Serviced ETA 2763 having erratic rate and amplitude

[VWatchie]

Before I start and if anyone wonders, these two images are from the PCTM software by @praezis with some additional graphs and data added by me.

The movement is a newly serviced manually wound ETA calibre 2763. As I have been working on watches for about six years as an enthusiast, I feel somewhat confident I have not overlooked anything too obvious. The movement was housed in a badly beat-up case that probably had seen a few decades of daily use. I picked it up as a non-runner (proved to be a broken mainspring) for £23 so I have no idea about how it would have performed before I serviced it.

The only adjustment I have made is to widen the gap between the boot and the pin, as it was very tight. I would assess the gap to have been about 1.2 to 1.5 times the thickness of the balance spring (it was clearly manipulated by someone else and the pin was leaning towards the boot). However, the spring was bouncing nicely and evenly between the boot and the pin. Hoping that widening the gap would make the movement less erratic I opened the gap to somewhere between 2.0 and 2.5 times the thickness of the spring. This manoeuvre improved overall amplitude by about 15° (and the rate dropped by about 60 s/d) but it didn't do anything for the erratic rate and amplitude.

I know it's impossible to pinpoint the source or sources of what could be wrong, but I'm out of ideas of what to investigate.

I did take quite a few pictures while assembling the movement and if you think these pictures could be of any help you can click on the link below to my OneDrive. Remember to sort the pictures by name in ascending order to get the assembly sequence right.

https://1drv.ms/f/s!AnVrKJ9agkNWkflUcMMeJq_v2ahgeQ?e=EnykA3

[ColdWind]

Sorry for the crappy photo, I don't have a better one in my phone. ETA movement with erratic rate hm, if it is the "consistent type of erratic" then I would suspect a faulty wheel from the chain but this wasn't it, yours is all over the place. I believe this must be a hairspring problem. Speak from my own experience if the distance (marked red) is somehow close enough the hairsrping could sometimes touch the stud making the erractic rate you are experiencing.

Edited March 13 by ColdWind
Typing errors

[VWatchie]

Thanks for the tip @ColdWind . Worth looking into and something I've never even considered before. I'll take a look now to see what I can find.

[Neverenoughwatches]

4 hours ago, VWatchie said:

Before I start and if anyone wonders, these two images are from the PCTM software by @praezis with some additional graphs and data added by me.

The movement is a newly serviced manually wound ETA calibre 2763. As I have been working on watches for about six years as an enthusiast, I feel somewhat confident I have not overlooked anything too obvious. The movement was housed in a badly beat-up case that probably had seen a few decades of daily use. I picked it up as a non-runner (proved to be a broken mainspring) for £23 so I have no idea about how it would have performed before I serviced it.

The only adjustment I have made is to widen the gap between the boot and the pin, as it was very tight. I would assess the gap to have been about 1.2 to 1.5 times the thickness of the balance spring (it was clearly manipulated by someone else and the pin was leaning towards the boot). However, the spring was bouncing nicely and evenly between the boot and the pin. Hoping that widening the gap would make the movement less erratic I opened the gap to somewhere between 2.0 and 2.5 times the thickness of the spring. This manoeuvre improved overall amplitude by about 15° (and the rate dropped by about 60 s/d) but it didn't do anything for the erratic rate and amplitude.

I know it's impossible to pinpoint the source or sources of what could be wrong, but I'm out of ideas of what to investigate.

I did take quite a few pictures while assembling the movement and if you think these pictures could be of any help you can click on the link below to my OneDrive. Remember to sort the pictures by name in ascending order to get the assembly sequence right.

https://1drv.ms/f/s!AnVrKJ9agkNWkflUcMMeJq_v2ahgeQ?e=EnykA3

What is the time elapse for the graph  watchie ? 12 minutes? 

6 minutes ago, Neverenoughwatches said:

What is the time elapse for the graph  watchie ? 12 minutes? 

Sorry watchie i missed the minute markers, so the fluctuations are quick.

What do lower amplitudes show when the spring diameters are smaller ?

[mikepilk]

5 hours ago, VWatchie said:

The movement is a newly serviced manually wound ETA calibre 2763. As I have been working on watches for about six years as an enthusiast, I feel somewhat confident I have not overlooked anything too obvious.

I've been fiddling with watches for about 8 years. We all get a bit over confident, then put in our place with a movement we cannot figure out. It happens to us all. I'm just trying to sort out a Smiths with a similar oscillating trace.

There seems to be a repeating pattern every 4-5 mins. Third wheel? Might be worth checking the teeth. 

There's a known problem on some ETA movements (I've mentioned it before)  - I think this might be one. The 'intermediate wheel (I think it's called) has too much end play, allowing it to touch the bottom of the barrel, and give readings like yours.

Kalle (Chronoglide) has described it as a design fault. I always check on ETA movements, and often have to reduce the end-shake

 

Edited March 13 by mikepilk

[VWatchie]

34 minutes ago, mikepilk said:

We all get a bit over confident, then put in our place with a movement we cannot figure out.

Oh yes, trying to get mechanical movements to work well is a humbling experience and any hybris is relentlessly punished. Thanks for sharing. It makes me feel a little less alone.

37 minutes ago, mikepilk said:

There seems to be a repeating pattern every 4-5 mins. Third wheel? Might be worth checking the teeth. 

Will do! I usually have a look at all wheels through my microscope, but not really in detail unless something happens to look suspicious. I'll have a close look at all the teeth.

38 minutes ago, mikepilk said:

There's a known problem on some ETA movements (I've mentioned it before)  - I think this might be one. The 'intermediate wheel (I think it's called) has too much end play, allowing it to touch the bottom of the barrel, and give readings like yours.

Yes, I saw your post about it and then saw Kalle's video about it. So, I was well aware when doing the service and tried to reduce the play a bit. Not sure it was needed but to make sure. However, it felt like the jewel in the main plate didn't budge (perhaps the anvil was too small, hmm...) so I ended up leaving it as it was. I'll have a look at that again.

I just had a look at the balance and recorded the following two videos. To my eyes the gap between the boot and the pin looks a bit excessive and there is a smudge of too little oil on the balance cap jewel, but I don't think any of that should cause the movement to be erratic. Speaking about the cap jewels, both were slightly scratched up but they had no potholes in them. Again, I don't think that would explain it.

One thing that just came to mind was that I did reduce the end shake of the pallet staff and perhaps I went too far. If the staff shoulders are touching the jewels on both sides simultaneously when swinging I guess that could cause friction and explain the erratic behaviour. Your thoughts?

Here are the videos:

[mikepilk]

4 minutes ago, VWatchie said:

To my eyes the gap between the boot and the pin looks a bit excessive

It does look a bit large, though the pins look straight, so maybe it's meant to be like this?

Also I've come across plenty of bent centre seconds wheels, which can catch on the bridge, but I'm sure you would have spotted that.

[nevenbekriev]

See what @ColdWind has said and check (look) if the hairspring touches the studd whhen the balance is rotated right to the max of it's movement. If so, this will explain the eratic rate. The distance changes much when the spring is not well centered to the collet

[VWatchie]

5 hours ago, ColdWind said:

Speak from my own experience if the distance (marked red) is somehow close enough the hairsrping could sometimes touch the stud making the erractic rate you are experiencing.

I had a close look, but there was plenty of room between the stud and the balance spring as the balance was oscillating, so unfortunately that didn't solve the problem. Anyway, your tip was much appreciated!

[nickelsilver]

The centering at the collet is actually a big deal. Out of center there can give all kinds of weird results. Not necessarily what you're seeing, but wouldn't surprise me.

[ColdWind]

ETA 2892-2 erratic trace I found this post with the erratic trace line seems similar to yours. Happens even on the so modern 2892-2 movement. OP did post a photo of the hairspring with the balance cock flipped upside down, look at it you may think there's plenty of room but no the problem actually lies in the hairspring. 

[Nucejoe]

2 minutes ago, ColdWind said:

OP did post a photo of the hairspring with the balance cock flipped upside down,

Even if he did , that doesn't show all of the spring coil, so details can  go unnoticed. 

Balance must be removed in order to show full details.

 

 

 

 

[nevenbekriev]

Actually, the traces are not so erratic, they are periodic. And the periode is something about 4-5 min as I see.

In the watchmakers text books, there is a head for timefrapher trace diagnostic, and there one usually finds the short simple answer that the problem is the wheel that turns with the same period.

Well, as H. Mencken says, to every complicated problem, a solution will be found - simple, clear and...wrong.

The tings that change periodically in the watch movement are more than entire turns of wheels. For ehample, if the 2nd wheel pinion has 12 leafs, 60/12=5 min, so for 5 min the second wheel moves exactly 1 pinion leaf ahead. This is for movements with 2nd wheel in the center and with 12 leaf pinion, which is not the case here, but...

Then, periodical rate changes are due preiodical amplitude changes due periodical torque changes, but this actially means that the escapement is not isochronic, so the rate strongly depends on torque. So, if in the first time, the movement is isochronic, then the rate will not so strongly depend on torque and thus on wheel radial trueing or teeth meshing problems.

This are only theoretical thoughts.

