Article: Predicting Crystal Drift Over Time

[Gramham]

Hello Everyone,

One of the more interesting sections of one of the books I am reading at the moment...

...is the section that discusses how manufacturers have to compensate for errors in the quartz oscillators they rely on to provide accurate timing signals to the rest of the quartz circuit.

As almost every quartzophile (is that a word?) knows, the quartz crystal package is supposed to oscillate at 32,768 vibrations per second:

https://en.wikipedia.org/wiki/Quartz_clock#:~:text=Most such quartz clock crystals,frequency of 32768 Hz.

What's interesting (and what I didn't know, but made sense as soon as I read it) was the fact that this isn't always true, and a certain amount of compensation (or regulation) must be  performed by the movement manufacturer, sometimes by by cutting certain traces on the circuit board:

After that, the rate of the crystal was then more or less assumed to be stable over time.

But that's not true.  Apparently, the timing of quartz crystals degrades over time according to a well-known equation (Arrhenius Equation).

https://www.allaboutcircuits.com/technical-articles/using-the-arrhenius-equation-to-predict-aging-of-electronic-components/

I am now quite curious to know if dynamic compensation was implemented (perhaps using a second crystal) to help account for "drift" in the primary crystal.  One could imagine very highly demanding timing environments requiring three crystals and some form of polling system to converge on an "agreed upon" rate of oscillation.  Wow.  Cool!

I am very quickly finding quartz movements ultra, ultra nifty as I learn more about how they work.  Revolutionary is not too strong a word.

(Aside:  One of our sister institutions, MIT at their Media Lab (the parallel to my Department at PolyU) has recently showcased a sculpture that attempts to make what's going on in a quartz movement more accessible to people.  It's called "32,768 Times Per Second" and a small write-up appears on the MIT Media Lab website at:  
https://www.media.mit.edu/projects/32768-times-per-second).

 

Sadly, I cannot find any photos of the installation 

g.
----