high frequency SSTC half-bridge MOSFET fails at highside. / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

high frequency SSTC half-bridge MOSFET fails at highside.

Move Thread

LAN_403

RayeR

Tue Aug 14 2007, 08:29AM

Registered Member #333 Joined: Mon Mar 20 2006, 06:02PM
Location: Czech Republic
Posts: 45

Hi,
I made this half-bridge for my micro TC operating at 2,5MHz, I redesigned PCB of older version but still have problems with MOSFET failures at higher supply voltages. See schematic below, there are following changes:
drivers replaced by microchip TC4450/4551, gate zeners are 16V/1.3W, gate rezistors 3ohms, MOSFETs are IRF630 and then better STP14NF10. It works nice up to 24VDC at bridge. When I go to 35VDC MOSFETs died.
So I tried to put safe rezistor 3.9ohm in series with supply and replaced MOSFETs with better STP14NF10 (they are 100V/15A/0.13R/only 500pF Ciss). I turned it on but it failed again at 35V within 1/2second. Then only highside MOSFET died (short at all 3 terminals). I watched Vgs on lowside before it has some ringing but still enough to open MOSFETs, output voltage at bridge middle node looks pretty square below tune, when it's tuned its distorted to some combination of triangle and square (sorry i didn't take a scope photo).

My question is how could highside MOSFET died? Average current shouldn't exceed 10A because of limitation resistor (It took about 1,2A at 24VDC from PSU at tune). MOSFET was only slight warm so TPD couldn't be exceed. Gate voltage is clamped by zeners (they are OK). Higher voltage at drain should go away through external free-wheel diodes (also OK).

Aghrr this little crap is quite silicon consuming. Maybe IU should finally try Class-E driver but I want to know whats wrong.

[Edit: 400 Pixels]

ragnar

Tue Aug 14 2007, 08:55AM

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

Rayer, you really want higher voltage MOSFETs, not lower. IIRC, the IRF6XX series are 200V, right? You've replaced them with 100V MOSFETs. You should really be trying something higher.

An IRFP450 has a 500V 14A rating, but this will be especially hard to drive at 2.5MHz. These are your limitations, really.

Experimentonomen

Tue Aug 14 2007, 09:33AM

Registered Member #941 Joined: Sun Aug 05 2007, 10:09AM
Location: in a swedish junk pile
Posts: 497

You should check your gate waveforms, its possible that thins high frequency cannot turn the fet on fully = high switching losses = the fet overheats internally and dies before any heat have ben dissipated to the heatsink.

Try the circuit on a lower Fres coil and see if the high side fet still blows up, if it doesent, then your problem is whats described above.

Welcome to the forum btw and enjoy your stay :)

RayeR

Tue Aug 14 2007, 03:01PM

Registered Member #333 Joined: Mon Mar 20 2006, 06:02PM
Location: Czech Republic
Posts: 45

>BlackPlasma
I want to operate this driver up to say 60V or less so I don't really need higher voltage MOSFETs because in bridge structure the voltage over MOSFET is clamped and shouldn't exceed supply rail voltage. IRFP450 and similar have much greater Ciss which I cannot drive. Gate driver ICs itself probably can do but tell me somebody how to make better GDT for this freq.? I use HQ ferrite core with trifilar winding but not fully satisfied with it - better idea?

>HFsstc-freak
Hm, maybe... Do you think it's possible even with serial 3R9 resistor between supply and halfbridge?
At lowside the rising edge time was less than 100ns.

Experimentonomen

Tue Aug 14 2007, 04:13PM

Registered Member #941 Joined: Sun Aug 05 2007, 10:09AM
Location: in a swedish junk pile
Posts: 497

IÂ´m not sure, check if you get any sharp voltage spikes over the high side fet.

hvguy

Wed Aug 15 2007, 09:54AM

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

Even though your gate capacitance is low I would stick with very low (~1ohm) value gate resistors or not even use them. At that frequency you need all the speed you can get. Your failure is almost certainly due to a bad gate waveform although I would not entirely rule out voltage spikes. If a bad gate waveform was the cause it could be due to the resistors or noise caused by Coss/Crss. If it was voltage spikes you should be able to use ~200v fets with better luck. Also do you have a bypass cap across the rails after the series 3.9 ohm resistor? The series resistor is most likely useless as the primary inductance in a system like this will limit the current to very reasonable levels. If the bridge was to shoot through it would probably fail anyway due to rapid die heating or excessive current sourced from the rail bypass capacitor.

Or maybe itâ€™s just an SSTC

RayeR

Wed Aug 15 2007, 01:32PM

Registered Member #333 Joined: Mon Mar 20 2006, 06:02PM
Location: Czech Republic
Posts: 45

hvguy wrote ...

Even though your gate capacitance is low I would stick with very low (~1ohm) value gate resistors or not even use them.

Yeah, I don't like gate resistors but in this case the ringing was simply too high. If I would feed more power to drivers I will blow up the zeners, If I replace zeners with large ones I'll have 2-times higher load capacitance and ri9nging will be greater so they will heat more and more in a loop :)

Yestersday I tried to redesign GDT but no success. I had tried 5 varoius toroid cores and one cylinder ferrite choke core from monitor. I used twisted enameled wire 0.4mm from 15 to 4 turns. But that damned ringing was there all the time when loaded with test capacity. I have no idea how to make it better. BTW in my older SSTCs working on low frequency (340kHz) I have perfect gate waveform with GDT on same core and same wire. But at 2,4MHz it is much harder.

hvguy wrote ...

At that frequency you need all the speed you can get. Your failure is almost certainly due to a bad gate waveform although I would not entirely rule out voltage spikes. If a bad gate waveform was the cause it could be due to the resistors or noise caused by Coss/Crss. If it was voltage spikes you should be able to use ~200v fets with better luck.

Yes I know what you are saying - high dV/dt induces spikes from drain to gate so low gate circuit resistance is needed. Well but harder to do... I also blowed IRF630 rated at 200V it doesn't matter.

hvguy wrote ...

Also do you have a bypass cap across the rails after the series 3.9 ohm resistor? The series resistor is most likely useless as the primary inductance in a system like this will limit the current to very reasonable levels. If the bridge was to shoot through it would probably fail anyway due to rapid die heating or excessive current sourced from the rail bypass capacitor.

I have 220nF bypass cap on PCB close to MOSFETs, resistor was placed between supply and PCB. I hoped that this 220nF doesn't have enough energy to destroy MOSFETs. When looking at supply rail there's no much noise/spikes just few volts, surely not exceed 100V. Maybe I could try put small transil across D-S of every MOSFET...

hvguy wrote ...

Or maybe itâ€™s just an SSTC

Yes, but if I would build it and it will work perfectly without any fails it would be boring ;)
Hm, but this low power FET failures are nothing interesting (silent, no effects) like MOSFETs in bigger SSTC shattered over entire room with loud bang :)

hvguy

Sun Aug 19 2007, 09:27AM

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

I notice your schematic has 10V zeners; is this the value you are using or did you chnage to the 16v ones? If your using 10v, why?The gate drivers should really be running at least 12v, preferably 15v. The zeners at the gates should be 18V or 20V as the higher gate voltage can make a big difference if there is any noise in the system. This has nothing to do with your problem but I see your using the Schottky bypass diode/fast diode combo. If youâ€™re using any modern, ultra fast FET itâ€™s internal diode should be fast enough, even at 2.5MHz, I would check the data sheet for what ever FET you happen to be using.

The nature of your failures combine with the GDT and zener issues, not to mention the very high frequency, really point toward the gate drive as the cause of your problem. If you can scope the gates and post a picture it would help. Also, it sounds like you may have too much leakage inductance between you gate drive and FETs. How long are the leads on your GDT?

Well, if the failures arenâ€™t explosive enough you could always try turning the variac up all the wayâ€¦real quickâ€¦with out a fuseâ€¦after the FETs fail.

RayeR

Tue Aug 21 2007, 08:21AM

Registered Member #333 Joined: Mon Mar 20 2006, 06:02PM
Location: Czech Republic
Posts: 45

Sorry this schema is little bit outdated, in 1st post I mentioned the changes. Zeners are 16V/1.3W they seems to have quite low capacity. Reverse recovery of STP14NF10 should be about 80ns I dont remeber but quite small. In IRF630 I used before it's 2-times longer. OK I will post some scopes. GDT leads are about 2-3cm long. Voltage on primary side looks pretty well under load but on secondaries it's ringed.

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