Hi.
Thanks for the reply and sorry for the long delay. My remote control had got reset at some stage --> time and date reset so I had to dig around in the logs to find the correct log file (I found the correct video first and from that I was able to identify the flight log). I enclose the log below for those interested in reading it. Today I got new MOS-fets and replaced them. The suspicion was correct. It turned out that one of the two MOS-fets was stuck (shorted) in conducting (ON) mode causing power to stay on with the power switch set to off ... I had tried to measure both and I thought that both were bad.
After replacement of the MOS-fets the drone behaves normally and I have done one short test flight after a compass calibration

.
My reading of the log is that something happened that caused the hardware error, the MOS-fet fault. The time from the first battery warning to the crash was extremely short (38 s). My interpretation is that the props were heavily iced causing the battery voltage to drop very fast. Notice that some ice may have flown off in the crash. An additional possible reason may of course have been low battery temperature.
My interpretation from the log is (the time of the day is badly off, it was in the middle of the day):
Drone switched on 22.14.42 idling while getting GPS fix
Take off 22.17.47
22.18.33 first compass warning error_flags=32
22.18.48 compass warning
22.18.56 compass warning
22.19.38 compass warning
22.19.50 compass warning
22.20.03 compass warning
22.20.13 compass warning + battery warning1 (error_flag=32+1=33)
22.20.35 compass warning + battery1 warning + battery2 warning (error_flags=32+1+2=35)
22.20.51 crash motor status=0 (all motors off) error_flags=compass+bat1+bat2=32+1+2)
It can be seen from the log that I almost was able to land before the crash. I was trying to find the correct position to grab the drone from the air.
An interesting detail is the very short time (38 seconds) from the first battery warning to the crash.
The whole episode was an interesting learning experience. Information in this forum allowed me to pinpoint the hardware problem very fast.
Getting compatible MOS-fets took some time though.
I still guess that the crash was mainly caused by very strong icing conditions.
/Lasse