[time-nuts] Question on crystal jumps

[Steve Rooke]

2008/10/24  <SAIDJACK at aol.com>:
> the jumps are very slow in their frequency change, so the crystal filters
> would just follow the frequency drift. I have seen these jumps take from a few
> seconds to 20 minutes or so. Of course in a Rb they would be compensated by
> the  Rb pulling or pushing the crystal onto the proper frequency.

But I'm not suggesting that the rubidium stage is used to discipline a
ocxo and then fed onto a xtal lattice filter, I meant that the
rubidium stage would go directly into the xtal lattice filter. This
way there is no xtal oscillator to jump off frequency.

> Also, the additional crystals in the filter will very likely jump too, so I
> would expect the overall performance to actually get worse.

If the jump is due to tiny physical effects on a crystal edge, as
previously described, I would doubt that every xtal would be exactly
the same wrt this and I very much doubt that all the xtals in a
lattice filter would jump at the same time and in the same way. This
type of filter works on the basis that each element suppresses signal
outside its centre frequency. If one xtal changes it's centre
frequency, all that changes is that the signal level at that stage
output drops (as in an off-tuned tuned circuit in a multi-stage filter
would do). Also the probabilities that every xtal would jump at the
same time would surely be very high and I'm not sure that using a xtal
in a filter circuit would be subceptable to this jumping in the same
way as using it in an oscillator circuit.

Now if the jumps were caused by something like a cosmic ray, I would
expect that the xtal would just wobble off frequency rapidly and very
quickly go back to its resonant frequency. Using a multi-stage xtal
filter should circumvent this as the particle would have to physically
strike all the xtals in the filter on it's path. Physical arrangement
of the xtals in the filter would easily guard against that.

If the jumps are caused by gravitational effects, it should be
similarly possible to design the filter the be less prone to these
affects by the orientation of the xtals.

73 - Steve
-- 
Steve Rooke - ZL3TUV & G8KVD
Omnium finis imminet