[time-nuts] Non electrical time-nuttery

[Lux, Jim (337C)]

>> BTW, Invar has a really low tempcoefficient, but tends to random
>> shifts in its structure, yielding to abrupt length changes. Can't
>> have everything, I suppose...
>

That's what my friend trying to build the 1ppm pendulum found.  He found
that the standard technique is to build the pendulum out of two metals with
different CTE, arranged so that as the temperature changes, the center of
mass of the pendulum stays the same.

Imagine the long shaft of, say, 1 meter, with a CTE of 10 ppm/degree. You
support the bob with a sleeve attached at the bottom of the main shaft of
30cm with a CTE of 30 ppm material (so the bob sits at 70cm from the pivot).
As the temperature rises, the main shaft gets longer, but so does the
sleeve, so it pushes the bob back to the proper location.   (Wow, is this
difficult to describe in words)