Now, what You have: In the video what i see makes me think that the hairspring is not well centered to the collet. This may lead not only to great periodical changing of studd-hairspring distance and touching when big amplitude, but in the same manner - to touching the hairspring inside coil to the collet too, which also leads to erratic but connected with amplitude rate changes.

Then, the other thing is that decrease of amplitude leads to decrease in rate. In such case, the normal thing to do is to close the distance between regulator pins, not to open it more. But it is after cleaning the problem with hairspring center, if it exists. By the way, hairspring off-center leads to 'balance not poised' behavior, is there such thing? Reduce the amplitude by releasing the mainspring and see if the rate/amplitude will get less erratic and if there are ' bad poise' traces ih vertical positions

[JohnR725]

20 hours ago, VWatchie said:

PCTM software

interesting feature of the software has I've only seen the free version.

As this watch makes you unhappy can you show us a watch that looks good on the software in this setting?

 

 

[VWatchie]

12 hours ago, nickelsilver said:

The centering at the collet is actually a big deal. Out of center there can give all kinds of weird results. Not necessarily what you're seeing, but wouldn't surprise me.

I don't even know what "the centering at the collet" means. I saw it somewhere else here on WRT but didn't get it. Obviously, I need more info about what it means, even if it isn't going to solve the problem. Is there a source of information you'd be able to recommend or can you talk a bit more about it?

3 hours ago, nevenbekriev said:

In the video what i see makes me think that the hairspring is not well centered to the collet.

Again, I can't quite figure it out. So, I know the collect should be dead-centre over the jewel hole, and that the spring coils should be concentric to the collet but I have a feeling that's not really what you're talking about, right?

3 hours ago, nevenbekriev said:

to touching the hairspring inside coil to the collet too

Do you mean the spring touching the collet?

Would it be helpful with a picture of the hairspring placed on the cock (upside-down) and attached to the stud without the balance wheel? Let me know what you need.

51 minutes ago, JohnR725 said:

As this watch makes you unhappy can you show us a watch that looks good on the software in this setting?

Yes, I'll try to include that in my next post.

[JohnR725]

24 minutes ago, VWatchie said:

Again, I can't quite figure it out. So, I know the collect should be dead-centre over the jewel hole, and that the spring coils should be concentric to the collet but I have a feeling that's not really what you're talking about, right?

as you're grasping you want to be very careful what the definition of? in other words you can have the collet centered over the jewel hole and not actually have the hairspring centered.

[VWatchie]

1 hour ago, JohnR725 said:

As this watch makes you unhappy can you show us a watch that looks good on the software in this setting?

Here are the equivalent graphs for a newly serviced Peseux 7040. This is a movement that keeps very good time and I was surprised to see the variation in amplitude. No difference really compared to the ETA 2763. Of course, the amplitude is considerably higher but the variation over time seems to be the same, that is about 30°. It's really a bit puzzling... The variation in rate differs a lot compared to the ETA movement though.

Well, I look forward to your comments!

[praezis]

16 hours ago, VWatchie said:

One thing that just came to mind was that I did reduce the end shake of the pallet staff and perhaps I went too far. If the staff shoulders are touching the jewels on both sides simultaneously when swinging I guess that could cause friction and explain the erratic behaviour. Your thoughts?

From experience, no thoughts: friction here can eat much power. It is a kind of hidden fault.

Test: without power the pallet must fall from one side to the other by its own weight.

Frank

[VWatchie]

6 minutes ago, praezis said:

Test: without power the pallet must fall from one side to the other by its own weight.

Wouldn't that depend on how the pallet is weighted? I used to do this until I ran into a movement where this test just didn't do it for me. @nickelsilver then suggested a very small puff from the blower and that's the method I've been using ever since when the pallet wouldn't fall by itself.

Anyway, thank you for the tip, I will absolutely look into it.

[VWatchie]

So, I put the problematic movement (ETA 2763) on the microphone after 24 hours to see the difference in rate and amplitude. I'm not sure what this brings to the table, but I just thought I would show you the results. The eye-catching bit is the huge difference in rate after 24 hours. The amplitude isn't exactly good, but it isn't catastrophic either. 

To make it easier to compare I'll repost the images from the first post in this thread.

4 hours ago, JohnR725 said:

you can have the collet centered over the jewel hole and not actually have the hairspring centered.

Still not sure I understand. I guess the hairspring could be non-concentric but still have the collet centred over the jewel hole, but I'm having a hard time picturing it. Anyway, is that what is meant by "hairspring is not well centred to the collet"?

I'll remove the hairspring from the balance wheel and mount the hairspring to the balance cock and take a picture of it upside-down. That will hopefully make the discussion easier.

