Findings with V0.0.4A
I have run AtomicGPSClock for 48 hours straight on 2 different systems;
my Dell laptop and a file server I have in my basement. The Dell has a terrible
problem keeping time, it is always several minutes off every day. While the
file server is only off about a minute every couple of months.
Updating the clock every 10 minutes I saw a maximum deviation of 1.670 seconds
with the average deviation being 0.550 seconds.
Updating the clock every 10 minutes I saw a maximum deviation of 0.047 seconds
with the average deviation being 0.006 seconds. Adding up the offsets over
48 hours resulted in a 1.624 second total clock error. Projected over a year
would be a 296.380 second error (almost 5 minutes!)
Excel data log
Theory of operation:
This isn't confirmed or documented anywhere, this
is just my findings. The eTrex has an internal clock which
is powered by a small lithium battery, this keeps approximate time on the
GPS to enable faster tracking of satellites. Once some satellites are tracked
the internal clock is continuously updated via the time information transmitted
to it. NMEA output is initiated at the turn of even numbered seconds, so I
can use the first character or several characters of data as a trigger to
update the PC clock.