OL-DRSSTC 11 - two

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: Terry Fritz <vardin@xxxxxxxxxxxxxxxxxxxxxxx>

Hi All,

Continuing from the last post...

I removed the 100 ohm gate resistors to see what would happen.  They 
had a pretty heavy 12.5kHz ring:

http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-011.gif

The CTs measure 27mH and the IGBT gates and wire must have about 6nF 
of capacitance and that LC is resonating.  I tried powering the coil 
up "a little" like that and it was spooky =:O  The gates were firing 
randomly...  So at a minimum, you need enough gate resistance to de-Q 
the CT-IGBT resonance.

Q = 2 x pi x F x Lct / R = 1 / SQRT(2)  = 2 x pi x 12500 x 0.027 / R

R = 3000 ohms

I made a little chart of the "D" constants for various resistors.

2200 ohms   D = 0.985
1500 ohms   D = 0.970
1000 ohms   D = 0.912

Since I want good margin with a D close to 1, 1500 ohms is the best 
value.  But I only have 1K 1/2W carbon resistors, so it is 1000 ohms ;-))

This is the trigger waveform now:

http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-012.gif

Perfect with zero ring and a D factor that gives 2.16X faster 
switching!!  This is the burst now:

http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-013.gif

Very pretty!!!!

The lower threshold on the gates discharges the system almost 
completely now!!  That is VERY nice!!  the whole burst is at 1mS now 
too so that heating on the IGBTs is less too!!  So forget the 100 ohm 
resistors and use 1000 or 1500 ohms...  Running at 100 amps 
peak.  The IGBTs go from 25C room temperature to 27 or 28C now 
:-)))  Thus, the thermal and heating issues seem solved!!

There are still some significant time lags but the changed gate drive 
makes it hard to compare signals directly now.

http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-014.gif
http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-015.gif
http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-015.gif

Here is a chart of the voltage in the loop and the current across one IGBT.

http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-020.gif

The yellow power line is Power/100 watts. you can see right of center 
where the system is driving power into the coil.  Then power is being 
taken back into the buss caps during the IGBT dead time.  there is an 
initial spike of power as the IGBT turns on and then the power as it 
is conducting.  When the current reverses, the power in the IGBT 
actually switches over to the reverse diode on the right.  One could 
stare at this graph for hours, but there is nothing to alarming at 
all here.  The initial power spike on the IGBTs needs to be watched, 
but that should do fine and the current is increased.

So right now there is only "one" issue left and that is the high 
frequency voltage ripple getting through to the buss caps:

http://hot-streamer.com/temp/OL-DRSSTC-2005-10-13-021.gif

The primary current is in yellow and the electrolytic voltage is in 
blue.  That is probably too much ripple and might not be good for the 
caps.  I will work on this next...

Cheers,

	Terry