My watch was gaining time. I thought I would adjust it. This was going to be the post about how I got an automatic running to COSC specs.

Instead, I think I broke it.

After noticing that I was resetting the time every week or two, I started keeping track of how fast it was gaining.

``  date             err (sec)  2/04/2016 10:29   25  2/09/2016  9:22  235``  2/15/2016 13:08  482``

It was gaining about 40 seconds per day. More importantly, the gain appeared to be consistent; about 42 seconds per day during the first interval, and 40 seconds per day during the second interval. Some more data points would have been helpful but I was anxious to get some use out of my case opener.

Prior to this project I had only the most rudimentary understanding of how a mechanical watch works: The user adds energy to an onboard power supply, which drives a series of gears to turn wheels at various speeds. Basically, the same mental model as a car.

A glossary of watch parts was very enlightening, explaining the form and function of the standard parts of a mechanical movement. Now I can look inside a watch and recognize the standard parts, like popping a hood and saying "Okay, that's probably the alternator, there's the air intake, the cylinders are in here and the exhaust exits there."

In a watch movement, the power supply is the mainspring, a coil of wire wound by the user (or a rotor in an automatic). In modern watches a clutch prevents overwinding. The mainspring provides the tiny amount of power necessary to overcome resistance in the moving parts.

The timekeeping is done by the balance wheel, much like the pendulum in a grandfather clock. This exquisitely-balanced wheel, connected to a delicate hairspring, rotates back and forth at a consistent resonant frequency, providing the ticks. The escapement is a tiny miracle of engineering that translates the oscillations of the balance wheel to the gear train, and provides a little power back to the balance wheel to keep it turning. From there it's just gear ratios to turn the hands and any complications at the right speed.

So I got the case open and located the timing adjustment, guided by others who had gone before me. The lever effectively changes the length of the hairspring, altering the period of the balance wheel. Very carefully with a toothpick, I gave it the slightest nudge to rotate it a few degrees toward the minus sign, holding the toothpick parallel to the case to avoid spearing the delicate innards.

All seemed to go well, and I closed the watch and began recording gain and loss again.

``  date               err (sec)  2/15/2016 22:06:00  16  2/16/2016 13:52:00   5  2/17/2016 13:01:00 -14  2/18/2016 10:30:00 -44  2/19/2016 10:30:00 -44  2/20/2016 10:00:00 -61  2/21/2016 10:50:00 -35  2/22/2016 10:28:00 -15  2/23/2016 11:06:00  -7  2/24/2016 10:24:00   7  2/25/2016 10:14:00  29``  2/26/2016 10:15:00  38``

I was more interested in precision than accuracy, and at first the watch consistently lost a little time each day. I thought I would be able to give the timing lever an even smaller nudge in the other direction. But then, after an unusually long walk, it started gaining time. If the error wasn't consistent, I didn't have much hope of tuning in a good accuracy.

Then, it stopped.

I wore it to bed, and in the morning it was showing 2:30, several hours behind. I wound it, but it lost considerable time through the day. I found that it would stop running when still and dial-up.

I opened it up again. Dial-down, it ran smoothly. I found a tiny loose speck of yellowish metal and removed it with a pencil eraser. I also noticed that the balance wheel seemed wobbly.

I hadn't noticed any wobble before, but might have missed it. My guess is that one of the fine pivot ends at the end of the balance staff, the axle of the balance wheel, has bent or broken off. It is perplexing how this would happen in the middle of the night. According to the spec, the movement has an anti-shock mechanism. Perhaps I had a bad dream about fighting crocodiles and banged it somehow. It's also possible that one of the jewel bearings was cracked and gradually wore down a pivot. Or maybe it's something else.

I thought I would try to live without a watch for a while. On the second day, someone asked me the time. I had to stand up and pull my phone out of my pocket, as if I were going to call someone who had a proper timepiece. This won't do.

Everything I read about amateur watch repair says that you will destroy at least ten movements while learning, and that's if you're sensible enough to start with large pocketwatches. I should take it to a watchmaker, but I get the feeling that they mostly work on heirlooms, and charge accordingly. I can get a replacement movement for about \$70, but would probably destroy it if I try to install it myself. Or I could buy a new, or nice used, watch. But this is the most expensive route of all, even if I can get a deal on a misspelled Jager.

bioemerl:

Why not get a decent digital watch?

posted by wasoxygen: 922 days ago