Hi

One of the weak points of a normal 10811 is that it is not hermetically sealed. Left
in storage for years, they “soak up” humidity. In some cases it can take a fairly long
time (weeks, months) for them to fully dry out. Until they have been run for at least a
week, don’t get excited about the drift.

Bob

Ulf wrote:

There is much more than thermal equilibrium goijng on.  Any oscillator
that has been disturbed in any manner [meaning, disturbed from a normal
state of being powered up and running 24/7/365 in a nice, quiet
location] will need more time to settle in.

A disturbance can be most anything -- mechanical shock, humidity change,
temperature change, adjusting the frequency, being powered off and back
on, etc., etc.

Each oscillator is unique about dealing with disturbances.  Some go
crazy, wandering around in both directions for weeks or months until
they settle and changing the direction of their aging drift.  Some
settle to a different frequency than they were on before, and need to be
adjusted (which is its own disturbance, so it may need to be done a few
times over the course of a month or more).

Where being powered down is the disturbance, how badly an oscillator
behaves when it is powered back on often depends on how long it has been
off.  If it has been off for a week or more, you could well be starting
all over again, just like with a new oscillator.  Also, oscillators that
have been off for a long time have often been disturbed in other ways
while they were off -- jostling, removal from equipment, being tossed
across the room (you'd be horrified at how salvage folks treat
oscillators and other delicate instruments, if you knew), being shipped
across the country (or around the world), exposed to temperature
extremes, etc., etc., etc.

Bottom line -- if a quartz oscillator has been powered off for weeks or
more, expect it to take at the very least a week, and in my experience
usually considerably longer -- a month, or even a year -- to settle to a
drift rate that a time-nut would consider acceptable.

My best 10811-class oscillator is a Symmetricom that looks and works
just like a double-oven 10811, which came as original equipment in an HP
GPSDO.  I thought it would never settle, and it took more than six
months to reach a "time-nuts-acceptable" drift rate.  It continued to
improve for another six months, until it finally settled in as my best
10811-class oscillator.  It has been powered up continuously for over
twelve years now, and it still is.

Be patient.

Charles

Hi,

For better oscillators I let them sit for at least a week, but
preferably a month, before making any judgements on them. The readings
taken is really just their warm-up behavior. Wider phase-noise can be
measured earlier, but very close-in and ADEV/MDEV stabillty, not so much.

Hadamard Deviation can be used to look somewhat below the drift, but in
practice it doesn't fully compensate it out, but it allows to peak below
the drift limit. A drift post-compensation in Stable32 for instance may
work, but loose the first heat-up part as it will confuse the estimator.

Anyway, for this reason I have a couple of oscillator heated on the
labbench so that they are ready when I need to play with them.

Cheers,
Magnus

On 11/08/2017 06:46 PM, Bob Camp wrote: