Re: [TCML] DRSSTC CM600 Ringing

[Steve Ward <steve.ward@xxxxxxxxx>]

Tyler,

The ringing is due to the rapid charging and discharging of the junction
capacitance which rings with the IGBT module (and bus) stray inductance.
Draw the circuit with small capacitors across the IGBTs and some inductance
in series with each IGBT and you will see that as one switch closes, you
charge up the capacitance across the other switch, but you do it through the
leakage L which serves to store some energy, and you get the usual overshoot
response of a 2nd order system.  "Proper" solution is a small RC snubber
across each IGBT to eat up this energy, but still this only helps and
doesn't completely eliminate the voltage overshoot *at the IGBT die*.  I
have not worked out what a proper RC snubber would be.  The design is partly
dependent on how much power you are willing to waste in the resistor, and
how much you want to damp the thing.  As a first guess i might try something
with a capacitance thats 2-5X the junction capacitance (at full voltage),
and set R equal to the impedance of the capacitance at the ringing frequency
so that its pretty damped.

I should mention that the junction capacitance of the IGBT will decrease as
the voltage increases, which makes the charging current less and the
overshoot less, id suggest careful measurement of the overshoot with respect
to bus voltage to see that the overshoot should in fact become less as the
voltage increases.  I don't understand the physical mechanism for this
decrease in capacitance, but look at the datasheet for real numbers (there
is usually a plot of capacitance vs blocking voltage).

So why is it worse at the beginning?  My best guess currently is that the
switching speed of the IGBT (mainly the turn ON speed) is significantly
faster without any load current, so the dv/dt across the switch is greater
which makes the peak charging current into the other switch capacitance
greater.  As load current increases, the switching speed becomes more like
what the datasheet says it should be (few hundred nS), and this is
significantly slower, avoiding excitation of the parasitic elements to such
a degree.

Steve

On Wed, May 11, 2011 at 10:15 PM, Tyler <tyler.c.c@xxxxxxxxx> wrote:

> I just recently finished rebuilding my coil with a new bridge built with
> CM600's, and a new controller that is fed by the hammertone prediktor phase
> lead.
>
> It is doing well, and can be adjusted for (almost) no ringing with the
> timing adjustments, definitely much less ringing than my previous designs.
> The problem is the first few cycles, no matter what I do I can't get rid of
> the ringing. No adjustment removes it, and I've never had a design that
> lacks this ringing.
>
> Can be seen in this scope shot (the horizontal line at the top of the
> waveform is the DC bus voltage):
> http://ctyler.scripts.mit.edu/IMG_0027.JPG
>
> Zoomed in here: http://ctyler.scripts.mit.edu/IMG_0030.JPG
>
>
> The only solution I've come up with that helped, but did not eliminate the
> problem, was putting in a big 25uF 1kv film capacitor directly across the
> rails at the IGBTs, which can be seen in my bridge photo:
> http://ctyler.scripts.mit.edu/IMG_0034.JPG
>
> Additionally, there seems to be a bit of extra ringing after the bridge
> turns off and the energy in the primary is going down, although it's
> possible this will go away when the secondary is in place?
>
>
> I've looked around quite a bit and haven't found a reliable solution to
> this
> problem, which is noted in a number of places. Has anyone found anything
> that helps out with this?
>
> Tyler
> _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla