[time-nuts] FEI 5680a status update
dansawyer at earthlink.net
Sun Mar 15 22:56:49 UTC 2009
After several weeks of slow evening and hour here and hour there the FEI
5680a is performing on the bench. I was able to get the serial interface
to operate for a brief period. The 'S' and 'F' commands worked long
enough to change the divisor to produce close to a 10 MHx output, it was
7 Hz low. After pressing the 'up' button several hundred times the unit
was very close to 10 MHz. I am not sure why the serial interface stopped
responding, I never tried the second documented 'packet' command set.
There is also a reset pin on the PIC, it is possible this is not
resetting on power up.
Now the search for a means to measure the signal error. I tried the
oscilloscope method of syncing of a 10 MHz signal produced from a GPS
locked oscillator, however it had enough jitter and back and forth to be
impractical use. This is where the thought to divide the 10 MHz from the
5680a by 1000 to get a 10kc signal and trigger that to compare with a
GPS produced 10kc signal. This produced good results, the FEI unit
drifts about 450ns per hour which calculates to be about 1.2**10-10.
This measurement is consistent half a dozen periods ranging for 1 to 4
hours and document range of results for using the F software command.
I have not measured the effect of a button up or down, however it was
significant enough to be obvious that either one up or one down was not
correct. There is supposedly a software command set to change the C
field. It is documented to have a narrow range; I am not sure of the
algorithm to calculate the 'check sum' for the data segment. I would be
interested in others experience with these units.
Thanks - Dan kb0qil
More information about the time-nuts