1

We have a 10kW inverter which is used to control a PMSM inverter and need some help to design a better inrush current limiter during startup.
The general approach would be to let the DC bus charge safely and slowly via a series resistor, and then bypass the resistor using a relay or a contactor.
In our application we are not allowed to use any mechanical parts like relay or contactors, so a MOSFET does the job of a contactor in this case.
Here is a rough circuit of the inrush controller.
..

The above setup works fine, but sometimes the MOSFET gets damaged. There is no heating or any physical damage, the drain and the source of the MOSFET are always shorted (We confirmed this by physically taking the damaged MOSFET out of circuit and checking its continuity).
Most probably is is some SOA region failure but don't understand why, as all the parameters are within its limits.
The gate of the MOSFET is ramped up using an off the shelf Inrush controller (LTM9100). We tried the following two cases.

  1. We Have not connected the series resistor (R1) and completely rely on the inrush controllers ability to ramp the gate slowly. We observed that in this case a few times the MOSFET fails.

  2. We then connect the series resistor R1, But even in this case we have observed the MOSFET failure, though the failure rate has reduced, but not completely gone. In this condition the potential at the drain and source are the same and the MOSFET should not fail, but it still does.

One more point, The MOSFET used is a switching MOSFET, and not a linear MOSFET. I understand that if the gate is ramped slowly, the MOSFET is more or less working in the linear region. But the reason we have opted for a switching MOSFET is that there are no linear MOSFETs that can withstand the high current requirement that we need.
We tried multiple MOSFETs but they all were damaged in the same way (STY145N65M5, MSC025SMA120B4).

Anyway to determine a possible root cause?
Is there a better way to limit the Inrush current in our case?

brhans
  • 14,723
  • 3
  • 35
  • 51
AK47
  • 95
  • 7

1 Answers1

2

The general approach would be to let the DC bus charge safely and slowly via a series resistor, and then bypass the resistor using a relay or a contactor. In our application we are not allowed to use any mechanical parts like relay or contactors, so a MOSFET does the job of a contactor in this case.

That's a sensible approach but then you don't follow your own sensible approach and, attempt to use the MOSFET like a current source. The MOSFET (particularly a switching MOSFET) will suffer from thermal runaway and it will be "game-end" in less than about 10 ms and you are left scratching your head with a cold MOSFET.

Anyway to determine a possible root cause?

MOSFET thermal runaway because you are using it incorrectly and inappropriately AND, you are not following your own guidance on how to use it (like a contactor).

Is there a better way to limit the Inrush current in our case?

Follow your own guidance and design it how you set out in your opening paragraph.

Most probably is is some SOA region failure but don't understand why as all the parameters are within its limits.

and

We have a 10KW inverter

Well, the worst case scenario for the MOSFET (as you have used it) is when half the supply voltage is across it (400 volts ÷ 2 = 200 volts). At this interim point, we can presume that half the full load current is meant to be flowing (12.5 amps). Here's where that point is on the SOA graph for the MSC025SMA120B4: -

enter image description here

Clearly you are in a totally dangerous area for that MOSFET = it would likely be dead within less than a millisecond. Here are the links I left you in comments on a previous question you raised: -

Andy aka
  • 456,226
  • 28
  • 367
  • 807
  • i Hi Andy, in my previous thread we did not have the series resistor,but now even after placing the resistor the MOSFET fails, why should it fail in this case?as the potential on the source and drain would be same.Is thermal run away still the reason even in this case? – AK47 Oct 07 '23 at 13:06
  • Are you talking about R1? If you are then the potential on the source and drain will not be the same <-- I mean how can it be until the inrush current is completed? Thermal runaway IS THE reason. – Andy aka Oct 07 '23 at 13:55
  • Yes R1,I am a little confused.The moment we power up, the DC bus starts charging via R1.We measure the DC bus constantly and only when it reaches close to 400v, we then only turn on the MOSFET to bypass the resistor R1.In this case won't the potential of the source and drain be the same?The moment the DC bus is close to 400v before we turn on the MOSFET, the inrush operation is already complete, but in this case we still see the MOSFET failure in some cases – AK47 Oct 07 '23 at 14:17
  • @AK47 how is this related to your question? Your circuit shows that you ramp the gate voltage up slowly (your words: The gate of the MOSFET is ramped up) and, the rest of your question is about that. My answer explains the folly of this method. Your proposed method is the right way to do things but, without details of how you activate the gate and the full circuit (including details of load current and input supply voltage variations) this cannot be answered. If you want to propose this new circuit raise a brand-new question and don't try to move the goalposts of your current question. – Andy aka Oct 07 '23 at 14:23