Here's some data that may be interesting concerning the accuracy of this clock thus far.
Clock with NIST, with the following results:
Clock Watch NIST Watch 8:00:00 8:00:00.00 8:10:00 8:09:59.38 8:01:00 8:00:59.89 8:11:00 8:10:59.42 8:02:00 8:01:59.99 8:12:00 8:11:59.39 8:03:00 8:03:00:00 8:13:00 8:12:59.24 8:04:00 8:03:59.90 8:14:00 8:13:59.39
Average of five measurements: 8:00:00 7:59:59.956 8:00:00 7:59:59.364
So the clock is now ahead of NIST by 0.956 - 0.364 = 0.592 seconds.
We started the clock speed test on 02/17/22 so now the results so far are:
03/04/22 02/17/22 Clock Time 07:59.59.956 12:00:00.176 NIST Time 07:59:59.364 11:59:59.774 Difference 0.592 second 0.402 second
So the clock is now ahead of NIST by 0.592 seconds and it was originally ahead of NIST by 0.402 seconds, so it has actually gained only 0.592 - 0.402 = 0.190 seconds in 03/04/22 - 02/17/22 = 14.83333 days. So the clock is running fast by 0.190/14.8333 = 0.0128 sec/day = 12.8 ms/day = 0.39 sec/month = 4.68 sec/year.
Your accuracy may vary due to climate conditions in your location, and time of year. We expect to do another calibration in a few months.
Our thermostat keeps our temperature very close to 74 degrees. So that probably helps the stability quite a bit. We did make the measurements visually, but for every measurement we took five and averaged to get a little better fractions of a second. We used a stopwatch for both the clock measurements and the measurements of the NIST time standard signals at https://time.gov
Wow, that is really awesome data. I am interested to know how you collected the data. Did you use instrumentation to get accurate measurements? I usually just look at the time when the second hand crosses a certain position, then quickly look at a digital clock. It is only accurate to about a second or two.
Your numbers are quite impressive. I expect it would drift a bit as the seasons change and the average ambient temperature shifts slightly.
I am starting to run low on the first batch of 100 circuit boards selling on Etsy. My goal is to re-design it slightly for the next batch so a small real time clock can be added. Calibration would no longer be needed. Accuracy would be about a minute per year with a precision RTC.
Steve