20KW DRSSTC / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

20KW DRSSTC

first previous 4 5 6 7 next last

Move Thread

LAN_403

hvguy

Wed Apr 04 2007, 02:15AM

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

The latest challenge:

I was planning on running the coil this evening so I though Iâ€™d grab some more waveforms before taking everything off the bench. All my testing is being done at 170V so I had not thought about this before but it looks like I still have a problem. I turned the bridge on for about 10sec, took a scope shot, and then shut it off. Just out of curiosity I reached over and felt a TVS string; it was HOT! This did not make sense as the peak voltage measured across the string (4 1.5KE250 is series) is only 285V. And that was measured with a calibrated TEK P5200 HV differential probe so I am confident the waveform was accurate. I hooked up the current probe and sure enough there was a nearly 600A 30nS wide current pulse through the TVS string occurring seemingly randomly. The string was turning on for no apparent reason. I thought about it for awhile then replaced the unidirectional string with a bi-directional string of the same parts. Sure enough this fixed the problem. As it turns out the forward voltage drop on these parts, which is not specified in the datasheet, is the standard .7V. This means that any reverse voltage is going to turn the string on and short the IGBT. This, I believe, was the real cause of my original failure.

I have some 1.5KE200CA parts (the bi-directional ones) on order and will use five of them in a string for the next run. Good news is, after all this testing and redesign I have a much superior bridgeâ€¦ Hopefully I will break the 20â€™ mark next run. I was also able to test the bridge up to 2.4KA with no problems.

Steve Ward

Wed Apr 04 2007, 03:16AM

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

Wow, thats a pretty serious discovery, and i will need to keep that in mind if i have any problems with my TVS. I suppose with only 4 in series, the 2.8V drop would be similar to the free-wheeling diode drop at a few hundred amps, so it makes sense. I think im using 7 (120VBR) TVS in series on my setup, so i should have some good headroom there, or i could always throw in a 30V bi-directional, but i havent had any trouble with them (though my strings of 4 of the 200V types DID melt on occasion!).

hvguy

Wed Apr 04 2007, 03:30AM

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

Ya, crazy problem to trouble shoot. Prior to this I had never even considered the free-wheeling diode drop. The good news is the problem was simple to fix. I will keep that in mind about the 4 200V TVSs melting, maybe I will parallel a few strings.

LarsE

Wed Apr 04 2007, 11:37AM

Registered Member #153 Joined: Sun Feb 12 2006, 03:44PM
Location: Sweden
Posts: 31

And you could also series more lover voltage TVS to increase the current capability and spread the heat.

hvguy

Wed Apr 04 2007, 05:49PM

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

I could, but I got a great deal on 800 or so 1.5KE200CA so I think they will be in everything I build for the next few years

Marko

Wed Apr 04 2007, 06:14PM

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

Hi guys...
Don't flame me for this... but is there actually a danger from DC bus ''voltage spikes'' in the end?
Steve blamed them as well in order just to find a more cardinal error in his gate driver!

Is there a DRSSTC (of any size) wich actually had a true and confirmed problem as such?

AFAIK free wheeling diode practically conducts only on the end of cycle, and not the full tank current, but still if the drop gets close to TVS's forward drop and they start to conduct, TVS will get hot much faster and their drop will go down.. very interesting problem you've found!

Steve Ward

Wed Apr 04 2007, 07:39PM

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

Don't flame me for this... but is there actually a danger from DC bus ''voltage spikes'' in the end?
Steve blamed them as well in order just to find a more cardinal error in his gate driver!

Good point! I sometimes forget that i dont actually have hard evidence to suggest that voltage spikes ever killed my IGBTs (except perhaps my earlier DRSSTCs that used secondary current feedback where hard switching and junk feedback caused the IGBTs to switch full peak current). Every failure ive had as of recently has turned up some other explanation (with very good supporting evidence) rather than voltage spikes... Of course, other people still have unexplained failures, which may very well be voltage spikes (or anything else, i suppose).

BUT, there is evidence (scope measurements) at low power that voltage spikes are very real, and that if they remained the same factor of the nominal bus voltage (2X or whatever) that they would be a real threat to the IGBTs. Though, as ive maybe mentioned before, these voltage spikes *appear* to diminish as i increased the bus voltage, and at a rate suggesting that they arent tooo bad. I stress "appear" because its possible that my scope probe (rated for 400VDC) is non-linearly attenuating the 10mhz switching transient noise that im speaking about (so they may in fact not be dimishing).

I still think the possibility of the bus voltage *resonating* up to higher voltage is a very real issue, but this is also very easy to see at low powers, and *doesnt* diminish as input powers climb. The cause of this (for any who arent aware) is the inductance of the bridge structure resonating with any low impedance poly-film or ceramic decoupling capacitors. It seems all to easy to get this L and C ringing up to dangerously high levels particularly if the bus inductance is large. But this problem is rather easily solved with an appropriate bus structure (i go with laminated copper sheets now).

HV Enthusiast

Wed Apr 04 2007, 08:31PM

Registered Member #15 Joined: Thu Feb 02 2006, 01:11PM
Location:
Posts: 3068

I have a question. What exactly is a 20kW DRSSTC? Are you specifying that 20kW as input power, or actually power delivered to the primary?

You should probably clarify that. A 20kW input power DRSSTC isn't hard to build. A 20kW output power DRSSTC (at least assume that much power is actually getting switched into the primary) would be much more impressive.

...

Wed Apr 04 2007, 08:53PM

Registered Member #56 Joined: Thu Feb 09 2006, 05:02AM
Location: Southern Califorina, USA
Posts: 2445

Is there really that huge of a difference? I mean, there can't be more than a kw going into that heat sink, and the caps can't be disapating more than a few % of the input power before they melt...

hvguy

Wed Apr 04 2007, 10:07PM

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

Dan, 20KW is the actual input power measured at the mains. Assuming 90% total line-primary efficiency thatâ€™s 18KW of through-put; unless you can think of a reason why the efficiency of this system is any less than that of other ZCS full bridge circuits...

I will be testing an IGBTs tolerance for over voltage shortly. I believe this and excessive over-current are the biggest killers of IGBTs in TCs so it will be nice to have some proof.

first previous 4 5 6 7 next last

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