How regular should I expect the rate to be?

[GPrideaux]

I've been working on a Russian Zim 2602 movement and got it from not running at all (caked in old grease and oil) to running pretty nicely. I thought I'd have issues with the mainspring being set (see pic) but I'm getting good amplitude, even after 24 hours.

My concern is that the rate as seen on timegraphing software has some regular variation. I've logged rate and amplitude and both show both a per-minute variation and a per five minutes variation, doubtless from irregularities in the third and fourth wheels. (The graphs show an hour of results recorded about 16 hours after a full wind.)

Rate:

Amplitude:

My question is whether I should expect to be able to get it running better than this? It's a 65 year old Russian watch and if I didn't have the timegrapher I would have happily re-cased it and called it job done. I might end up doing that anyway, but I'm curious what those of you who are more experienced would recommend. Should I take it apart again and try to polish up the pinions again? Or would I get negligible improvement?

If I do try to improve it, will I be able to get away with just re-cleaning the wheels and bridge, or will I need to strip and re-clean the whole movement?

 

[Neverenoughwatches]

I remember something very similar to this...a question that @VWatchieposed about a year or so ago. Again using timing softwear that was detecting rapid changes in rate. The answer was put down to micro power fluctuations from the mainspring, more so when the rate has no set pattern of change, since this mainspring is old and starting to look set...this reason could stand. If I remember in VWatchie's case the rate changes were more rapid than this. I think it's something that occurs in most movements that balance out over longer periods of time.  Setting a timegrapher time plot over a longer period you'd expect these variations to smooth out. The last 15 seconds look curious showing a regular pattern of around 5 seconds, that almost ties in with an 18.000bph escapewheel making a 6 second full revolution could the escapewheel be slightly out of round ?

[AndyGSi]

15 minutes ago, Neverenoughwatches said:

The last 15 seconds look curious showing a regular pattern of around 5 seconds

The scale is minutes.

[VWatchie]

I have some experience with Russian movements (Vostok, Raketa, Poljot), although not specifically the Zim 2602, and my impression is that Russian movements are generally not manufactured with the same precision as Swiss and Japanese movements.

A few years ago, I spent an enormous amount of time trying to make these Russian movements as precise as Swiss ones, but the reality is that it's not possible. Despite significant variations in rate, they can, of course, still be regulated to run very accurately and are therefore practically just as functional as Swiss movements.

Personally, I believe you’ve done a good job and don’t have much to gain from performing another service/inspection. The results you’re getting are probably quite close to what can be considered optimal under the circumstances.

A few years ago, I wrote a more detailed post about this, which I think you might find interesting.

 

[Neverenoughwatches]

2 hours ago, AndyGSi said:

The scale is minutes.

Ah ok Andy .I was looking at the 01 then thought the 02 was minutes and 45 was seconds

So the first and last 15 minutes recordings appear to have quite  regular accurate 5 minute cycles.

For amplitude so the rate would correlate with that.

[nevenbekriev]

Looking at amplitude graph, we can see that it can change from 290 to 250 and again to 290 in 3 sec. Do You guess what I will say here? Well, this is not true. For sure it is not real. Do You really have to use this software???

3 minutes ago, nevenbekriev said:

Looking at amplitude graph, we can see that it can change from 290 to 250 and again to 290 in 3 sec. Do You guess what I will say here? Well, this is not true. For sure it is not real. Do You really have to use this software???

OK, I se I am wrong here, as it is one hour long record, not one minute

[Kalanag]

10 minutes ago, nevenbekriev said:

Looking at amplitude graph, we can see that it can change from 290 to 250 and again to 290 in 3 sec. Do You guess what I will say here? Well, this is not true. For sure it is not real. Do You really have to use this software???

Hard to see: The scale of these graphs is day/hour/minute

Edited January 14 by Kalanag

[nevenbekriev]

Thanks, I figured it out.

Well, it is normal for the amplitude to vary a little, but if this is real, it is much. I have grown with russian movements and know them well. 2602 has small wheels with small teeth count, but should not have such big variations. One of most precise soviet movements ever is 2603 of ПЧЗ (Raketa), which differs only by shock devices presence. May be there are bad wheels here. The main suspect is 3th wheel. It will advance about 1 pinion leaf ahead for one minute

[Neverenoughwatches]

3 hours ago, VWatchie said:

I have some experience with Russian movements (Vostok, Raketa, Poljot), although not specifically the Zim 2602, and my impression is that Russian movements are generally not manufactured with the same precision as Swiss and Japanese movements.

