weird stop-and-go timegrapher readings after balance staff change and pallet stone refitting -- Vostok 2414

[Knebo]

Hi all!

Wow, with this watch, I've done several things I'd never done before.

1. I changed the balance staff (because a pivot was broken off)
2. refitted the exit pallet stone as it had come loose
(3. also corrected the hairspring, which was out of flat AND non-concentric -- but that wasn't the first time).

Obviously, lots of things that could have gone wrong and that could explain the issue I have. I had my challenges, but I think the results are good!

The watch was fully cleaned (as usual) and oiled like I had done on several 2414's before. I'm sure that's not the issue.

So, the watch actually seems to be running fine. The watch runs in all horizontal and vertical positions. 

 

But let me jump to the problem / weird timegrapher reading:

So, what's happening? The timegrapher picks up the correct beat frequency, but shows the dreaded "++++ s/d". It displays rather nice straight lines for 18 or 19 full oscillations. THEN it goes off to "beat rate auto detect". And starts again, correct beat, shows "++++", straight lines, 18-19 oscillations... OFF.... again... OFF.... again. And so on. Just to be clear, the watch/balance keeps ticking the whole time.

Also, the amplitude reading is CORRECT, as confirmed by a slow motion video, see here:

 

BIZARRE, no??

 

 

Before you ask, the balance or hairspring don't touch anywhere. The hairspring is flat...

...and concentric:

 

When fully wound down (mainspring let down), the balance comes to a stop in the almost perfect neutral position:

 

The depth of the pallet stones seems to be ok too -- maybe not perfect and it was my first time touching them (well, only the exit stone).

Here are the four fork positions (forgive me if I'm using terminology incorrectly, I was just reading up and trying to understand it last night...):

1. drop lock (impulse)

 

 

2. total lock (impulse)

 

 

3. drop lock (exit)

 

4. total lock (exit)

 

ANY IDEAS what's going on??

 

 

[LittleWatchShop]

I would put a known good watch on the timegrapher and see if it performs as expected.  If not, the timegrapher could be on the fritz.

[nickelsilver]

Locks look great, if anything a little light for such a watch. Rotate the balance by hand and check drop lock on every tooth, on both stones. After drop lock rotate the balance a few degrees more and push the fork so the horn contacts the roller jewel, the escape tooth should remain locked and not go on the impulse face.

 

Could you get a good image of the roller table and its interaction with the fork from the side (balance stationary)?

[mbwatch]

Ruling out the timegrapher:

Put a minute hand on it and let it run for several hours to find out if it is running on time vs >+999s/d

I did two Raketas recently that my Weishi would not lock onto unless the movement was directly against the metal clamp on the microphone. The movements clamped in the holder, clamped in the microphone would not hold steady readings. I think mine showed correct rates but very wrong beat error (9ms instead of 1ms)

[Knebo]

1 minute ago, mbwatch said:

Ruling out the timegrapher:

Put a minute hand on it and let it run for several hours to find out if it is running on time vs >+999s/d

 

That's exactly what I'm currently doing It's a definite possibility. 

 

8 minutes ago, LittleWatchShop said:

would put a known good watch on the timegrapher and see if it performs as expected.  If not, the timegrapher could be on the fritz

I did that. Timegrapher is ok. Other watch reads fine. 

 

8 minutes ago, nickelsilver said:

Locks look great, if anything a little light for such a watch. Rotate the balance by hand and check drop lock on every tooth, on both stones. After drop lock rotate the balance a few degrees more and push the fork so the horn contacts the roller jewel, the escape tooth should remain locked and not go on the impulse face.

 

Could you get a good image of the roller table and its interaction with the fork from the side (balance stationary)?

I'll look into this. But all teeth of the escape wheel lock. There's no slipping. But let me check better later. 

 

THANKS!!! 

[praezis]

I see two lines, wide apart. That doesn‘t match the nearly perfect beat error!

Frank

[AndyGSi]

Have you tried fixing the beat rate and see what results it shows.

[Knebo]

7 minutes ago, praezis said:

I see two lines, wide apart. That doesn‘t match the nearly perfect beat error!

Frank

Indeed, I noticed that, too! But my picture of the balance in neutral position (original post) shows that the 0.3ms seems more correct than the distance between the lines.

I'm getting the feeling that it's just running extremely fast and that the lines that look parallel are actually super-steep... I'll move the regulator to slow in a minute.

[mbwatch]

39 minutes ago, Knebo said:

I'm getting the feeling that it's just running extremely fast and that the lines that look parallel are actually super-steep.

if this turns out to be the case.

[Knebo]

UPDATE:

After an hour of monitoring, it really looks as if the minute hand completed the full turn two minutes too fast. That could imply ~+120s/h * 24 = +2880 s/d. 

I demagnetized, but it didn't help. I also put the regulator to the slowest position -- as expected, that also wasn't enough. 

