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.

TPS62175: Hot plugging boards with this part to 24v line may burn out: never on 12v line

Part Number: TPS62175
Other Parts Discussed in Thread: LM3478,

Am prototyping with a series of boards using RJ12 connectors for power in and TPS62175DQCR to convert either 12v or 24v in to 5v.

3 times when plugging a board into a 24v circuit the TPS62175DQCR has burnt out (on the power in pins). This has never occurred plugging to a 12v source.

Starting to not seem coincidental: is some kind of input protection advised?

(The 24v source from a switching supply is quite stable with +-100mv and certainly well below the 28v advised or 30v peak maximums for the part. Downstream the current draw at 5v is never more than 100-200mA.)

  • Hi Rupert,

    Have you already looked at the Vin net on the PCB for hot plug overshoot? This blog explains it: e2e.ti.com/.../what-is-that-giant-tantalum-cap-on-the-input-of-the-evm
  • Perfect response. Thank you!

    Seems that the corresponding evaluation board uses a 22uF 50v tantalum giant with a 100mohm ESR. That low ESR did not come cheap.

    www.ti.com/.../slvu743.pdf

    Our setup involves 24v power over 28AWG cable - so there's plenty of impedance. However the 24v output of the supply (LM3478) is quite smooth (also thanks to some capacitance on local boards) so perhaps I can go for a higher ESR (lower cost) tantalum? And since the ESR will be higher (eg 600mohm) perhaps get away with a lower capacitance for the snubber. Double cost saving. Any comments?
  • Hi Rupert,

    I would recommend confirming the theory by using the scope to look at the Vin waveform.

    If it is going too high on the hot plug, then you can simulate the circuit and select a suitable bulk cap.
  • Good advice.

    Was quite shocked by what I saw. When the TPS62175 was not placed the overshoot and ringing (on hotplugging RJ12 connector) could be tidily managed with a 1uF & 100R in series. However with the TPS62175 present it took a full 6*4,7uF = 28.2uF ceramic 50v 1206 capacitance (resistors not helpful despite lack of ESR) to ensure the overshoot when hotplugging at 24v from lab power supply remained under 30v.

    Therefore suspect that the "blame" lies not so much with the inductance of the wires as with the Inductor of the convertor itself.

    Propose to place 3 * 10uF 50v 1206 ceramics on boards which require hotplugging. 

    Thanks.

  • Thanks for looking at this in your lab. Are you able to share the waveforms?

    I'm a little surprised that you saw a difference without the IC present. Perhaps you can retest with the EN pin held low and the IC present. Once EN goes high, there should be about a 1 msec delay before the IC even starts switching.
  • Only just seen your message. Didn't think to test with EN disabled because all our PCBs have VIN and EN sitting together on a single block of copper. In the future might do a little microsurgery to try. Waveforms were very much like the blog post (with higher peaks) and, as I remember, the action took place in the sub ms timescale.

    Subsequently tested the circa 30uF ceramic capacitance on different power supplies and in a variety of real circumstances within our setup and found that combining with 3ohms of serial resistance produced, on average, optimal protection.

  • Thanks for sharing the results of your testing!