[time-nuts] means of calibrating/verifying Ru with GPS

Bruce Griffiths bruce.griffiths at xtra.co.nz
Mon Dec 24 04:31:27 EST 2007

Angus wrote:


>> Tim
>> The biggest problem with your proposed measurement scheme is that the
>> 10kHz signal is no more stable than the PPS output.
>> The 10kHz output is phase jerked every second to realign it with the UTC
>> second.
>> This behaviour is clearly documented in the receiver datasheet.
> All the docs I've seen are rather functional in their description, and
> short of detail. I was curious how exactly the 10KHz is generated, how
> it's "aligned", and the precision/position of each rising edge -
> anyone got any info on this?
> Thanks,
> Angus.

It should be easy to measure both the PPS jitter  and the PPS edge to
next 10kHz edge (connect PPS to START input, and 10kHz to STOP input)
with a time interval counter.
The PPS pulse is derived from the 10KHz pulse and is N (256) 10KHz
pulses wide.
The jitter in the delay from the PPS leading edge to the next 10KHz edge
should be very small.
The jitter in the time interval between successive PPS pulses will be
several tens of nanosec.
Using a spectrum analyser to look for phase modulation of the 10kHz
signal as some have done isnt very effective as the modulation is buried
in the noise of most spectrum analysers.

There are many possible ways to implement the feature/function, one of
which is to use a divider to generate 10kHz that can be synchronously
set the divider (divide by N) to N-1 once a second just before the UTC
second to realign its output transition with the UTC second.


More information about the time-nuts mailing list