[Geotex]

Watch movements are very complex and inter-related mechanisms. If a tooth is bad, then the periodicity of a problem can simply be related to the time taken for one revolution of that wheel. However, if two interacting wheels have slightly bent pivots, worn jewels or are otherwise less then perfectly centered, it can theoretically be more complex. There can be some points where those problems would add together to bog down the movement, but other times where they could cancel each other out and give better performance. As they rotate at different speeds, those points of deflection would come and go and set up a resonance (for lack of a better term) that would repeatedly occur with a timing different than either of their rotation rates but related to their tooth ratio.

Not sure that this would be relevant to your problem, but it seems to me that an issue that occurs somewhat rhythmically over the course of many minutes would be more related to the wheel train than a hairspring problem.  But then again, my experience with hairspring manipulation rounds to zero, so you should listen to others!

[nevenbekriev]

6 hours ago, VWatchie said:

I don't even know what "the centering at the collet" means. I saw it somewhere else here on WRT but didn't get it. Obviously, I need more info about what it means, even if it isn't going to solve the problem. Is there a source of information you'd be able to recommend or can you talk a bit more about it?

Again, I can't quite figure it out. So, I know the collect should be dead-centre over the jewel hole, and that the spring coils should be concentric to the collet but I have a feeling that's not really what you're talking about, right?

Do you mean the spring touching the collet?

 

See, the hairspring has a imaginable center. It doesn't exist as some of the spring has been cut to make room for the collet, but even when manufactired, the spring didn't go to it's center.  The idea of centering is to adjust the collet position so the collet hole center coincides with the hairspring center. Nothing to do with the spring on the bridge and the collet above the stone. This is done by bending the internal spring coil near the place where it is pinned or attached to the collet. You can imagine the well centered spring as a groove of vinyl record. If the record is centered on the platter, then the arm moves gradually and only to the center when the record plays. If the record has big center hole (45 RPM one) and it is placed not centric on the platter, then the arm constantly moves in/out. Such is the behavior of not centered spring. The centering of the spring to the collet is usually well described in the books where the attaching and pinning the spring to the collet is described.

And yes, I mean the spring can touch the collet and this can 'short' the entire internal coil if the spring is not well centered to the collet. Also, not centered to the collet spring brings bad poising to the balance.

[Neverenoughwatches]

3 hours ago, VWatchie said:

So, I put the problematic movement (ETA 2763) on the microphone after 24 hours to see the difference in rate and amplitude. I'm not sure what this brings to the table, but I just thought I would show you the results. The eye-catching bit is the huge difference in rate after 24 hours. The amplitude isn't exactly good, but it isn't catastrophic either. 

To make it easier to compare I'll repost the images from the first post in this thread.

Still not sure I understand. I guess the hairspring could be non-concentric but still have the collet centred over the jewel hole, but I'm having a hard time picturing it. Anyway, is that what is meant by "hairspring is not well centred to the collet"?

I'll remove the hairspring from the balance wheel and mount the hairspring to the balance cock and take a picture of it upside-down. That will hopefully make the discussion easier.

The hairspring does have 2 mounting points. At the stud and at the collet, how the spring also leaves the collet has an effect on the hairspring's shape. Bent or twisted hs at the collet can be seen when collet is on a broach or balance complete in truing calipers .

[JohnR725]

6 hours ago, VWatchie said:

Well, I look forward to your comments!

whenever I have timing problems at work usually the first thing I look at is the plot like this and the oscilloscope. I use it to explain why I'm having timing problems. But based on my past knowledge of you and a timing machine I would throw all of this away because you're going to lose your mind and become obsessed about a problem that you may or may not ever be able to fix or a problem doesn't actually exist sort of.

1 hour ago, Geotex said:

Watch movements are very complex and inter-related mechanisms.

yes one of the problems with the watch is a set of gears that in order to produce absolutely linear perfect power ideally shouldn't even be gears at all as gears will always produce pulsations of power. Then if everything isn't exactly perfect like the manufacturing the wheels are slightly out around pivots that don't fit absolutely perfectly or even if somebody furnishes the pivots and are slightly too small soup of flopping around in the whole not by very much conceivably will cause power surges as the tiny bit of the wheels moving around binding up with the gear train.

Watches are really interesting things in that the average out a lot of their problems. Timing machines though will show all of these problems that may or may not really be a problem. For instance with this watch does it actually keep time with the hands on?

