DRSSTC, yes another! / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

DRSSTC, yes another!

previous 1 2 3 4 next

Move Thread

LAN_403

Marko

Sat May 26 2007, 10:34PM

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

Hi, I have been lurking around, seen tons of new members and threads... although I don't post much lately. This coil really looks better and better.

Yes, I did. I stole that idea from EVR... it seems to work quite well. It's a 10uF 35V IIRC.

Hey, can you explain what are you doing there? Is that some kind of (really odd) bipolar cap?

That's just it... I couldn't figure that out either. I mean, the only thing I could come up with were voltage spikes related to the D-S spikes (which at the time were just a little over 160V) or I happened to smash off the insulation when I built the GDT. The latter is much more probable, but I was careful this time 'cause it had happened once before. Still, I'm not believin' that 15/30V ate through two thicknesses of even smashed insulation. I'll bet it has something to do with Steve C's comment, I guess I'll be doing the Chris Hooper epoxy bit on the next series.

Really.. if that happened to me, especially two times in series, I would be really pissed off.
Voltages present there are way too low to do anything serious to the insulation.
Any voltage pulse energetic enough to do that would kill the IGBT much sooner.

Identically constructed GDT's have passed with 800V DC + transient spikes so I must say this is really a ridiculous failure.

And even if you completely broke the insulation in many places, it's still virtually impossible for it to happen on two adjecent wires and two times in a row.

Before putting new IGBT's in, I would do serious autopsy of that GDT and probably some crash tests.

I have wound one of my GDT's woth magnet wire and after screwing it (directly!) to perfboard, it showed a short between two windings (apparently I damaged the insulation in process).
After that I put some PE between transformer and the board it was OK and then I ran itup to 325VDC without problems. I was probably very lucky, but still.

This needs to be fixed right now.

Part Scavenger

Sun May 27 2007, 03:45AM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

Ok, here's the workup on the GDT failure.

The primary reason the GDT failed is because of the below picture. You'll note that in the top right hand corner of the picture on the green/white wire, I have some "smashing" there was an alarming amount in the GDT actually. I guess I did have cheap CAT5e. It didn't get hot, so I don't know what it's problem was.

In the case Steve Connor was mentioning, I would have had a short from secondary-primary; but in this case the blue and the brown wires were shorted indicating secondary-secondary shorting. This occurs because the voltage potential between GDT1 and GDT2 (see schematic) is:

plus whatever voltage spikes I may have. In this case that would have been 0-80V bus + about another 80V. So, basically, 80V+ found it's way into the GDT across Q2 explaining why Q2 failed. I presume it found its way through the IGBT gate. I guess that means I need to replace my diodes even though they test OK. C2 did not discharge through the GDT because of the MMC, and I was puzzled why Q1 did not let go but then considered that it's drive signal was most likely loaded down below the active region. This shed a bit of light on my last failure, I was not so lucky last time, Q1 shorted along with Q2 dumping C1 & C2 into the GDT causing total destruction of the GDT, the 33V zeners completely vaporized, and literally killed everything in the bridge. At that point I decided it was time to build this new DRSSTC.

NEXT: I'll make a GDT that can stand up to 600V. I will make the primary, and both secondaries separately, wind the primary first, and the secondaries on top of it in an attempt to get them away from the core a bit. Then I'll hook it up to my HV power supply, probably test to 1000V and stop if something doesn't end the test sooner. If it fails, I'll dip it in epoxy like Chris R. I don't have any more CAT5e cable, but I went "diving" behind the phone company yesterday and scored a couple hundred feet of CAT3 cable!!!

Tell me, why do they throw stuff like that away? My phone bill is high enough! But, the insulation looks much better on this stuff.

Oh, and about the tantalum? It's polar all right. Maybe EVR can explain this one a little better 'cause I'm not sure. I just know it works.

The positive side goes toward the UCC37321, so I'm guessing that there's no problem with pumping negative current up the little tant's backside. I would have thought it would overheat, or die because tant's are quote: "rabidly polar." But, nope the little guy's been humming along just fine.

Steve Conner

Sun May 27 2007, 10:50AM

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

If you managed to blow the GDT out that way with only 80V on your DC bus, you probably have a lot of spikes and ringing going on. You might want to take a look at your bypass capacitors and the layout of wiring between the DC bus cap and the bridge. Unless the insulation was just so badly smushed that it gave out.

Marko

Sun May 27 2007, 11:51AM

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

Scavenger (really, I haven't figured out your real name yet), I think you are really being overly paranoid about the insulation. Crappy magnet wire GDT works just fine to 600V and even more
considering insulation isn't damaged (problem is that in most cases it *does* get damaged).

