UCC21739-Q1: 2 level turn-off not working correctly

Andy Foggin

Part Number: UCC21739-Q1

Hi, we are using UCC21739-Q1 in an inverter driving an IGBT, and during double pulse testing we have noticed that the 2 level turn-off is not working correctly.

CH1=input to driver

CH2=Vge

CH3=Vce

CH4=IC

The first waveform is just before the OC event

The second waveform shows the OC event when the OC pin reaches 0.7V which is 7us into the waveform, the fault pin is also responding correctly, but the gate seems to get stuck high for over 2us. At 9us into the waveform it looks like the driver is doing it's 2 level turn-off all be it at the wrong voltage, but the duration seems to match T2LOFF as per the datasheet.

Any support is appreciated.

over 1 year ago

0 Nancy Hidalgo over 1 year ago

TI__Expert 3710 points

Hi,

Could you take another scope capture with the OC pin probed? This will help us determine the cause more accurately. Could you also measure the stray inductance in the circuit, as pictured below?

Thanks,

Nancy

0 Andy Foggin over 1 year ago in reply to Nancy Hidalgo

Prodigy 10 points

Hi Nancy, any chance you can send me your e-mail address, I have a great deal more information to share.

0 Andy Foggin over 1 year ago in reply to Andy Foggin

Prodigy 10 points

Hi Nancy,

Total stray inductance is 25nH based on measurements.

For OC pin see waveform below.

CH1=Vge

CH2=input to driver

CH3=OC pin

CH4=IC

0 Nancy Hidalgo over 1 year ago

TI__Expert 3710 points

Hi,

After further inspection, it looks like the device is behaving as it is supposed to. After the OC threshold is met for a period longer than the deglitch filter can handle, the device initiates the 2-level turn off. The gate voltage drops to approximately 9V (V2LOFF can be as high as 10V) and holds for up to 1 microsecond. Note, this 8.3-10V V2LOFF is higher that the threshold voltage of the switches, which have a typical Vth of 6.2V. The gate then discharges. The second plateau here is likely the Miller plateau that is typical for IGBT devices. In fact, the IGBT device may behave this way because it is not going into desaturation. Here is an annotated capture to help explain this.

This behavior may not be exactly what you need for your system, in which case I’d recommend taking a look at our other smart driver devices here: (+) [FAQ] UCC21710: What are the differences between UCC217xx and ISO5x5x? - Power management forum - Power management - TI E2E support forums

Best,

Nancy

0 Andy Foggin over 1 year ago in reply to Nancy Hidalgo

Prodigy 10 points

Hi Nancy,

Our expectation is that after TOCOFF the device would pull down the gate to 10V very quickly, it seems it takes a little over 1us to pull the gate to V2LOFF, can you explain this please? It seems the device is limiting the gate pull down current during this region.

Regards,

Andy

0 Andy Foggin over 1 year ago in reply to Andy Foggin

Prodigy 10 points

Hi Nancy,

Regarding the second plateau not sure this is due to the Miller effect, as the current is already zero, hence the IGBT is in the off state when this plateau occurs.

Regards,

Andy

0 Nancy Hidalgo over 1 year ago

TI__Expert 3710 points

Hi,

I see. For the two-level turn off the gate driver uses the soft turn off current (ITL3 which is typically 500-1200mA) which constrains how fast the gate voltage decreases. The initial ramp down is also pulled down with the soft turn off. Here is an example waveform from another device with the same behavior:

Best,

Nancy

0 Andy Foggin over 1 year ago in reply to Nancy Hidalgo

Prodigy 10 points

Hi Nancy,

Thank you for the waveform, it looks like this test was using a SIC MOSFET, do you have any test results using an IGBT? This would help us make a comparison to what we have on the bench.

Please advise,

Regards,

Andy

0 Nancy Hidalgo over 1 year ago

TI__Expert 3710 points

Hi Andy,

While the schematic looks like a MOSFET I believe that this was tested with an IGBT (hence the Vce measurement). Let me double check and get back to you.

Best,

Nancy

0 Andy Foggin over 1 year ago in reply to Nancy Hidalgo

Prodigy 10 points

Hi Nancy,

Looking at the lack of a Miller plateau on the gate waveform, we suspect this was tested with a SIC MOSFET. If you do have any testing with an IGBT, it would be great to see the waveform.

Regards,

Andy

+1 Nancy Hidalgo over 1 year ago

TI__Expert 3710 points

Hi Andy,

Here is a better waveform. Hope this helps!

Best,

Nancy