Big mo-fo ferrite HV transformer driving (another CCPS/SLR inverter) / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Big mo-fo ferrite HV transformer driving (another CCPS/SLR inverter)

1 2 3 4 next

Move Thread

LAN_403

uzzors2k

Tue May 06 2008, 09:25PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

I'm about to use the cores I was so lucky to win on ebay a while back, and I thought it would make an decent project thread. So far I've made the transformer and fullbridge (my first!) so most of the fun is already over. I should have posted earlier to create some suspense, but then I guess this project is nothing new. Control electronics will be a garden-variety TL494 driver, maybe with over-current protection. SLR inverters are constant current sources unless I'm mistaken so the over-current protection is useless really, or am I mistaken?

For the fullbridge I'm using IRFP450s for the initial tests, permanently if they hold up. I've never touched an IGBT before but at a first glance they seem to far out-preform mosfets, where's the drawbacks? The transformer is wound with 0.3mm wire, about 250 turns per winding. With 10 primary turns that a step-up of 50, so around 16kV peak from + to -. The windings were actually cut form a secondary I made a long time ago. 0.3mm wire on a 5 cm diameter former doesn't make for a nice fres.

And finally some pictures of my ghetto H-bridge and the biggest ferrite I've every possessed.

1210109079 95 FT0 Fullbridge Electronics

Arcstarter

Tue May 06 2008, 10:17PM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

Sweet! I cannot wait to see this thing run. I loved steve wards so i would like to see new ones. The only drawback i know comparing mosfets to igbts is igbts are MUCH slower than mosfets. I have never owned one but i know the big ones somethimes are 200khz or less(unless i am wrong).

teslacoolguy

Tue May 06 2008, 11:00PM

Registered Member #1107 Joined: Thu Nov 08 2007, 10:09PM
Location:
Posts: 792

That looks like it is going to be one nice ccps.

Finn Hammer

Wed May 07 2008, 06:59AM

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

Uhzsaurs wrote ...

0.3mm wire on a 5 cm diameter former doesn't make for a nice fres.

Doesn`t matter.

Fres is determined by the leakage inductance and the series capacitor. Short the secondaries and resonate the tank with a function generator. Monitor peak with scope. From this, calculate the leakage inductance.

With these data, you can calculate power troughput, as a compromise btwn. frequency, leakage inductance and capacitor size.

Cheers, Finn hammer

Marko

Wed May 07 2008, 10:00AM

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

Eirik that transformer is a monster, it completely dwarfs the puny bridge of IRFP450's.
I don't think you'd be able to get a significant fraction of this core's rated 10+'ish kVA with IRFP 450's.

In your place I'd look for some IGBT modules, 100 amp or so modules and a suitable large heatsink. 1200V devices also have advantage that you could run them from rectified 3 phase supply (which you would definitely need if you want to fully use that core, as it looks to me.).

IGBT's are also proffered in SLR because of high peak currents which occur in capacitor charging due to their relatively constant voltage drop instead of constant resistance.

You could as well parallel a whole bunch of IRFP450 inverters, but I'd really advise against that - it will end in a huge mess, be hard to heatsink, hard to provide separate gate drive to all mosfets and in the end you'll surely regret why haven't you just bought some larger devices.

Look at finn's thread about what he got around. I'm sure he can help you much more...

Doesn`t matter.

Fres is determined by the leakage inductance and the series capacitor. Short the secondaries and resonate the tank with a function generator. Monitor peak with scope. From this, calculate the leakage inductance.

With these data, you can calculate power troughput, as a compromise btwn. frequency, leakage inductance and capacitor size.

Cheers, Finn hammer

Finn, I think he meant that the secondary coils now on the transformer were a part of tesla coil secondary before, which was useless to him as it was wound with too thick wire and had too high Fres. Nothing to do with CCPS.

uzzors2k

Sat May 10 2008, 10:12PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Marko wrote ...

Finn, I think he meant that the secondary coils now on the transformer were a part of tesla coil secondary before, which was useless to him as it was wound with too thick wire and had too high Fres. Nothing to do with CCPS.