A few years ago, I spent an enormous amount of time trying to make these Russian movements as precise as Swiss ones, but the reality is that it's not possible. Despite significant variations in rate, they can, of course, still be regulated to run very accurately and are therefore practically just as functional as Swiss movements.

Personally, I believe you’ve done a good job and don’t have much to gain from performing another service/inspection. The results you’re getting are probably quite close to what can be considered optimal under the circumstances.

A few years ago, I wrote a more detailed post about this, which I think you might find interesting.

 

Not that post you've added here H, you questioned something very similar regarding very short rate fluctuations about a year or so ago. The final opinion was mainspring power fluctuations, I'm sure John had a lot to say about it. I'll try find it.

1 hour ago, nevenbekriev said:

Looking at amplitude graph, we can see that it can change from 290 to 250 and again to 290 in 3 sec. Do You guess what I will say here? Well, this is not true. For sure it is not real. Do You really have to use this software???

OK, I se I am wrong here, as it is one hour long record, not one minute

I tbought the same Nev until Andy corrected me.

44 minutes ago, nevenbekriev said:

 The main suspect is 3th wheel. It will advance about 1 pinion leaf ahead for one minute

Might that be a 6 or 8 leaf pinion possibly resulting in a six minute revolution ? Its quite easy to see the 5 minute repeating fluctuations. 

17 minutes ago, Neverenoughwatches said:

Not that post you've added here H, you questioned something very similar regarding very short rate fluctuations about a year or so ago. The final opinion was mainspring power fluctuations, I'm sure John had a lot to say about it. I'll try find 

 

Found it !

https://www.watchrepairtalk.com/topic/28929-serviced-eta-2763-having-erratic-rate-and-amplitude/?do=findComment&comment=244717

This thread made interesting reading again. A fairly deep discussion of  gearing power fluctuations.  I take that to be the gradual meshing of teeth and pinion leaf interaction. The increase and decrease of friction as the modulus of teeth/leaf come together and then seperate. Better designed and manufactured teeth would have less friction so less fluctuations .

Edited January 14 by Neverenoughwatches

[GPrideaux]

Turns out I should have checked the watch in vertical positions before thinking about small improvements to the DU and DD running. This is what tg-timer shows when it's pendant down:

Urgh .

Different vertical positions give different results, so I let the mainspring down until I had a little less than 180 degrees amplitude and watched what happened to the rate as I rotated it. I got rates that varied between around +700 and -300. I think this means the balance is out of poise, something I am not currently equipped to correct. I might have to shelve this movement and move on to the next (another 2602).

[Neverenoughwatches]

2 hours ago, GPrideaux said:

Turns out I should have checked the watch in vertical positions before thinking about small improvements to the DU and DD running. This is what tg-timer shows when it's pendant down:

Urgh .

Different vertical positions give different results, so I let the mainspring down until I had a little less than 180 degrees amplitude and watched what happened to the rate as I rotated it. I got rates that varied between around +700 and -300. I think this means the balance is out of poise, something I am not currently equipped to correct. I might have to shelve this movement and move on to the next (another 2602).

I'm sure you understand that poising error does not affect horizontal rates. There is also a sweetspot of amplitude in vertical positions where poising errors are unrecognised, that being 220° . So heavy spots on the balance wheel produce negative and positive rates at opposing positions. Most of the variation can be reduced with static poising, though dynamic poising in the watch is more accurate. 

But anyway these rate differences seem much too high for poise error.

[nevenbekriev]

3 hours ago, GPrideaux said:

Different vertical positions give different results, so I let the mainspring down until I had a little less than 180 degrees amplitude and watched what happened to the rate as I rotated it. I got rates that varied between around +700 and -300. I think this means the balance is out of poise, something I am not currently equipped to correct. I might have to shelve this movement and move on to the next (another 2602).

What equipment You think You need? You are able to register the rate and it's differences in vertical, then You have all needed.

We need to see if the pivots are not bent and also picture of the hairspring from above first, also You can check this balance in another 2602 movement to be sure, but be brave and go ahead

[Neverenoughwatches]

3 hours ago, GPrideaux said:

I might have to shelve this movement .

