Rate increases after regulating - diagnosis?

[MikeAdvicePI]

Over the past six months I've been learning a bit about watches and I've managed a keyless work repair and a couple of services. It's all been very enjoyable and mainly successful (now I've learned just how easy it is to lose parts with bad tweezer technique). I'm pretty well equipped with the necessary tools but I lack the experience to diagnose a frustrating issue I've come across.

I've reassembled a Sea Gull 3620 (an ETA 6498 clone) after cleaning and lubricating, except for the mainspring. I adjusted the etachron-style bosses on the balance to minimise the positional variation and regulated the rate as follows:

Position                Rate       Amp      Beat error

Face up              +4           256         0.0

Face down          +3           264         0.1

Crown left           +2           250         0.1

Crown down      +3           244         0.0

Crown up            0             233         0.0

I was pleased with that and decided to leave it a while.  About half an hour later the rate had risen to  this:

FU          +20         275         0.0

FD           +19         279         0.0

CL           +23         251         0.0

CD          +22         256         0.0

CU          +21         249         0.1

This is a pattern that has been repeated several times. I've wound the movement, left it a while, regulated it to gain a few seconds and found it had speeded up some time later. The same thing happens when it is partially wound - that is, when I've re-regulated the speedy watch after more than twelve hours, it still reverts to a higher rate after an hour or so.

I first noticed this after I had cased the movement up and to begin with I thought it happened when I screwed the case back down. This morning I was planning to test that, but since it happens outside the case too, I thought it was time to call in the more experienced heads.

I plan to strip and clean the movement again, but any insights into what I should look out for while I do it would be very welcome.

[syfre]

No explanation for your case; but i remember a watch which had the same symptoms, it was a watch from the 50's and finally found that the regulator sleep back to it's original position, like it had a "memory" of the position it had held during the last 50 years and decided to come back by itself.

[jdm]

I recommend that you do not strip the movement again if the timegrapher pattern is regular. I see pretty much the same with Seikos, large variations depending on state of winding, and sometime even if inside or outside the case, which has no easy explanation. Just regulate it to the best average and confirm with daily use. A mechanical watch like that is made to give a reliable time indication not to compete against a measurement instrument.

[MikeAdvicePI]

Thanks for your input, gents. The movement isn't that old, so I doubt that it has a nostalgia issue. Mind you, the regulator is excessively sensitive around the sweet spot, so your suggestion might be related, syfre. 

I will do as you suggest, jdm, and wear it for a month to see what happens in practice now I've serviced it. The timegrapher trace looks much cleaner than before and once it settles on a rate it maintains it very steadily. I hope I've found a reasonable setting now that will make the watch a reliable enough teller of the time.

[StuartBaker104]

Is the timing consistent once it has settled? Does the hairspring move freely through the regulator pins when you move the regulator?  This could cause it to settle if this happens over time.

S

[jdm]

I notice that the amplitude is a good 10% greater an half hour after full winding, and consequently the rate is faster. It can be that the mainspring / barrell can't give max torque at the beginning because of some friction, that would be very difficult to locate and address. I would then be happy with it.

[MikeAdvicePI]

The timing is consistent once the movement has settled, yes. The hairspring is centred in the regulator pins and the pins have been closed up to restrain the spring on each back and forth oscillation without trapping it. It doesn't deflect either way when I move the regulator. It's concentric with the free coil next to it and the beat error doesn't change even with large movements of the regulator. The section of spring between the regulator pins and the fixed end can be seen to "breathe" when the balance is swinging. My understanding is that is how it ought to be. 

Do you mean that if the hairspring is free between the pins, the rate will settle slowly?

[MikeAdvicePI]

@jdm - I noticed the amplitude difference as I typed the post. I have records of the timings of this movement where the amplitude didn't show any significant difference but where the rate was at least doubled. I did wonder if the mainspring was the culprit, but that seemed to suggest otherwise. Hence my idea of dismantling and rebuilding. Maybe I left an eyelash in there somewhere!

