This is in continuation of this question:
https://www.eevblog.com/forum/projects/measuring-clock-quartz-frequency-accurately/I've one quartz clock that has a dedicated adjustment trimmer pot (suppose it's a cap, but not sure) and I adjusted it comparing the 32.768 signal of the clock to a references on a scope and it's now pretty accurate (off by 1 minute per year)
Now in my other (wall) clock there's no adjustment, it's just a chip on board (epoxy blob) connected to the magnetic coil, a crystal and the battery.
So I wonder
1 - How do I measure the crystal without disturbing the frequency (the other clock has a buffered output, so it's not an issue)
2 - If it's possible to retrofit a pot to adjust the frequency somehow (e.g. a trimmer cap across the crystal, or between the crystal and the chip or something like that)
You can lower the frequency slightly by a parallel capacitor/trimmer. However, near all those 32kHz crystals have an undesirable temperature coefficient such that critical adjustment of frequency is a waste of time.
A real world example is the case of my two clock radios, cheap items but amazingly accurate. One has a green LED display and the other has a red one. With the red one sitting on my nightstand and the green one on top of it, the red one runs faster than the green one at about one second a day. With the green on the bottom, they are almost at the same speed, and don't change much at all from day to day.
The slight temperature difference, with the lower one warming the upper one, is enough to change the frequencies by that much.
If the clock is running fast, which they usually are, I put some extra capacitance across the crystal formed from 2 pieces of thin enameled copper wire tightly twisted together, about 2 cm from memory.
The idea is to first cause the clock to run too slow then gradually decrease the capacitance by untwisting and sometimes cutting off the wires at the free end.
I'm able to quickly check the speed by comparing the stepper motor drive pulses to the Pulse-per-second signal from a GPS module.
Solder flux residue can cause problems.
I have described one solution that worked for me in the old thread.
If your clock IC is junk enough the trick may work again