Odin The All-Fragger: a large DRSSTC / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Odin The All-Fragger: a large DRSSTC

1 2 3 4 next last

Move Thread

LAN_403

Steve Conner

Mon Mar 20 2006, 12:37AM

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

Update Oct 2014: Odin had a successful weekend at the 2014 Cambridge Teslathon. The electronics ran fine with no issues and output was again limited by building size

5th test run was a success. Biggest spark is now 9ft

Latest (4th) test run was a success! I got an arc of 8ft (2.4m) which is a new personal best.

Output was limited by ceiling height again.

After the last blowout, I rebuilt the bridge using surplus SKM400GB123D bricks with 3.3 ohm gate resistors, and increased the current limit to 1kA. I carefully tuned the phase lead and deadtime while running at high current with the primary coil only. Then I kludged an extra turn onto the primary and made a second topload (8" x 36" approx) from aluminium ducting.

I then rolled it out for another test run with my evil henchman Jim, and Alan, one of the original pioneers of solid-state coiling. We started with the same configuration as the end of the previous run: single topload and 5 primary turns. Even though the primary current was increased, the spark length stayed much the same. Fiddling with the primary tapping point didn't seem to have much effect. I also wanted to try the lower pole frequency, but the PLL refused to lock to it reliably for some reason and fiddling with the primary tap didn't help at all.

Then we added the second topload and the new 6th primary turn. This made a dramatic difference to the output, I now got regular arcs to the floor and ceiling 8ft above. It also hit a strike target positioned 8ft away horizontally. We ended the tests due to worries about burning the floor or damaging stuff inside the suspended ceiling.

The output from the Rogowski coil was quite hard to analyse due to noise that was getting into it from somewhere. But it looked as if the PLL driver's control of soft switching was just "meh". Not so bad that it would blow something catastrophically, but not tight enough for really low switching losses. I think a phase lead driver could do better, so I won't recommend the PLL for ordinary DRSSTCs any more. For higher impedance, tighter coupled coils, and for QCWs and SSTCs where you want to modulate the phase shift, it works great.

Anyway, to the moon!

1406055696 30 FT4896 Odin Test Burntisland A1

1406055696 30 FT4896 Odin Test Burntisland1

1406055696 30 FT4896 Odin Test Burntisland2

Update July 2014: The third test run was somewhat of a disappointment. The sparks were very bright and loud at longer burst lengths, but not very long. We could not hit a target at 2m horizontal distance, and one of the SKM300GB12T4s failed after about 5 minutes runtime. :( I didn't get any photos or video footage myself, but several other people were filming. I will post pictures when I get any.

I set the system back up in the lab, as a half bridge with the remaining good IGBT. I broke the top off it so I could probe it at the die level, and I also added a small Rogowski coil around the emitter terminal. The following scope shots show the collector voltage (probed as close to the die as possible) and the Rogowski coil signal showing the emitter current of the same device. The PLL driver settings were as close as possible to the last fatal run, I just adjusted the frequency to account for the fact that it was running with no secondary coil. The DC bus voltage was reduced as I didn't really want a face full of IGBT goop if anything went wrong. :)

These results suggest that the PLL driver was set up with too much phase lead. The funny thing is that even with this large hard switched current, and turning off the IGBT as fast as possible with no gate resistors, and probing as close to the die as possible, the voltage waveform is very clean. I can't see any spikes that would endanger the breakdown voltage of the device. If I tune the phase lead for perfect ZCS according to the Rogowski coil signal, the spikes get somewhat worse, and if I tune it for leading current, the spikes get terrible.

My next move will be to set it up with SKM400GB123Ds, gate resistors, current limit increased to 1kA, and the PLL retuned for less hard switching. I will try to log the Rogowski coil waveforms while making sparks, and that should settle once and for all whether the PLL driver is good enough for the job.

1404900606 30 FT4896 10496059 10152608278251584 2505759111616666150 O

1404900606 30 FT4896 10454371 10152608278231584 3861835156770858505 O

Update May 2014(4): Second test run with smaller 0.3uF tank capacitor and SKM300GB12T4 bricks was a success!

We hit the HV lab ceiling with power to spare.

Test video!

Update May 2014(3) First test run ends in disaster! Not exactly sure why, see last post for discussion.

Update May 2014(2) Successful dummy load test at full bus voltage and >1000A peak. (the needle pegged at 1000

)

Update May 2014: Completed the primary

Update April 2014: Trial assembly of everything. It all seems to fit together OK

The overall height is an imposing 1.8m.

Update April 2014: Winding the secondary coil:
315mm dia X 1m height, 1175 turns of 0.75mm wire, coated with 2-part epoxy.
After overheating the cordless drill motor, we changed to an old Black & Decker mains powered drill connected to a variac, with a vacuum cleaner for cooling.

Update March 2014 (2): Constructing the front panel for the power electronics box. First in a series of timelapse movies

Update March 2014:
An enormous bank of GTO snubber capacitors was procured. 0.75uF total capacitance at eleventy billion volts.
The bridge has been tested to 750A peak with a dummy primary coil.
The bridge and driver have been bolted together and mounting rails added.
A 12" x 39" coilform made of grey PVC air duct has been procured, as well as 6kg of magnet wire and some low viscosity epoxy resin for coating.
A new interrupter has been constructed.

1395511425 30 FT4896 Win 20140316 125439

1395511425 30 FT4896 Win 20140321 195719

1395511425 30 FT4896 935523 10152031541521584 1934577150 N

1395511425 30 FT4896 1393820 10151999009056584 773641908 N

1395511425 30 FT4896 1459979 10152186540611584 1329840340 N

Update, July 2013-
The 600V test was a success. Output waveforms still looked great so I adjusted the interrupter duty cycle to draw the full 1kW output of my power supply, and soon the dummy load (a steel coffee can full of water) was hot enough to make tea!

The duty cycle ended up at 2% so the peak power must be around 50kW and the RF current about 100A peak. The coil is still untuned, so the current will be a triangle wave and the full 100A will be hard switched. Nevertheless, the heatsink was still cold when the tea was ready. No fan installed yet.

Update, July 2013-
This project has been going so long that I gave up on it several times, and lost or sold almost all of the original parts. Nevertheless it is back from the dead!

I was motivated by meeting a local artist who works in an old factory that would be ideal for firing giant Tesla coils. I ended up selling Mjollnir to him, so now I need another DRSSTC.

I've just completed the bridge and test fired it at 150V DC bus voltage into a dummy load composed of a steel can with wire wrapped round it. There is no resonant capacitor: I'm hard switching to get a feel for the bus inductance and voltage spikes. The waveforms look excellent and the water boils nicely.

The test details are:

Operating frequency 60kHz
Interrupter: 400Hz, 20% duty
DC bus: 150V, 2.6A

The three large black blocks are 50uF, 800V metallised film caps from Panasonic. They cost about Â£12 each. I think they are intended as high reliability DC bus caps for solar inverters and other kinds of SMPS.

I originally intended to use the SKM300GB123D IGBT modules from Semikron, but when I first tried to test the bridge, I discovered that I only had one good one, all the others I had kicking around were DOA! An anonymous benefactor donated some SKM200GB128Ds, a newer generation using "soft punch-through" technology. They seem about as fast as the 123 series but with lower voltage drop.

I also ordered some SKM400GB125Ds from a Hong Kong seller with good feedback. These are rated for resonant operation up to 100kHz in induction heaters. Once everything is proved out with the SKM200s, I'll swap them in.

The next step is to increase the DC bus voltage to 600V and try to brew tea in the dummy load.

Update, November 2007-

Unfortunately the sale fell through due to me not being able to get it built on time, not to mention worries about reliability. All I can say is that it will "Maybe" get finished "sometime" :|

Update, September 2007-

Well, I've had a fairly serious offer from a sponsor who wants to buy the drive electronics from Odin and make their own resonator to go with it. So, I've decided to get back to work! If it falls through, I'll try to bring it to Cambridge instead, and maybe hook it up to someone else's resonator.

I finished rewinding and wiring up all the drive transformers, and all four drivers appear to be good. I connected two of the drivers to a spare Semikron 300A halfbridge IGBT brick for testing.

The drivers hooked up. Yes those are Cat5 cables.

Gate waveforms, check. The top waveform is the trigger reference from the signal generator, the lower two are the gate voltages. We are driving the two 300A IGBTs CW at 25kHz, to +24 and -12V.

I hooked the halfbridge up to a ferrite transformer from an electronic NST, and fed it with 20V DC. It made a nice spark. This is possibly the most overkill flyback driver ever.

Oh noes! The transformer is on fire.

What's the matter, little transformer? 300A IGBT brick too much for ya?

This proves that the gate drivers work, at 20V DC bus voltage at least! The finished coil will operate at 600V DC bus, so I still have some way to go...

Update, June 2007-

I tried Hi-pot testing the new gate drive boards, and the results were dramatic. The isolation transformers were destroyed completely! :(

So I tried a new GDT design using Kynar wire on Teflon wrapped cores. It easily survived a zapping with a MOT! The scope trace shows the voltage across it taken with a 100x probe.

I bought some more Kynar wire and made eight of these new GDTs, but haven't got round to fitting them to the boards yet.

I also have some plans for a fibre optic link using Toslink parts, inspired by jrz126. I plan to use three fibres to transmit the interrupter signal, the DC link voltage control signal, and the PLL fine tuning signal. I was going to multiplex the signals using chips designed for digital audio, but it seems less hassle (and less chances for EMI to crash things!) if I just use three copies of jrz126's circuit.

1182026231 30 FT4896 Img 1900 Mot Destroy

1182026231 30 FT4896 Img 1902 500vperdiv

Update, March 11 2007-

After a long time off this project, I got back on it by building another two gate drivers. I now have a complete set together and working. I'll hi-pot test them tomorrow with a MOT or whatever.

Things I still need to do: Build the "hub" that connects all the Cat5 cables to the PLL driver, with the inverter in it that supplies power to the gate driver boards. Hook the SCRs in the voltage doubler up, and modify the SCR driver for more current. Hi-pot test the dodgy looking insulation in the H-bridge. Wire up all the power wiring and meters. Hook the IGBTs up and test the drive. Fix the secondary. Build a new primary. Sort out the tank capacitors. Get some kind of fibre optic system to control it all. Etc.

Update July 8 2006-

I built another brick driver and compared the outputs from both of them.

1152463181 30 FT1630 Imgp0814 Brickdriversx2

1152463181 30 FT1630 Imgp0815 Bdx2 Mindead

Outputs at 200kHz with deadtime control turned to minimum.

1152463181 30 FT1630 Imgp0816 Bdx2 Maxdead

Same with deadtime turned to max. The deadtime looks a little uneven, but that shouldn't be a problem at the lower frequencies I would use bricks with. I'm just using 200kHz because that's what the PLL unit from my other coil is set to, and I can't be bothered opening it to change it.

Update July 1 2006-

I got the H-bridge and meters bolted onto the old wooden frame from the OLTC 2.

I also replaced the OLTC 2 single turn primary with a rough 9 turn coil, but I don't have a picture of that. (It looks terrible.) I wound a couple more of the signal and power isolating transformers too.

Update on June 17 2006-

I finally managed to get one of the boards built and hooked up to a power supply and gate drive. It seems to work as planned, though I bought the wrong ribbon cable connectors to hook it to the IGBT properly. Also, I can't test the undervoltage lockout, because I forgot to get the right zener diodes for it.

The power isolating transformer's core gets a little hot, so it probably needs more turns.

other minor oops-ups:
-Got the cores mixed up in the Cat5, so the HF power is twisted together with the UVLO signal.
-Never noticed that the MOSFETs I bought aren't actually in D2pak footprints. They still sort of fit, anyway.

finished board

scope screenshot shows the gate of a 600A IGBT being driven with +24 and -12v at 200kHz. >_< The second trace shows the zero volts level.

board had to be kludged onto IGBT with two short pieces of wire.

Update on May 14 2006-

The boards are back from Gold Phoenix, omg omg! There will now be a short interlude while I build a set up, bolt them onto some IGBTs, and test them. Assuming they work, I plan to sell all the extra boards I don't need to you guys

Update on April 13 2006-

I finished the outboard driver PCB, got the safety clearances as big as I could, and generated the Gerber and drill files for it. I just need to revise the main PLL board to get rid of all the bugs that were in the last version. I should have the board order placed with Gold Phoenix by the end of today.

1145117820 30 FT1630 Img 1696 Driver Mockup

a paper mockup of the board to check component footprints

Now I've nearly got the outboard drivers done.

schematic. Note the optoisolators that act as undervoltage lockout, Richie Burnett persuaded me to put these in.

preliminary PCB layout (still got to check safety clearances)

The first step was to build the H-bridge.

