Protecting ZVS driver MOSFETS / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Protecting ZVS driver MOSFETS

1 2 next

Move Thread

LAN_403

Zenador

Thu Sept 03 2009, 12:47PM

Registered Member #1733 Joined: Thu Oct 02 2008, 03:17PM
Location: Hamilton, ON, Canada
Posts: 100

Hey all,

Been a long time since I posted here (or worked on any of my projects...)

I'm building a FBT driven TC. I have 2 working ZVS drivers and one that I killed last night. After getting everything into the project box, I attached my secondary and primary. The coupling was a "little" too tight, and there was some beautiful flash over to the primary. The SG fired and the sucker worked fine till the flash over.

My Mosfets didn't explode, burst into flame, or perform any other wonderful tricks at their time of death. I put my scope on them, and they appear to be latched open. I have more IRFP250's, so replacing them is not a problem. I just don't want to keep replacing them.

Input is 26V AC through a 50V 8A bridge rectifier. I've followed the original Mazilli schematic. One of my working ZVS drivers is on my Jacobs Ladder running at 40VDC. I had it on and off in the backyard last night for over an hour with no issues, so I know my build is good, parts are of good quality, etc.

Is there any "feasible and realistic" method of protecting the fets from a backwash of overvoltage?

Cheers, Zen

Dr. Dark Current

Thu Sept 03 2009, 12:54PM

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

Hi,
if I get this right, your FETs died after the flashover? Well then, wind the primary better so it does not flash over.

Otherwise: your problems might be
-overheating - use bigger heatsink (I don't think this is the problem)
-overcurrent
-if none of the above, the circuit might have started oscillating at high frequency through the diode feedback (this is usually the cause when everything runs cool and the circuit suddenly explodes).
Fix - use 1k pullup resistors instead of 470R.

Zenador

Thu Sept 03 2009, 03:26PM

Registered Member #1733 Joined: Thu Oct 02 2008, 03:17PM
Location: Hamilton, ON, Canada
Posts: 100

Thanks Dr. K.

I just replaced the fets, and ran the primary without a secondary. It ran for about 45 seconds, then failed. Under the same setup last night (getting the SG right) it ran for several minutes with no issue. The fets and the bridge rectifier are on the same heatsink. The heatsink came from a RS/6000 power supply, and had 6 TO-220 fets on it. It was warm to the touch when it failed, but not hot.

My frequency counter measured 220kHz on the ZVS. It only fails when run through the MMC tank and SG. If I just pull arcs fron the FBT, nothing heats, nothing fails. The MMC is 6 2kV 47nF WIMA Polypropoleyne caps (2 strings os 3) , SG is 3-4mm.

At $3 per fet, these failures are going to get expensive...

Z

raff

Fri Sept 04 2009, 12:37AM

Registered Member #2315 Joined: Tue Aug 25 2009, 02:35AM
Location: Leyte, PH
Posts: 161

Hi Z

try using 680R 5 watts and even UF4003 works fine for ZVS.. what flyback make are you using?

since I started using the zvs circuit, I have *never* destroyed any, my 1st circuit was eevn using a lowly IRF630A but it didnt die or anything.. now Im using IRFP250 for that circuit ans is working fine (even with 40V@8A).. Ive been using it to drive my SGTC :)

is it built on a dedicated circuit board?

Arcstarter

Fri Sept 04 2009, 01:46AM

Registered Member #1225 Joined: Sat Jan 12 2008, 01:24AM
Location: Beaumont, Texas, USA
Posts: 2253

How about the isolation from the MOSFETs to the heatsink? Perhaps you are having an insulation failure? This could short the drains, and short the supply, making you think they were dead. Check the fets to make sure they are dead always! You might not expect what you see

. I have done this, i forgot the insulation and i had the power supply short (it was capable of around 30 amps with no heat, quite a large current through the fets when they shorted) through the fets/secondary. Then i pulled them out, tested them, and they worked. I even tested them in a sstc (where they died, it was inevitable :P).

I have never had a 'cool' failure of a MOSFET in my ZVS. I was using 1 watt zeners, a big 1kohm Dale 100 watt resistor, tapped in the middle (which heated!) for 500 ohms for each pullup... Always a cold death, and the only reason i know it is dead short is because the supply transformer will hum, and the lights will dim (even more, that is :D). Never a boom, despite the filter capacitor being shorted.

I have had huge arcs jump to the fet drains, and they would just avalanche and be fine, and continue to operate. Not only that, the tank cap would sort of kill most of the HV, i'd think. HV spikes should not kill a mosfet, unless they avalanche for a long time and heat the die too fast to dissipate it. That would be a cold death. That, and overcurrent.

Could gate ringing do it, due to a small pullup resistor? As in, it rings enough to fry the gate. I am not sure if the ZVS would be capable, however i know you can ring the gate to death on other things with powerful drivers. Or, perhaps the gate was ringing at too high frequency, too high for the MOSFET to switch, making it linear, and frying itself? That does not sound likely.

Zenador

Fri Sept 04 2009, 04:39PM

Registered Member #1733 Joined: Thu Oct 02 2008, 03:17PM
Location: Hamilton, ON, Canada
Posts: 100

The insulation is good. It's 1mm thick, and covers the whole mount section, large enough for 3 fets per side. The fets are mounted with a plate and screw. Plate over the top of 2 fets, screw between them and into the heatsink. This would allow (I believe) more heat transfer away from the fet directly and into the heatsink.

My zenners are 20V 65mA. Pullups are 470r, 2 watt. Drain to opposing Gate are 1N4009. Source to Gate resistors are 10k 1/2 Watt.

After taking everythign apart and testing each component (other than the fet) everything checks out. I've tried the DMM MOSFET test, but I can't get it to do squat, even on a known good fet.

After the failures, if I leave the input xfmr on, it starts heating after about 25 seconds. I'll take some photos of my working set to see the build and heatsink. I'll rebuild the circuit with my suspected failed fets and post the scope trace.

To Raff : this is a point to point build, no PCB. I could design and etch one, but I don't see the point. My jacobs ladder has an AC FBT, and the TC setup is on a DC FBT. Both types of FBT will kill my setup if I add the TC Tank circuit.

I don't have any bigger, suitably rated resistors or diodes on hand, I'll have to pick them up tomorrow.

Cheers guys...

raff

Sat Sept 05 2009, 02:05AM

Registered Member #2315 Joined: Tue Aug 25 2009, 02:35AM
Location: Leyte, PH
Posts: 161

you forgot to mention MICA(or any other) insulation between MOSFET and heatsink.. how is your setup? both MOSFETS on single heatsink? or one for each?

about your diode.. do I read correctly 1N4009? is that fast enough? I thought the 1N400x series was max only to 1N4007(1kv)

Myke

Sat Sept 05 2009, 05:38AM

Registered Member #540 Joined: Mon Feb 19 2007, 07:49PM
Location: MIT
Posts: 969

Even thought the 1n4009 is fast enough, it doesn't have a peak reverse voltage rating anywhere close to what you need. The 1n4009 has a PIV of 35V while you need something >400V. I would suggest using the UF4004 or UF4007 because those work well for the ZVS flyback driver.

You may also want to look for something less than 20V for the zeners. 15V is fine but 12V is what's suggested.

Your MOSFETs (IRFP250) shouldn't be dieing at 26V... Also, is you power supply filtered because filtering it may help with the performance of it.

Is the voltage between the drain and source of the MOSFETs going above 200V? You may want to try using a 200V TVS to clamp the voltage so it doesn't exceed the D-S voltage rating.

Proud Mary

Sat Sept 05 2009, 10:35AM

Registered Member #543 Joined: Tue Feb 20 2007, 04:26PM
Location: UK
Posts: 4992

If you put fast low resistance Schottky diodes of suitable amperage in each of the source leads, this will prevent destruction of the device by negative swing.

You can also prevent over-voltage with an avalanche diode in parallel with the device,(or MOVs which are cheaper) and/or put a snubber network in each of the drain circuits.

Frankly though, I would advise not only implementing some of these measures but moving to a much high voltage device.

I had exactly the same problem as yours when following the Mazilli circuit diagram as is. At 12V it worked very well, but at 24V it would blow out one or other of the MOSFETs within seconds. The circuit as given is, I think, too simple, like a skeleton circuit in a textbook, and needs several levels of protection added to it if it is to run 24/7.

Zenador

Mon Sept 07 2009, 04:10PM

Registered Member #1733 Joined: Thu Oct 02 2008, 03:17PM
Location: Hamilton, ON, Canada
Posts: 100

Hey guys...

Some corrections - The cross gate/drain diodes are 1N4007, not 4009. The Zenners are 18V, not 20V.

This is a photo of my stable Jacobs Ladder setup. Have not blown a single component. This is running on 40V in, no smoothing cap.
3896958786 F0210ceab0

These are the Meter Readings from a new build. Zenners are 12V, gate/drain diodes are 600V 1A fast switching (the UF4007's were not available at my store. I have added a smoothing cap to this board, and brought the pullup resistors to 1k ohm 2W. This is 25V in. All readings are under load - pulling 1 inch arcs. The small meter is measuring source/gate voltage @ 4.1V. The large blue meter is actual MOSFET temp, after 2 minutes of run time - 36 C. The yellow meter is operating frequency in kHz - 66.4 drawing arcs (drops to 24.1 if I break the arc.)
3896180485 63c82b1fd0

Scope @ 100V 5ms/div (Channel 2 inverted, Both showing - chop)
3896180523 C67ecde266

Scope @ 100V 1ms/div (Channel 2 inverted, Both showing - chop)
3896180555 A7c80ef253

I have some non-hobby work to do, but I will add the TC tank circuit with the SG and take more readings later today. If someone has any other suggestions before I possibly blow 2 more fets, I'd love to hear them.

***Edit/Update***

I believe I have figured out the cause of the failures. The insulating/isolating mylar did not extent to the edge of the heatsink. The left Mosfet drain tab was next to the mounting hole used for the bridge rectifier, and the right fet was about 2mm from the edge of mylar. Under full load last night, the 2 drain tabs were arcing to each other through the heatsink. The arc was very small and faint, I had the lights dimmed, so it was a fluke I saw it. I have since repositioned the fets and added the TC tank circuit, small secondary and primary, and ran it successfully for several minutes.

Thanks for the support and ideas...

Cheers guys,

Z

1 2 next

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