Hi

Crystals most certainly do still jump. Like a lot of “features” of
crystals this does not get a lot of attention in the form of published
data. Manufacturers are not very excited to talk about problems.
Doubly so if they happen to have an in house solution for this or
that issue :).

Academics are mostly interested if they can propose a new solution
(or at least a new source for the problem ). Most theories relate to
mechanical stress in the part. Micro fractures in the blank are a
classic “answer”.

Either way, you don’t get a lot of papers or published data. That
said, there are tons and tons of data out there. It simply is not
in the public domain. Manufacturers test thousands of parts a week
for this sort of thing ….What they do next is very much in the
“we don’t talk about that” category.

Yes it does depend a bit on just what you call a jump. Steps in
the ppb range were pretty typical on AT OCXO crystals back in
the 80’s. Steps in the < 1 to 0.1 ppb range showed up on SC cuts in
the 90’s. As the decades ticked off, the typical magnitude dropped.

If you drill down to parts in 10^-11, you will see steps. Are they the
same thing? How sure are you when you measure a 5x10^-12 “step”
in an OCXO aging curve? If a crystal is aging at +3x10^10 per day
for a couple days and switches to -1x10^-10 per day. Is that
a step? ( even if there is no abrupt change ). At some point things
become more than a bit unclear.

Data wise a classical frequency jump ( or phase hit ) looks really boring.
The crystal is moving along on some curve due to aging or whatever.
On the next reading, it has taken a step up or down in frequency. It then
continues on the same curve as before.

Time wise, the classical steps tend to show up on new crystals. The
spacing between steps might be days, weeks, or months. They tend
to space out more and more as the crystal ages. ( making them tough
to see on older parts ).

These are not the same as temperature perturbations. Those happen
at very specific temperatures ( typically in TCXO’s ). They are related
to pretty well known physics. They repeat again and again as the
temperature is changed.

Bob

Hi,

yes, they do jump.

Some years ago our DSL-modems fell out of sync when they were exposed to
a temperature jump., ie. in the warm exhaust of a vac cleaner. Sure, all
xtals have a temperature dependency, but with this batch of crystals
there was a frequency change of 50ppm or so for a temp change of
2Kelvin. These are perceived numbers, I do not have exact data. The
modems' PLL could not track the frequency change fast enough and the
modem fell outta sync. . We had a lot of these crystals in the field and
that was a pretty severe problem.

Cheers

Detlef

Am 02.09.2022 um 16:25 schrieb iovane--- via time-nuts:

wow. Great story Detlef.

I have had crystals 'jump' but this was at the part per billion level...
mainly hysteresis effects. It causes my pulling loop to go mad. There
might have also been some non monotonic behaviour also.

I was told by the crystal mfr (this about about 30 yrs ago) the better
crystals exhibited lower hysteresis, but it was present in all crystals
to some degree. .

On 4/09/2022 6:26 am, dschuecker via time-nuts wrote:

I just wanted to clarify that crystals are quite capable of
jumping without any help from temperature shifts.  The E1938A
proved this.  With a thermal gain in the hundreds of thousands,
the crystal temperature never budged.  Yet the crystals (which
were essentially the same as what was in the 10811) did jump;
every one of them from time to time.  There were no crystals
that "never" jumped, depending on how you define "jump".
A crystal might appear not to jump for while, but if observed
long enough you would always see a jump sooner or later.
It wasn't like you could sort them for "non-jumping" units.

The situation reminded me of the so called "smart clock"
concept, where the clock would "learn" what its aging was
and then compensate it out during hold over.  The trouble
with this was that the aging curves have "knees" in them
and you are hoping that a knee doesn't occur during holdover.

With the advent of small, low power, affordable atomic frequency
sources, we now have a way to get rid of jumps.

Rick N6RK

On 9/3/2022 1:26 PM, dschuecker via time-nuts wrote:

Hi

There are a number of very different types of jumps. It depends
a lot on just which sort you are worried about. Hysteresis typically
comes in looking very different than a jump. Temperature driven
jumps can occur for a number of reasons ….

Bob

I would be surprised to find a crystal that did not jump at some level.
Remember that a crystal in use is undergoing continuous stress and strain.
Any small defect in the crystal structure could be changed resulting in a
small change in physical dimension thus changing its resonant frequency.
Thump on any rock hard enough and it will change shape.

Bill, NL7F

On Sun, 4 Sep 2022 08:17:55 -0800, Bob kb8tq via time-nuts wrote:

might have also been some non monotonic behaviour also.

to some degree. .

cleaner. Sure, all xtals have a temperature dependency, but with this batch of crystals there was a frequency change of 50ppm or so for a
temp change of 2Kelvin. These are perceived numbers, I do not have exact data. The modems' PLL could not track the frequency change
fast enough and the modem fell outta sync. . We had a lot of these crystals in the field and that was a pretty severe problem.

Bill Beam
NL7F

Hi

All frequency sources move around. ADEV is one way to
demonstrate this. ADEV is normally driven by continuous
processes in the device. Jumps are something that happens
very rarely.

The term jumps is normally used to describe a change in frequency
that is not driven by the external environment. It might be used
to describe a frequency perturbation that occurs over temperature.
It would be more common to see the issue described as a
perturbation.

Once you get past those two areas, it is uncommon to see the term
used that way. Thus a bit of confusion when it simply gets tossed
up with no context.

Bob