[time-nuts] Aging rate of crystals

Mike Fahmie mpfahmie at lbl.gov
Tue Feb 19 18:07:00 EST 2008


Aging is a composite of phenomena sometimes resulting in a negative rate 
and sometimes in a positive rate (most common). Some causes that haven't 
been mentioned in this thread are the slow release of stress in the 
crystalline structure created by shock or temperature excursion, and the 
diffusion of contact metalization into the crystal. A transistors gain can 
change with time and temperature, thus changing the phase shift and forcing 
the oscillating frequency to move up or down the phase plot of the crystal 
to compensate. Other changes in passive components can cause the same 
effect. I suppose one might also expect that atoms may be slowly shed from 
the bulk over time and (probably) increase the frequency.

Somewhere, I have a NASA pub that describes a technique using Gamma Rays to 
accelerate aging so that later aging is much reduced, can't seem to find it 
now. I would guess that the radiation cause the stresses to relax. NASA was 
evidently interested because satellites experience a lot more radiation 
than we receive here on the surface and it probably caused a rapid 
frequency shift in otherwise unradiated crystals.

I have two HP quartz oscillators that I've been running and logging for 
well over a decade. They run 24/7 and are on battery backup. The HP-103 
shows about 6 parts in 10^-12/day and the HP-107 about 2 parts in 
10^-13/day. I attribute this to steady temperature and solid mounting in a 
heavy rack.

-Mike-

At 03:41 PM 2/18/2008, you wrote:
>The long term aging rate is due entirely to the crystal,
>for all practical purposes, for any well designed oscillator
>circuit (or even a mediocre design).  The aging of the
>crystal is basically not predictable.  It's like the famous saying
>by J P Morgan when asked what the stock market will do:
>"It will fluctuate."  About the best you can do is test
>the aging for a few months and hope it won't get worse
>in the future.  You have to be careful about getting
>an oscillator that simply got lucky during your aging
>test and put in much better than typical numbers.  This
>can happen if the direction of aging changes (not unusual)
>in the middle of the aging test.  You
>shouldn't overpromise aging compared to what you know
>your process can support.  Other than holding the temperature
>constant, there is nothing else you need to do to get
>the best aging the crystal can do.  If you get a lucky
>oscillator that has really good aging, it might continue
>to be really good, but there is no guarantee.
>
>Cheap crystals might have more predictable aging due to
>outgassing processes.  However, this will be a large
>amount of aging.  As you eliminate known causes of aging,
>it gets less predictable.
>
>Rick Karlquist N6RK
>
>iovane at inwind.it wrote:
> > I learn from this discussion that the aging rate claimed by 
> manufacturers would refer to the
> > aging of the whole assembly, not the crystal alone. And for practical 
> purposes that is correct.
> > And even in the case of sealed assemblies, components other than the 
> crystal itself may affect
> > the overall measured drift.
> > So my original question on this subject seems to lose any sense, 
> because we will never be able
> > to measure the aging of the crystal alone (if any, at this point) and 
> hence variations in the
> > aging rate either.
> > Anyway some doubts of mine are not yet fully answered by this 
> discussion, and I would appreciate
> > your opinions.
> > Given a good quality sealed OCXO running in constant ambient 
> temperature, what kind of aging
> > curve  should one expect, a fluctuating one? (I understand that this 
> might be the case, due
> > to the interaction of known "intrinsic" aging factors having different 
> timescales, as I've just
> > learnt on this list. A "regular" curve would be hard to get).
> > May it happen that fluctuations in frequency due to "external" causes 
> such as tides, geomagnetic
> > storms, or so, and not actually affecting the "aging rate", are 
> interpreted as fluctuations
> > in the aging rate?
> >
> > I'm running a simple test comparing an OCXO (option 04E on a military 
> Racal 1992 counter) to
> > rubidium (LPRO), the counter being counting the LPRO. The test is 
> running since about two weeks,
> > and I started recording three days after power up. In the first days 
> the OCXO showed a decreasing
> > drift starting with some 3x10e-10 per day until it reached a stability 
> within +/- 1x10e-10 in the
> > last 5 days (that is, since 5 days back, the counters reads always the 
> same value +/- the occasional
> > uncertainty of the rightmost (11th) digit (10 seconds gate time). The 
> OCXO specs are <= 5x10e-10
> > per day. I didn't notice whether it is sealed, and won't check right 
> now. I don't expect that the
> > counter will always stay there, and I don't know what to think when the 
> drift (aging rate?) will
> > change.
> >
> > Thanks,
> > Antonio I8IOV
> >
> >
> >
> > _______________________________________________
> > time-nuts mailing list -- time-nuts at febo.com
> > To unsubscribe, go to 
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> > and follow the instructions there.
> >
> >
>
>_______________________________________________
>time-nuts mailing list -- time-nuts at febo.com
>To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>and follow the instructions there.





More information about the time-nuts mailing list