[StuartBaker104]

11 hours ago, MikeAdvicePI said:

The timing is consistent once the movement has settled, yes. The hairspring is centred in the regulator pins and the pins have been closed up to restrain the spring on each back and forth oscillation without trapping it. It doesn't deflect either way when I move the regulator. It's concentric with the free coil next to it and the beat error doesn't change even with large movements of the regulator. The section of spring between the regulator pins and the fixed end can be seen to "breathe" when the balance is swinging. My understanding is that is how it ought to be. 

Do you mean that if the hairspring is free between the pins, the rate will settle slowly?

Ok - that all sounds good!

If you leave the regulator alone, does the rate vary over a long period, such as if the barrel were distorted? Is there heavy wear on the barrel arbour bushes?

DeCarle describes in one of his books that if you get a rate variation which occurs over a long time period you should try and match that up with the rotation of one of the train wheels.

S

[oldhippy]

12 hours ago, StuartBaker104 said:

Is the timing consistent once it has settled? Does the hairspring move freely through the regulator pins when you move the regulator?  This could cause it to settle if this happens over time.

S

I would also make sure the whole watch is de-magnetised and pay attention to the h/spring, make sure it is spotlessly clean.  

[jdm]

12 hours ago, MikeAdvicePI said:

Hence my idea of dismantling and rebuilding. Maybe I left an eyelash in there somewhere!

Go ahead if you whish but I bet there will be no change, It's just the way this particular MS releases energy. Seiko recommends to  regulation 10-60 mins after full charge, And since you have attained excellent positional variance I would congratulate myself and move to the next.

[MikeAdvicePI]

Thanks, everyone. I cased it up this morning after a thorough examination for eyelashes and dandruff (there were none I could see). It was running at the same rate I left it last night, so I called off any further work , as jdm suggested. I can still take the advice of oldhippy by demagnetising the watch now. Then, after it's been in use for a while, I could let svorkoetter's Watch-o-Scope monitor it overnight as StuartBaker104 suggests.

I think I've figured out what to do to balance all the variables now. For a £30 movement that's been stripped and reassembled by a novice, I think it's behaving very nicely.

Thanks again to everyone - it's been very helpful and reassuring to have so much expertise and experience to call on.

[MikeAdvicePI]

As a follow up to this I'm posting an image of an overnight run with svorkoetter's Watch-O-Scope. This is the movement back in its case, face up and in a quiet room. There seems to be a lot of regular variation in all three parameters, but I don't have a reading from a known good watch to make a comparison. I would have expected a lot less variation in amplitude and beat error.

Any comments welcome.

 

[jdm]

37 minutes ago, MikeAdvicePI said:

As a follow up to this I'm posting an image of an overnight run with svorkoetter's Watch-O-Scope. This is the movement back in its case, face up and in a quiet room. There seems to be a lot of regular variation in all three parameters, but I don't have a reading from a known good watch to make a comparison. I would have expected a lot less variation in amplitude and beat error.

I do not see any variation, to the contrary very regular patterns and as mentioned before, a watch that is working very, very well.

[MikeAdvicePI]

So I should ignore the regular bobbing up and down of the amplitude and beat error traces? I wasn't expecting it to be so lumpy. 

Thanks for the input.

[jdm]

I don't see any "bobbing up and down". All values are nicely consistent within the reasonable range of a mechanical watch.. 

[MikeAdvicePI]

Ah. I thought the periodic variation in amplitude, for example, which rises and falls over a ten degree range almost twenty times an hour could be described as bobbing up and down. If that is actually " within the reasonable range" I will be able to interpret the traces better in future. Thanks.

[jdm]

10 degrees in 240 is less than 4%, small variation. And you can see the amplitude decreases slightly as the mainsrping unwinds. All seems really good.

To check for defects look at the regular timegrapher plot over 1 minute. There are various telling patterns described in a thread here, sorry I don't have the link at hand.  

 

Edited April 23, 2016 by jdm