I'm going to try attaching about a dozen pictures to this and see if it works.

four big inverter grade capacitors (3300uF @ 420V). The tops are wrapped with electrical tape to reduce the chance of the metal can coming into contact with the metal case I'm using to hold them.

two 300A 1200V half-bridge IGBT modules and one half-bridge SCR module for soft-start and DC voltage control

Don't ask what the insulation is made out of.

These funny shaped busbars connect the IGBTs to the capacitors with minimum stray inductance.

like so

The HF output terminals on the bricks aren't connected to anything yet.

woot!

This thread continues from

on the old board.

ragnar

Mon Mar 20 2006, 04:09AM

Registered Member #63 Joined: Thu Feb 09 2006, 06:18AM
Location:
Posts: 1425

I love it Steve!

How are you driving the bricks, and what model are they? I've got some semikron bricks that look like that... muahaha.

Steve Ward

Mon Mar 20 2006, 05:50AM

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

Excellent. You always go out of your way for these ultra low inductance designs but... shouldnt it be soft switching anyway?

. Hey, where are your low inductance decoupling caps? Or ar those lytics so good that you dont need them? I expect at least 10 feet from a coil driven with that monster of an inverter! Good luck.

Marko

Mon Mar 20 2006, 09:21AM

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

The last picture is the best...
Nice huge heatsink, you can build entire DRSSTC on it itself

Im troubling myself to find anything larger than small CPU cooler, maybe few rusty TO3 heatsinks. Not to talk about 40N60s, TVS supressors...

Good luck then

Steve Conner

Mon Mar 20 2006, 12:54PM

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

Hi guys, thanks for the feedback! The "Nice large heatsink" was donated by Finn Hammer, its name is Finn's Fins.

blackplasma: The bricks are Semikron part# SKM300GB123D. I managed to get hold of the 400A version too, but I'm saving them till I've pop popped all the others

I don't know if anybody has used these in a DRSSTC before but I reckon they should work fine. If not, I tried to make sure the layout would fit the Powerex F and H series bricks too. I'm driving them with my PLL driver through a set of outboard gate drivers that I documented on my site.

steve: I was discussing this with Richie a while back. He said that commercial SMPS often just use the electrolytics on their own, and also I had trouble with resonances caused by the plastic film bypass caps on my last DRSSTC. So I'm going to try leaving them out this time. That's why I wanted the low inductance busbars, to get the electrolytics "closer" to the IGBTs so they would be effective. I think they have about 25nH ESL which is about average for inverter grade caps. I'll check the DC bus for spikes when I test it, and if necessary I'll add some caps in.

vasil

Mon Mar 20 2006, 01:25PM

Registered Member #229 Joined: Tue Feb 21 2006, 07:33PM
Location: Romania
Posts: 506

woot
If it will run OK, I will buy my own semkrons to build another drsstc.

Good luck Dr Conner!

Hazmatt_(The Underdog)

Tue Mar 21 2006, 11:51PM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

You might want to switch out your cap stud screws from black phosphate SHCS for Stainless.

Steve Conner

Wed Mar 22 2006, 12:09AM

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

Why? Everyone should be too busy looking at the sparks to notice that my cap screws aren't shiny.

All kinds of steel are lousy conductors of high frequency current, so I arranged it that every joint has direct copper-to-copper contact (or copper to aluminium in the case of the capacitor teminal posts) and the screws are just for clamping force.

Non-magnetic stainless isn't too bad- they make radio antenna whips out of it after all- but it's still not as good as brass, copper, or aluminium.

Hazmatt_(The Underdog)

Wed Mar 22 2006, 06:11AM

Registered Member #135 Joined: Sat Feb 11 2006, 12:06AM
Location: Anywhere is fine
Posts: 1735

*shruggs* I was just trying to consider how much current will go through the stud as opposed to the terminal face as opposed to current going through both. Not that I was worried about shiny, or cost, or corrosion or whatever.

Steve Conner

Wed Mar 22 2006, 10:12AM

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

I was kind of assuming that it would more or less all go through the terminal face since copper conducts so much better than steel. But I'll keep an eye on the screws and see if they do get hot, if so, I'll change them for some other material like non-magnetic stainless steel or brass. Not that I've ever seen a socket headed cap screw made of brass

1 2 3 4 next last

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