Dual High power IGBT gate driver

[mcray]

Hi,
I need to drive 2 parallel inverter (12 IGBTs 1200V 1000A), I found a lot of industriels IGBT gate drivers but I want to build my own one. I tried to use bipolars and mosfets transistors (push pull) in order to provide a current output driver Ipeak=30A but I didn't succed.
Does any one have a schema or suggestion that could help me to build a power high driver?
Thanks

 

[moffy]

I guess you know that you need seperate floating 12v supplies for the high side IGBTs. I would use just mosfets, because of the peak current reqirements and speed, but then you have an extra high side FET (assuming they're both n channel) which needs 12v? You don't want punch through so you'll have to make sure they're both not on at the same time.i.e. some anti cross conduction logic/timing. Then you still need isolation for the high side.

I would do my own low side driver and purchase a commercial high side driver if it existed.

 

[tcmtech]

In industrial applications where groups of very large Mosfets or IGBT's need to be driven its most common to use an individual driver for each switching device.

Basically if you can afford 1200 volt 1000 amp switching devices then you can also afford to buy the correct driver IC's for them as well.

 

[mcray]

thanks moffy,
the problem is that I still can't have a Ipeak=30A and I have a cross conduction. How can I amplify my current and solve thecrossing conduction problem?

 

[mcray]

thanks tcmtech,
thats what i'm trying to do,an individual driver for each switching device but I can't provide 30A.
do you have any suggestion?

 

[tcmtech]

What are the devices you are driving that needed that much gate current capacity even when driven at their upper switching speed limits?

I run 600 amp 1200 volt IGBT's off of common 2 amp rated driver IC's without problems so now I have to ask the more obvious question of what are they switching and at what frequency meaning what is the device they are in and what does it do that requires that large of switching devices in the first place?

 

[mcray]

I'm driving a 1200V 1000A IGBT's at 2 or 3KHz, I need a power peak current in order to drive more than 2 IGBTs simultaneously because I may drive 2 or 3 parallel inverters.
Do you have a schema of your driver?

 

[tcmtech]

These are the most common gate driver IC's I use and they have no problems running a 600 amp IGBT well past 3 KHZ in normal operation (provided the IGBT is rated to run that fast of course.)

600 volt View attachment IR2110 - IR2113 HALF BRIDGE DRIVER.pdf
1200 volt View attachment IR2114 HALF BRIDGE DRIVER.pdf

 

[moffy]

To amplify the current:
1. You'll probably need a better floating 12v supply for the high side, with some good surge caps.
2. You buffer the drive signal with higher current MOSFETS. Say 30v-50v/30A devices.

To avoid cross conduction:
1. You don't want cross conduction in the driver FETs and you DON'T want cross conduction in the IGBTs. The normal way is to introduce a delay for the switch on device that ensures enough time for the device switching off, to be fully off. Since turning on is from 0 to +12v you just need to delay that edge(you still need to cleanly switch this device,no long switchon), while not delaying the +12v to 0v edge. If you're driving 1000A worth of 1200v IGBTs, you should know how to do that.

 

[ronsimpson]

So you have 12 IGBTs with their gates tied together.
You could have 12 small gate drive ICs, drive each gate separately.

 

[mcray]

I solved the crossing conduction problem in the driver using the attached schema. I'm using LTspice for simulation, I use a capacitor to simulate the igbt gate, without the capacitor (gate) I have a high current in the resistance but when I add the capacitor, the current drop I dont undersatnd why? What do you think?

 

[moffy]

The circuit isn't correct. You should swap the n channel FET to the -12v and the p channel to the +12v. You need different drive signals for each gate. I don't think you have enough of an understanding to be attempting something so ambitious. Such a project would give very experienced engineers pause.

 

[tcmtech]

I still don't follow this assumption that you need a 30 amp gate drive circuit for a basic 100 amp switching system.

Given the specs here for a 600 amp 1200 volt IGBT block I just cant see the reasoning or logic for it. At a 2 ohm gate resistor value the turn on and turn off times are still under 700 Nanoseconds which is still around 500 times faster than your 3 KHz switching frequency.

What IGBT's are you using anyway?

View attachment 600A 1200V IGBT BLOCK.pdf

 

[mcray]

Yes Moffy,
I don't have enough of an understanding I'm still a biginner in power electronics, could you please explain me why I need different drive signals for each gate.
thanks

 

[mcray]

I understand what you mean Tcmtech, here is my IGBT.
thanks

 

[mcray]

IGBT that i'use

 

[tcmtech]

That IGBT has an internal gate resistance of .97 ohms and a peak drive voltage of +- 20 volts which in any practical application I can not see any rational justification to need a gate drive system that has a 30 amp capacity especially for a low 3 KHz switching frequency.
Even if ran with a +- 20 volt gate drive voltage the IGBT itself would never get much into a 10 amp absolute peak gate current level being it has a gate capacitance of only 120 nanofarads at most and a internal resistance of .97 ohms on top of that.

To even come close to a 10 amp peak gate current you would have to have an incredibly short gate rise time duration that that still would have to be several hundred times shorter than what the IGBT's inherent rise and fall times are.

What we need a more accurate description of what exactly are you building that needs or justifies this level of power handling along with a schematic of it as well. Reasons being this all seems suspiciously grossly over designed unless you are building a custom VFD unit that runs in the megawatt+ power levels which I highly doubt.

 

[moffy]

"I don't have enough of an understanding I'm still a biginner in power electronics, could you please explain me why I need different drive signals for each gate.
thanks "

Please refer to my previous post, I explain it there.

 

[mcray]

I meant why should I swap the n channel FET to the -12v and the p channel to the +12v while it works well even if I swap the n channel FET to the +12v and the p channel to the -12v? I'm just trying to understand, you're right in all the drivers they swap the n channel FET to the -12v and the p channel to the +12v I don't know why.

 

[moffy]

The n channel goes to the -ve because:
1. You need a positive gate voltage, which is already available.
2. The source is fixed, so you don't need a floating drive to track the output.

Same but reverse goes for the p channel device.
Also for high RDSon FETs and low voltages, the same gate signal can drive both, as their RDSon limits the cross conduction currents to safe levels. But once the RDSon becomes low, it is not safe to do that so you need seperate drive signals.