LMG3411R070: UVLO limits not found in datasheet

Hi

So seeing -10.5V on VNEG after a while and with a whining coil and a FET that is getting warm ( 40degrees) without switching anything is obviously a cause for concern.

Hi Martin,

The Vneg pin voltage is internally regulated by the internal buck boost.

What is the inductor you are using for the buck-boost converter?

Thank you,

Kyle Wolf

Hi Kyle

To be clear; This solution has worked but i have had a few products that have failed in the field.

The inductor used is BRC2518T220K as suggested from the datasheet.. Only one of the shown digital isolators are mounted. A SIP7 isolated unregulated 24 to 12 converter is used to drive each of the 4 legs of the H-bridge

It seems like the buckbooster has failed due to how i run the bridge. 

// Martin

Hey Martin,

A few more questions regarding the scenario around the devices that failed.

• Which inductor in the system is whining?
• Is the failure occurring on the high side or low side device?
• What topology are the devices being used in?

Thank you,

Kyle Wolf

Hi

The whining comes from the BRC2518T220K. On the board in front of me, it is the BR (Bottom Right) leg of the H-bridge, so that is low side. But i have also seen it on the high side.

Topology: The H-bridge is connected to a transformer with 40:60 turns. The output is connected to a lossy serial LC oscillator (the cap is intentionally lossy) where one the transformer outputs is connected to protective earth. This is an extremely simplified schematic.

The system is then driven as a PSFB starting a a very low duty cycle 

It there anyway that the buck boost can be broken in this way?

Best regards

Martin

Hi Martin,

Could you try to supply a regulated 12v supply to the VDD pin on the low side FET that is showing the issue and see if the whining coil or heat increase is still observed. Once using the regulated 12v supply could you also check the voltage on the VNEG pin.

Best Regards,

Kyle Wolf

Hi Kyle and thanks for taking the time to look into this

So I connected 12V from my lab supply and it draws 50 mA. Bare in mind that this also supplies one side of a ADuM2286CRIZ from the 5V LDO.

The VNEG pin is still at -10.5V. I have also attached some scope images. First VNEG

And then two images of the switchnode.

Best regards

Martin

Hi

I can answer some of these questions without testing as i have already done them on a previous board.

The scope pictures in my previous post are with the LMG3411 on but not switching. As /FAULT is active i am not sending pulses to the LMG3411.

I will need to do some modifications to allow my system to switch while /fault is asserted. I can do them on Monday.

This is the third board i have encountered this problem on a board. The last two times however, we have not time to look into it deeply and just replaced the LMG3411 directly. After that, the board worked again. To be clear, the LMG3411 has been working and then after a couple of hours, of switching this problem has occurred. I have seen it on both high and low side LMG3411 in my H-bridge. I realize that the problem might be the use case i am applying, but the way that the device appears to break with only VNEG being out of spec seems a bit strange.

Best regards
Martin

Hello,

I would first like to understand more of the waveforms you sent, it appears that the devices are switching according to the switch node voltage. Maybe I am interpreting this incorrectly, please elaborate if you could.

Based on your additional information, and the fact that this does not appear to be an unorthodox use case, I am lead to believe that part of the circuit is running at near peak values, or some transient event is causing the device to fail after a couple hours.

If the VDD pin were to experience a transient event, it could easily damage the VNEG pin since the VDD voltage supplies the circuitry used to generate VNEG. Could you share more information about where the "12V_HVDC_TL" pin is coming from?

Thanks,

Zach

Hi Zach

The device is not switching in the images scope images in previous posts. I have just applied 12VDC from a regulated bench supply and it reported that it sourced 50mA. I don't think that you are interpreting the scope images wrongly. They show VNEG and the switchnode while the IN signal is held low.

I could share design files if you had the time to investigate this further.

12V_HVDC_TL comes from isolated supplies as follows. One for each leg of the H-bridge

The HVDC to PGND rail is 360VDC so there is no its not near the peak values. I myself is consider a transient event the culprit, but it is still strange that device should break in this way. I am currently hardswitching the current and i am looking into if i instead can change into softswitching.

// Martin

Hi Zach

I have sent you the material.

Best regards

Martin