20KW DRSSTC / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

20KW DRSSTC

first previous 2 3 4 5 next last

Move Thread

LAN_403

Steve Conner

Tue Mar 27 2007, 09:16PM

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

Hi Aron,

I have no idea if my driver is as reliable as Steve Ward's, and a coil that throws 18' sparks is probably not a good place to start the testing!

I have never had a single IGBT failure with it, and neither has Finn, AFAIK. But Steve's is pretty much the "industry" standard now.

If you want to know more about the PLL, see

Tom540

Wed Mar 28 2007, 01:56AM

Banned on 3/17/2009.
Registered Member #487 Joined: Sun Jul 09 2006, 01:22AM
Location:
Posts: 617

Speaking of your PLL Steve. I was thinking of using it to rebuild one of my coils. I have a question about your schematic. What are the parts labeled as LCL1-5? They look like two series resistors and a cap.

Steve Ward

Wed Mar 28 2007, 03:00AM

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

Looks like you have gotten yourself in the same place i was at, with the inherent IGBT delays.

For the CM600 bridge i built for Chris R, i put RCD snubbers (using film and ceramics in parallel for the C) on each IGBT. This helped a little... The other thing was to slow down the turn ON of each IGBT, since its really the diode recovery that kills ya in this case (since the switching lags the "active" current, and takes place during the recycle portion of the current flow).

In the end, for my own coil, i decided that i just had to reduce the input voltage to a safer level, which meant no more than 700VDC on the 1200V silicon (and even that might be pushing it). Ever notice that all of the tests in the datasheets list the supply voltage at 600VDC? There must be a reason for all that headroom... and yes, that sucks.

hvguy

Wed Mar 28 2007, 07:16AM

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

Worst case voltage spike was found to be 2.3X the rail. This definitely explains why the 1KV TVS string exploded. Here is a scope shot at 170V:

Although the caps are already very close to the terminals, about 2â€, that is still room for improvement. This design was not originally intended for 20KW or 800V. I actually built it for a commercial application at about 6KW at 350V. At that voltage and power level the leakage spikes where of no concern. However, that project fell though so I decided to use the driver as-is for the 16â€.

I managed to get the fast spikes down by about 50% with an RCD snubber. The problem is the snubber was dissipating about 20W at 170V. That means well over 100W at full power, per IGBT, if I go that route; not cool. Unfortunately, I think a mechanical redesign will need to happen in order to get these spikes down. Hopefully this will allow me to hit the 800V/24â€™ mark. I will still include some form of protection, hopefully a TVS string will do it provided the spikes are occasional rather than almost constant as they are now.

I agree Steve; it is becoming ever more obvious why these large bricks are rated more than 1KV. BTW, what values did you use for your snubber? I ended up with 25ohms and 220nF.

Hazmatt_(The Underdog)

Wed Mar 28 2007, 07:29AM

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

Can you not snub that HF transistion by putting 1 or 2 turns of one of your power busses through a huge ferrite core?

I might be way off, but that really HF switching noise is going to be almost impossible to eliminate unless you use some kind of choke.

Steve Ward

Wed Mar 28 2007, 04:43PM

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

Can you not snub that HF transistion by putting 1 or 2 turns of one of your power busses through a huge ferrite core?

I might be way off, but that really HF switching noise is going to be almost impossible to eliminate unless you use some kind of choke.

Eeek no! thats the exact opposite of what is needed since any added inductance (while lowering the frequency) would store more energy, making the spikes bigger yet. Chokes are fine if you need to keep noise from getting in, but they dont actually stop the noise from being generated. I cant blame you if you dont immediately understand the problem Aron has, since its more of a power electronics issue (something i certainly havent learned from school yet), and you probably wouldnt ever think about it unless you start messing with these big machines.

Aron, your spikes of 2.3X Vbus arent *that* bad really. I found that as i increased my voltage input, the spikes became less severe (of course, this might just be my cheap scope probes to blame). If you are only going to 700VDC on 1700V silicon, then you should be OK i would say. Anyway, if you snubber is dissipating that much power, you need to redesign it. For the CM600 snubbers, i think i had a .68uF film cap plus .1uF in ceramics. The resistors are not wired to completely drain the caps, but rather to reset their voltage to Vbus. If you wired your RCD to completely drain the caps every half-cycle, then yes, thats a lot of energy to dissipate. I found the performance of the 2 different schemes to be similar really, and the first scheme disspates very little power overall (i was able to use 5W resistors). Of course, unless your snubber has lower inductance than your buss structure, you wont gain anything.

