About six months ago I got a self-winding mechanical watch. I thought it would be cool not to have to worry about replacing the battery. Since it wasn’t expensive, I wasn’t expecting accuracy on the level of the U.S. Naval Observatory, and accepted the fact that, unlike with quartz-regulated electric watches I’d owned in the past, I would have to adjust this one every few days.
But with this watch, it can be every day. I originally assumed that this was happening gradually and continuously over the 24 hours of a day, but I’ve noticed now that the gain sometimes happens overnight. That is, it’ll be correct when I go to sleep, but when I wake up it’s five or even seven minutes fast. When I mentioned this to an acquaintance, she said I might want to get a watch winder, which is designed to keep an automatic watch running. I was skeptical, thinking that if my watch is getting ahead of itself, the problem can’t be that it’s running down overnight. But is my assumption incorrect? Is it possible that such a watch, having nearly run down, would jump ahead? You’d think it would just stop, or at least slow down.
Watches that are running down may be subject to short swings in their ballance mechanism, which uses both a pendulum like regulation, and a spring. (Not the main spring.) The short swing makes the watch run faster.
Try placing it in a different position, either on its back or on its side. Expensive watches are supposed to run at the same speed regardless, but cheaper ones will go a bit faster or slower in different orientations. I take advantage of that to keep my watch accurate, but then it only gains about 20 seconds overnight.
No, I do keep it on my wrist all the time, unless I’m showering or swimming. But it’s unlikely that I move the watch around nearly as much when I’m asleep, so it may not make much difference.
It might be that the watch is predominantly in one position while you’re asleep. Easy enough to test: take it off at night, and see if it runs differently depending on the position. My Seiko runs fast on its back, slow on its side. Maybe that’s not it, but it’s probably worth a try before forking out for a watch winder that might not help either.
I had a self-winder for about 30 years (until it, literally, fell apart on my wrist) and never noticed a problem like this. Indeed, it rarely ran down overnight (and I would take it off overnight), only if I left it off for a day or so. It was a fairly cheap watch in the first place too, nothing fancy. Your watch may have something wrong with it, but I doubt if it has anything to do with it being a self-winder.
I have noticed the behavior described by the OP in my Omega watches, but not my Rolex Sub. Perhaps it’s because I never let it wind down as much or it might have more reserve.
My theory has been that as the mainspring unwinds, its force imparted on the mechanism decreases, causing the balance wheel to be thrown with ever so slightly less vigor, and possibly finishing its cycle in a few split microseconds shorter time.
But that’s only a theory. I have noticed my watches gaining time as they “ran on empty” so to speak.
What if you got one of those little gizmos that is supposed to continuously move your watch while you’re not wearing it? Would the time stay accurate then?