But I don't quite understand the conjecture that "the system is using an internal clock until it syncs with the gps satellites", because prior to getting a GPS fix the flight recorder shouldn't be recording locations into the flight log either, no?

Yes, the system seems to have a fix on the satellites and the Lat/Lon/Alt seems to be consistent from one record to another. It certainly doesn’t take several minutes for the system to get a gps fix. The question is why does it record an incremental time for a few minutes and then resets the time by a couple of seconds. It would seem the system should record the time stamp directly from the gps. If it did that, how could sequential time records change? Is it possible the gps starts with an internal clock number initially and then after a few minutes makes a comparison check with the clock from the gps? It gets back to Kensell’s point about how the gps system is handling the data and or the firmware is using the info inside the computers. These seem like good questions for vendors which I hope to get soon.