Right.

I'll buy some IGBTs if I'm sure I can keep them alive, but for now I'll have to do the testing with the hundreds of IRFP450s I have. Their constant voltage drop will really reduce power loss, by at least 75% if I find some good ones.

I finally got everything ready for some real tests today, and I even got some arcs! But before I can take any pictures I need to solve something- I put an over-current shutdown circuit in since I plan to use this inverter for other things later, and with all the mosfets involved it's a good idea anyway. When running from a 50V supply the current waveform looks good, and when the transformer is open circuit the current drops. However when running offline I notice this: The second current spike grows while the first one decreases, this happens as an arc is drawn out. If I leave the transformer open-circuit or without an arc the over-current protection trips. So I assume the current rises high with no load. How do you guys pull arcs with a these things?

2A div, 5Âµs. Transformer shorted, run from mains. Btw, I am using an external 40ÂµH inductor. It sort of swamps the transformer inductance of 20ÂµH, but that might be good. Why is it that Steve recommends using an external one?

Electroholic

Sun May 11 2008, 12:05AM

Registered Member #191 Joined: Fri Feb 17 2006, 02:01AM
Location: Esbjerg Denmark
Posts: 720

go to the tesla coil page and read the ccps thread from start to finish, that ought to get you up to speed.

Steve Conner

Sun May 11 2008, 11:05AM

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

Sounds like the higher voltage is making some of your magnetics saturate when you run offline.

Also, have you tried lowering the drive frequency a little and reducing the duty cycle slightly? This might help by making sure that each cycle of resonance is completely finished before the next one starts, so everything will be totally reset with no chance of subharmonic or chaotic effects.

Finn Hammer called the shape of this waveform "buttocks", so I guess what I'm saying is that it's not good for them to be clenched tight together all the time. Doing what I suggested should move them apart and open up a slight period of zero current between them.

The reason why Steve Ward suggested using an external inductor is that it allows a high ratio of magnetizing inductance to leakage inductance. If you start messing with the transformer to add leakage inductance, it can lower the magnetizing inductance too, which brings the danger of unwanted resonances when the secondary is open-circuit.

uzzors2k

Mon May 12 2008, 10:21AM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

Thanks for the tips. I haven't been able to do any actual tests on the transformer over the last two days becasue I was improving the gate drive. Everything looked ok when run without any power applied to the bridge, but with just 50V applied the gates started ringing like crazy. It was so bad that low duty cycles would result in a mess of ~500kHz oscilliations in the linear region, which probably caused problems when running it offline. After some intensive work I fixed it though, and now I can hardly see a difference in the gate waverforms with 50V applied and without.

1210587486 95 FT44737 Waveform Comparison Slr

Now it should be possible to take it under 50% duty, which I doubt I was able to before.

uzzors2k

Wed May 14 2008, 04:58PM

Registered Member #95 Joined: Thu Feb 09 2006, 04:57PM
Location: Norway
Posts: 1308

I rewound the transformer for better coupling, now the primary inductance is 5.5mH with 17.8ÂµH of leakage inductance. I tried playing with different frequencies and duty cycles, but I can't find a point where the waveform looks really good. The closest I get is the "tight" waveform seen earlier. I ran a 25W bulb from the output of the full-bridge, but even with just a sliver of a gate pulse the output still seems around 25%, so maybe I can't get the duty cycle low enough.

Anyway I asked myself, which benefits does SLR have when you just want to pull arcs? Constant current is nice, but good open circuit performance is even nicer. Therefor I'm considering other soft-switching techniques. Are there any that are easy to implement into a full-bridge?

Just to get some arcing action I added a primary side ballast inductor of 40ÂµH, increasing the apparent leakage inductance. The short-circuit current can then be regulated by the swithcing frequency and kept under control. I know it's hard-switched but at least I got to see it arc. I wasn't able to take pictures of the arcs unfortunately, just Lichtenberg figures.

1210782130 95 FT44737 Lichtenberg Secondary

1 2 3 4 next

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