After some digging, it seems that a $GNRMC message has an additional field than a $GPRNC message. The code in loragw_gpc.c is written to accept either a $GPRNC and a $GNRMC but rejected the $GN message because it has 14 fields rather than 13. I changed the code to allow either length since the relevant fields are identical. There was also a test in the “UBLOX” message handler that checked for a divide by 0 when computing the slope error for internal clock adjustments. This could happen if they were in sync, a very good possibility if the clock is accurate. I changed this area to accept the fact that the clocks are the same. Don’t understand how this could have ever worked properly if the message were not $GP specific. Guess the engineers at SEEED didn’t test as they should.