I always get a little upset when I see these words . I understand why, but in a lot of cases it makes me think someone is giving up when I know that there is more they can do and more importantly such a lot that can be learnt from doing it. We just need a little nudge to keep going, like the proverbial stone.

[GPrideaux]

I appreciate the feedback and encouragement. I think what I most lack is knowledge and experience. I've seen a video or two at some point of someone using a Poising Tool, which I don't possess, but principally I think I need to read more about dealing with positional error. I'm not giving up on it, though, and while you're willing to keep offering help and encouragement (nudging) I'll keep going. I'm just not going to let it keep taking all my attention as it has the past few days.

I'll post some images of the balance and spring shortly.

Here's the balance spring from above, in motion in the running watch (with not much wind in the mainspring, hence the low amplitude).

Here's a static image with no power in the train:

A view from the side:

The imagery above is using my digital microscope. The ones below are iphone through the eyepiece of my stereo microscope.

With the balance removed, a top view:

Not perfectly concentric in places, but I wouldn't have thought that would be bad enough to explain what I'm seeing.

Pivots:

(The black ends on timing screws are where most of them have had their tops filed off at some stage.)

The pivots look ok to be, but ... am I imagining it? The impulse jewel doesn't look quite straight. It's not loose, though ... I touch it gently with tweezers and it doesn't move.

And I think that's a crack in the safety roller.

The focus isn't great (I need to learn to take better pics with the microscope) but that's definitely a crack.

 

[VWatchie]

5 hours ago, GPrideaux said:

Turns out I should have checked the watch in vertical positions before thinking about small improvements to the DU and DD running. This is what tg-timer shows when it's pendant down:

Urgh .

Looking at the numbers (I don't understand the graphs) I don't get what is "Urgh"!? So the fastest rate dial-up is about +20 seconds and the slowest rate in the pendant down position is about -20 seconds, right? So that makes a delta of about 40 seconds. I'd say that's not too excessive, perhaps even what should be expected, or there about?

5 hours ago, GPrideaux said:

Different vertical positions give different results

That too is to be expected even for a perfectly working Rolex movement.

6 hours ago, GPrideaux said:

so I let the mainspring down until I had a little less than 180 degrees amplitude and watched what happened to the rate as I rotated it. I got rates that varied between around +700 and -300.

Despite the Breguet hairspring, when the power reserve is almost completely depleted, all bets are off. However, with a good number of hours remaining in the mainspring, then yes, that seems excessive!

6 hours ago, GPrideaux said:

I might have to shelve this movement and move on to the next (another 2602).

As hard as it may be, that is sometimes the best thing you can do. Let your subconscious mind process the problem and before you know it your conscious mind will serve up a solution or something new to try out. Watch repairing should be enjoyable. Never let it bog you down. And, since you have another 2602, it might let you discover some differences that could help you with your current 2602.

Can I be so bold as to suggest that you wear the watch for a week or two (while working on the other 2602) to see how it actually fares? That is, wind it fully every morning, wear it during the day and then, if you prefer, let it rest on its back during the night. That would be very interesting! Perhaps it would prove to be a reliable timekeeper!?

[tomh207]

@GPrideaux it might just be the photo (or me) is the upper balance pivot worn? It looks smaller in diameter to me and not polished and smooth. Most I have poked around the pivots have been the same diameter both ends and, at least to me, these look different.

 

Tom

[GPrideaux]

7 minutes ago, VWatchie said:

I don't get what is "Urgh"!?

The "urgh" is a comment about the graph. A straight line shows a watch that is running at a constant rate. In the roughly two minutes shown on the graph the rate varies between about -80 and -10 seconds per day.

11 minutes ago, VWatchie said:

6 hours ago, GPrideaux said:

Different vertical positions give different results

That too is to be expected even for a perfectly working Rolex movement.

Yes, but not like this, I think. I've got more than two minutes a day difference from different vertical positions.

