My first thread (Also, on my Casio F-91W)

[Ronnie the Rotter]

I started writing the numbers down again on Tuesday. The air pressure stayed at 1016 hPa for too long. That matters because quartz behaves differently when it thinks the sky is heavy. The object is a standard Casio F-91W. Nothing modified. Factory battery, allegedly 7 years but that is optimistic if you actually observe the device instead of trusting marketing literature like a farm animal. Everyone knows the backlight is LED but nobody talks about the sub-frame drift. You only see it if you press the light at exactly the same second every minute for several hours. The human eye averages it out unless you train it not to. So I stopped averaging. At first the delay was around 0.004 seconds between button depression and illumination. Normal. Mechanical switch lag plus controller wake cycle. But after the 113th press the light stopped being consistent. The delay stretched to 0.006. Then 0.009. Then it returned to 0.004 but the digits looked slightly thinner. (Not dimmer, thinner) Like the segments were trying to save space. I thought it was fatigue so I used a tripod and recorded it. Frame analysis confirmed it: every 47 presses the illumination curve spikes and the LCD segments redraw half a millisecond later than expected. This is where it gets strange, although maybe not strange if you’ve been paying attention to any quartz drift tables from the late 1980s. At press number 188, the seconds digit rolled from 3 to 4 before the light fully activated. Which is not supposed to happen because the LCD refresh is synchronized to the oscillator. I repeated the test. Same thing happened again after 47 presses. I checked online. Only two mentions of this phenomenon exist online and both are from 2009 and written in Romania. The hesitation is real. I can see it now without the camera. Last night I ran a longer test: 500 presses. Around press 423 the watch beeped even though the alarm function is disabled. Just a short chirp. Dry, like an insect snapping its legs. I opened the case to check for corrosion. Nothing unusual except the quartz can had a faint oily fingerprint on it that I don’t remember leaving (TBF I was exhausted). The oscillator frequency printed on it is 32,768 Hz like normal. That number is important because it divides cleanly down to one second. But if you divide 32,768 by 47 you get 697.19…, The decimal repeats. Which means every 47 presses the timing error stacks slightly off phase. That explains the drift. What it does not explain is why the LCD segments sometimes redraw as if they are deciding which number should be there. At 03:17 this morning the watch displayed 88:88 for one frame when I pressed the light. That pattern requires all segments active simultaneously. The controller does not normally do that. I checked the footage. It happened, just once. I’m continuing the log tonight because the battery voltage is still stable and I want to see if the cycle completes at 752 presses (47 × 16). If it does, the oscillator probably just degrades in a weird harmonic pattern. If it doesn’t, then the digits are not lagging.

Anyway if anyone else has an F-91W please try this:

1. Set a timer for 3 hours.
2. Press the backlight exactly every minute.
3. Watch the seconds digit when it rolls from 3 -> 4

Don’t blink during the transition.
If the segments hesitate you’ll know immediately. It’s subtle at first but after a few hundred presses the display starts to feel crowded.

 

[Gobnull]

dnr very very long