Re: DRSSTC-3 VCO drive test (long)

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: Steve Ward <steve.ward@xxxxxxxxx>

Hey Guys,


On Fri, 04 Feb 2005 18:16:01 -0700, Tesla list <tesla@xxxxxxxxxx> wrote:
> Original poster: "Jason Judd" <JJudd@xxxxxxxxxxxxxxxx>
>
> Hi Steve and all,
>
> good to see you are making very similar observations to me.
>
> Switching spikes on the IGBTs with a VCO drive have always been the problem
> for me and I would say responsible for at least 90% of my failures.

Well, im not sure if switching spikes are the cause of my failure.  I
fixed my coil and returned to secondary feedback, but just to test out
my new current limiter.  After some fiddling, it works like a charm,
can set it from 200A to 600A or so.

Anyway, i found out something *really* bad that i cant believe i didnt
catch earlier!  While running secondary feedback i noticed i was hard
switching on EVERY cycle.  The IGBTs were switching about 45-60
degrees too late!  This means that i was hard switching a good 250A or
so at times.  I traced the problem back to my gate driver... several
things i did wrong a long time ago and never fixed them.  Firstly, i
have 1n4007s clamping the output of my gate drivers (not good with the
super slow recovery time).  Second, im using way too much resistance
on the IGBT gates (10 ohms).  Thirdly, i really should double up my
gate drivers for more ummph.  Anyway, im seeing about a 700ns delay
for turn on, and a 200nS delay in turn off... thats really bad when a
half cycle is only ~2.3uS long.

>
> The primary feedback idea that Terry is using seems to be the most fool
> proof, using some of that "noise" as the drive.

Yep, after giving it more thought, im thinking primary current
feedback might just be the best way to go.  At least it ensures soft
switching, which should really eliminate most IGBT casualties...
except in my case when you have absurd driver delays!

Interestingly enough, my giant DRSSTC is already perfectly set up for
primary current feedback (already built in an Over Current detect
circuit :-)).

More updates later.

Steve W.

 I am hopping primary
> feedback will be able to produce the normal notched or "Antonio" mode, but
> it will probably be a little difficult to achieve (if possible). The
> notched tunning with low coupling has given me the best results so far in
> terms of spark output for IGBT current.
>
> Jason.
>