Some explanation of the images below:

• Top line: stats. Should be self-explanatory. I have lift angle set to 42 degrees for these.
• Left panel: graph of watch rate. Red horizontal lines are a minute apart; green horizontal lines are 10s apart. A healthy watch running at a constant rate produces a straight line here. A non-zero beat error produces a pair of parallel lines. The blue oblique lines have an angle that matches the rate shown in the top line (which is averaged over several seconds). These disappear if we're more than about 70 seconds fast or slow. The trace tracks down the screen in much the same way as if it was printing on paper tape so the most recent measurement is always at the top.
• Right panel: traces of the sounds picked up from the microphone: tic, toc and both. The last sound, when the lever strikes the banking pin, is set to zero on the upper two graphs so that the amplitude can be read off the scale to the left from the start of the first sound (when the impulse jewel contacts the fork). The third graph shows the full period, tic and toc, with the regions shown in the graphs above shaded in blue. This is particularly useful when running with less than about 100 degrees amplitude when the software will treat the second sound (unlocking?) as the first and give inflated amplitude values that must be ignored.

As well as the numbers in the top line I'm mostly looking here at the graph on the left to see how steady the rate is. I showed the pendant down results above; here are the other five positions. These were recorded about 8 hours after a full wind.

DU - not straight but not awful:

DD - a bit faster but still pretty straight:

Pendant Up - more than a minute slower than dial down, but still nice straight traces:

6 Up (or whatever we call this) - this is troubling, not just because it's around two minutes a day slower than the horizontal positions, but it's also irregular, fluctuating between about -110 and -60:

6 Down:

 

2 hours ago, VWatchie said:

Can I be so bold as to suggest that you wear the watch for a week or two

Yes, I'll do that. I was holding off because the dial has a foot missing and I'm waiting for some low temperature solder I've ordered to arrive before I attempt a repair, but I have another (uglier) dial it can use, so I've put that and the hands on, found an 18mm band and will wear it for a bit and see what happens. I've got nothing to lose!

1 hour ago, tomh207 said:

is the upper balance pivot worn?

Perhaps. It does appear smaller in the photo, but it's hard to be sure because the focus is poor and the lighting is different on the two sides. But if the pivot were really smaller than it should be I would expect that to be evident in excessive balance side shake. When I checked balance side and end shake, side shake was negligible (as in I couldn't see/feel any) and end shake was just enough that I wasn't concerned.

[praezis]

My 2ct: 
pictures are too blurred to judge the condition of the pivots. But both don‘t look good at all (worn). The upper, right, looks tapered in 2nd pic (very bad). The lower pivot looks slightly bent, if true will put the balance out of poise.

With such issues I would change the staff.

The hairspring seems not belonging to this watch: its collet was forced open unduely. Consequently all weight screws had to be filed down excessively.

I agree, it is not much fun working on such a botchered watch.

Frank
 

[fellerts]

12 minutes ago, praezis said:

The hairspring seems not belonging to this watch: its collet was forced open unduely.

4 hours ago, GPrideaux said:

And I think that's a crack in the safety roller.

Maybe the balance staff was replaced at some point, and the repairer chose a staff whose roller table arbor and balance spring collet arbor were too large? This being a non-shock protected balance, I think re-staffing following a broken pivot is likely. That could explain both of the quoted problems.

If that is the case, other dimensions might be off too...

I'm working on a movement with similar symptoms: cracked roller table and hairspring collet. Old staff had a broken pivot so I replaced the staff, but I'm worried I based my replacement on measurements from the incorrect staff. Interested to heat what the expert detectives have to say.

[Neverenoughwatches]

3 hours ago, praezis said:

My 2ct: 
pictures are too blurred to judge the condition of the pivots. But both don‘t look good at all (worn). The upper, right, looks tapered in 2nd pic (very bad). The lower pivot looks slightly bent, if true will put the balance out of poise.

With such issues I would change the staff.

The hairspring seems not belonging to this watch: its collet was forced open unduely. Consequently all weight screws had to be filed down excessively.

I agree, it is not much fun working on such a botchered watch.

Frank
 

I must admit... that collet looks well stretched over the staff.

[nevenbekriev]

I can not judge about pivots condition from this photos. The hairspring is repaired, not perfectly, but more or less OK. Not well centered (internal end to the collet). The amplitude is normal (if timegrapher readings are correct), so  friction in balance bearings is not increased. Clues like marks on the roller, stretched collet will say that the staff has been replaced. Probably staff from different factory is used and this is why the collet is stretched. So it is no wonder that there are big position errors. If the pivots are not bent and badly worn, it is worth to try poising - static and then dynamic. No poising scale? No problem at all, use tweezers. Dynamic poising - You have timegrapher, no other things needed.