Tip / Sign in to post questions, reply, level up, and achieve exciting badges. Know more

Intelligent Power Modules (IPM) Forum Discussions

abdullah
Level 1
5 sign-ins First solution authored First reply posted
Level 1

We are using IKCM30F60GD for a 220V 200Hz 9A motor drive circuit with 8kHz sinusoidal PWM. We have the PCB ready and is driving the motor as desired.

However, we need to have protection for shorted motor windings, i.e. U-V or V-W or U-W shorted.

We have created a current sense circuit that creates a voltage well above 0.47V (limited to 3.1V) at I_TRIP pin in case of a high current(33mV/A), and this circuit rises I_TRIP voltage to 0.5V in 1.2usec. That trips the IPM successfully and shuts the output successfully. We have tested the circuit with down to 2 ohms instead of direct short, and it worked flawlessly.

However, when we go down near 1 ohm, or short, the IPM shuts down the output, however it dies. After it is dead, we have measured inbetween P,Nx,U,V,W and it seems like IGBTs are fine, we measure the same values as a non-faulty IPM. However, there is a diode reading of 1.7V on VDD to VSS with the multimeter(VSS on cathode lead).

And when we power the faulty IPM using only VDD and VSS pins with +15V, we have 200mA current drawn. And the heat on the package is coming from the point just in the upper side of pin 7.

Why is this happening?

Attached is the part of the package that heats up.

0 Likes
1 Solution
abdullah
Level 1
5 sign-ins First solution authored First reply posted
Level 1

Hello Pablo,

Sorry for the delayed reply. 

Yes, we have made the measurements and as you mentioned, absolute maximum rating for the leg currents is passed easily. However, the rating is noted as 60A for 1ms. Also there is a "short circuit withstand time" which is noted as 5us with V_DC below 400V. In our measurements, we are reaching about 300A peak and the short circuit conduction lasts  for about 2.1us. I don't feel like we are passing IGBT ratings 😔

The other measurements are within the limits. However, we have also measured VSS (pin 16) to Nx (pins 17,18,19). We have found out that US1M diodes we are using to surpass negative voltage on Nx pins are not much effective at all, due to their reverse recovery time and high forward voltage. So we have replaced them with SX34 40V 3A Schottky diodes and wow, the waveform changed a LOT!

Since then we have made about 100 shorting tests (with the cable, without the cable, with the motor, without the motor and combinations of them, min. 5 sec between tests). And we still don't have a single faulty IPM since 😁.

Nominally, chip sense resistors' parasitic inductance is somewhere near 2nH. Our wire sense resistor (L-KLS6-ASR-16-10-5J) is about 8nH. However, the VSS(pin 16) to GND path was also a little narrow and was on the leg current's path. So we have replaced it as follows, also. The voltage on this path was getting a little high due to high dI/dt from time to time and was destroying the gate drivers.

With my appreciation and best regards,

Abdullah

PCB.png

View solution in original post

0 Likes
9 Replies
Pablo_EG
Moderator
Moderator 250 sign-ins 250 replies posted 25 likes received
Moderator

Hello Abdullah,

Thank you for posting on Infineon Community.

We understand that your device experienced a breakdown while testing your custom winding short protection.

In order to understand your setup better, could you send us a schematic of your application?
It will help us point out any problem there might be.

Looking forward to your answer.

Best regards,
Pablo

Pablo_EG
Moderator
Moderator 250 sign-ins 250 replies posted 25 likes received
Moderator

Hello Abdullah,

Thank you for sharing the schematics.

Firstly, we would like to know what parameters were measured in the IGBTs to conclude that they are not damaged.

I am asking this due to the Id of the IGBTs being surpassed by your testing.
As there is no input filter in your design, when causing the winding short, the current can reach:

Id = Vrms*√3/Resistance = 220*√3/1 = 381 Apeak

This is extremely high, compared to the peak maximum current of the IGBTs, which is 60 A.

As there is no winding L considered in your test methodology, the current can freely increase up to the calculated peak value.
Even in a real winding short, there is some inductance in the system which attenuates the di/dt of the current.

In addition, when a current as high as 381 A is suddenly cut, we can expect to have a voltage surge in the DC line.
In your case, however, as there is no input inductor filter, this surge can affect all the way to the AC line.
This surge can affect the flyback converter and its output, which is connected to the VDDD of the IPM.
This could be the reason for the controller side to have broken down, as in this scenario VDDD could have passed the 20V max rating.

If possible, I would like you to introduce an input inductor filter, so that the current rise can be maintained low.
We would also like to have Voltage and current readings.

With the current setup, Id_max = 381 A, I am afraid that we cannot guarantee the integrity of the IPM, as the max Id rating is likely being passed.

Best regards,
Pablo

0 Likes
abdullah
Level 1
5 sign-ins First solution authored First reply posted
Level 1

Hello Pablo,

Unfortunately, adding an input filter to the design is not possible as we are really tight on space. However we did our tests on an isolation transformer for the sake of limiting current. This transformer's output dips after 8A and we do not think it will be possible for it to supply more than 15A.

However, there is also the output caps that can supply high currents, so we have tried adding a long wire (2m) to U and W instead of shorting them with 20cm of wire. There is no change on this, neither..

We have already burnt 13 IPMs..

For the IGBT health tests, we have measured the diode drop between pins P, U, V, W, NU, NV, NW, VB(U), VB(V), VB(W) after the failure. We have tried all the different combinations that can be made amongst these pins, for example P-U, U-P, NU-U, U-N, V-W, W-V etc.. The readings are exactly the same with a new non-used IPM.

