[TCML] 1700V DRSSTC Capacitors

[Tyler <tyler.c.c@xxxxxxxxx>]

First few weeks of summer after my first year of college yielded lots of
late nights and a 1700V DRSSTC. It's pretty much done and ready to be hooked
up to a big boost converter that has also been built. But first, one
question regarding if anyone has any suggestions for more DC bus-caps for
these big bridges. These big bricks sure make it hard to keep inductance
low.

Upon testing yesterday, there was massive DC bus drop at the IGBT's, which I
mainly solved today by reducing the electrolytic bank inductance as well as
adding some larger value film snubbers at the IGBT's. It could still stand
to be a bit better though, I believe. Here is what the waveform looks like
from ground to one of the output legs:
http://ctyler.scripts.mit.edu/GND_To_Output_Leg.JPG

The electrolytic bank is a 4x series string of 450V, 10,000uF caps, all very
low inductance inverter-grade caps. The local snubbers are each 20uF 800V,
configured in a 2S-2P pack.

 Quite a bit of bus-voltage drop at the IGBT terminals is occurring which
causes it to ring up above bus voltage, which will lower my potential
operating voltage. I'm not worried about the tiny hair-line rings at the
beginning, that can all be easily tuned out by turning some trim-pots, just
haven't gotten around to that yet until everything else is worked out. More
worried about the larger ring-up across the full cycle.

A lot of voltage drop can be seen in the DC bus when measured near the
IGBT's, which can be seen here (there is quite a bit of hair and noise in
this, this scope shot is an average take over 30 seconds so that's not so
visible): http://ctyler.scripts.mit.edu/DC_Bus_Drop.JPG


The bridge bus wiring can be seen here:
http://ctyler.scripts.mit.edu/Bus_Plates.JPG

Bus-cap mount can be seen here: http://ctyler.scripts.mit.edu/Cap_Mounts.JPG

Any suggestions on affordable caps to get for this, or some layout change to
help things? I imagine ideally I would drop a few hundred dollars on nice
multi-hundred-uF snubber grade film caps to put directly on the IGBT
terminals, but that's a bit more money than I'd like to put into it.


Another interesting thing is that I had a bunch of CDE 940C .22uF 1600V film
caps and I thought I'd put around 15 of them on the DC bus to act as
snubbers. In fact this was what was in place of the new 20uF black-box film
caps. But the thing is, they actually made the ringing much worse than
having none at all! Even adding one in to the system now makes it ring quite
a bit more, and having them 15 makes it ring to nearly double bus voltage.
My only real thought here is that maybe the capacitor, leads, and crimp
terminals, and bus plate added up to a resonant frequency right around where
the system is switching, although according to the datasheet numbers, it
should be resonant at around 2mhz, and it's hard to believe the leads and
bus would add that much. Just out of curiosity, any thoughts on why these
make awful DC-bus snubbers that are better to not even have?


Tyler
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla