Post Go back to editing

LTC4162 blows up at power disconnect

Hi, 

our customer has reported that the LTC4261 blows up at power disconnect. The case happens on testing the demokit DC2038A-H with input voltage 28V, battery charge voltage 24V (2x lead-acid 12V) and load 6W.

We expect that the demokit DC2038A-H should withstand up to 32V continious voltages. The transients due to hot-plug effects are estimated below this value. 

After blow-up the LTC4261 is short to ground at input.

  • Hi Dmitry,

    The demo board can withstand well above 32V (tested to 40V). I would expect that the transients due to the hot-plug events are indeed responsible for blowing up the circuit.

    You should also be aware of the LTC4162's +/- 0.3V differential abs max across both sense resistor inputs. Battery insertion blow-ups are not uncommon with any charger, but I have not heard of any damage to this board as a result of disconnecting a supply.

    Can you please give more detail about the event which causes this damage? Which source is disconnected? And what is the state of the board before disconnection?

    Regards,

    Zack

  • Hi, Zack,

    we have investigated the case with the customer, and it looks that they neglected the protection of the sense pins. We think from conversation with the customer that they removed the protection from the demokit, to check their design (which does not carry the protection). 

    Is it possible, if the overvoltage on the sense pins occurs, the ic fails and starts to behave as a short at the input?

  • Hi Dmitry,

    That is one of the more common reasons for failure of this part. The vast majority of the time, it happens on the battery side just due to the nature of batteries, but it can surely happen on VIN. That said, I'm not as familiar with the failure signatures of a VIN sense input differential overvoltage.

    If this causes frequent failures, they should try out the simple protection method used on the DC2038A (the Schottky diode / resistor across the CLP/CLN pins). This should prove fairly quickly whether this is the reason for failure.

    Regards,

    Zack