We cannot access the gates of the IGBTs directly, so we have no clue on that. 

However, we were also told by Infineon that the malfunction area, the area that heats on the IPM package is gate driver.  We were advised that we should take precautions so that there is minimal inductance between VSS and Nx pins. 

So, for this, we have soldered 3 US1M diodes from VSS pin (pin 16) to Nx pins (pins 17, 18 and 19) with cathodes being on the Nx pins.
 
We also removed the GND connection on the left side of the PCB and made a single solid GND connection from the IPM cap (C28) to the GND legs of sense resistors, which is very close to surge cap (C25).
 
In addition to these, we have added a TVS(SMBJ15CA) in between VDD and VSS rails, onto the legs of the cap(C28) that is near the IPM.
 
However, our problem continues.. 😔
 
Could you please guide us on this as we have already bought a big stock of the IPM as we trust Infineon brand more than others..
0 Likes
Pablo_EG
Moderator
Moderator 250 sign-ins 250 replies posted 25 likes received
Moderator

Hello Abdullah,

I am sorry to hear that the problem persists and for the loss of the IPMs.

Could you please send the voltage and current measurements of the tests you have conducted?
Specifically: -Drain and source voltages of the arms being shorted,
                          -Voltages of all voltage source pins (such as VDDD),
                          -Gate output voltage,
                          -Current waveforms of legs being shorted.

This will help us greatly in debugging your setup.

Bets regards,
Pablo

0 Likes
abdullah
Level 1
5 sign-ins First solution authored First reply posted
Level 1

Hello Pablo, 

Thanks for the quick reply.. Could you elaborate a little more on your questions?

Let's say we will short U and W:

Drain and source voltages of the arms being shorted: You mean V_P-U (voltage between pins  23-22) and VW-N_W (voltage between pins 20-17)?

- Gate output voltage: You mean voltages on pins HIN_U(pin 7) and LIN_W(pin 12) with respect to VSS?

- Current waveforms of legs being shorted: You mean the current that flows from U to W (or vice versa) ?

0 Likes
Pablo_EG
Moderator
Moderator 250 sign-ins 250 replies posted 25 likes received
Moderator

Hello Abdullah,

Sure, let me elaborate.
All the data I have requested is oscilloscope data.
The waveforms should be acquired at the time of the failure (when windings are shorted).
If you do not feel comfortable repeating the 1Ω test, you could take the waveforms with the 2Ω test so that we have somewhere to start from.
We may find out that even in the 2Ω test the current/voltage exceeds the absolute maximum values given in the datasheet.

-Drain-source voltage: Voltage of P referenced to U/V/W.
 The voltage of U/V/W referenced to NU/NV/NW.

-Gate output voltage: My mistake, I forgot that this is an IPM and the gate output cannot be monitored.
  As the gate driver part seems to have faulted, the gate output monitoring would greatly help.
  Alternatively, having HIN(U)/HIN(V)/HIN(W) referenced to GND and
  LIN(U)/LIN(V)/LIN(W) referenced to GND would help.

-Current waveforms: Current waveform acquired using a current probe on at least one of the windings being shorted, e.g. cable between IPM U pin to Motor U winding.

-Voltage of voltage source pins: VB(U)/VB(V)/VB(W) referenced to VS(U)/VS(V)/VS(W)
  P referenced to GND.
  VDD reference to GND.

Please check on your side that none of the absolute maximum ratings given in the datasheet are surpassed.
After checking this, let us know about your findings.

Best regards,
Pablo

0 Likes
abdullah
Level 1
5 sign-ins First solution authored First reply posted
Level 1

Hello Pablo,

Sorry for the delayed reply. 

Yes, we have made the measurements and as you mentioned, absolute maximum rating for the leg currents is passed easily. However, the rating is noted as 60A for 1ms. Also there is a "short circuit withstand time" which is noted as 5us with V_DC below 400V. In our measurements, we are reaching about 300A peak and the short circuit conduction lasts  for about 2.1us. I don't feel like we are passing IGBT ratings 😔

The other measurements are within the limits. However, we have also measured VSS (pin 16) to Nx (pins 17,18,19). We have found out that US1M diodes we are using to surpass negative voltage on Nx pins are not much effective at all, due to their reverse recovery time and high forward voltage. So we have replaced them with SX34 40V 3A Schottky diodes and wow, the waveform changed a LOT!

Since then we have made about 100 shorting tests (with the cable, without the cable, with the motor, without the motor and combinations of them, min. 5 sec between tests). And we still don't have a single faulty IPM since 😁.

Nominally, chip sense resistors' parasitic inductance is somewhere near 2nH. Our wire sense resistor (L-KLS6-ASR-16-10-5J) is about 8nH. However, the VSS(pin 16) to GND path was also a little narrow and was on the leg current's path. So we have replaced it as follows, also. The voltage on this path was getting a little high due to high dI/dt from time to time and was destroying the gate drivers.

With my appreciation and best regards,

Abdullah

PCB.png

0 Likes
Pablo_EG
Moderator
Moderator 250 sign-ins 250 replies posted 25 likes received
Moderator

Hello Abdullah,

Thank you for your response and detailed explanation.

It is great news to hear that with the discoveries you made and their respective fixes, the winding short test does not fail anymore.

With your response, I understand that the issue is now fixed.

I hope your product, using Infineon IPM, is very successful in your application.
Best regards,

Pablo

abdullah
Level 1
5 sign-ins First solution authored First reply posted
Level 1

Yes, with your help, I guess it is now fixed, we will do field tests now.

I really appreciate your attention.

With my best regards,
Abdullah

0 Likes