Hi All,

I'm having a few issues servicing a Longines 23M movement and perhaps members can offer insight. I've successfully serviced several manual vintage movements but this is my first with a flat hairspring - the others had an overcoil design.

There are three issues:
1) Way too much amplitude. I replaced the mainspring ("white alloy" type) because the blue steel mainspring that came with the watch had "set". This replacement is supposed to be made for the 23M but the I'm getting over-banking in the horizontal positions even half-wound. Fully wound the vertical positions have an amplitude around 300 degrees. Could it be vintage Longines had been designed with more friction in mind (e.g. in the form of lubricants) or is it that my new mainspring simply too powerful?

2) Hairspring movement. The hairspring is a flat design but curiously the there is no "terminal curve" in the region of the regulator. It is simply the spring spiral going through the regulator (see picture). The spring is bent close to the stud (can just be seen in the picture below). The regulator is not a pair of pins but a pin on the inside and a larger block on the outside. If I leave the pin vertical then the second spiral hits the pin during oscillation. If I bend the pin towards the outside (closer to the block, reducing the gap) the pin is no longer vertical but at least the second spiral doesn't rub against the pin.

3) Hairspring movement. Again because there is no terminal curve, while oscillating the second spiral hits the outer spiral near the stud. Here's a link to a video that shows the motion of the hairspring. The file is about 20MB and recorded at 120 fps, so if you are using VLC to view it you can reduce the playback speed to have a slow-motion view.

I'm not sure what is happening with the hairspring. Was it a replacement that was never properly formed? or a design that I haven't figure out...

 

So first question is what is the size and type of end on the mainspring you installed? Based on Bestfit books, the spring should be 1.20 X .10 X 343. Not sure if you simply ordered one from a supplier and asked for a spring for that movement, but you should note that when suppliers don't have the spring you need, they will often send the "closest available" and even a small change in strength in particular can make a huge impact on amplitude. This is why when I email a supplier and ask for pricing, I always tell them that if they don't have the exact size I'm looking for, let me know what they are proposing as a substitute.

Second is, are you sure what you are getting is rebanking? If you have a timing machine, has the lift angle been set properly, and if so what is the amplitude like before you get what I assume is a very fast rate? Have you demagnetized the movement as part of the service? I typically demagnetize twice - once before cleaning and again when the movement has been reassembled. Is the balance spring free of oils and dirt? All this need to be checked.

As an aside, although most of the time you fight to get more amplitude on a vintage watch, there is nothing quite as frustrating in my view as fighting too much amplitude, if that's what this really is...

For the regulating pins, they needs to be parallel and properly spaced. What you have done in bending the one pin up high will cause dial up and dial down rate differences. So you need to put a double bend into the pin so that you can close up the gap if needed, and also keep the pin parallel to the boot that can turn. Here's an example of the opposite, where someone bent the pin out at the bottom:



And it had to be bent back to be parallel again to eliminate the large positional variance from dial up to dial down:



So in your case it's tough to see in the photo you have posted, but you may need to bend it in towards the boot up high, and then put a second bend in just under the first bend in order to keep them parallel.

As for the balance spring shape, can you post a clear photo of the spring at rest? I've not serviced a 23M (although I have a 23Z in the shop right now) so I'm not sure what the spring is supposed to look like on this one specifically, but I really need to see a shot of the balance at rest to see what is going on here.

As I've said before on the forum and to you in PM's as well, learning how to fix things like automatics and chronographs should only come after being able to really get a proper job done on a simple manual wind watch. 90% of the skills you need are used on a manual wind...

Cheers, Al

 

Thank you for sharing your knowledge. I'll double bend the pin as you suggest to obtain a uniform gap where the spring is located.

The eBay listing from whom I bought the spring shows that it should be the correct dimensions (see seller's picture). I'm assuming he sent me a spring with those dimensions. Maybe I can source a micrometer to measure the spring thickness with the required accuracy.



I followed your suggestion to demagnetized the movement but I did not notice a difference in behaviour.

I am quite certain it is rebanking at high amplitudes. See attached screenshot of the timegrapher plots with the watch slightly wound (lower amplitude) and more wound (higher amplitude). I also re-verified the amplitude calibration of my instrument (i.e. setting of the lift angle). With the lift angle adjusted to 47 degrees, rebanking occurs around 315 degrees. on my machine.



Here's a picture of the balance at rest.




Here's a screen grab of the video where the balance is at its peak amplitude . We see the outer coils of the spring touching.

 

Looks like the correct mainspring, and I see it's a tee end spring. Not easy to find others that are weaker just to try as it would be with a regular tang end as those are more plentiful in different strengths.

