A power supply design has a significant surge when the battery power source is connected. A simulation can estimate the magnitude of the current and power, but the accuracy of the result depends on the accuracy of the data used for the components. Most of what impacts the surge current and power are parasitics that don't have good numbers. Connectors for example are specified with maximums while the important number for this calculation would be the minimums. Same with the FETs, they offer typical and max, but no min for the channel resistance. This is also impacted by temperature. A unit that was in a cold place gets a battery plugged in and the FETs will be a lower resistance than typical.
Even after entering reasonable data, but not worst case, the resulting power appears to be right on the curve the FET manufacturer indicates as the max, 275 W for 250 us, an average of the sloping power curve at possibly the worse part. Even adding a 0.1 ohm resistor to spread the surge doesn't do a lot because it increases the time, while dissipating 1.6 watts in operation.
I'm having a hard time convincing the board designer that anything needs to be done to mitigate this issue. He insists he has designed similar boards without problems. How do you argue with that sort of reasoning?
A very simple solution is to use a connector I found that incorporates a resistor which makes contact before the main part of the connector, XT90-S. This has a 5.6 ohm resistor and is commonly used in RC devices with lithium batteries. Put that in the cable to the battery and the problem is solved... as long as that is the connection made last... Another approach is to simply provide instructions to charge the caps prior to connecting the battery. The surge with the DC power source should be much less or the battery can be connected initially through another connector on the board which does have the resistor. The hazard is if they leave it on the wrong connector.
Of course adding circuitry to deal with this via a soft startup is too much to ask someone who doesn't think there is a problem.
I noticed a spec on the capacitors for leakage current that seems to indicate they will dissipate in a few minutes. "I=0.01CV or 3μA, whichever is greater." Am I misinterpreting this spec?