RE: [usa-tesla] internal diodes in IGBT
- I thought a straightforward way to do it would use enough inductance in series with the cap to ring down to near zero voltage in the storage cap and turn off before the current rises again. That would need a very low Z (Low resistance and low distributed inductance) source cap, and some way to turn the switch off before the current reverses. An SCR might work, especially in series with a diode (that would keep it current starved for around a half cycle, but I am not sure if it will turn off fast enough. This would limit the tank to a +/- the DC voltage on it's initial cycle. Not sure but I think you can tap into say a third of the coil to use 100VDC to drive a 300VAC into the tank.
Wouldn't you want the primary resonance to be around a few KHz rather than a MHz? (Easier to turn off with at least tens of microseconds).
Interesting. I just looked up the Wiki on the device. Says you likely will need a series diode to prevent reverse current in the device (unless your particular selection has this built in?), and that it's forward voltage is basically diode limited. This means you will get two diode drops in the forward direction. A high current MOSFET, however, can have a way low forward voltage, even into the millivolts when you are well under it's high current limit. Oh, BTB. I do recall a big deal about what are called snubbers circuits, designed to limit and dissipate the device stress when you slap a high voltage on or off. Should be worth a bit of research.
From: firstname.lastname@example.org [mailto:email@example.com] On Behalf Of Chris Swinson
Sent: Monday, January 11, 2010 11:04 AM
Subject: Re: [usa-tesla] internal diodes in IGBT
also I forgot to mention that design I have actually built already, but I used a SCR and 70uF tank cap. The idea was not to use resonance between the pri-sec at all but to provide a higher power pulse for a "single shot".
Easier said than done! Actually not sure what the frequency was, but I think it was too low for any resonable secondary coil to be built. but its on hold for now, its why im building the same version again, but with a IGBT not a SCR...
Drifting off a little on the SCR setup, I also had the idea that if I cant discharge fast enough, then what about the turn off speeds ? the SCR takes 400uS to turn off, but when it actually does, it happens very fast which would be in the low mhz range speed wise. I almost got it to work once, but it involved 1,000uF tank cap! I am not sure if the turn off idea would work, but if you can dump 5uF into a primary fast enough for 100khz, they why can't you dump 1,000uF into the primary and just turn off at 100khz rates..
It was next up on my "to do list" but I arranged the SCR and primary and tank cap incorrectly so I could not get the turn off pulse, cant remember why, think the tank cap was always charging via the primary so it never actually could turn off the way I wanted... really it was just dumping one hell of a charge into the primary (very slow with 1,000uF!) then when the SCR turned off the actual turn off speed would only be limited by the capacitance of the wiring and such... I think a 1nF capacitor in parallel with the primary was 1MHz, so providing the SCR turn off wasn't too slow then it would allow the primray to ring at 1MHZ, so would need a 1MHz secondary coil, which would have a fair amount of energy from 1,000uF tank cap!
I dont know if it would work, I gave up talking about it on the TC groups and just gave up with it... and decided to just build the one I am now..
Yahoo! Groups Links
- Hi Bert,
I just thought, with 600uS and 10mH, There wouldn't be any point in
monitering the current as in theory the tank would have rang down long
before then anyway.. There would not be any way to quench at the first notch
as that would happen more like at 50uS to 100uS or there abouts. So low
value inductor will be needed to push enough current into it in 100uS or
less, otherwise it makes having the inductor almost pointless for voltage
Of course 10Amps is at full power which it probably will never be run at. So
for a lower end power 2amps will probably be only there anyway. This is at
300uS too so the overall current will be less once the timing is set
correctly. So overall it may not even be a problem.
There only worry is if the timing was set to low, such as 10uS, where the
IGBT would have to turn off half way though the ring down waveform and that
probably would be bad for its health. Though I did set this to be 50uS -
300uS so that could not happen as much. Though this is why I need to check
on JavaTC to see what my ballpark figures are for this timing. If the
transfer time take 100uS for example, then 80 - 120uS would be the fixed
range of adjustment. So at least the ring down would have compleated a
couple of cycles before the circuit attempted to turn off.