Two mutually twisted cat5 wires seem to break down at 15-20kV DC depending on luck.
And your cable looks pretty good to me.

If you managed to blow the GDT out that way with only 80V on your DC bus, you probably have a lot of spikes and ringing going on. You might want to take a look at your bypass capacitors and the layout of wiring between the DC bus cap and the bridge. Unless the insulation was just so badly smushed that it gave out.

So overvoltage (this is about 100th time people blame ''voltage spikes'' for failures) just can't be the cause. For those low voltages wires really need to be in touch or really dirty.
For those, just put a TVS over supply rails, but I assure you you won't see any difference.

I really believe there had to be something else that damaged the insulation.

Your core, your heatsing or something else touching the wires could have been building up voltage via capacitive coupling and buning the insulation down.

Form your pic, it even looks like you really had arcing from the core on that GDT.
You can try winding entire cable with outer insulation or putting heatshrink around the wires (like I did).

Still, all of those causes are very unlikely, and you are probably the first person ever to have CAT5 GDT to fail.

If your soldering was crappy, and you really heated a hell out of those wires, they could have had displaced the insulation as it softens and melts and come into short.

Part Scavenger

Mon Jun 04 2007, 03:25AM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

Excuse me! My soldering job was perfect thank you very much... LOL

I wound another toroid with the stuff and hooked it to my HV supply, I don't remember where it arced, but it wasn't very much. Good ways under 1000V. The setup was along the lines of 0-1000V with some capacitance in parallel with the GDT so the arc would cause enough damage for a good inspection and/or wouldn't go un-noticed with my HV supply's output of > 10mA. So, I made a new GDT out of CAT3, 'cause that's all I have now, and it passed the 1000V mark just fine. For this GDT, I wound the primary first, and then the secondaries on top trying my best to keep it all separate. Tests revealed a VERY nice waveform @ 200kHz.

I'm holding off on replacing the bus capacitors, I've got friends coming over tomorrow, and I didn't have time to get all that worked out before then. They'll be impressed by 10" @ 50V sparks anyway.

Soldered the GDT in... replaced all other components in the bridge, let's go! It's prolly important to note I've switched to using FGA40N60UFD's,(which are like $3 IGBTs) now just to see how far I can push them now that I know what I'm doing. Ok, MORE about what I'm doing. LOL.

These are some scope shots from today, I was wondering if you guys would help me out a bit.

This is a dual trace of the voltage across the bus vs. my primary current waveform. Most of the time it's ZCS-ing pretty nicely, but sometimes I get a trace like this. I can't figure out why, and it usually comes from turning up the variac some more. No amount of tuning fixes it, and I haven't figured out what makes it go away yet, other than it "just does." Any experience anyone? Do you think I am just getting interference from the secondary into my waveform? When it's doing this, it does seem to closely line up with the secondary's output wave... I'm just not too keen on blowing up any more silicon.

EDIT: After inspecting the cause of this problem some more, I have concluded that it was interference from the secondary. In fact I didn't have my brand new, nice O'scope probe connected properly so I was getting ALL secondary waveform and NO primary current waveform!!!!!! YIKES! Ok, fixed now, and my bridge is running quite well. I'm all but ZCSing at 200V now!!! After a 5 or so minute run, the heatsinks were still cool to the touch!

This is a look across the bus. Can anyone tell me what is going on at the end here? First it drops down to like 0.5 the bus voltage, at the end and then rings out. From looking at other people's stuff, the ringing out is normal, but can anyone tell me what it is? How about the diminished switching just before it?

EDIT: I think this has to do with my crappy bus capacitor setup(which also makes it FREAKIN' hard to trigger the O'scope 'cause it's bouncing around so much), and should be corrected when I get my new bridge built. I'm guessing that it has to do with when the line voltage charges the capacitors.

EDIT:
I had a good run today, my cousins and their friends and my relatives were all at my house for the day... I can't tell you how good it feels to be able to stand up and give a demonstration, turn up the voltage 2x past what I have tested for and without any failures!!! I really like these $3 IGBTs, they are alot less trouble than some more expensive ones I have used. I was getting 20" ish sparks at less than 100A (I think. That's what my OCD was set for, but the O'scope said 150, I was busy at the time and didn't investigate) @ between 200-300W. This thing SERIOUSLY needs tuning, it was tuned but as I upped the voltage and the sparks got longer it is now "un-tuned." :D I have also started getting breakout at places other than my breakout point which confirms my suspicions that it is too small! That's ok, I just made a REAAAAALY nice SSTC toroid and I was dying to make a new one for my DRSSTC anyway. This project may get a bit slow for the next week. I have to install an airconditioner in our house, and depending on how Wednesday goes, I might have to work on my new CNC machine!!!!! :P

Part Scavenger

Wed Jun 06 2007, 03:25AM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

Excuse me, I'm kinda on an O3 high right now... LOL.

