[time-nuts] Orbiting crystals

Magnus Danielson magnus at rubidium.dyndns.org
Sat Jun 27 13:39:53 UTC 2009

iovane at inwind.it skrev:
> Time-nuts,
> this post wouldn't be merely speculative, I have an actual 
> interest in knowing if what I'm going to ask is possible.
> As it has already been discussed here, crystals may jump in 
> frequency. If I recall, about 20% of good quality crystals are
> prone to jumping, but a crystal that used to jump might not 
> jump anymore, a crystal that never jumped might jump in the 
> future, and so on.

To the best of my knowledge, it is due to strain defects which releases 
themselfs. Annealing is used to reduce stress. I am sure we have some 
that could comment on it's use in (deep) space programmes.

> The trace of a disciplined (atomic or GPS) jumping crystal 
> shows spikes (because it is soon steered), while the trace of 
> its control voltage shows permanent changes in level. 
> Well, I thought it is legitimate to assume that orbiting 
> crystals, namely the ones disciplined by caesium aboard GPS 
> satellites, are not extraneous to the issue.
> And here is the question: may a time-nuts grade equipment 
> detect this?
> The answer would not be simple. In the majority of cases we 
> get from our multichannel GPSDO the 1 PPS pulse. How may the 
> pulse be affected by the spike in frequency of a single 
> orbiting crystal?
> I know this list is also populated by qualified professionals,
> and hope to get for sure one of the following conclusions:
> -crystals aboard GPS satellites do not jump at all;

I think there could be some interesting articles to dig up on that.

I expect them to be conservative and use annealed crystals, so if there 
is jumps, they are rare occasions.

I know that effects of the time-pieces have been reported.

You should recall that there is several individual clocks. I expect the 
telemetry data to include a myriad of parameters of the clocks, so an 
oscillator frequency jumped should not be missed.

The time-constants of the lock-up loops should give the oscillator some 
time of it own before it is pulled back into frequency, so that 
short-term deviation should be detectable.

> -if they do, time-nuts can't detect this;
> -time-nuts can detect jumps, but can't identify the jumping 
> satellite;
> -time-nuts can detect jumps, and can identify the jumping 
> satellite;

Depends on how you are rigged up. If you have propper L1/L2 receiver 
locked to a cesium and do propper RINEX dumping and analysis of the 
result, you should be able to see the resulting phase shift-dance when 
it occurs. However, you must be able to rule out other effects such as 
multipath and scintilation. Multiple observations of the same effect 
aids in that process. The resolution available in these receivers should 
be more than sufficient to detect such a jump, if it occurs.


More information about the time-nuts mailing list