Keep in mind that your 1uF cap, without any sort of diode in series, will allow ringing. Id be curious to see what happens if you removed that cap (in some cases i found it helpful!). I was also surprised to see how much a dinky .1uF ceramic disk cap (the ones that are the size of a nickel) did to help reduce the ringing... Of course, if you cant get your snubber down to <20nH, its almost useless

Finn Hammer

Wed Mar 28 2007, 06:24PM

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

Aron,

You coil is awesome!
I have been working on a way to predict the zero crossing in advance, on a level where there is no PLL or other strange stuff involved.
I have gotten it to work in Orcad, this is how it is:

At the resonant frequency, let`s assume that when the current has fallen to 1/10th of the peak current, then there is 1ÂµS to the zero crossing.
I`m not using real numbers here, just some nice numbers to help draw the picture.

Now, if you take the signal from a current transformer monitoring the primary current, and rectify it, you have a pulsating DC signal that indicates the primary current.
You use this signal to do 2 things:
You sample the signal, and hold it at the peak. Divide it with a 1:10 resistive divider. Feed this into a comparator.
The other input of the comparator receives the undivided signal from the transformer.
When this signal gets smaller than the 1/10th of the peak signal, the comparator changes state, and there you have your "let`s initiate the transition" signal.

From this signal, with RC delays and schmitt triggers, you can derive the precise timing for the turn on and turn off of the bricks.

In a real circuit, there are diode drops to compensate for, source followers to add servo boost etc. but this is just the basic idea.

In orcad, this has shown, that within reasonable tuning points, the IGBT`s are not forced to switch more than 45A even if the peak current is 1200A.

This scheme will not adapt to the change in resonant frequency caused by streamer loading, but it will allow you to tune the coil to produce beat envelopes should you desire so.

I can send you a Orcad file to play with, if you are interested.

Cheers, Finn Hammer

hvguy

Wed Mar 28 2007, 10:39PM

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

Thanks Finn. If you don't mind I would like that file, you can send it to **link**. I like the concept. Sounds like it should be simple and relatively effective. Compensation for the frequency shift would be nice, but a shift of only 2KHz or so should not have a significant affect on things since the ZC will only change by a 100ns or so.

While testing various snubber options I notice another problem. Every time my girl friend turned the hot water on in the house, 150â€™ away, my driver would begin to oscillate at about 2Hz and trash the GD waveforms. Ok, this was strangeâ€¦ I figured the most logical reason for the oscillation was due the lower line voltage when the 17KW instant hot water heater was on. I measured the line voltage, and sure enough, the voltage dropped from 117V (already a little low) to 113V. While this was clearly causing the problem I was confused as to why a 4% line voltage change would have such a profound affect on the logic. Just to verify the results I hooked the logic board to a variac and lowered the voltage until the problem started, which was at about 115V. My 15V rail is regulated and showed no change even at 90V on the line. The 160V rail for the GD dropped, as it should, but a change of 7v is divided down though the GDT to only 1V so there was no significant affect on the already 30V gate waveform. At this point it occurred to me the only non-regulated logic device on board is the PC817 opto coupler between the remote ON/OFF switch and the logic. I checked the voltage at its emitter and found it to be right at the turn on threshold for the 2N7000 it was driving; thereâ€™s my problem. Simple enough, I changed the 1K resistor on the PC817s LED to 560ohms and the 1K pull up on the 2N7000 gate to 4.7K. This fixed the problems but it makes me wonder how much of an affect this had on the TVS failure. I know the coil drops the line voltage out here in the shop by at least 10V at full power. This may have been the reason for the failureâ€¦ Either way its fixed and the voltage spikes are still to excessive for my liking, so back to the bench.

BTW, Steve, it turns out my CDE snubbers where causing more harm than good. I am trying different combinations of caps to kill the HF spikesâ€¦. Weâ€™ll see how it goes.

Tom540

Thu Mar 29 2007, 03:15PM

Banned on 3/17/2009.
Registered Member #487 Joined: Sun Jul 09 2006, 01:22AM
Location:
Posts: 617

Hey Aron,

Wow that coil is 1337. I'd love to see that thing run without a breakout point. Anyway I thought I would do some research here at work Before I quit in two weeks. We(Elgar)make some big beefy power supplies here. Some are in the 30KW region. I'm gonna browse through the schematics and see if I can find a super snubber design that can tackle your problem. Theres gotta be something.

hvguy

Thu Mar 29 2007, 09:25PM

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

That would be great, thanks. I recently purchased two surplus Elgar AC supplies to sell on eBay. I opened them up before listing them; very well built equipment.

I am currently redesigning the full bridge mechanical layout. I think the design I have come up with is about as low inductance as is humanly, or mechanically, possible. I will post an update with pictures and test results later today.

first previous 2 3 4 5 next last

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