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

Re: SSTC idea - DRSSTC ?



Original poster: "jimmy hynes by way of Terry Fritz <teslalist-at-qwest-dot-net>" <chunkyboy86-at-yahoo-dot-com>


Hi Stephen,

I realize there "shouldn't" be any ringing, but there is. I know in 
Justin's SSTC, the voltage clamps on the gates aren't even needed, in my 
case they are a necessity (I did manage to kill an IGBT when the clamp fell 
off). I dont see what I could do to lower the leakage inductance besides 
moving to a toroidal core. In most cases, it is easy to just twist the 
primary and secondary together, but I neeed a turns ratio of more than 
2.5:1. when I tried to use multiple primaries in parrallel, each around 
part of the secondaries, the inductances were slightly off. Because the 
twisting worked so well, I had a shorted transformer. If I made my own gate 
drivers with 50 volt MOSFETs, I could twist the wires. That is what I would 
do if I were to do it again. In this case, the voltage is coming up from 
-40 volts instead of -15 or -20. That means that the same transformer 
running at 20 volts could be fine because the ring! ing is proportional to 
the voltage. In the gate drive transformer itself, isn't the ringing 
independent of the frequency? The magnetizing inductance is bigger with 
lower frequency transformers, which implies a bigger leakage inductance.

I am not sure what exactly happened when the gate drivers exploded. I do 
know that half wave rectified mains voltage with no series impedence found 
its way to the output pin of the drivers (I put the current limiting 
resistors on the low side because it was easier, and the diode bridge 
"should" have been completely isolated from the gate drive circuit). I also 
know that it did go through the transformers, because some of the windings 
had burned open internally. I don't think that is their fault though 
because it works now, and everything else was because two wires that 
weren't supposed to touch, did.

The gate driver chips do have around 2 ohms each, so the resistance isn't 
that low. the microprocessor also gets the offset very tiny. It's another 
thing I would probably design in next time, but for this time it turns out 
it isn't needed.

I havent tried programming PLL yet, and I know I cant do it by myself, 
because I am not a programmer. If it turns out to be so tough my dad cant 
help me with it, then the guy that programs stuff for my dad will do it.

By the way, ebay usually has some good deals on those massive IGBT bricks.

  Tesla list <tesla-at-pupman-dot-com> wrote:
Original poster: "Stephen Conner by way of Terry Fritz "

At 17:21 02/04/03 -0700, you wrote:
 >Original poster: "jimmy hynes by way of Terry Fritz "
 >
 >
 >
 >Hi Stephen,
 >
 >In order to use one transformer for the whole bridge, I would have to add
 >resistance to damp out the ringing, to prevent the IGBTs from turning on 
early.

With a well designed transformer, there shouldn't _be_ any ringing. Most
SSTCers run an entire H-bridge off a single transformer, at frequencies
many times higher than you're proposing, with no problems. Judging from all
the other trouble you've had with exploding driver chips and the like, your
transformer design probably needs some work.

 >The DC blocking capacitor is just one of those "good prac! tice" things and
 >not needed if the mean voltage is zero, right?

It would not be needed if the mean voltage were _exactly_ zero. In
practice, there will always be a small offset. The DC resistance of the
transformer is very low, so it only takes a tiny mean voltage to saturate
it (and/or blow your driver chips) So the capacitor (usually a combo of one
plastic film and two back-to-back electrolytics) is essential.


 >I want to keep with the microcontrollers because it is easier to do PLL. I
 >dont want my IGBTs trying to turn off at several hundred amps each cycle.

Have you tried writing the PLL code yet? It could be a bigger job than you
think! Plus, if it crashes/glitches you'll probably blow out all your
IGBTs. Analog chips don't usually crash! These days I believe they make
resonant-mode SMPS controller chips that ought to pretty much plug and
play. Or you could do like feedback SSTCers do and feed a s! ample of the
secondary base current (with adjustable phase shift network) back to the
sync pin of your TL494/whatever chip.

 >The micro also allows me to do cool stuff like sweep the break rate, play
 >songs (low fidelity) and do double bursts easily. To do double bursts I
 >would have to switch phase right after the first burst, which could be
 >challenging w/out a microcontroller.

It'll be challenging even _with_ a microcontroller ;) I believe that phase
reversal at the notch only happens if the primary is tuned at/below the
secondary frequency.

 >I could also do several bursts in a row followed by a break, like staccato
 >SSTCs and VTTCs. Didn't you mention that you were thinking about doing
 >that with your cool little OLTC?

I originally meant it to run at 400-600bps continuously, or up to 1200bps
on half-rectified supply. However it heated less than I thought and I was
able to run it a! t 1200 for 10 second bursts. That suggests that if I used a
proper staccato with very short bursts, I could go for some ridiculous
break rate like 6000bps and get a lot more spark. Assuming the wee stubby
secondary can handle it without flashing over.

What's more promising are the research opportunities. With fine enough
control, and a one-shot facility, it should be possible to grow a single
streamer made of 1, 2, 5, 10... bursts and observe/photograph how it
evolves. But since the control circuitry is all analog it'll be a pain in
the butt to implement. I might leave it for the Son Of OLTC, if I ever get
a good deal on some of those railway traction IGBTs.

Steve C.





Jimmy