This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TPS22810: Part sometimes fail on power-up

Expert 1570 points
Part Number: TPS22810

I designed a battery-powered LED driver and USB power supply, and this part was used to disconnect the load from the batteries when they get discharged to prevent over-discharge.  Most of the time the part works fine, but 20-40% of the new boards that are assembled have this part fail immediately on power up.  Another ~10% fail sometime after first power-up, but again when powered up.  

The original design is attached.  The EN pin is used to enable the output when the input voltage rises above a certain threshold, similar to the example in Fig 18.  Thinking that a slower rise time would help, I added a 10nF cap on the CT pin (to ground) to give a 2ms rise time.  I noticed on a scope however that the rise time didn't change, and then I noticed in the datasheet that the Vin needs to be stable before EN is asserted in order for this rise time function to work.  I increased the cap C27 to 1uF and this caused a larger delay before EN goes high, so then I would see a ~2ms rise time.  Unfortunately the chip continued to fail.

The total load on the output rail is about 100uF.  The input (at +9V_SW) is typically 9V from batteries, but can be as high as 12V from a wall adapter when the batteries are charging.  

The datasheet states: "The device has a thermal protection feature. Due to this device protects itself against thermal damage due to over-temperature and over-current conditions. Safe Operating Area (SoA) requirements are thus inherently met without any special design consideration by the board designer."  But this doesn't seem to be the case for me.

Any thoughts or suggestions?

Thanks

  • Hi AQ,

    This is definitely strange behavior, especially with the delay on the ON pin and 2ms rise time. It doesn't look like you are exceeding any voltage maximums either. A few questions:

    1. When you say the device fails on startup, how does it fail? Does the switch fail to turn on, or does it try to turn on and stop? 
    2. How much current is the load on +9V_OUT pulling?
    3. Do you have reason to believe the device is damaged before turn on? What I mean is, do you think the action of inserting the battery may be the cause of the damage?

    Thanks,

    Alek Kaknevicius

  • Thanks for your reply Alex.  When the failures occur, they do so immediately or within a few seconds.  Sometimes there is a pop and smoke, and we have even seen flames.  Once failed, I see a short across Vout to ground.  To confirm, after removing the part and checking the circuit, there is no short - it is shorted within the part.

    The output draws about 1A under full load (split between an external LED strip and USB device limited on the board to 1A @ 5V) but these failures occurred with no loads connected.  Only the load of the caps in the circuit and any quiescent power to operate a couple of switch-mode regulators is active at that point.  We are also seeing this before any batteries are connected.  The initial steps of our test is to plug in the 12V charge supply (with no batteries), and turn on the (physical) switch, which brings that 12V supply to this IC.  It should then soon turn on the output, which supplies that power to the remaining circuits on the board.  Of the various boards we had this component fail on, after replacing the part we found no issues with any other part of the board.

  • Hi AQ,

    This is definitely not expected device behavior, especially for the loading conditions you have described.

    Is there any way you could supply a scope shot of the 2ms turn on? I would like to see if the turn on behavior looks normal (linear form start to finish).

    Also, is the 12V charge supply active when it is plugged into the device? If so, this hot-plug event may be causing high voltage transients that the input of the TPS22810 can't handle.

    Have you tried repeating the test on a known good unit to see if it fails after repeated testing? It may be that only 20-40% of the devices fail initially, but I am curious to know whether or not they would all fail under this testing eventually.

    Thanks,

    Alek Kaknevicius

  • Hi Alex, the F0000TEK.JPG image was taken when C27 was 0.1uF with a 10nF cap between CT and ground.  The rise time is is around 20us.  After changing C27 to 1uF, F0003TEK.JPG was taken.  The rise time looks more like 2ms as expected.

    The 12V supply may be active when plugged into the board, however that power passes through a physical switch before reaching U5, so the voltage was stable when switched on in the failures we saw.  But there is still potential of transients from the switch.  I could try keeping Vin tied directly to the power supply (which would still expose it to those potential hot-plug transients) and using the switch only to enable the device.

    We got a report today of a unit that was already in the field for several weeks (sent before we started seeing this problem appear) which had the same failure (photo attached).  So I'm not sure if every part that passes initially will eventually fail, but it appears that at least some do.

  • Hi AQ,

    Thank you for the information, the scope shots do not show any concerning behavior. After discussing with our team, we have a few additional questions that can help us debug this issue:

    1. Is reverse current expected through the TPS22810 for the application? Since there is a backup battery involved, we want to confirm whether or not this is the case. If so, then perhaps the reverse current across the body diode of the device (when turned off) may be uncontrolled and cause the device to heat up very quickly.
    2. Do you have an input current waveform of the TPS22810 turning on that you can share? If it is difficult to get a current probe to measure the current here, perhaps you can measure the current coming from the adapter itself during the load switch turn on. This will help us eliminate excessive current draw as a cause for the issue.
    3.  Can you share the schematic/layout for the switch so we can see if there is any configuration that may be causing the damage?

    Thanks,

    Alek Kaknevicius

  • Thanks Alex, I can send you the schematic - is there an email address I could use?

    I will also try to set it up to capture the current measurement.

  • Hi AQ,

    Feel free to send me any information through a-kaknevicius@ti.com.

    Thanks,

    Alek Kaknevicius