CM300DY-12H Failure, Looking for some info. / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

CM300DY-12H Failure, Looking for some info.

Move Thread

LAN_403

Killa-X

Sat Mar 16 2013, 05:58AM

Registered Member #1643 Joined: Mon Aug 18 2008, 06:10PM
Location:
Posts: 1039

Well, I assure everyone I'm not dead, from my lack of being around the forums or IRC in .. a long time.

So I know someone out there may have had similar experience and could say what they have found. But here's my little story.

I had my DRSSTC in storage for 5 months. Prior to storage, I ran the coil in front of a class room ( Lab as its my 7ft DRSSTC ), At The Geek Group, and at my house. I've had well more than an hours worth of spark time EASILY put into this thing due to the amount of times ive ran it, longest run, maybe 15 minutes solid. The waveforums on the bridge have always looked great with proper phase-lead adjustments. All is well.

So. I pulled it out of storage to make some modifications. My midi interrupter that I've built with the assistance of Eric Goodchild allowed me to remove the variac from my DRSSTC allowing me to run it on full mains and simply adjust the duty cycle ( Or primary current cycles ) and bring up the spark length. I didnt check the waveforums as well, Storage...why would it change? I got multiple tests on it, I even had a wrong burden on my CT and found I ran my CM300s at 2100A for 6 seconds. Well, Today I played 6 minutes worth of music through it, polyphonic, with a tin can in the center. Worked well. Later on I came back to it and plugged in a new song which had a few areas where it was REALLY short note on times ( 0.123 secs ) very rapid climb.... Well, The coil was running maybe 200A 330VDC ( lower than what ive been running, 600A ) and while it was on i noted a little crrr noise every now and then, and my lights (CFL) would do a quick flick, like if i blinked. Well, 5 seconds into the song and that crr got louder and BANG. Big bright spark at my voltage doubler capacitor. It wasn't dead tight. Well, I turned the coil back on and nothing... After some chats with Kizmo, I took it all apart and Ripped open the CM300.

One IGBT died, the other one still works.

So, I took BOTH IGBTs out and plugged in some CM300DY12-NF's just incase the other IGBT maybe a zombie brick. I replaced the dead 5819 diode on the gate board, wired it all together and the thing ( induction test ) works again.

So, That's my story. I just wondered if anyone out there may have some experience enough to say "Oh that sounds like something that happen to me before, make sure _____"

Also I found the dead CM300 never had the TVS diodes installed from +330VDC to center.

My Theory what might have happen:
I think the notes played were just way too bloody fast and quick, enough that it broke the limit of my code and possibly shot out a long on-time burst, causing my system to basically over-current draw really damn fast, and blew up the IGBT. Kizmo gets credits for helping with this idea. The fact the gate protection is shot, says it was a damn hard death too. Now I do have a PW Limiter, which has never been set before ( with my luck its at infinity duty cycle ) As i never knew a safe place to set it. How this one works, its ran with a 555 timer. If you set the limiter to 10% duty and feed a 120hz 50% duty signal, it spits out 120hz 10% duty. Well, Yeah.. I never set it and its set to infinity im sure. Also the OCD does work as i've tested it to act at 1000A.

As always, Thanks for your time to read my thread ^^

Extra information:
Full Res Dead IGBT:

Primary Induction:

Bridge Picture 1:

( Note the GDT is just out of site, 3cm leads )
Bridge Picture 2:

New 12NF's insteadded

teravolt

Sat Mar 16 2013, 04:32PM

Registered Member #195 Joined: Fri Feb 17 2006, 08:27PM
Location: Berkeley, ca.
Posts: 1111

Hi Killa-X, are your current mesurements correct? Cirtianly
higher interupter rates and currents will lead to hot spots on the dies in your CM300's. I asked Steve Ward once and he doesn't run his past 1200 Amp. It sounds like you are on the harry edge considering how much play you got with it.

HV Enthusiast

Sun Mar 17 2013, 12:26AM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

1200A is a lot of current for those CM300s. We generally never put more than 800A peak current through them.

As teravolt said, high PRFs and currents will lead to hot spots on the dies or transient heating conditions (as opposed to the steady state baseplate temperature)

BTW, very nice looking bridge. Nice workmanship!

Killa-X

Sun Mar 17 2013, 03:42AM

Registered Member #1643 Joined: Mon Aug 18 2008, 06:10PM
Location:
Posts: 1039

I have my CT as a 1:1000, Ive been running around 200A during the time it died. The day i hit 2000A is when I had a 5 ohm burden where 5V is 1000A and my waveform on the scope was showing 10V with 1 primary turn as induction heating.

I know 1000A is an everage setting ive seen people do on CM300s, with that set i havnt actually ran it at 1000A during sparking as it doesnt get that high with 330V at the tuned frequency.

But yeah, it basically died at 200A oddly enough. May have been a delayed death due to the accidental 2000A that happen a few days ago

Goodchild

Sun Mar 31 2013, 09:10AM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

Looks as if the IGBTs cooked and the body diodes are intact. The NF series although faster have a smaller IGBT die size than the H series. Consequently they are not nearly as hardy as say the CM300DY-12H or 24H. I generally don't run 300's over 1000 amps but the H series have been know to run reliably up to 1500A with little issue.

Not sure if this helps much, but just some curios insight.

Dr. Dark Current

Sun Mar 31 2013, 10:14AM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

I would say it was simply overheating of the die.

Another thoughts: It is not often realised that these IGBTs in a DRSSTC duty suffer great heat cycling of the dies, which can result in weakening/cracking of the soldering work/ceramic insulators etc. over time. And I wonder how many people take the care to calculate peak power loss in the IGBT and correlate with the transient thermal impedance graph, to make sure they don't exceed the maximum peak teperature of the chips...

Steve Conner

Sun Mar 31 2013, 12:54PM

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

Steve Ward, Jimmy Hynes, Richie Burnett and I all did this. I've also seen an interesting paper where an IGBT brick was removed from its case, painted matt black and filmed by a high-speed thermal imaging camera.

As far as we know, the thermal cycling in a properly functioning DRSSTC isn't as bad as in railway traction applications, which do see a great deal of thermal fatigue failures. (I'm no expert, but I've heard rumours that Powerex are the most reliable brand.) A traction inverter has less losses than a DRSSTC due to the lower switching frequency, but it handles far more average power.

Conundrum

Mon Apr 01 2013, 12:12AM

Registered Member #96 Joined: Thu Feb 09 2006, 05:37PM
Location: CI, Earth
Posts: 4061

Maybe we need to keep the IGBTs super cooled, like they do with multi GHz overclocked CPUs now..

Anyone want to test this theory?
Gonna test it myself when these parts turn up, in principle even a relatively low power ie 60A 100V unit should be able to survive much higher currents.

-A

Goodchild

Mon Apr 01 2013, 03:52PM

Registered Member #2292 Joined: Fri Aug 14 2009, 05:33PM
Location: The Wild West AKA Arizona
Posts: 795

Supper cooling probably won't work due to the fact that most IGBTs have higher conduction losses at lower temps. The body diode will also experience this. Both due to a negative tempco. Even if an IGBT was found that did have a positive tempco chances are it's body diode would still have a negative one.

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