# Can I control IGBT linearly with gate voltage?

I want to make a dummy load using IGBT. Is it possible to control the Ice (collector to emitter current) using the gate voltage? It would be really great if it is possible to explain the answer. The following are the parameters.

Voltage and current need to be controlled: 85V 50A ( I want to control the current from 0A to 50A as required by the application).

Gate voltage: Anything which IGBT requires. Following is a sample circuit only.

• Have you chosen or ear-marked an appropriate IGBT yet? You should post a general circuit idea showing what the load is and how it is connected. You should also explain what other DC power rails can be made available and what level of control accuracy you need. Commented Sep 26, 2023 at 13:33
• Is there any specific reason of selecting an IGBT for the application? Have you checked the datasheet(s) of any of the IGBTs you plan to use? If you have, have you noticed any graphs for SOA and $V_{GE}$-vs-$I_C$? Commented Sep 26, 2023 at 13:53
• IGBTs aren't usually designed for linear operation. There are a few--I think Toshiba makes one--but they're few and far between. Commented Sep 26, 2023 at 13:58
• @Andyaka Yes, IKZA75N120CH7 this one. Commented Sep 26, 2023 at 14:21
• That device cannot do what you want. See the reasons in my answer that I have added and compare the device you stated with the types of graph added to my answer. They haven't even shown a safe operating area graph in that device's data sheet because they don't want you using this in linear applications. Also look at the typical transfer characteristic graph in your device = it will fail in milliseconds. Commented Sep 26, 2023 at 16:42

Voltage and current need to be controlled: 85V 50A ( I want to control the current from 0A to 50A as required by the application).

This doesn't side like a problem that is conveniently solved by an IGBT (or MOSFET for that matter). You wish to control voltage and current to a load it seems but, when the voltage is (say) at the halfway point (42.5 volts) and the current into the load will be 25 amps (assuming a linear load resistance), the power dissipated by the device will be 1062.5 watts and that's a lot of heat to get rid of.

So, unless you have a cunning plan that involves water cooling the transistor I would suggest you walk down the path towards using a high-power buck regulator. Even then, it's not likely to be better than 95% power efficient so, you might expect to have to deal with a heat power dissipation of 50 watts. However, this can be solved by decent heatsinking and possibly a fan.

@Andyaka. This circuit is just a dummy load... It won't supply power to any other circuit/device.

If it's a dummy load, your power dissipation is going to be 85 volts × 50 amps = 4250 watts. That's 4 times greater than what I stated earlier. If you insist on going down the IGBT route here's the type of problem you could easily face that can leave you with dead devices and a lot of head scratching. Consider the IXXN200N60C3H1 from IXYS just as a quick example of something that you might come across. It has decent enough ratings: -

However, the devil is in the detail such as this graph: -

I've added the coloured lines and text. As you can see, at 85 volts and 50 amps, you cannot operate the device for any longer than about 3 or 4 ms. In other words, not much of a dummy load for continuous operation.

But it does get worse than that regarding the control of the gate voltage and collector current; if your 85 volt supply is capable of delivering hundreds of amps, using an IGBT like this (intended for switching applications) but in a linear application then you run foul of thermal runaway. Look at this graph: -

At 25° C you nominally need to drive the gate with about 8.6 volts (with respect to emitter) to achieve 50 amps collector current. Then, as the device inevitably rises in temperature (very, very quickly on the die of the device) the current can rise to 70 amps in a few milli-seconds and, unless your feedback circuit is very, very quick at responding, you'll end up with a dead device in maybe less than 20 milli-seconds. The problem here is that the body of the IGBT won't even feel warm to the touch and, no amount of heatsinking would have saved the day.

It's up to you what route you go but, I bet that the least-painful route is not linear control.

• It's a load, so the heat dissipation is a given. Even if it's rated for 50A and 85V, it doesn't need to be rated for both at the same time. The device may additionally have a wattage rating.
– Drew
Commented Sep 26, 2023 at 14:23
• 1000W isn't too hard to dissipate. Just use like 4 transistors instead of 1, and bolt them onto a huge heatsink with forced air.
– Drew
Commented Sep 26, 2023 at 14:24
• Actually, I want to make a dummy load. Input is 80V and 50A, and I want to dissipate the current I wish. i.e some time 10A , sometime 15A etc.. I've added a sample circuit for clarity. Commented Sep 26, 2023 at 14:24
• @Nithin, so you do sometimes need to draw 50A at 80V? That's a 4kW load. That can be done, as long as you're aware that this will be a large device. You may want to use water cooling with a large bucket of water as the heatsink.
– Drew
Commented Sep 26, 2023 at 14:27
• @Nithin no, MOSFETs in linear operation are no better than IGBTs. Unfortunately (despite what drew says) the problem doesn't necessarily lie with heatsinking but with finding devices that are rated for linear operation. IXYS make a few and, you need to use several in parallel each driven with its own control loop to ensure reliability. Please read this answer and the links for the same problem with MOSFETs: electronics.stackexchange.com/questions/657868/… Commented Sep 27, 2023 at 7:52