I have a interesting book recommendation is that expensive book I have a video of a book review. It has a nice job of reviewing it it's not a perfect book but it sometimes has its moments. Like for instance there's a section on gears. then it even is a section on the faults of gearing fortunately that's only a couple of pages I could probably scanned that in. Where I'm going with this is watches are mechanical devices gear trains produce power fluctuations watches will still keep really good time with power fluctuations up to a certain points. Be careful not to get obsessed about timing machines software or if you think your first experience with a timing machine was a nightmare this will send you straight to the psychiatric ward. Understand why you're using the software and why you're looking at it.

 

 

58 minutes ago, nevenbekriev said:

See, the hairspring has a imaginable center. It doesn't exist as some of the spring has been cut to make room for the collet, but even when manufactired, the spring didn't go to it's center.  The idea of centering is to adjust the collet position so the collet hole center coincides with the hairspring center. Nothing to do with the spring on the bridge and the collet above the stone. This is done by bending the internal spring coil near the place where it is pinned or attached to the collet. You can imagine the well centered spring as a groove of vinyl record. If the record is centered on the platter, then the arm moves gradually and only to the center when the record plays. If the record has big center hole (45 RPM one) and it is placed not centric on the platter, then the arm constantly moves in/out. Such is the behavior of not centered spring. The centering of the spring to the collet is usually well described in the books where the attaching and pinning the spring to the collet is described.

 

3 hours ago, VWatchie said:

the collet centred over the jewel hole

it's amazing how a few words can instantaneously cause of flashback to something in the past which I found to be a pain in the ass. One of my friends had centered his hairspring exactly as you describe and had it perfectly centered exactly as you described. But physically centering without the balance wheel in this case with the Hamilton 992 result of the hairspring that was not flat and not right at all with the over coil. I spent hours of time analyzing trying to figure out where he bent looking at another watch to put the hairspring back right because of what I perceive a stupid method of doing things.

The preferred centering is with the balance wheel and hairspring together. Exactly what is described in the first quote above. 

it's one of the objections I have when people have hairspring issues and everyone says take the balance out of the watch hairspring issues with balance wheels typically occur in the watch you have to be able to see the problem in the watch you don't always have to fix it in the watch but you have to see the problem in the watch balance wheels out of the watch a lot of times a look really nice but not in the watch and collets perfectly centered aren't actually centered at all necessarily.

 

 

 

[nickelsilver]

23 hours ago, VWatchie said:

I don't even know what "the centering at the collet" means. I saw it somewhere else here on WRT but didn't get it. Obviously, I need more info about what it means, even if it isn't going to solve the problem. Is there a source of information you'd be able to recommend or can you talk a bit more about it?

 

A basic check one does when checking a hairspring is how well centered and flat it is at the collet. Out of flat is visually quite clear in the watch, out of center is harder to describe, mainly the inner 1/2 of the coils look "jumpy" as the balance oscillates. Adjustment is done with the balance out, hairspring on, in ideally a set of hairspring truing calipers. These permit a better view of the spring, and better access, than regular calipers, though those work as well. Adjustment is done on the first 180 degrees or so of the spring from the collet. The caliper is adjusted so the balance can spin with no play. Just slowly turning the balance, follow the first coil/s of the spring from its exit from the collet, and it should make a nice progressive spiral. If it is out of center, it will seem to spiral out, then cease, then out, or even go back-and-forth sort of.

 

As with everything it takes a bit of practice to figure out where and how much to adjust, but the basic rules of hairspring adjusting still apply. Adjustment in the flat is done 180 degrees from maximum error, in the round, 90 degrees. When checking the flat in the caliper, you have to focus on the first 4-5 coils, as the rest are influenced by the weight of the stud and will naturally look low. When all good, those first few coils will look distinctly flat viewed from the side, and the whole spring will look like one of the optical illusion spirals as the balance slowly spins (viewed from above).

 

The effect on timing from a spring out of true at the collet can vary from almost unnoticeable to rather extreme. I was just working on a new-build, first time on the timing machine. I saw there was a little truing needed at the collet, but ran a cycle anyway. In my case, the verticals were slow compared to the horizontal, with an apparent poise error equaling 60s. I trued it at the collet, and the difference in both h/v and the "poise error" were cut in half. Another reason not to chase dynamic poise if everything isn't as perfect as possible.

 

The effect of centering at the collet is used by good regleurs sometimes by introducing an error. Sometimes you find the verticals just want to run slow, with regulator pins adjusted perfectly etc. Like, perfect poise, but 15s slower than horizontal. Sometimes throwing the centering out at the collet by say half a spring thickness can resolve this. It is pretty much unnoticeable visually. Also, you have to guess which way to "throw it out", haha.

 

A Levin hairspring truing caliper and a Swiss simple caliper that works well too

 

 

 

 

Edited March 15 by nickelsilver