I don't know what to think about my toriod anymore. Experience tells me it's too small, it looks too small and it breaks out too early. However, the impedance match calculation comes out exactly on the actual value, and this is the toroid Scan Tesla suggests. Hmm, I'll have to think on this one a bit, but for now I think I'll keep it 'cause it's really cool to have a DRSSTC that can run without a breakout point!!!!! The primary current isn't all that high after it breaks out, and it dances all over just like my old SGTC!

I'm not sure who to trust at this point, my CT says I'm running 300A peak, but my current limiter is set to 130A. To be honest, I wasn't very interested in figuring this one out tonight, I know these IGBTs can take either reading. I'm leaning toward my CT being wrong 'cause I still haven't fixed it properly, but if that's true, it's odd that I'm getting such spark lengths at such low currents.

An odd thing happens when you turn off the interrrupter without turning it breakrate/pulse width down. The coil sometimes does not turn off completely and the bus current draws almost exactly 1A pretty much no matter what the input voltage is. Needless to say, the IGBTs do not like this AT ALL. Also the primary current goes all funky. The heat sinks get hot quickly, and I must cool them off. Weird, I'll have to dig out the schematic and have a think, unless someone wants to just tell me the answer?

Because of my primary tuning setup, which is supposed to look really nice and does, it's a pain to tune it. So I used a jumper for initial tuning. Soon after switching out the jumper with my permanent tap, I realized that my tuning was WAAAY off, so the initial tuning was pretty useless other than I learned a good bit about tuning the DRSSTC with the scope. The best results were obtained when the envelope was set to reach maximum current before the first notch. It was amazing the difference between not tuned and tuned!

I got my best results at 80Vin, 24" somewhere under 300W (with a jumper that was getting stinkin hot). Efficiency got alot better after I switched to all copper. I after that I could get 20" at 200W, but no longer than that 'cause I grew tired of trying to tune it. I think it notched twice before reaching maximum current, so I'm pretty far off from "tune". I was further amazed at how far off raising the input voltage and increasing streamer length dragged the tuning. This testing stuff is fun now that I actually know what I'm looking at!

All in all, even though I didn't collect any hard data, I was really happy with performance this afternoon! I ran it continuously for almost an hour, total about 2 1/2hrs and I haven't fried a single FGAxx IGBT yet! I even ran for a while at some insane breakrates and nothing died. My soft switching looks excellent, as far up as I can see without my 10x probe (around 100V), my voltage spikes are under control... etc. AND there is ABSOLTELY no corona in places it ought not to be!!! FINALLLLLLLY! Unless I am running at over 200BPS, at this (off) tuning the coil refuses to draw more than 300W. Oh, and for the most part, the heatsinks don't go over "warm to the touch."

Oh, I can't wait to get this thing tuned and see what it will do!!!

Part Scavenger

Thu Jun 07 2007, 12:03PM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

So I was running it last night and my variac shuts off because I ran it over the variac's breaker rating. Why did that kill my IGBTs?

CT2

Fri Jun 08 2007, 12:02AM

Registered Member #180 Joined: Thu Feb 16 2006, 02:12AM
Location: Ontario, Canada
Posts: 187

Don't know about the IGBT's, but maybe the breaker tripped because the IGBT's died first? And watch out for that light! My flourescent light above my bench no longer works because of my coil haha...

Part Scavenger

Fri Jun 08 2007, 02:40AM

Registered Member #79 Joined: Thu Feb 09 2006, 11:35AM
Location: Arkansas
Posts: 673

LOL. I'm pretty darn sure the IGBT's didn't die first. When that happens the effect is totally different. I swear I heard the variac pop just before the sparks stopped, and I was running right at 5A, which is the breaker I put in there. Usually, when the IGBT's die I have time enough to unplug the variac before the breaker trips. It's SLOW, like 4-5 sec.

I'm not saying its impossible you're right, but I wanted to know if this was a failure mode I couldn't figure out.

Steve Ward

Fri Jun 08 2007, 05:13PM

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

Does this same breaker supply power to your control logic? If so, then that is likely a problem. The fix (for my case) was to install a 120V coil relay across my interrupter signal. If my control logic loses power, the relay grounds out the signal (the interrupter has 100 ohms output impedance for safety from this short). This is how i lost about 4 CM300s awhile back

previous 1 2 3 4 next

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