: Terry Fritz <teslalist@xxxxxxxxxxxxxxxxxxxxxxx>

Hi All,

I played with the "freewheeling" thing (theoretical/models) where one closes the two lower rail IGBTs to help keep the primary current in the coil system during turn off. Although it seems to be a good idea, it just does not seem to contribute much to the streamer power.... I do wonder if having the primary circuit shorted "restricts the freedom" of the secondary system to ring freely... It was easy to modify the gate drivers to do that. You just run the lower ones from "-rail to +rail" so they are powered for like years on end ;-)) Probably a "good idea" anyway... No reason to power them from the 100kHz +-170V mess... The top IGBT drives have to follow the top IGBT emitters, so they are "going for a ride" anyway...

Freewheeling would pretty much need a PLD to implement which is out of "my area"... The TIBPAL16L8-15CN seems like a real nice chip to do it until one tries to figure out how to program it "these days"! I am not sure I trust those new fancy "retains programming for 20 years" microscopic die things... Whaaa, like in 20 years it forgets a gate, and the thing blows a hole through the roof!!! I still use TC stuff I made 30 years ago! Rather have like 10,000 years of data retention ;-))) Those <15nS delay times sure are nice!! ;-))

But some type of "permanent" PLD does seem nice for this DRSSTC control stuff. I am using 5 ICs now and I might make a four layer board when I am "feeling lucky". It needs to be four layer and the cost/"try" is about $150... So screw ups in the board layout are pretty painful!! But a nice CMOS style PLD might make it all real simple. It really is just a decoder type of thing... Hard to say what is best since once you have the "board", the cost of the little ICs is cheaper than the PLD... But if this ever goes to a "single board solution", the PLD will avoid the four layer board stuff!! Also must beware of the reports that 5V logic and DRSSTCs don't get along well... My way of thinking is that if noise is getting into the logic, the problem is getting rid of the noise rather than changing the logic... But I respect Dan and Steve's thoughts here and they are pretty capable too!! But I think I am keeping the RF stuff pretty well contained in the "little" bridge loop too...

So I will continue to explore the freewheeling and PLD things on the "side". For now they are not needed but it might be cool for the future. I know three digital gurus so I will ping them for info and clues... But for now, this seems to be the circuit I am going with:

http://drsstc.com/~terrell/schematics/Controller.gif

I threw all kinds of weird stuff into a model of it today and it always performed gracefully... I think R7 and R8 might need to be trimmer pots to adjust the dead time directly (C10, C11 in the 330pF range)... The timing on U5 might have to be shorter on higher Fo coils... Might b4e better digital solutions there...


I see DigiKey went on backorder for my favorite IRG4PF50WD IGBTs :-( So I got 32 of them off supplier "#2" >:o))) NOT(!!) that I am planning on blowing any up, but if I do, I don't want to wait until March...



The delay in my digital CT seems to be due to the diodes in the digital phase detecting CT:


http://drsstc.com/~terrell/schematics/CurrentMeasurement.gif

I use the gate capacitance of the big 1N5337 to sustain the rail voltage. Might be able to go to smaller diodes but if one fails it would be "BAD"! Lots of peak current runs through them... The ones I use now won't let me down ;-)) The delay only gets better with higher currents so I think I will just stay with it and be happy ;-)) I am a little concerned about the cheap CTs being able to handle say 500+ amps. I can't test that well right now since my high current test machine is in dis-repair... I will probably have to fix it though to be sure...

So "thinking", while waiting for parts....

Cheers,

	Terry