IGBT gate drive

[Styx]

I am working on an active gate control for a IGBT : 1000A device, 1200V.
This IGBT is using infineon die and its IGBT switches extreamly fast

From day1 I have been tasked with building an active-gatecontrol

Basically for a link of 350V and switching 600A I am getting 300V overshoot (ie 650V at Vce). Although the device is rated at 1200V this massive overshoot needs to be slugged.

Usual method is increase gate resistor but we need it to switch as fast as possible (turn-off in 400ns for 350v,600A), plus the device already has swithcing losses of 1Kw we cannot have that much more

I have sucessfully measuring the dv/dt of Vce and am in the process of modifying my output stage of the 1st iteration of the gateboard.

I need to use a P-type MOSFET as a linear amplifier. It needs to be able to source 11A peak, switching freq abt 10Mhz (main switching freq is 10Khz but the active side needs to gatevoltage shape very fast).

I have found the MOSFET needed (power handling, current rating...) I am after any advice on using a MOSFETin its linear regon.

Thanks

 

[bmcculla]

I'm a little unclear about what the MOSFET linear Amplifier is doing. Is it driving the base of the power transistor or is it clamping the output voltage to reduce overshoot?

Most of the linear amps I've studied operate in the saturation region and not the linear region. My first though is to use an opamp to close the loop around the transistor and drive it into the linear region when required. You migh have bandwitdth trouble though as the op amp will have to be able to drive a fair amount of current to pull the charge off the FET Gate. Adding extra smaller transistors between the op amp and the FET to increace th current drive might solve this. You also could set up a feedback scheme with discrete transistors.

Just some thoughts
Brent

 

[Styx]

I'm a little unclear about what the MOSFET linear Amplifier is doing. Is it driving the base of the power transistor or is it clamping the output voltage to reduce overshoot?

I am using a standard P-type N-type output leg to drive the IGBT gate region. They are bing driven into saturation at turn-on (so +15V is provided to the gate of the IGBT) and turn-off (so -15V is provided to the gate of the IGBT).

However during turn-off instead of just turning the P-type off and the n-type on (so +15 -> -15V) the dv/dt of the IGBT Vce is measured and the P-type is again turned-on BUT in the linear region so not the full +15V is supplied to the IGBT. Due to this the IGBT is not turn-on fully but is in the "linear" region, this way giving the current a change to decay before the device turns off.

Most of the linear amps I've studied operate in the saturation region and not the linear region. My first though is to use an opamp to close the loop around the transistor and drive it into the linear region when required.

That is what I am doing

You migh have bandwitdth trouble though as the op amp will have to be able to drive a fair amount of current to pull the charge off the FET Gate. Adding extra smaller transistors between the op amp and the FET to increace th current drive might solve this. You also could set up a feedback scheme with discrete transistors.

Just some thoughts
Brent

Yep yep exactly what I am doing, although I am using a lm7171 (very fast and 100mA output drive) not enough to ensure that the P-type Gate-MOS switches fast enough (gate capacitance chage/discharge).

I have all the power side sorted, I was just curious is there anything that need to be strapped to teh P-type MOSFET (ie cap's around it) to aid it since it is will be working in the linear region for 2us out of 50us (50us it is HARD-ONm 2us is active turn-off)

cheers

 

[Russlk]

I don't see what the advantage is of this circuit over a gate resistor?

 

[Styx]

DEL
double post

 

[Styx]

because the IGBT that I have even with 100Ohm gate resistor the overshoot is still huge but now I have slowed down turn-off and my switching losses have gone through the roof.

It have been released my Infineon that this particular die WILL need some form of "gate-pulse shaping during turn-off"

Thus this ciruit is needed.

 

[IRQ57]

IGBT Drive

Hello Styx,

There is a simple way to obtain different speeds to turn on and off an IGBT. Use a diode with a low value resistor for the charging circuit of the gate, and in an antiparallel configuration, another diode with a higher resistor value for the discharge path.

In this way you can obtain a very high turn on speed, limiting the losses during the pass thru the unsaturated region, and a moderated turn off speed, decreasing the over shoot voltage to a reasonable level.

Regards

 

[Styx]

thakyou for your reply. I have this already. - I am not using diode I am putting the resistors in the Drain/Source of the P-type/N-type so:

P-type
Turn-on resistor
-------------------------------------------> IGBT
Turn-off resistor
N-type.

This way no diodes and I have means to limit shoot-through current on the gate drive stage

I have 3x 8.2R 2512 1W SMD resistors in parallel for turn-on (3W at 2.7Ohm). I also have 3x 42R 2512 SMD resistors in parallel for turn-off (3W at 14Ohm).

Even with 14Ohms for turn off I am getting 200V of overshoot (still too fast turning off). And even for a double pulse test the gate resistors are getting very hot.


From simple capacitor charging:
Turn-on Res=2.7Ohm
Gate capacitance = 122nF (no current flowing ie not swithced)

it takes 1us for the current to decay to a few 1/10 of an amp. This puts power dissipation at 1.9W in the Turn-on gate resistor. Equally for turn-off at 2.7Ohm 1.9Watts.

I practice this device has a huge millar capacitance region, so the gate voltage hovers at 10V too long for higher gate resistor to be used.

I have done a simulation of just using turn-on/turn off reisistors and 100Ohm turn-off resistor will control the overshoot. But:
1) far too much power dissipated in gate resistor
2) IGBT turns off far too slowly.

 

[IRQ57]

IGBTDriver

Hi,

You are using -15V for turn off , it is not usual to apply so high voltage, most circuits that I have seen use -5V max.. This way the power dissipated in the turn off MOSFET resistor is 9 times less than with -15.

Anyway, there is a document that may be of your interest:

https://www.electro-tech-online.com/custompdfs/2004/05/IGBT20overshoot20protection.pdf

I hope this will be usefull for you.

regards

 

[Styx]

Re: IGBTDriver

lol got that 1, but thanks anyway. Here is a ref so you lot can follow what I mean

https://www.electro-tech-online.com/custompdfs/2004/05/epe2003-active_gate_control.pdf

anyway good point abt the -5V since that would be enough negative supply to ensure that and miller turn-on does not happen.

However,
1) that requires another rail to be generated.
2) in generating the -5V that will reduce the power loss in the turn-off MOSFET (and more importantly the turn-off resistor) it will also drastically slow down the turn-off of the IGBT (less forcing volts).
3) the only reason to us teh -5V would be to use a high gate resistor but I cannot have a high gate resistor I need to switch this IGBT as fast as possible due to the switching losses (600A continous at 350V - with overdrive at 800A)

thanks again

Anyone have any tips on using a P-type MOSFET as a linear amplifier