Half Bridge Induction Cooker

[resonant]

Good Day Everyone

I'm making a half bridge resonant inverter, I'm using the AC line voltage which is rectified and filtered with a 2.2uF cap. The problem is, whenever I push the AC voltage above 100V (I'm using a variac to control the voltage) it seems as if the IGBTs cross conduct causing a short and destroys both the switches. I watch both the switches voltage and current and it seems fine. I'm using a l6384E half bridge gate driver and 600V 30A igbts (STGW20NC60VD). My switching frequency is 20kHz at the moment but I'm going to vary it up to around 60 or 70kHz to control the temperature. My output power is designed to be 1kW. For my gate resistors I used 56 ohms in parallel with a diode and a 12 ohm resistor. The the gate driver can source 400mA and sink 650mA. My bootstrap capacitor is 1uF. Please can anyone help with a solution.

Regards

 

[mtwieg]

That gate driver has dead time settings, and you're also implementing dead time one the gate side, so dead time shouldn't be a problem, unless the switching times on the igbts are very slow.

Before failing, do the IGBTs get hot at all? Do they fail only when driving a resonant load? Does the failure only happen when driving the load on or off resonance?

Also a 2.2uF cap is very small for a AC line filter cap.

 

[resonant]

Thanks for the quick reply mtwieg, the igbts do get a bit hot but I have them on a massive heatsink. I set the deadtime to about 4us incase the igbts are slow. The cap is just to filter the transients so I dont get huge spikes over my igbts.

---------- Post added at 12:23 ---------- Previous post was at 12:21 ----------

Didn't test it without a resonant load. Also I have 47nF snubber caps across my IGBTs

 

[mtwieg]

Thanks for the quick reply mtwieg, the igbts do get a bit hot but I have them on a massive heatsink. I set the deadtime to about 4us incase the igbts are slow. The cap is just to filter the transients so I dont get huge spikes over my igbts.

You should probably have a bigger cap. Keep in mind that the DC filter caps have to endure at least the same rms current as the resonant load.

Didn't test it without a resonant load. Also I have 47nF snubber caps across my IGBTs

You mean just a capacitor in parallel with each IGBT? That could very well be your issue. You should post a detailed schematic.

 

[resonant]

I've added the circuit diagram in pdf. The resonant capacitors are 680nF, input capacitor is 2.2uF, snubber capacitor is 47nF, the inductance is 22uH and the resistance is around 2 ohms (according to my calculations). when i set the switching frequency to 28kHz at around 50V the output current is more or less sinusoidal. A small amount heat is transferred to the pot so I know it's working its just when I crank up the voltage it shorts.

 

[FvM]

You mean just a capacitor in parallel with each IGBT? That could very well be your issue.

I agree, that this is a rather bad idea.

 

[resonant]

I have them there to suppress transient voltages. Could they be the source of my problem?

 

[mtwieg]

I have them there to suppress transient voltages. Could they be the source of my problem?

Those caps will definitely increase the dissipation in the IGBTs. Every time either IGBT turns on, it will dissipate all the energy stored in those "snubber" caps. At 50V, that's 58.75uJ and at 28KHz that's 1.645W per IGBT. At 100VDC that will quadruple, and so on.

There are a few ways to properly snub a half bridge. First is to put good snubber caps on the DC rail, right next to the IGBT terminals. Second is to fit the half bridge output with a good RC snubber. The sizing of the snubber components must be done carefully in order to get good performance and not cause significant heating.

However, when driving a resonant load, snubbers should hardly be necessary, since you should be switching the IGBTs at zero current. I've worked on solid state tesla coils (just a dual resonant load) and we switch currents over 1KA easily at ~30KHz, and no snubbers are necessary as long as you are driving on resonance. Doing so greatly reduces the stress on the IGBTs, allowing us to push the current far beyond its ratings.

 

[FvM]

The excessive capacitive by the "snubbers" load causes at least rather high peak currents. At worst case, they bring up oscillations with magnitudes exceeding the transistor voltage ratings.

 

[resonant]

Ok Thanks. Think it will be better to just remove them from my circuit. Also, from my switch current and voltage waveforms how will I be able to see if I'm getting ZVS/ZCS? And what size cap should I use as my filter?

 

[mtwieg]

ZVS switching is much harder to achieve, but isn't necessary at such low frequencies. To check for ZCS, probe the load current and bridge output voltage, and check that the output voltage switches just as the current waveform crosses zero amps.

For the DC filter cap, it should be much larger than your resonant caps (I'd say at least a factor of ten) so that they don't actually detune the resonant load. And they have to store enough energy to prevent excessive drooping of the DC bus in between AC line cycles when loaded. So that's kind of just your judgement call.

 

[necati]

**broken link removed**
**broken link removed**
Induction Cooktops?Home Appliances/Industrial Systems?TOSHIBA Semiconductor & Storage Products Company
**broken link removed**

 

[resonant]

Thanks. So I'll try removing the snubber caps tomorrow and check if it's stable. Anything else I should check incase removing those snubber caps don't make any difference? Should I increase my deadtime between the gate signals just to be safe?