Every so often on this forum we discuss how to adjust the rate on quartz watches. One movement that I have discussed is the Seiko 7A28, 7A38 and 7A48 series of chronograph movements. I had convinced myself that the adjustment was by a variable capacitor via an adjustment mechanism that used discrete steps.
While the adjustment is certainly done via by turning the adjustment mechanism (in 0.26 sec/day steps), I have come across some information that leads me to believe that the adjustment is digital rather than the analog method of using a variable capacitor. This would explain why the correction is not consistent unless the adjustment is made in the required step.
The following information was found in the following forum post:
"Some watches have a real trimmer capacitor which allows for fine adjustment, but has it's own aging and temp drift problems.
The 7A series is equipped with a multi-stage switch that allows adjustment in steps of 0.26s/day. The 7A28 technical guide calls it a 'rotary step switch'.
Adjusting is done by changing the number of pulses to be counted for a time span of 10 second. Normally, the watch counts 32768 pulses per second, or 327680 pulses in 10 seconds.
If you count one pulse more or less (i.e. making every 10th second a pulse longer or shorter, for a total of 327681 or 327679 pulses), you adjust the timing by a fraction of 1/327680. For a day with 24*60*60=86400 seconds, this fraction is about 0.26 seconds. For a month (or 30 days) the adjustment 'step size' is 7.9s. So you're very lucky to have a 7A that can be adjusted to 1s/month.
PS: There really is no trimmer capacitor. I've taken apart a dead 7A28 a few years ago (no pics taken, sorry). The switch is just a piece of metal with two contact 'tongues' that connects different pairs of a circular pattern of gold plated pads on the circuit board, depending on it's position."
If you scroll down in this post far enough, you will see a 7A movement taken apart and the adjustment mechanism exposed.
Mr. Rothe gives additional information (origin unknown):
Note what Rothe says about the "raw" rate of a typical quartz crystal. 4.3 sec/day!! This is certainly in the range of what I have measured in movements that use the inihibition method over a 60 second period (typical non thermocompensated ETA movements). The actual rate is the "raw" rate corrected by the inhibition jump and puts the effective rate in the tenths of a second per day range.
"There are several ways to adjust a quartz watch to a desired rate. First you can adjust the quartz frequency itself, or you can adjust your way of counting.
The quartz frequency can be adjusted by manufacturing the quartz to very high specs (expensive), or by using a rate trimmer condenser, which 'loads' the resonating quartz oscillator and allows to pull the frequency a little. Adjusting the rate trimmer is a manual process which is not desired in mass production.
Since the mid 80s, digital adjustment (also used in the Seiko 7A series) is the favored method.
A normal watch quartz is supposed to generate 32768 pulses per second. The watch electronics counts the pulses, and every 32768 counts, it advances the second hand (or counter). If the quartz is off, 32768 pulses are no longer exactly a second. Instead of using a trimmer condenser to pull the frequency to 32768Hz, we could adjust the counting, i.e. count a pulse more or less every now and then.
Doing it the same way every second (i.e. always count to 32767 or 32769) would lead to a very coarse adjustment (in steps of about 2.6s/day, or about 30ppm). If we modify the counter limit only every 10th second, the adjustment gets finer. Counting one pulse more (make the watch go slower, compensates for a fast crystal) or less (the other way round) once every 10 seconds leads to adjustment steps of 0.26s/day (3ppm), which is fine enough.
Current quartz watch testers have a feature called '10 second gate', allowing to measure the rate over a 10s period to account for that.
Typical quartz watch specs say the watch will keep to +-15s/month (0.5s/day) within a temperature range of 10-40°C.
Just to recall the accuracy parameters of a normal 32768Hz watch quartz:
The quartz is fastest at a temperature of about 25°C (+-5). The initial frequency error from quartz factory at that temp (T0) is usually 50ppm, or about 4.3s/day. If a quartz is made to higher specs, the inital accuracy may be 20 or even 10ppm (1.7 or 0.86s/day).
The quartz frequency is also temperature dependant, with about -0.04ppm*(T-T0)², meaning the quartz will be slower if the watch is colder or warmer than T0. A temp difference of about 15 degrees C from T0 makes the watch run about 9ppm (or 0.8s/day) slower. Also the quartz will age a few ppm/year, the older the less.
So, during production, the watch needs to be adjusted to compensate the initial frequency error. Since watch production is usually done near T0, the quartz will be near its fastest speed. Therefore, the watch will be adjusted a little on the fast side, say, near +0.3s/day (remember, we are doing it in steps of 0.26s/day). Thus, even if the watch is cold or hot on the edge of the temp range, running 0.8s/day slower, it will still be within the 0.5s/day limit. BTW, this also explains why quartz watches that are stored at room temp usually run a little fast.
The correction value will be programmed into the watch chip fully automated in the factory, using contact spots on the mov'ts circuit board.
Now, to compensate for aging, high quality watches need to supply a means to adjust the correction value after the watch has left the factory. E.g. the Seiko 7A chrono series is equipped with a rotary multi stage switch, allowing to adjust between -2 and +3 steps (or -6 and +9ppm), which should be enough to compensate for many years worth of aging (check the 7A28A tech guide, page 11)."