The coils of the balance spring look concentric, so that's good.

When a watch rebanks, the rate usually jumps up to a very fast rate, so when this happens I often see rates go from just a few seconds fast to hundreds of seconds fast per day, so 800+ fast per day. Yours goes from +5 to +7, which to me does not indicate rebanking. This could be a difference in timing machines I suppose, but rebanking will cause the watch to run very, very fast. What does the movement sound like when this is happening? When rebanking happens, it will sound like a galloping horse - in fact I've seen this referred to as galloping. This is a video I made a long time ago where I take a cheap movement and force the winding past a normal spot, which increases the balance amplitude and causes it to rebank - note the sound:

I am assuming these screen shots were taken with the watch in a horizontal position? If so, when you see whatever it is that is problematic in a horizontal position, does it go away when you turn the watch vertical?

The amplitude reading you see on your timing machine is an average of two amplitudes - one in one direction and one in the other. If you have a scope function on the timing machine you can sometimes read the amplitude in each direction, but many machines do not have this function. Rebanking happens when the roller jewel travels so far around that it strikes the outer portion of the fork horn, and rebounds back the other way quicker than the balance spring would normally return it (hence the faster rate).

For now assuming it is rebanking, the next thing I would suggest you do is reduce the beat error as much as possible, because it's currently far too much. Beat error is often described in terms of time, so that the beat of the "tick" is longer then the beat of the "tock." However it's not just time, it's a difference in amplitude (travel) in one direction compared to the other, which in turn causes a difference in time. When the beat error is too high, it's possible that this can cause the roller jewel to travel too far in one direction, while the other direction is fine. So if you are close to rebanking or have very high amplitude, having the beat error as close to zero as possible will help prevent one side from going a lot further than the other. In other words the amplitude in each direction is closer to the average.

Cheers, Al

 

Yup. Exactly.

My timing software doesn't show a much faster rate.It could be how the rate is derived as you mentioned.

Yes, it behaves normally in the vertical positions, even at full wind. Though amplitudes are high for such positions (around 300).

The software does show the acoustic profiles and also shows the amplitude in each direction (blue vertical line in each profile). So in the case where the watch is partially wound, the beat error is 1.5 ms, the average amplitude is 278 degrees, one direction is about 283 degrees and the other about 273 degrees. Reducing beat error to 0 ms (no easy feat in these vintage designs! ) would give 5 degree more headroom before re-banking occurs. I wonder if it is enough...



I dug out a graph from a lesson the Chicago School of Watchmaking that has plots similar to mine. A) "over-banking", B) normal.

 

Even if it isn't, it should be closer than it is now. It's actually not that hard...

Let down the mainspring and look at how the pallet fork is centered between the bankings. You should be able to see that it's a little closer to one side than the other. Remove the balance and balance cock, and place it on a balance tack. Use a small screwdriver or tool made for the purpose for rotating the collet - turn the collet CW if the roller jewel needs to go CCW, and vice versa. Just like when dealing with a modern watch that has an adjustable stud carrier, you don't need to move it much. In this case just barely precepting the turn would be where I would start - more feel than seeing it turn.

Place the balance back in the movement and check it. Should be able to get it close in 2 or 3 adjustments if you are careful. This is routine work, and all part of getting a simple movement running properly. I always shoot for zero in the dial down position, knowing that it will often drift a bit in other positions, but if you get below 0.5 that is a good start at least. Most brand specs allow for 0.8 max, but to me it's a little too much.

Cheers, Al

 

Just seen this and it's getting late here but, a few thoughts and apologies if I'm not following everything exactly:

I have the same mainspring dimensions as Al and your eBay picture so, you just have to deal with the result. As Al says, these T ends are a pain to find in varying strengths. First thing I'd do is check the dimensions of the spring you've been given.

You need a boot (the outer part) in the regulator when you have a flat spring so that when it's closed, the next coil in can't get caught in the regulator when the watch is running. Of course, you don't need that for a Breguet overcoil so you just see a pair of pins but always these pins should be parallel as Al says. You'll see in Al's pictures that he has a screwdriver slot in the boot to open it. Yours has that hole marked in red below so with the cock/balance mounted on the plate, use an old fine/superfine oiler through that hole to rotate the boot open, undo the stud screw and then release the cock from the plate to leave the balance/hairspring behind. I assume you did this anyway to oil the top jewel.


I'm surprised there is no obvious terminal curve on your hairspring but, on some calibres this is subtle. On a lot of modern calibres, you see a big change in the shape but there is no reason to see this here, just a gentle change in the curve - as an example, the spring below is a '50s Omega bumper. I don't see any change at all in your pictures. As Al says, your's looks concentric but, how does it look when released from the regulator? Try opening the boot and lowering the stud a little to see where it lies relative to the regulator - if it is hard up against the inner pin then the outer coil is missing a profile change.



