Reginald Neale wrote:
>
> I've just been re-reading a twenty-year-old article in SCIENTIFIC AMERICAN
> on fitting a pendulum clock with a crystal-controlled device that tweaks
> the pendulum period to make the clock super accurate. The author doesn't
> use an oven; he uses a "tuning capacitance" that has a special tempco,
> along with a selected 3 mHz crystal. Is anyone familiar with this article?
> Does anyone know whether the author's crystal recommendations are still the
> best available? He claimed an accuracy of about one second per year for his
> setup. That seems pretty incredible.

Seems reasonable. It also sounds like something that wouldn't work if
you made 1000 of them. That "tuning capacitance" is nothing more than a
cap that had a tempco exactly matching, but in the opposite direction,
that of the xtal. To the degree that you can get all your tempcos to
match, you don't need an oven.

I'd bet that this is not a commercially available part, but something he
made.

The basic technique sounds a lot like injection locking, where you can
hold a 10 Ghz source stable by feeding in a little of a stable lower
frequency.

Your power line is at least that accurate, if not more. They tweak the
generator speed by hand, so that in the long term, it's dead on to wwv.
In the short term though, it could be dramatically off.