Vintage watches always out of beat? Do hairsprings change shape over time?

[Levine98]

Hi,

I've serviced a dozen or so 1950s era Bulovas (17 and 23j, manual and automatics) the past couple of months and in almost all cases the reassembled and demagnetized movement beat error was way high.

The watches do not run at all upon receipt (that's the way I purchase them), so I can't baseline how they are performing before I strip and service them (and repair them if needed). As an aside, some end up with new mainsprings, others I just reuse if in good condition.

I do regulate them before putting them back for sale on eBay, but I'm wondering if this is typical and if so, does anyone have any thoughts about this?

All of these movements are fixed studd, so I have to adjust the hairspring collet to get the beat error down.

Do hairsprings change shape over time? I'm trying to figure out if this is just normal for older watches (i.e., hairsprings aging and changing shape) or if it is something I'm doing during the service (since I assume they left the factory in beat). I do take great care when stripping and cleaning the movement, and try to follow best practices with respect to handling and cleaning the balance complete. 

Edited December 10, 2022 by Levine98

[nickelsilver]

How out of beat is "way high" ?

 

Up until the 90s pretty much every professional watchmaker used a timing machine that had a paper tape, and printed the dots of the escapement ticks. Lines together, watch in beat, lines apart, watch out of beat, lines pretty close, pretty in beat. Even with the advent of Witschi and Greiner/Vibrograf machines with LCD display that showed amplitude and beat error, most watchmakers stuck with their paper tape machines as those were already expensive and the modern new stuff was REALLY expensive. Beat error was a judgement call, as was amplitude.

 

In the 70s some machines could show amplitude, usually with an accessory unit between the microphone and the timing machine. At least one, the Vibrograf MU700, would show beat error in ms as well as amplitude. Vibrograf stated in the manual anything between 0 and 3ms is acceptable.

 

Nowadays, most modern movements have mobile stud holders, and  anyone can get timing technology that only a few professionals might have had even up to the availability of Chinese timing machines a few years ago (the price of new Witschi or Greiner stuff just isn't in the budget for many professionals).

 

So pieces that were timed on paper tape machines, with fixed stud holders, were judged ok if the lines were kind of close. Hairsprings really can't change shape on their own. They may get caught on something, but it would take an enormous shock to possibly cause a change in shape (I even doubt it then). Poor handling during service will certainly change its shape. Restaffing, requiring the removal of the spring, might end up with it out of place. Whether the watchmaker was concerned enough to get it really close was their decision- but now that everyone can read beat error in tenths of milliseconds there is a bit of obsession with getting it really small- just remember, Vibrograf (really, really smart folks who know about timing) say under 3ms is fine.

 

The really high end companies I've worked for do ask for 1ms or less in 6 positions on repairs and restoration work, but they also don't mind paying 130 bucks an hour for as long as it takes to do a full service, haha.

 

 

[Levine98]

33 minutes ago, nickelsilver said:

How out of beat is "way high" ?

 

Up until the 90s pretty much every professional watchmaker used a timing machine that had a paper tape, and printed the dots of the escapement ticks. Lines together, watch in beat, lines apart, watch out of beat, lines pretty close, pretty in beat. Even with the advent of Witschi and Greiner/Vibrograf machines with LCD display that showed amplitude and beat error, most watchmakers stuck with their paper tape machines as those were already expensive and the modern new stuff was REALLY expensive. Beat error was a judgement call, as was amplitude.

 

In the 70s some machines could show amplitude, usually with an accessory unit between the microphone and the timing machine. At least one, the Vibrograf MU700, would show beat error in ms as well as amplitude. Vibrograf stated in the manual anything between 0 and 3ms is acceptable.

 

Nowadays, most modern movements have mobile stud holders, and  anyone can get timing technology that only a few professionals might have had even up to the availability of Chinese timing machines a few years ago (the price of new Witschi or Greiner stuff just isn't in the budget for many professionals).

 

So pieces that were timed on paper tape machines, with fixed stud holders, were judged ok if the lines were kind of close. Hairsprings really can't change shape on their own. They may get caught on something, but it would take an enormous shock to possibly cause a change in shape (I even doubt it then). Poor handling during service will certainly change its shape. Restaffing, requiring the removal of the spring, might end up with it out of place. Whether the watchmaker was concerned enough to get it really close was their decision- but now that everyone can read beat error in tenths of milliseconds there is a bit of obsession with getting it really small- just remember, Vibrograf (really, really smart folks who know about timing) say under 3ms is fine.

 

The really high end companies I've worked for do ask for 1ms or less in 6 positions on repairs and restoration work, but they also don't mind paying 130 bucks an hour for as long as it takes to do a full service, haha.

 

 

Great background, thanks!

To your question, these old movements have ranged between 4-6ms after servicing. Most of them have watchmakers marks on the inside of the case, so they’ve been serviced several times over the years.

When adjusting, if I can get it below 2ms I’m happy because I’ve decided mucking with it any further only risks damaging the hairspring or collet or both.  50% of the time I can get it below 1ms—then I’m really happy.

[dadistic]

I've worked on a fair number of 10BPAC movements, and the beat error varies considerably. If the movement is marked as adjusted and is in good shape, I expect to see a couple of ms. In a non adjusted movement, around 5 ms is typical.  I don't expect these non adjusted movements to be great timekeepers, so I would only mess with the collet if the beat is so far off that the movement won't self start.

Be very careful if you do adjust, as I've had collets crack on me, but that was when I was trying to move it by twisting with a tool in the slot.  

Cheers!

[Nucejoe]

This tool ( shown by Nickelsilver) should be quite useful. 

Rgds

[Levine98]

3 hours ago, Nucejoe said:

This tool ( shown by Nickelsilver) should be quite useful. 

Rgds

What is this tool? 

[Nucejoe]

Its use is explained here,

https://www.watchrepairtalk.com/topic/20300-adjusting-beat-error-on-a-movement-without-adjustable-stud-carrier/page/3/

Rgds

[JohnR725]

12 hours ago, Levine98 said:

When adjusting, if I can get it below 2ms

What are you using to regulate your watch with?

10 hours ago, dadistic said:

as I've had collets crack on me, but that was when I was trying to move it by twisting with a tool in the slot.  

What sort of tool?

13 hours ago, nickelsilver said:

Up until the 90s pretty much every professional watchmaker used a timing machine that had a paper tape, and printed the dots of the escapement ticks. Lines together, watch in beat, lines apart, watch out of beat, lines pretty close, pretty in beat

Not only the watchmaker but for instance the B200 timing machine was the timing machine of choice. I remember When I was in Switzerland if there was a timing machine that's what it was typically. So this brings up the problem of even the factories because they didn't have a newfangled machine with numbers in their face how closely did they regulate the watches too?  Is a lot easier to please a timing machine with a paper tape versus a machine with numbers in your face.

A problem with any watch that's ever been serviced by anyone else or several other people is things happen and we just end up with all sorts of strange mysteries of watches including Rolex watches that are hopelessly out of time adjustment and all sorts of things where you question what happened maybe somebody had a bad day.

[dadistic]

15 hours ago, JohnR725 said:

What sort of tool?

A modified screwdriver.  I think in the recent case of the cracked collet, the screwdriver was not modified enough, and was still too thick. 

[Levine98]

22 hours ago, Nucejoe said:

Its use is explained here,

https://www.watchrepairtalk.com/topic/20300-adjusting-beat-error-on-a-movement-without-adjustable-stud-carrier/page/3/

Rgds

Thanks. Why don’t you still use the tool—sounds like the safest way I’ve heard to do this adjustment. BTW does the tool have a name? I’ve not seen anything like it online before.

Also, if I read it correctly, you lay the balance complete flat, then pull the balance cock up away from the balance wheel slightly to get at the collet—why don’t you remove the balance wheel from the balance cock first before adjusting the collet (and remove the risk of twisting the hairspring)?

 

[JohnR725]

3 hours ago, Levine98 said:

why don’t you remove the balance wheel from the balance cock first before adjusting the collet (and remove the risk of twisting the hairspring)?

Personally I always prefer to have the balance wheel by itself not attached to the bridge. Some hairsprings really do not like to be stretched and become very unhappy and do not go back to wherever they came from.

[Nucejoe]

4 hours ago, Levine98 said:

 Why don’t you still use the tool—sounds like the safest way I’ve heard to do this adjustment. BTW does the tool have a name? I’ve not seen anything like it online before.

This probably a proprietary tool which has been in some manufacturer's assembly line in its time. 

Rgds

 

[mikepilk]

On 12/10/2022 at 2:07 PM, Levine98 said:

When adjusting, if I can get it below 2ms I’m happy because I’ve decided mucking with it any further only risks damaging the hairspring or collet or both.  50% of the time I can get it below 1ms—then I’m really happy.

For fixed stud holder movements, that's my criteria too - about 2ms, for the reasons you mention. 

When stripping these watches, I fit just the balance/cock to the mainplate and check where the roller jewel sits in the banking pins. I then put a mark on the balance if a slight tweek is needed. If there is no shock protection jeweling on the cock, the balance+hairspring has to be separated anyway, so adjusting the hairspring is easy.  

Just by line of sight, it's easy to get within about 2ms.