[time-nuts] Antique pendulum clocks

[Tom Van Baak]

Hi Philip,

For low rate measurements like a pendulum clock the timestamping method 
works well. A number of us do it that way. The same method works for 
GPS/1PPS-like signals and also mains (raw 50/60, or divided down to 1) Hz.

Correct, if your measurements are precise enough, you should see rate 
modulation due to uneven mechanical effects like slightly unbalanced 
hands or gears with uneven wear. You might also see the effects of 
winding. The more complicated the clock the more rich the spectral peaks.

You might want to fix the seconds hand so that it's balanced. Most clock 
hands are long and thin on one side and short and fat on the other so 
that rotational balance is maintained.

Note that both TimeLab and Stable32 work with pendulum data. In 
Stable32, you'll find both PSD (Power Spectral Density) and ACF 
(Autocorrelation function) very useful when analyzing data from 
mechanical clocks.

You asked if anyone else is interested in this level of detail. Yes, 
you're not alone. There are active discussions like this right now on 
the Synchronome list:

https://groups.io/g/synchronome1/

Also, check out 30 years of wonderful HSN (Horological Science 
Newsletter) articles:

http://www.hsn161.com/

Bryan's Microset website is a great resource for clock data and analysis:

https://www.bmumford.com/microset.html

Some nice examples of my own analyzing pendulum data:

http://leapsecond.com/pend/shortt/
http://leapsecond.com/pend/clockb/
http://leapsecond.com/hsn2006/

We tend not to discuss the subject of mechanical clocks and watches here 
on time-nuts; mostly because those topics are already well represented 
with magazines, journals, books, and multiple web forums or email lists. 
But if you have specific questions about performance or measurement 
technique that's probably ok. See also the picPET timer chip, which does 
sub-microsecond timestamping [1]. And the TAPR TICC, which does 
sub-nanosecond timestamping [2].

/tvb

[1] http://leapsecond.com/pic/ and http://leapsecond.com/pic/pp06.htm

[2] https://www.tapr.org/kits_ticc.html



On 11/19/2019 5:10 PM, Philip Gladstone wrote:
> I've started to monitor the individual ticks on a grandfather clock from
> the 1790s. Essentially I timestamp whenever the pendulum breaks/restores a
> light beam.
>
> The data that I get is surprising in that the pendulum swing varies
> according to the position of the hands on the clock. It appears that the
> amplitude of the swing depends on the driving force imparted by the
> escapement. Since the second hand is not counterweighted, there is slightly
> more energy available to drive the escapement during the first half of each
> minute and slightly less in the second half. There is much bigger effect at
> the end of each hour when the mechanism has to move a lever to trigger the
> strike mechanism. This 'end of hour' effect changes the pendulum swing
> enough so that the period is noticeably affected (maybe by 300ppm)
>
> Anyway, my google-fu did not reveal anybody else interested in this
> stuff... Anybody here interested?
>
> Philip
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.
>