[time-nuts] HP 5345A
cfmd at bredband.net
Fri Dec 29 18:07:45 EST 2006
From: "Rick Karlquist" <richard at karlquist.com>
Subject: Re: [time-nuts] HP 5345A
Date: Fri, 29 Dec 2006 14:39:09 -0800 (PST)
Message-ID: <10058.192.25.240.225.1167431949.squirrel at webmail.sonic.net>
> Hal Murray wrote:
> > I thought aging was generally uni-directional and reasonably predictable
> > if
> > you had enough data.
> > Does it wander in both directions?
Allways enjoy your insightfull postings.
> This is probably on the list of the "10 greatest myths about
> crystal oscillators". Many decades ago, there were systematic
> aging effects such as you speak of. I remember learning as
> a youth that glass crystals age up and metal crystals age down.
> Over the years, any such systematic effects have been analyzed
> one by one to understand the root cause, and then the process
> has been fixed to get rid of that aging effect.
I assume we are talking about fairly top of the line units here, such as the
10811 and friends. I'm sure that the really cheap AT crystals all over the
place still show some of these old behaviours even the average quality may have
> What we are
> now left with are tiny cracks and crevasses that grow sporadically
> like a crack in an auto windshield. At least that is what we
> think is going on. The process people, like my friends Charles
> Adams and Jack Kusters, have worked themselves out of a job
> had taken retirement, because, like the "efficient stock market"
> theory, there is no predictability to the aging data. It is truly
> a "random walk down Wall Street" or in this case a random walk
> in time. Oscillators will age in one direction for a while but
> may then age in the opposite direction for while for no particular
> reason. Not only that, but crystals will jump a part in 1E^9
> or so every so often. I've never seen a 10811 crystal without
> jumps if you wait long enough. I don't know of any other crystal
> makers who claim to not have jumps.
I've seen some measures of crystals being held at a constant temperature for
several years and they show only a decaying glide as a continous process over
all those years. I think they where 3-4 years. I could dig the reference up
but it might take some time. It was a rather old experiment thought. You should
have seen it.
More information about the time-nuts