I have an outdoor GFCI outlet which powers a pond pump. The pump stopped working over the winter and testing parts seems to indicate the pump itself is the problem. When I reset the GFCI the pump works for about 90 seconds and then the GFCI trips. This seems really strange to me. Why would it work for a little while and then trip?
There’s probably a filter in the GFCI that takes awhile to integrate the current imbalance. I would guess you have a very small leakage, and it takes that long for the GFCI to react to it.
Thanks Beowulf. I do not understand how they work and never thought they were complicated. Replaced the pump and it works just fine now.
I did a little research on this. Many GFCIs will simply trip within 25 ms when it senses a leakage current between 4 mA and 6 mA (or greater). But UL943 allows longer trip times for low levels of leakage current. The formula is
t = (20/I)^1.43
where I is the leakage current (mA) and t is the maximum allowable trip time (s). So according to UL943, a GFCI receptacle must trip in less than 5.6 seconds when it senses a leakage current of 6 mA. I find that interesting.
So… I guess it’s possible the OP’s GFCI is a “smart” unit that allows longer trip times for low levels of leakage current. According to the equation above, a trip time of 90 seconds would indicate a leakage current of 0.86 mA.
Another possibility is that the pump only develops a leakage current a certain amount of time after power is applied to the pump, perhaps due to a temperature or vibration issue.
Personally I would just replace the GFCI. They’re cheap. If the new GFCI trips then you know there is a problem with the pump or cable.
Oops, just saw this. Good deal. 
It’s also possible that the pump must be generating pressure for a while before before water is forced into wherever the ground fault is occurring.
Or it took 90-some seconds for the pump to become sufficiently primed that the load on the pump motor increased enough to cause the fault.