[time-nuts] 8040c specification clarification, help for a time-nut beginner

[Pee Akka]

I measured the stability of GPS disciplined 8040c using the three cornered
hat method (THC). The other two clocks were rubidium standards that were
not locked to GPS. The measurements were done using 5115A phase noise test
set.

The calculated ADEV result of the 8040c+GPS is an almost constant
horizontal line from tau=1s to tau=1000s located between 9E-13 and 1E-12
ADEV. After tau=1000s it starts to trend downwards (increasing stability),
almost tracking the stability of the GPS receiver. The 8040c's tau=1s
calculated stability is about the same regardless if it's disciplined to
GPS or not (this is normal of course as a rubidium standard should be more
stable in short term than GPS). The results are radically better that what
the 8040c specifications state for its short term stability.

I have not measured the GPS 1PPS output, the numbers I quoted are the
specs. Unless specifically said otherwise, I always assume that the specs
are stated with a coverage factor of two which is the GUM recommendation
and my anecdotal experience of manufacturers' habits.

Basically what you two are saying is that the 8040c's 1E-12 GPS disciplined
spec is worthless and it shouldn't be used under any circumstances when
determining the uncertainty of the whole system? My personal take is this:
The 8040c determines itself whether or not the disciplining 1PPS signal it
is receiving is of good quality and it does not lock to a bad quality
signal. The specifics of the locking mechanism are described in the
SA.22C-LN designer's manual. Therefore, if the 1PPS signal is qualified by
the 8040c and we make the assumption that the device is working as
intended, we can justifiably trust the spec and classify it as a Type B
uncertainty. We can make the assumption that the 8040c's internal
oscillator is working as intended and it is good for qualifying the 1PPS
input because the three cornered hat *without* 1PPS lock yields good
results (TCH with and without 1pps lock are basically identical up to
tau=1000s).

My master's thesis goal was to determine the frequency accuracy and
stability of the GPS+8040c and the conundrum being that it seemed to be
best clock at the lab. Therefore I did not have the possibility to measure
it against a better - for example cesium - standard. I use past tense here
because at the moment I'm just writing the thesis and all the measurements
are done. I do not have the possibility of doing more measurements as I
don't have access to the equipment anymore.

I apologize but I would like to preserve some anonymity so I'm hesitant to
declare my school. My major is measurement technology but this frequency
standard field is a little bit outside my expertise (which is good for
learning I guess). I realize now that I should have come to this place
sooner as you seem to know your stuff pretty well and have great ideas that
I hadn't thought before :)