218 TimeHub 5500 User’s Guide
097-55501-01 Revision M – January 2009
Chapter 5 Maintaining the TimeHub 5500
Troubleshooting the TimeHub 5500
If an SNDRIFT event(s) occur, the following considerations can be helpful in
determining what action, if any, you should take:
If there is more than one input enabled, is the SNDRIFT event reported for more
than one input?
– If yes, and assuming that the inputs are from independent synchronization
paths, this could be an indication that the internal reference oscillator in
the clock module is drifting. The reasoning is that the internal reference
oscillator is common to all the measurements; it is unlikely that
independent inputs would all be drifting.
– If no, it is likely that the input is drifting.
Does the other clock card report the same condition?
– If yes, it is likely that the input is drifting. The clock cards independently
assess the inputs. It is unlikely that internal reference oscillators in both
clock cards are drifting; hence the input is the likely drift source.
– If no, it is likely that the internal reference oscillator is drifting on the clock
that is reporting SNDRIFT. There are some special considerations here:
If one clock is stratum 3E (55512-02) and the other is stratum 2 (55514-02) the drift
limits are different (1E-9 vs. 1E-10 per day) so it is reasonable that the stratum 2
clock reports a drift event while the stratum 3E clock does not due to the tighter
limits in the stratum 2 clock. In this case, the stratum 2 clock should not be
particularly suspected since the input could be drifting in the 1E-10 to 1E-9 per day
range. On the other hand, if the stratum 3E clock reports SNDRIFT but the stratum
2 clock does not, that would be an indication of possible drift problems on the
internal reference oscillator in the stratum 3E clock.
The drift assessment is based on long-term observation of the reference clock -
input frequency relationship. Each clock re-assesses vs. the drift thresholds every 8
hours. Since the clocks are independent, they are generally not on the same 8-hour
schedule, so it is likely that the clock cards would not report SNDRIFT at the same
time. Hence, one clock card might report a drift alarm up to 8 hours before the other
one also reports it.
Because the SNDRIFT event is a long-term assessment, the potential for setting or
clearing of the SNDRIFT event is at 8-hour intervals. Hence, a sudden change of
drift on an input will not result in an immediate reporting of SNDRIFT, but could
affect the next update (up to 8 hours later).
Use the decision tree in
to resolve a persistent SNDRIFT event. An
occasional SNDRIFT event could occur on a Clock card due to external
temperature variations or, in rare cases, a frequency hop. The following decision
tree is for an SNDRIFT situation that persists for several days, or comes and goes
repeatedly.
