20KW DRSSTC / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

20KW DRSSTC

first previous 7 8 9 10 next last

Move Thread

LAN_403

Steve Ward

Wed Apr 11 2007, 09:03PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

At high pulse currents like this most IGBT's will go PTC because of stray ohmic resistances.
Stll, splitting the MMC bank is much more *right* way to paralell IGBT's!

Yep, we are just lucky enough that its possible to do that with the DRSSTC.

I suppose ideally you would go for the biggest, single-die base IGBTs you can find (like the 40N60 mini-blocks) and build up as many H-bridges as seems necessary. If you want to play within the spec sheet rating of 300A, and run them at only 350V, you would need 10 such bridges to build my DRSSTC-2. This is just not practical.

It seems like the same story as our MMC designs. Who can really afford to run within the *AC* rating of the caps? Surely the lifetime goes up to the n'th power when you do run within spec... and perhaps the same is true with the IGBTs. Depending on how long you want your IGBTs to actually last, you might just save money by going with more IGBTs to start with, but for most of us, its not worth it.

Marko

Wed Apr 11 2007, 10:56PM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Hi steve

Regarding paralelling, I ran around something

wich is actually 'standard speed' IGBT but is still much faster than CM600's (except maybe delay time) and 12 of those would make 1350A peak SOA!
They cost about $30, expensive but still much cheaper than new bricks of same rating!

I guess it's all about what you run at on ebay.

Steve Ward

Wed Apr 11 2007, 11:41PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

Those are some really cool looking IGBTs... the 450Apk rating is really impressive. Too bad they dont co-package them with a diode. Where did you find them for 30 dollars? I found them at digikey for ~77 each .

Marko

Thu Apr 12 2007, 12:01AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

Oh, at rell, non-stock unfortunately..

There are lots of good IGBT's around, especially considering that even standard speed isn't 'all that slow' for bigger coils.

This one has a diode and 600A current rating, but is only for 600V.

Adding a diode shouldn't be too big problem (it conducts a bit only at the end of cycle) if IGBT's are worth of it...

hvguy

Thu Apr 12 2007, 03:21AM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

I definitely think I can â€œmakeâ€ a better high speed, high current IGBT then can be bought in a single package. I priced out a bridge made from IRGPS60B120KDP IGBTs. Since the peak current rating is what we are interested in it would take about 80 of these, 20 per leg, to make a full bridge for my current DRSSTC. This would give me a rated 4.8KA at 1.2KV. In my opinion the ultimate tesla coil switch is a parallel string of smaller IGBTs. They are relatively cheap, easier to obtain, uber fast, and can easily be replaced. The catch is still the price. At $22ea from rell the bridge would cost $1,760 not including the PCB (or what ever your connecting medium) and hardware and the time it takes to stuff the thing. This is much cheaper than a comparable bridge of bricks, say the CM600DU-24NFH which are $293.12 from rell. It would take 16 of those to match the IR parts and at $4689.92 it would be nearly three times the cost. Another nice thing about the smaller parts in what happens during a failure. Based on my experience with the bricks only a few of the parts should fail under normal circumstances. Itâ€™s a lot cheaper to replace a few $22 parts than one or two $293 parts.

The parts with out the co-pak diode are a little cheaper. I wonder if we can use external diodes and save some moneyâ€¦. I think I will test the idea on a relatively small scale some time, maybe with a 6â€ coil. This idea really needs some research.

sparky

Thu Apr 12 2007, 04:09AM

Registered Member #530 Joined: Sat Feb 17 2007, 07:56AM
Location: Victoria BC, Canada
Posts: 178

Very intriguing idea, Aron

teravolt

Thu Apr 12 2007, 04:21AM

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

hvguy do you have any dead time in betwean switching.

hvguy

Thu Apr 12 2007, 08:17AM

Registered Member #289 Joined: Mon Mar 06 2006, 10:45AM
Location: Conroe, TX
Posts: 154

No dead time. This is because the gate drive transformer can not create a shoot through scenario, there for it gives me the closest possible zero current transition and no complex gate drive circuitry.

Thanks sparky, though Iâ€™m sure this isnâ€™t the first time itâ€™s been mentioned. My Dad will be in town this weekend so I will probably try to put something together next week for a test.

Marko

Thu Apr 12 2007, 08:59AM

Registered Member #89 Joined: Thu Feb 09 2006, 02:40PM
Location: Zadar, Croatia
Posts: 3145

IRGPS60B120KDP

I don't think they are actually worth $22, but this could be a baked idea.

I've been looking for microsemi IGBT's wich could match or beat these...
only 225A 1200V for $19, in stock at rell.

BUt, even with those 'faster' IGBT's delay time isn't all that much shorter, so if you want to pull out with less $$ slower IGBT's could be used.

But it's all up to you and your wallet size.

Steve Conner

Thu Apr 12 2007, 09:56AM

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

I was always worried about running the larger bricks without deadtime, when you consider that they can have a turn-off delay time of nearly a microsecond. In other words, they don't actually stop conducting until maybe 500-700ns after you remove the gate voltage! So even if your gate drive waveforms aren't such as to cause cross-conduction, the IGBTs can maybe "fix" that for you

That turn off-delay time is specced for hard switching, where the IGBT itself has to interrupt its full rated current. In our application, with near zero current switching, it should be shorter, but I still don't believe it can be neglected. Also, due to delays in the feedback loop, Aron's coil is nowhere near zero current switching!

This is another reason why I went with the PLL circuit: it's easy to cancel out delays and add deadtime. Once I get my gate drivers up and running, I can hook up a couple of bricks and try to test some of my hand waving ideas. I can use a Rogowski coil to look at shoot-through currents, diode recovery transients and the like. Unless nobody cares, in which case I'll just go ahead and try to make sparks, and scratch my head like the rest of you when things blow at random...

Steve: I think the situation with us and IGBTs is exactly the same as with Terry Fritz and MMC caps. The currents we run them at give levels of reliability that are acceptable to the hobbyist.

PS: Do you think you could modify your driver circuit to give a phase lead to the feedback signal? All you need is a high-pass RC filter somewhere after the current transformer. If you make the "R" of the RC filter adjustable, you may be able to adjust the lead to cancel out the switching delay in a similar way to what my PLL does.

first previous 7 8 9 10 next last

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