[time-nuts] Using Thunderbolt to Discipline FE 5680A?
Arthur Dent
golgarfrincham at yahoo.com
Thu Dec 22 18:39:52 UTC 2011
Although it isn't using the 5680A Rb that is cheaper, perhaps the
Datum/Summetricom X72 Rb oscillator would be a much better
unit to try to disipline. I'm going to try that scheme with one of my
X72 oscillators. You could use the 1PPS output an unmodified
T-bolt or any GPS timing receiver, like one of the inexpensive
Motorola units, to control the X72, which is very simple. Below is
an except from the X72 manual. I also have an older Datum 9390
that came with a Rb for the oscillator that works quite well, except
for the week 1024 date display problem.
-Arthur
"X72 qualifies 1 pulse per second inputs by analyzing the time difference
between the X72’s 1PPS output and the external 1PPS input. This is
referred to as the HoldOver state. The X72 determines whether the 1PPS
input is useable by calculating the rate of change in timing measurements
that are taken once per second.
Once a 1PPS input is qualified, the X72 1PPS algorithm determines if it is
necessary to adjust the counter that produces the 1PPS output (JamSynch
state). The algorithm then begins to adjust the output frequency of the X72
to keep the 1PPS output aligned with the 1PPs input. This is the disciplining
state and the control method is a Proportional Integral Derivatives (PID)
scheme.
The amount of frequency change and the length of time required to
reach 1PPS accuracy can be adjusted by setting y1 (damping factor) and y2
(time constant) parameters. During this disciplining state, the timing of each
1PPS input is compared to the expected value. If the offset exceeds 333
nanoseconds, the algorithm changes to the HoldOver state and the process
begins again."
More information about the time-nuts
mailing list