You can use Al's method or gravity to determine which way the collet needs to turn. For the gravity method, align the balance and pallet jewels as shown below by rotating the movement, measure the beat errors and that will tell you which way to turn the collet. Try it with a movable stud carrier watch and you'll see the logic. Let me know what answer you come up with for the one shown below and I'll confirm.



For me, solve the amplitude issue and the contact between the coils will probably go away. Follow Al's advice in getting the beat error down and, if you're still a little high on amplitude, then we have ways to advise how to reduce the amplitude a little as when it is cased, this will get worse... By the way, with a 47 degree lift angle, I doubt you really have rebanking at 317 degrees amplitude, it would need to be higher and is probably an error in the measurement of the amplitude.

Hope this helps a little, Chris

 

Hi Chris,

Thanks for joining. I really like the gravity method to figure out which direction to turn the collet. I have to think about it to figure out how it works!

I just took a picture of the balance wheel. One can clearly see there is no terminal curve where the regulator is located (which would be at around 1-2 o'clock on the picture). I can attempt to "fix" this by removing the collet from the balance staff and placing the collet on the cock to slightly reshape the hairspring. Is this a foolish idea? It seems to me this step should be done before addressing the amplitude/rebanking/beat error.

 

Hey Chris and Al,

I think you guys will be proud of me. After two iterations of adjusting the collet I obtained this:


The beat error is quite invariant with position.
I think I found a solution to the hairspring spirals touching each other during oscillation: by adjusting the height of the stud in the cock.
The rebanking persists though...

 

Very interesting read even if too dense for me at times.

I have a 23M in the shop which ran consistently ran like 2 minutes per day fast within the first six hours, and then slowly ran closer to 0/per day after about a day of it running. Sounds like the movement isn't so straight forward, hope my watchmaker figures out whats going on. I've already brought it back to him to work on more.

 

Send it to me!

 

I miss it.



What are you servicing?

 

I'm pleased you didn't try to add a shape to the outer coil as all that will do is move the collet relative to the balance staff axis and then when it's mounted, you'll see the spring is no longer concentric. Nicely done with the beat error, you get a feel for this after some time. If your outer coil is in the middle of the regulator pin gap (pin and boot in this case) then the beat error will not change as you adjust the rate.

I'm not sure what you've done to stop the spring contacting because the stud should be adjusted so that the hairspring is completely flat and parallel to the plate all the way from the stud to the collet - you shouldn't be kicking up/down close to the stud to avoid an interference. Like I say, if you solve the excess amplitude, the spring issues you're having will probably go away.

Thanks to Witschi, you can see the lift angle easily here although the amplitude is not actually calculated from the impulse jewel contact with the fork. So, how high is the amplitude in this case? I'm a bit confused but, from full wind and let down one turn of the barrel, what is the amplitude in horizontal and vertical - plus is this movement cased yet? If you're only slightly in the galloping area, there are ways to drop the amplitude a little.


Are those screenshots of the measurement above normal? With slightly wobbly lines like those, I'd be looking at the cleanliness of the pallet jewels and escape wheel teeth but am not sure if your device plots always looks a little like that.

Going back to your questions about the hairspring shape, especially close to the stud, if you look at figures 215 and 216 here, you can see that you have the second one and it's a viable configuration.

This calibre is straightforward as it only has one complication - the central sweep seconds. Sometimes, one movement will give issues but this is a simple design and shouldn't give odd results. If you wind once and get variable performance as the watch runs down, then isochronism is the problem - probably the regulator pin gap.

Cheers, Chris

 

What does total lock on your escapement look like?

 

We are getting there Ashton - trying to keep the noobs from jumping too far ahead!!!

 

Sorry!!

 

LMAO! Does this help?

 

The plots are quite typical of the vintage watches I service (I've had much worse in some very old Elgin wristwatch movements). Youtube videos of watch adjustments of modern movements look much cleaner on their timegrapher. Is it because of my timegrapher software, old movement design and manufacturing, worn movements, too much lubrication, wrong lubrication, noisy environment, etc I don't know.

 

I have a 23M movement, LXW-mark, as was American-market watch. It works fine with a clean dial that was meant for building a new watch. Then I had a kid. I'd be happy to send it to you for shipping cost only. Then you have a spare to muck around with! I have often though about trying my hand at watchmaking. Then I read fun posts like this and realize I'd better stick to my easier day job as a surgeon. This watchmaking stuff looks hard