Every (admittedly el cheapo) digital (LCD display)watch I’ve owned loses time. I often listen in to the radio news during the day and evening and that starts exactly on the hour (CNN usually). I set my watch by it and then several (5 or so) days later my watch is 5 seconds slow. It seems to lose about 1 second per day, maybe a little less.
I know about atomic clocks and have some table models but I like my cheaper wrist model(s) because they talk to me or play musical alarm/hourly tones. I haven’t found a talking atomic model.
Why do they lose time? It’s not because of weak batteries as it occurs even when new or with new batteries. Any authoritative reasons?
Your watch uses a quartz crystal oscillator for its time-base. It isn’t uncommon for cheap quartz crystal oscillators to be off-frequency by significant amounts (100 ppm or more). Their frequency is also affected by temperature and the effects of age. To make things worse, the ones used in watches are designed for low cost and low power consumption, not accuracy. The battery is not the problem.
I have always figured, but do not know for a fact, that the frequency of a crystal is not an invariant property of the material but determined experimentally when using them for watches. The ones that test within tolerance would be used. I have something like 25 quartz watches, and the two that keep the best time (off by maybe a second a week at most) are one nice chronograph that cost maybe $400, and a Mickey Mouse watch that I bought at Disney World for maybe $30-40. I guessed that the more expensive watch maker threw away a lot of crystals, and that the Mickey watch was just a lucky fluke.
The operating frequency of a crystal is NOT a property of the material (quartz). It is mechanical oscillator. Imagine a xylophone bar with an electrical interface. Not exactly, though, as different vibrational modes are used. The most temperature stable is known as transvers sheer. Imagine holding a sponge between your palms and sliding your hands in opposite directions. The physical dimensions, and weighting due to plated electrodes establish the operating frequency.
Nearly all watches use a 32,678 Hz crystal. Given the small size of the crystal, and the low operating frequency. “normal” crystal resonant modes are not used. Instead, the quartz is in the form of a tuning fork. This is not a very temperature stable configuration, but a watch should either be near body temperature, or near room temperature, so it usually works out OK. For consistant operation the watch should either be worn at all times, or for the same amount of time each day.
Circuit parameters DO influance the operating frequency, and provisions for trimming the nominal frequency are normally included. If you knew the circuit diagram, there is most likely a test point that will provide pulses at what should be 1 second intervals. Timing these with a well calibrated electronic timer/counter would allow straight forward bench calibration. Failing those resources, you could resort to trial and error, just as is used in a mechanical watch. Turn the trimmer 20 degrees or so in one direction, and observe the effect. When the correct direction is determined, work your way down to 10, then 5 degree adjustments, 'till you have the desired accuracy. Keeping written notes will speed the process a lot.