Really big brick - Off-line SISG coil? / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

Really big brick - Off-line SISG coil?

1 2 next

Move Thread

LAN_403

J. Aaron Holmes

Wed Oct 11 2006, 03:56PM

Registered Member #477 Joined: Tue Jun 20 2006, 11:51PM
Location: Seattle, WA
Posts: 546

It was an impulse purchase. I saw some Powerex CM1000HA-28Hâ€™s on a shelf at my favorite local surplus shop. I knew they were IGBT bricks, but nothing immediately told me what they were good for. I bought one anyway. $50 + tax. I donâ€™t even know if it works. Briefly, the ratings are: Vce=1400V, Ic=1000A, pulse current = 2000A. If it *does* work, seems like it was probably a really good buy!

Hereâ€™s the datasheet:

Initially, I thought it would just be a fun thing to tinker with, but seeing the 2000A rating, Iâ€™m left wondering if this would be a good choice for a single-element SISG, e.g. for an "off-line" coil. Certainly the Vce is more than enough! What do those of you more familiar with these bricks think? Did I waste my money? If itâ€™s broken, Iâ€™m sure the answer is YES!

I suppose I need to wire up some kind of test rig now.

In the OL-SISG scenario, I suppose Iâ€™d use voltage-doubled 240V to charge up a few dozen uFâ€™s and mate this with a single SISG element using this brick and two 300V SIDACs. Doubtless some of the other component values (resistors, cap) need tweaking to accommodate the brick. I havenâ€™t thought too much about that. Will it be able to drive this brick? If not, perhaps a more â€œclassicalâ€ OLTC design is in order, but Iâ€™m not sure Iâ€™m geeky enough to pull one of those out of my hat

Lastly: Can anybody suggest a relatively safe-n-sane method of putting this thing through its paces that'll give me fair confidence that it's in good shape? I'd hate to burn a lot of energy planning a coil around it and then find out it's a goner and that a working one will be $100's. That's a trap I'd prefer not to fall into!!

)

Best Regards,
Aaron, N7OE

Steve Conner

Wed Oct 11 2006, 04:07PM

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

Hi Aaron

I used two of the slightly smaller CM600HA-24H in parallel in my OLTC2. After a lot of tinkering I got it to run at over 4kW of power and produce 80" sparks. Your brick is probably good for somewhat more. The peak current rating seems to be just a guideline, since the peaks in a Tesla coil circuit are much shorter than the 1ms pulses that Powerex specify them for. Mine were rated at 1200A peak, but I think I ended up running them at something like 4500A peak

When I got my CM600HA-24Hs, I built a test fixture with some MO caps and a single turn loop inductor to pulse current through them:

The OLTC is really not too different to the SISG. As far as I'm concerned, the main difference is that I developed the OLTC to use DC resonant charging. This makes the most efficient use of the components, but Terry's SISG trigger circuit can't cope with that (it would fire just once and then stick on forever) I'm working on that, but I am stuck for a way to do it without ruining the simplicity of Terry's circuit :(

J. Aaron Holmes

Wed Oct 11 2006, 10:18PM

Registered Member #477 Joined: Tue Jun 20 2006, 11:51PM
Location: Seattle, WA
Posts: 546

Thanks, Steve!

4500A is a lot of amps!

How do those Geek caps fare at those hideous currents?

A simple ~100W design might be:
- Voltage double my 240V, 60Hz mains
- 600V firing voltage
- Terry's "Piranha" charging topology (second edition)
- This all suggests a 60bps firing rate, yes?
- 10uF + 600Vpk + 60bps => 108W coil power
- ~1uH single-loop primary => ~1900A peak primary current!!!
- ~50kHz resonant frequency (seems "ok")

All this for what will surely be some pretty wimpy sparks!!

But hey, if it works at all, it'd certainly be one of the simpler OLTC-ish..."things"...out there

The primary peak current makes me worried for the tank cap, though. Even the 600VDC-rated CDE942C's top out at around 1500A. I'd prefer not to have them going off like firecrackers! Would the 600VDC ones survive? They come in 2.5uF, so I'd only need four of them! That would sure be convenient.

EDIT: Actually, I'm thinking about the cap current all wrong, aren't I? Haven't done much MMC'ing. The per-string current is the peak current divided by the number of strings, yes? So with four "strings" (of one cap each

), I only see ~500A per cap. Ok, maybe that's not so scary after all.

Regards,
Aaron, N7OE

Steve Conner

Wed Oct 11 2006, 11:42PM

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

Hi Aaron,

I used between 14 and 32 parallel "strings" of one 1uF cap, so they didn't see too much current each

Terry Fritz

Thu Oct 12 2006, 03:30AM

Registered Member #393 Joined: Tue Apr 18 2006, 12:30AM
Location:
Posts: 297

Hi,

First to test it.

Measure the resistance from the gate to the emitter and collector in both directions (positive to negative and negative to postive). In all cases, the gate resistance to the other terminals should be "infinite" or look like a say 200nF capacitance (!!!). If there is resistance, it has been damaged but "might" still work.

Take a 9V battery and charge the emitter to gate to -9v. The collector to emitter should look open. Then charge the gate to +9v and the collector to emitter should look shorted. If it does that without any trouble, it is good

Typicaly if these bricks fail, there is case damage as in cracking. But the electrical test is really good for detecting damage. Surplus parts like these are typically removed when another brick blows and they replace all of them. In many cases the other bricks are fine.

