Twin System Build

[fnordest]

Hi Everyone,

i'm a Tesla addict since 3 years now. Most of the things i have learned regarding tesla coils is from you guys so big thanks to all of you.

My latest project is a twinsystem inspired by Gaos Model82. For now it runs on a IXGN400N60 fullbridge but i will upgrade it as soon as i have the money 
Some specs:

MMC           300nF  /  6kV
BusCap      9000uF 450V
Snubber     2x 2uF / 2kV
Secondary 20cm dia  / 1m length
Torus          60cm outer dia  /  20cm inner dia  / 33mm tubing
Controler   UD2.7
F0               70kHz

Some pics from the build. The case and support for the primary  is made from laminated plywood (Siebdruckplatten)


The bridge is not ideal in terms of low inductance. I will change it to a different design with stacked busbars (turning it inside out  ) but for now the ringing is acceptable.



Our first full power test had to be aborted due to bad grounding. We had very weird sounding interferences and huge sparks randomly popping at about 150VDC. At some point we had 2kV on ground potential between the coils  .


Here we are testing the musical setup at 100VDC

As soon as i find a new place to test i will follow it up with some pics in action. If you have any suggestions regarding running two coils at once i would be happy to hear them .

-Cheers

[Mads Barnkob]

That is a couple of really good looking coils, I especially like your hexagon base and the way the lids are mounted in a slot in the supports, I would like to steal that idea for a future coil

What did you have as "bad grounding"?

Usually when I have interference issues, it has more to do with my interrupter/computer being too close to the coil, I mean if you just have a wire shorter than 5-10 meters to a metal construction, something in the ground, that would not be my first place to look for issues. But for sure, use thin ribbon or multiwire cables with a good large gauge for ground connection to a solid grounding rod in moist soil.

[fnordest]

Thanks a lot i would be happy to see some some parts of the design adapted in other coils.

About the grounding issue: Now i also think it was more of an interrupter issue. We have been maybe 5 meters away from the coils. Usually it is not a problem even that close. After those weird interference i measured the base of the coil and got those huge voltage spikes. I disconnected the interrupter case from the ground but still had the same issues.
Later at home i checked the interrupter again and found one of the poti who control pulselength was kind of broken and one pin had connection to the case. I have another chance to test it on Monday in an old transformer building. I will bring some extra chickenwire with me just in case.

[oneKone]

These coils look slick! I also have to follow mads.... The clamping method for the clamping plates is an awesome idea. Looking forward to updates.

[fnordest]

Hi everyone,

the last days have been quite interesting. I have found a new place to set up my coils. It's an old transformer building turned into science education playground. There are a lot of metal structures where i can clamp on but i still have a high impedance to ground.  The solution to the interference problem was to disconnect the controller housing from hf-ground. I really don't like the idea of having a floating potential but it seems to work.

For the first power test i went with a relative low coupling, 100us ontime and 200 BPS. Everything run nice with a steady buzz. I hit 400A with 230Vac. From there i set the OCD to 600A and slowly increased the ontime.  At around 550A there was a bright flash and a flame shooting out of the inverter followed by a couple of small explosions.     

Two IGBTs and their gate circuits got completely blown apart. All other components seem to have survived. The IGBTs are rated for 400A hard switching so i think they didn't just die from overcurrent. I noticed the hv end of the mmc was very close to the GDT who drove both IGBTs .
I changed the setup so that the mmc connects to the outer part of the primary, set the OCD back to 400A and run some midifiles.

[Netzpfuscher]

It's possible that they died from overvoltage. I don't see any snubbers and a high inductive path to the capacitors. If you draw current paths over your busbars you will see a big surface. And there is not much safety margin to the 600V.

[fnordest]

You're right, the layout is not ideal in terms of inductance and there are no snubber in the pictures  . I have  2x 1uF  snubber installed. I got the values by following Mads guide, i measured the bridge output without snubbers and from the ringing on the bus i calculated the stray inductance, which was around 100nH. That gave me the 2uF capacitance, never double checked  that.   

The bridge output was quite ok. There were some switching spikes i could not eliminate with phaselead. As i turned the voltage up, the spikes got smaller relative to the voltage and also smaller with longer pulses so i didn't worry too much.   

The scope shot is without phaselead and  ca. 120Vdc

[oneKone]

Do you have any thermal transfer medium between the igbts and the heatsink? Many of me igbt deaths have been from over temperature on the die.

[Mads Barnkob]

Unfortunately you seem to have hit the limitation that everybody else using newer generation of SOT-227 miniBlocks have seen, above 500A they die, nevermind what their pulse rating is, which is not that much different since it same die size, actually the same as you find in TO-247, I think they simply die due to heat dissipation, the temperature rise between pulses can not be cooled down and suddenly you have a die at 150 degrees and from there on its only a matter of time before boom.

[fnordest]

thanks for all the suggestions  . I checked the IGBTs again and there was not much thermal paste but enough to cover the whole base surface and make the bricks stick to the heatsink. So i think this should be ok. Maybe i should stop pushing those little buggers so hard  and go for something bigger .
I am still not really happy with the phaselead. When i go for the smallest switching spikes i have a voltage dip to zero right at pulse turnoff when the phase jumps. When i use a CT and try to switch exactly at  zero crossing i have bigger spikes and a huge voltage dip.  If i go for a clean transition at turnoff i get the biggest spikes and the current seems way behind.
Maybe i am overthinking this already and should just move on and make a better layout for the next bridge.

[Bambinz]

The thermal paste and the heatsink is not sufficient in some case, expecially when you use short pulse duration but with high current (low mean current).
You should check also the thermal impedance.

[fnordest]

On paper the thermal impedance looks just fine. R_(th,JC)(single pulse)  for 200us PW is about 0.008 K/W and for continuous mode it is 0.125 K/W which is in the same region as a CM300.
The pulse rating is 1500A and the maximum switching current 430A. With a maximum dutycycle of 0.026 for high current pulses those bricks really shouldn't die at 600A.

[Mads Barnkob]

On paper the thermal impedance looks just fine. R_(th,JC)(single pulse)  for 200us PW is about 0.008 K/W and for continuous mode it is 0.125 K/W which is in the same region as a CM300.
The pulse rating is 1500A and the maximum switching current 430A. With a maximum dutycycle of 0.026 for high current pulses those bricks really shouldn't die at 600A.

I can not find the post I read, some 10 years ago, by Steve Ward, that came up with the same conclusion about the good old 40N40 and the new 60N60, where the 60N60 looked better on the datasheet, but could take much less abuse than the 40N40.