ZCS resonant driver idea for (HV) ferrite power transformers / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

ZCS resonant driver idea for (HV) ferrite power transformers

previous 1 2 3 4 next

Move Thread

LAN_403

Dr. Dark Current

Thu Nov 01 2007, 05:35PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko wrote ...

Jan, have you tried measuring your primary current?

If you did, you may have been surprised. You may have noticed that if you keep the output of your transformer shorted, it doesn't generate much heat, and your DC power consumption is also very low at that point, comparable to open transformer. And you probably also noticed that your power consumption is biggest of all when arc is most stretched out just before it extinguishes.

I watched it on a scope and it was indeed the highest with the sec. shorted. It went down a bit as the arc was drawn. However I didn't measure the absolute value.

Just some thoughts of my driver that has exploded: I was using 40Acont. rated transistors, and lets simplify it, a transformer with 25 primary and 2500 secondary turns. (actually they were 2x2 transformers in series-parallel). From the laws of transformers, secondary current is primary current multiplied by turns ratio. So if the turns ratio was 1:100, and there was 40A average current through the primary (which I highly doubt because the 8A fuse in series with the filter cap didn't blow), there would be nearly half an amp on the output! It really didn't look like that (from how the arc looked like, I guesstimate ~100-200mA at most) . So why did my IGBTs explode?

I am really no expert when it comes to inductances and transformers, so please excuse me if what I am talking is completely off.

I have problems understanding the current-fed inverter. From what I have found, it consists of a current source, a choke in series and your standard bridge.
Now, where do you get the current source? And what is exactly the role of the choke?

1193938302 152 FT33515 Appnotes Overview Current Fed Inverter B

Marko

Thu Nov 01 2007, 06:58PM

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

Where exactly did you put the 8A fuse? Schematic?

If it was in path of reactive current (bridge output/primary) then I don't know why your IGBT's blew. Maybe steve does.

In any case your IGBT's should have cooked before they died. It can happen fast and without notice if you are distracted

Make sure they, and diodes, are heatsinked heavily.

I have problems understanding the current-fed inverter. From what I have found, it consists of a current source, a choke in series and your standard bridge.
Now, where do you get the current source? And what is exactly the role of the choke?

Here we are talking about resonant current fed inverter like royer, but converted to halfbridge.

Hypothetically it's transistors would see no more than pi/2 DC voltage and make the circuit plausible to run from 230V mains with 600V switches. It would probably be most efficient topology known for >100kHz.

I don't know how to build it nor anyone who did. I found no information about it anywhere apart from one CFL driver circuit and most of this comes out of my head.

THere should be a reason why I didn't see it anywhere.

Choke is what acts as a current source; because of it's large inductance, it keeps current relatively constant during cycle.
Choke needs to be split for each transistor, unlike your hard-switched schematic.

Dr. Dark Current

Thu Nov 01 2007, 08:46PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko wrote ...

Where exactly did you put the 8A fuse? Schematic?

If it was in path of reactive current (bridge output/primary) then I don't know why your IGBT's blew. Maybe steve does.

In any case your IGBT's should have cooked before they died. It can happen fast and without notice if you are distracted

Make sure they, and diodes, are heatsinked heavily.

The fuse was between the + of the 330uF filter cap and the top IGBT collector.

Heatsinks never went above ~50C or so.

The reasons I can think of why the IGBTs died:
-I reused one igbt from a dead halfbridge
-too fast gate voltage fall and rise times (because of overkill gate drive) or the 1.2uS deadtime. (I highly doubt this was the cause)
-the RC network (3R3 in series with 10n) that I connected to the output of the bridge to decrease dv/dt on turning-off transistors.

I'm not buying another 2 IGBTs just to see if they blow again

, this is too expensive fun for me.

I still fail to understand how do you exactly want to convert the Royer circuit to half-bridge.

How do you think a push-pull style drive method (hard switched) would compare to a half bridge?

Marko

Fri Nov 02 2007, 11:55AM

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

If you didn't use decoupling caps after fuse I really can't comment on why IGBT's died, if it wasn't overheating. You could have tried one in series with primary, but doesn't matter now. One of steves may tell you more.

If I was rebuilding the circuit I would definitely go for better gate drive and overcurrent protection circuit. Just a simple SCR based circuit saved me a halfbridge lots of times, but I think you should go for a comparator there.

I don't see the purpose of your RC network apart from wasting power, but still it shouldn't be responsible for the failure.

Dr. Dark Current

Fri Nov 02 2007, 02:45PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Of course I used decoupling caps, but they were "before" the fuse.

Marko wrote ...

If I was rebuilding the circuit I would definitely go for better gate drive and overcurrent protection circuit. Just a simple SCR based circuit saved me a halfbridge lots of times, but I think you should go for a comparator there.

So are you saying that too fast (rise/fall) gate drive actually isn't good?

Marko wrote ...

I don't see the purpose of your RC network apart from wasting power, but still it shouldn't be responsible for the failure.

The idea of this was to take away some heat from turn-off loss from the transistors.

I'm posting a schematic of the exact circuit I used; I'll be no more asking on the topic of my exploded driver unless someone has a clue why it failed.

What about the push-pull driver? You would need to use 2x higher voltage FETs which would have lets say 2x higher on resistance, but they would only need to switch 1/2 the current so overall dissipation should go to half.

Marko

Fri Nov 02 2007, 03:00PM

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

So are you saying that too fast (rise/fall) gate drive actually isn't good?

No, just that your level-shifted solid state driver isn't closely as bulletproof as a good GDT. Simpler circuit with no additional silicon on bridge side is less likely to fail and you really aren't driving something at 5MHz not to be able to use GDT.

You'll easily achieve rise and fall times much smaller than IGBT delays anyway.

GDT will not do something unpredictable like pulling one of transistors permanently on and causing shotthrough.

Your best chance is to kill most possible problems in one shot.

Snubber is also usually put across the load, although I think it's fairly safe to exclude it.

I wonder if someone can help you more...

Marko

Steve Conner

Fri Nov 02 2007, 08:09PM

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

JMartis' circuit is the one I was thinking of. AFAIK, you can't make a current-fed inverter as a halfbridge, it has to be a fullbridge.

The current source is usually implemented as a buck converter with current feedback, so the driver IC adjusts the duty cycle to keep the current constant on a slower timescale. The choke keeps it constant cycle-to-cycle.

That's roughly all I know about current-fed inverters. I haven't really a clue what is causing your IGBTs to explode either. :(

Your original idea would have worked with a voltage-fed inverter too. If you think about it, this is how DRSSTCs work, they are a transformer (air cored, though) with a capacitor in series with the primary, and feedback to make them operate at resonance. The only problem is that the resonant frequency of a ferrite transformer might change a lot depending on whether it's unloaded or loaded by an arc, but I think Marko covered this already.

Dr. Dark Current

Sat Nov 03 2007, 12:23PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Marko, could you please give me any details on your SCR overcurrent protection? Thanks

uzzors2k

Sat Nov 03 2007, 02:25PM

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

This is my multi-purpose inverter, which I've used for flyback driving and induction heating so far. Low freq SSTCs would also work. The over-current shutdown is designed for 12A peak. I'm sure you can figure out how to adjust it, if not I have a formula on my project PC which I can share.

The current transformer steps the current down by 35 so the voltage over the 12 ohm resistor is R*(I/35). Subtract the diode voltage drop and you get the voltage at the comparator input. The 15k and 10k resistors set the triggering voltage. An extra op-amp and bi-LED will give you the status of the inverter.

1194099939 95 FT33515 Et Multipurpose Inverter

Dr. Dark Current

Sat Nov 03 2007, 02:56PM

Registered Member #152 Joined: Sun Feb 12 2006, 03:36PM
Location: Czech Rep.
Posts: 3384

Uzzors, the problem of your over current detection is that if one fet fails, the other one dumps the filter caps to it, ultimately destructing it. I'm going to make an OCD that rather samples current on one of the power rails. This way if one FET fails and the other one starts turning on, the OCD will (hopefully) switch it off before it explodes.

previous 1 2 3 4 next

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