It could be used as an SISG easily and you can probably just add another SIDAC to make it fire at 1200V. Then just use it as a OLTC type thing. The 200nF needs some thought though since that is 40X normal for an SISG IGBT. I will have to think on that

You probably can just eliminate the 100nF cap and hope the turn-on is "slow" enough to allow it to reach say 24V on the gate before the IGBT takes over.

Or, go to a 1uF cap...

Cheers,

Terry

Finn Hammer

Thu Oct 12 2006, 03:19PM

Registered Member #205 Joined: Sat Feb 18 2006, 11:59AM
Location: SkÃ¸rping, Denmark
Posts: 741

Neve C. Stoner wrote ...

I developed the OLTC to use DC resonant charging. This makes the most efficient use of the components, but Terry's SISG trigger circuit can't cope with that (it would fire just once and then stick on forever)

Could you explain why it will stay on forever?

I thought the SISG circuit could work this way, if it was fed trough a charging choke with a voltage lower than half of the Sidac`s breakdown voltage:
A Thyristor, or an IGBT, is connected to short out 3/4 of the Sidacs. The first ringdown is initiated when this SCR or IGBT is triggered.
When the double voltage, due to the charging choke arrives at the primary cap, it is stil not enough to force breakdown of the Siac`s, so the trigger elements still have to be operated for another ringdown.
This would enable full remote triggering of the coil.
I have an idea for a trigger circuit.

But of course, if it doesn`t quench after the first ringdown, this won`t work.

Cheers, Finn Hammer

J. Aaron Holmes

Thu Oct 12 2006, 05:03PM

Registered Member #477 Joined: Tue Jun 20 2006, 11:51PM
Location: Seattle, WA
Posts: 546

Terry Fritz wrote ...

Hi,
First to test it.
(verbose testing instructions snipped)

Thanks Terry, Steve. Wish I'd read this last night! Now I'm at work and itching to get home and try it out!!

Terry, one SISG question: Where does the 785A/us figure in the SISG PDF come from? (this in your 250kHz 500A Ipk example). I'm trying to understand this section of the doc a bit better now so that I can work out the implications of the capacitor (e.g., if 1uF were used).

Thanks!
Aaron, N7OE

Steve Conner

Thu Oct 12 2006, 09:36PM

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

Hi Finn, all,

These are the problems I expected to happen with DC resonant charging:

1: Terry's SISG won't be able to run with discontinuous current unless it's modified to trigger on command. This is because it's triggered by voltage, and once the inductor current falls to zero, the voltage will quit increasing. Therefore, the closest you can get is to tune the voltage such that it fires exactly as the charge current stops. This won't be possible to do exactly due to circuit tolerances, therefore in practice the SISG would always end up firing before the charging current stopped, and hence the charging current would never stop.

2: The average voltage across an inductor is always zero. (not counting IR drop, but at the high voltages and low currents in Tesla coil circuits, it may be possible to neglect this.) Put another way: Inductors don't ballast DC!

3: Terry's circuit stays on for rather a long time, during which time the inductor is shorted across the power supply and the current will ramp up quickly. This could cause boost converter action which would be self-reinforcing.

Due to the interaction of these three factors, I'd expect a SISG running off DC to run away, with the breakrate increasing and inductor current ramping up until the SISG stuck on permanently and blew the power supply out. It might be possible to find a power supply voltage setting that gave stable running, but I'd expect it to be very touchy. I thought of some ways to tame it:

1: Use command resonant charging to force discontinuous current. Unfortunately this needs a high voltage switch that can be triggered, and if we had a reliable triggerable HV switch, we could just have...

2: Achieved discontinuous current operation by making the SISG itself triggerable. This would make it more complicated.

3: Add resistive ballast in series with the charging choke. This wastes power, though, and would probably just tame the instability a little rather than getting rid of it altogether.

4: Use hybrid DC/AC resonant charging with a full-wave rectified unfiltered HV transformer. Run this off a lamp dimmer type of circuit to cause zero periods in every half cycle where the SISG can have a chance to recover. This is really just command resonant charging with the HV switch moved to the primary side of the power transformer and the breakrate fixed to 100/120Hz as a consequence. (You should find a dimmer setting where the breakrate is exactly 100/120.) The charging choke could be replaced with an ordinary ballast on the primary too.

I got a couple of SISG modules from Terry, and I hoped to try this stuff out, but the more I look at his original design, the more it seems to "just work fine" as it is.

Terry Fritz

Fri Oct 13 2006, 03:00AM

Registered Member #393 Joined: Tue Apr 18 2006, 12:30AM
Location:
Posts: 297

Hi,

Terry, one SISG question: Where does the 785A/us figure in the SISG PDF come from? (this in your 250kHz 500A Ipk example). I'm trying to understand this section of the doc a bit better now so that I can work out the implications of the capacitor (e.g., if 1uF were used).

See this:

It is the change of current vs. time equation. A calculus derivative.

Stever McStoner Mc Connerr... said :o))

Something like the SISG might not run well off a pure DC source. But a little resistance will probebaly drop the source "enough"... Think a simple RC filter there for isolation. In the PIRANHA, the MOT's inductance really is the only isolation along with minimal curent limiting from the MOT.

It woorks good!! That giant 165nF primary cap is not "trivial" to charge!!! The "DC" supply would have to charge it back to say 6800V in 500us for it to "hang on".

Cheers,

Terry

Steve Conner

Fri Oct 13 2006, 10:18AM

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

Terry, I never quite understood your voltage doubler charging arrangement. As far as I can tell, it has to either fire at 60bps, or else only charge half of the tank cap for each bang. In the latter case it would waste 3/4 of the energy capability of the cap?

Steve

1 2 next

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