"Inazuma" Mk.1 DRSSTC / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

"Inazuma" Mk.1 DRSSTC

previous 1 2 3 4

Move Thread

LAN_403

Steve Conner

Mon Jul 30 2007, 07:15PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

If it's my PLL driver James is using, it should have a current limiter already, unless he left it out for extra spark length

thedatastream

Tue Jul 31 2007, 07:38PM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Yep, current limit circuit seems to work OK - provided the CT / burden is as calculated (it can't be that far out)

thedatastream

Tue Aug 14 2007, 07:22AM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Well everything seems to be in order now. I've now assembled the new IGBTs, diodes and surge arrestors into the bridge. The bridge itself is no longer the old bit of PCB you saw in the first post but a bit of double sided copper clad board with extra reinforcing provided by copper foil soldered to the top and bottom. HF decoupling has been doubled to 2 x 470nF foil caps (same as in the MMC). I've added some bidirectional surge arrestors across each IGBT as well.

A quick tune this morning with 30V in and 3" sparks out allowed me to set the phase control for minimum noise. I shall be taking the unit home tonight and waiting for a free evening (with nice weather) when I can take it out into the garden and crank it up.

thedatastream

Thu Aug 23 2007, 11:25AM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

And crank it up I did. Current limit set to 100A, 25Hz rep-rate, 50us on time. Got to about 230V input when one of the surge arrestors across one of the bottom side IGBTs exploded with a nice bang. I haven't had time to figure out what has gone pop or to fix it yet, but when I have I will get a scope on it, run it up and see what is what. There were some noises coming from it so I haven't killed it completely.

I did play with the tuning before I took it home and noticed that I did get spikes on top of the switching waveform. The position and relative magnitude of these spikes changed relative to the amount that the coil was de-tuned by me moving around it. This could imply that there wasn't enough signal for the PLL to lock on to and me detuning the coil caused a change in the operating frequency relative to the unlocked drive frequency. This phase difference could have caused non ZCS and resulted in a bang. I need to check that the PLL locks properly!

Also, I haven't played with the tap position. I have only calculated the theoretical optimum tap to get the same fres on the primary as the secondary when uncoupled. I then adjusted the drive frequency to tune to the upper pole when the two were coupled. Should I play with the tap point, could this gain me anything? Since the PLL driver should control the ZCS, changing the tap point would only affect the power transfer and not affect the switching - correct?

I need to investigate the deadtime setup and possibly remove the capacitor in the RC network that controls the deadtime in case that is causing added deadtime (which may have killed the first bridge). This should set the deadtime to 0, meaning the only deadtime that I get is from the time delay in the gate bipolar drive signals. I can then see how this affects the noise on the bridge.

Another possible problem is that gate drive is insufficient due to the fairly long wires (approx 125mm / 5") from driver to GDT. I need to measure the gate waveforms with a voltage present on the bridge to evaluate this. One option could be to stack UCCs on top of each other on the control board to get more current. The other option which I'm leaning towards is to have a local gate driver board to the bridge and carry power and signal to it using screened or ribbon cable. Assuming that this is the problem.

With all this, I'm tuning the performance so that I get the least noise on the bridge. It should then follow that I can crank the coil up knowing that the only thing that will blow my IGBTs up is them getting too hot from the resonant current and not avalanche energy!

My wife recently tidied the cellar up - time to make it messy again?
Sorry for the long post - any and all comments welcomed!

Thanks
James

Steve Conner

Thu Aug 23 2007, 12:00PM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

Hi James,

Bang is not too good =(

One thing you can do is look at the VCO control voltage on the 4046 pin 8. To avoid introducing noise with your scope probe, I'd mount a BNC (or SMB or whatever) on the driver circuit enclosure, and connect the centre pin to pin 8 via a 1k resistor, to make a permanent test point.

If you look at this voltage with a scope synchronised to the break rate, you should see it change as you move around near the coil. It normally sits at half the supply voltage, and will ramp up or down over the course of a burst.

Even if you don't do this, you may be able to see the driver change frequency slightly as you move near the coil. With an analog scope looking at the gate drive waveform, triggering on the rising edge, and set so that one cycle fills the whole screen, you should see some fuzziness of the falling edge, which indicates the waveform is changing period.

If it seems to change a little, but not enough, you might not have the loop gain turned up far enough. I always cranked it full up.

The tracking is not perfect, as it takes a little time to converge. If the frequency changes quickly during the burst (due to a big streamer suddenly developing, a ground arc shorting the secondary, or whatever) the VCO will lag behind slightly. However, it's always been close enough to work well for me: I've seen waveforms from self-resonant drivers that look a lot worse.

steve

Marko

Thu Aug 23 2007, 01:40PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

TVS going bang is very bad thing.. If your IGBT's are fine, It could be the same thing as with hv guy's giant DRSSTC; TVS's forward voltage becomes less than of CE diode and it starts conducting freewheeling current, overheats and Dies.

Still, Aaron's bang energy was tremendous and from fact that everyone uses single TVS's without problems I think that it's not all, you may be hard switching all the time.

TVS thing can be solved by using two of them antiseries, more lover voltage TVS's in series or simply a Schottky diode in antiseries.

Still I fear for your IGBT's.

Steve's circuit always worked extraordinarily well and I have no clue what could be now.
Don't worry about ringing on your supply voltage - I don't believe it can do anything to IGBT or TVS, not at that power level at least.

thedatastream

Thu Aug 23 2007, 02:12PM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Chaps, thanks for the comments.

Steve, I'll check the tuning, just as soon as I can get a scope at home.

Marko, unfortunately I was using 2 series bidirectional TVS across each IGBT. You are correct in that I am probably hard switching :(

When I said least noise on the bridge, I actually meant least noise on one leg of the full bridge - sorry.

James

previous 1 2 3 4

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob