Problem with GDT driver, voltage peaks, TVS-Diodes hot / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Problem with GDT driver, voltage peaks, TVS-Diodes hot

1 2 next

Move Thread

LAN_403

mikrohenri

Wed Aug 01 2012, 09:14PM

Registered Member #4467 Joined: Wed Apr 18 2012, 09:31AM
Location: Germany
Posts: 3

Hello!

I have been busy for some time to build induction heater. In the beginning there were more smaller structures with an output power of 50-200W, then I came back in the area of ca.1kW and now after many tests and experiments, I would now like to build a 2kW model.

The basic structure consists of an IGBT half-bridge and a parallel resonant circuit of HF-Cap and the induction coil.
As a control I use a GDT, the ratio of 1:1, bifilar wound on a ferrite toroid (10Wdg.) and was a square wave signal of + / - 20V out there (ca.40kHz). At the gate of the IGBT's are 22V TVS diode (1.5 KE22CA, bidirectional). The IGBTs have following electrical characteristics: 1200V, 570A, 30nF input capacitance.

Now the problem is that the TVS diodes become very hot. After a few test's I've found that the TVS diodes are only hot when the gate of the IGBT's are connected. After further experiments, in which I have the input capacitance of the IGBT's simulated by foil capacitors, I can see on the oscilloscope, that this will build a resonant circuit (see figure) and caused a surge that will make the TVS diodes so hot. Despite further measurements I can still find no solution to this problem. Since GDT's so often used to control, are also used by big IGBTs, I wonder whether something fundamental I'm doing wrong? Have you possibly experience in this regard?

Thanks for any help!

greeting
Dennis

HV Enthusiast

Wed Aug 01 2012, 09:47PM

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

The TVS are hot because they are conducting all the time. I'm guessing you may need more gate resistance to help knock down some of the parasitic oscillations and also your gate transformer may need some modification as well.

Ben Solon

Thu Aug 02 2012, 12:54AM

Registered Member #3900 Joined: Thu May 19 2011, 08:28PM
Location:
Posts: 600

Yep, evr is right, the secondary inductance is resonating with the hate capacitance. Try dampening it with a gate resistor.

Plus that's the worst gate drive signal I've ever seen minus some saturation of ferrites. Notice the gate voltage rating of the fets/igbts? Typicaly they're rated for 20v continuous maybe 30v peak-peak. You have 40 volts so you're probably killing the gate layer and possibly latching the device. And notice how the voltage doesn't just swing back into the linear region, it drops down bellow 0... You have --40v pk-pk-- on that gate where there should be a nice square wave. Do a bit of research on gate drive.

Wolfram

Thu Aug 02 2012, 06:24AM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Do you have the parallel resonant tank circuit connected directly to the IGBT bridge output? If so, that's bad and you need a series inductor.

mikrohenri

Thu Aug 02 2012, 04:56PM

Registered Member #4467 Joined: Wed Apr 18 2012, 09:31AM
Location: Germany
Posts: 3

Hello!

Thank you for your replies!

@EasternVoltageResearch:
I've already experimented with the gate resistors, but there was the problem that the resistors (each 10W) have also become very hot. But the signal was so much better. Its no solution with these resistors, because the control-mosfets are also hot. No wonder because the resistors make a lot of heat...

@ben123324:
Without the voltage-peaks I have exactly 20V. The gates are charged with +20 V and discharged with -20V. I haven't got 40V!?? Do You think the gate signal will become better with an other ferrite-core?

@Wolfram:
No, the parallel resonant tank circuit is not directly connected to the IGBT-output. There is a matching inductor and a DC-blocking capacitor is series. I think this is ok.

Does anyone have ideas, how I can remove / filter the voltage peaks?

Thank you!

greeting
Dennis

HV Enthusiast

Thu Aug 02 2012, 06:26PM

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

Yes, the resistors WILL become hot especially if you're running them at CW.

Next question. Is this going to be for a pulsed SSTC such as a DRSSTC? If so, and you're testing them at CW (100% duty), then during normal use (pulsed operation at low duty cycles), the components will not get very hot.

Any ideas how to remove the voltage peaks? Easy. You already found the answer. You need to add the resistor.

Actually, two things you can do.

1. Twist the wires as best you can between the driver and the transformer and the IGBT. This will keep parasitic inductance down and limit the magnitude of those inductive spikes you're seeing. If its a board layout, then make sure the RTN for the gate drive is on a plane underneath the traces going from gate drive to transformer.

2. Put those resistors back in. If they are getting to hot, you need to make further evaluations to see how to reduce the heat. (are you operating at the correct frequency, with the correct capacitive load, correct duty cycle, etc...)

Finally, the total heat dissipation in the gate drive circuit is going to be divided up primarily between the gate resistor and the impedance of the gate driver. If the gate resistor is 10 ohms and gate impedance 1 ohm, then the gate resistor will see approx. 10x the amount of heat dissipation.

Dr. ISOTOP

Thu Aug 02 2012, 09:28PM

Registered Member #2919 Joined: Fri Jun 11 2010, 06:30PM
Location: Cambridge, MA
Posts: 652

I know from experience that driving bricks CW at 40 KHz in induction heater duty means you need a *very* large resistor. I was running +/-15V drive on a full-bridge at 65 KHz in mine, and even the 250W resistor I used at one point got blazing hot (mind you, this was with some inefficiencies and no heatsinking).
I eventually took the gate resistor out of mine, and somehow it still worked.

Steve Conner

Fri Aug 03 2012, 09:28AM

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

I think the problem is that your IGBTs are too big and slow for the job. You can get 2kW from TO-247 devices with far less gate capacitance.

mikrohenri

Fri Aug 03 2012, 10:54PM

Registered Member #4467 Joined: Wed Apr 18 2012, 09:31AM
Location: Germany
Posts: 3

Thanks for the replies!

@EasternVoltageResearch:
The driver is for an induction heater, this should mainly run with CW.
The wires are already twisted good.

@Dr. ISOTOP:
A gate-resistor with 250w is a bit much. I dont't think it's the best solution.

@Steve Conner:
You're right, with TO-247 devices, it is also possible to get 2kw. With less input capacity (~5nF) i have also these peaks.

I will try to change the ferrite material, maybe this is the problem!?

greeting
Dennis

Steve Conner

Sat Aug 04 2012, 06:19PM

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

Yes, you'll still get the peaks with smaller IGBTs, but they'll contain less energy, so your resistors and TVS won't get so hot.

I've never seen large IGBTs driven directly by GDTs, outside of amateur DRSSTC work.

1 2 next

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