• TI thinks resolved

TPS92515HV: chip damage when hot-plug

Part Number: TPS92515HV

Hello,

I'm developing 2 different boards with TPS92515HV.
First board has 12V 6W LED, 48V as Input voltage
Second board has 30V 26W LED, 48V as Input voltage

Sometimes happens that TPS92515HV blows-up (it's phisically broken) when board is connected to powered-on lab power supply (hot plug) with 30inch cables,.

It happens also using evaluation TPS92515HVEVM at 48V (replacing TPS92515 with HV one).

I read this:
https://e2e.ti.com/blogs_/b/fullycharged/archive/2015/05/20/what-is-that-giant-tantalum-cap-on-the-input-of-the-evm

I'm worry about this problem because boards have some precautions:
   * 56V fast transient voltage suppressor on input line (SM15T56AY)
   * Limited input inrush current using MOS on GND line (MOS is driven by RC network). It give Vin similar to Figure 36 of datasheet.
but it happens anyway: sometimes TPS92515HV (and few times also MOS transistor) blows-up when I connect the board to 48V.

Do you think that it should be caused by voltage spike of power supply?
Do you have some suggestion to give me to make boards more strong?

Thank you a lot!

  • Hello,

    An input transient with a hot plug is generally the issue, but it could be both positive and/or negative overshoots, it all depends on parasitics. A 30 inch line between the supply and board will introduce a lot of parasitics, but layout could also be a factor. It may be as simple as increasing your input capacitance, but it could also have something to do with your RC input limiter. Is it possible to share the schematic at a minimum? The BOM and layout would really help too as well as any waveforms you might be able to capture. A TVS on the input can help, but depending on other factors may not clamp sufficiently at the IC itself or may not protect against negative transients.

    Any extra info you can give will help but I have also asked the systems engineer if they have any input and will let you know.

    Regards,

    Clint

  • In reply to Clinton Jensen:

    Thank you for answer Clint,

    I probe Vin only few times today but I can see that voltage supplied by my lab power supply reaches more than 60V... (D1 is not so reliable)
    I hope that with "consumer" power supply this doesn't  happen.

    I could share power stage, it's here:
    https://unsee.cc/nizageru/

    Little description:
    NMOS T1 switches on slowly all board, it's 3A, 60Vds, 20Vgs
    R3, R4, R5, R6, C5, C6 drives Gate of T1
    R1, R2, D2 helps to discarge C6 fast when board is powered off

    There is someting wrong on schematic?

    Next week I'm going to probe and test better power stage, I hope to find solution to make circuit more strong

    Regards

  • In reply to TecnoLAB:

    Hello,

    I'm not able to access a cloud due to security concerns chrome itself is giving me, is there any way to attach a picture or a pdf? I would be happy to look over it then. Do you by chance have waveforms of VIN and also PWM when this happens?

    Thanks,

    Clint

  • In reply to Clinton Jensen:

    Hello Clint,

    Next week I'm going to see waveforms of Vin and PWM, today I make only few test.

    PWM is not driven: it's pulled up to VCC pin with 12kohm and has 100nF to GND.

    Regards

  • In reply to TecnoLAB:

    Hello,

    Two things I notice:

    The first is that the TVS where it is located will not help with overshoot at the board. I would measure the input at the IC to check for overshoot. But if you are applying power in a controlled manner this may not be the issue.

    The second is using an RC on the PWM pin. This input is meant for a fast TTL type signal and as such it doesn't have hysteresis. So if the slew rate on it is too slow it can actually toggle very fast and damage the IC. This would be easy to test, just removed the 100nF capacitor to ground. Ideally if PWM isn't used you connect it directly to VCC.

    Regards,

    Clint