Upon a closer look, it may be that the hairspring coils as sticking together a bit. I'll try to clean it again. Unfortunately, I can't do it right now... have to run do groceries and go out for dinner with my wife. 

Edited February 7 by Knebo

[Knebo]

Ok... 

It's definitely running fast. Beyond the +999s/d that the Weishi can capture. The traces look parallel, but are actually super steep @mbwatch. That also explains the mismatch between beat error 0.3ms and the distance between the traces @praezis. 

But why?? 

I demagnetized again (but since amplitude is high, I wasn't really expecting this to be an issue). 

I cleaned the hairspring again and it's not sticking. 

@nickelsilver, all the teeth of the escape wheel lock fine. It's not skipping anywhere (jf that's what could happen). I also took a slo-mo video of the balance wheel/pallet fork for several full turns and it looks correct (no skipping). 

I'm not exactly sure what I'm looking for here:

On 2/7/2025 at 4:17 PM, nickelsilver said:

Could you get a good image of the roller table and its interaction with the fork from the side (balance stationary)?

Could you explain? Maybe that will help me take a useful picture. 

 

Then, one possibility is the following. I realised that I mixed and matched balance wheel and hairspring from two movements.. both 2414A, of course. This could certainly cause issues with the rate (shorter hairspring with lighter wheel). But that much?? In a highly industrial movement like a modern-ish Vostok? @VWatchie - any experiences with this? 

[praezis]

36 minutes ago, Knebo said:

I realised that I mixed and matched balance wheel and hairspring from two movements.. both 2414A, of course.

I wonder why the essential info always comes after all tried to solve the puzzle? 

[Knebo]

27 minutes ago, praezis said:

I wonder why the essential info always comes after all tried to solve the puzzle? 

Sorry about that.... I had myself forgotten about it. There were so many things I did for the first time (balance staff, pallet stones) and did so over a period of two weeks. I only remembered when I was thinking hard now about possible causes. 

So you are convinced that this must be the issue? 

Edited February 8 by Knebo

[praezis]

39 minutes ago, Knebo said:

So you are convinced that this must be the issue? 

100%
Most members here know it, often found out the hard way

[Knebo]

Hmm. I experienced it before with a 1930s Omega T17 movement. It was like 150s/d off or so (from max position of regulator). Fixed it with timing washers. For a movement that old, it made sense to me. 

But for an industrial movement like this to be like 2000s/d off really surprises me. 

[Neverenoughwatches]

1 hour ago, Knebo said:

Hmm. I experienced it before with a 1930s Omega T17 movement. It was like 150s/d off or so (from max position of regulator). Fixed it with timing washers. For a movement that old, it made sense to me. 

But for an industrial movement like this to be like 2000s/d off really surprises me. 

It would be interesting to manually count the full oscillations for 1 minute if possible . On time would 165, but at the rate you are seeing you'd be counting almost 169.

[Knebo]

17 minutes ago, Neverenoughwatches said:

It would be interesting to manually count the full oscillations for 1 minute if possible . On time would 165, but at the rate you are seeing you'd be counting almost 169.

Seconds like fun.. I hope it won't put me to sleep 

Will try at the next opportunity. 

[VWatchie]

19 hours ago, Knebo said:

In a highly industrial movement like a modern-ish Vostok? @VWatchie - any experiences with this?

Nope, I'm afraid not 

19 hours ago, Knebo said:

I realised that I mixed and matched balance wheel and hairspring from two movements.. both 2414A, of course.

So, is this the famous "balance and hairspring must be vibrated together"? One would think that a movement that was once produced by the millions per year wouldn't need that, but I know way too little about it.

[Neverenoughwatches]

41 minutes ago, VWatchie said:

Nope, I'm afraid not 

So, is this the famous "balance and hairspring must be vibrated together"? One would think that a movement that was once produced by the millions per year wouldn't need that, but I know way too little about it.

I thought the same, 2000s/d out of time seems a lot for a mis-matched hairspring and wheel for a relatively modern mass produced watch. Though I figure the Soviet manufacture ethos was to churn them out thick and fast then re- adjust the ones that were under par. As seen with the brass balance cock shims.  My initial thoughts were an incorrect movement frequency, the smallest increment is half a beat or quarter oscillation which calculates to 66 minutes of hands movement in one hour.

[praezis]

Balance wheels and hairsprings cannot be manufactured to <0.07% tolerance (=1min/day)!

So the factory will measure hairspring (elastic moment) and balance wheels (moment of inertia) separate, sort them into classes and combine matching ones.

@Knebo mixed parts from different classes and logically got a huge error.

Frank

[Neverenoughwatches]

21 minutes ago, praezis said:

Balance wheels and hairsprings cannot be manufactured to <0.07% tolerance (=1min/day)!

So the factory will measure hairspring (elastic moment) and balance wheels (moment of inertia) separate, sort them into classes and combine matching ones.

@Knebo mixed parts from different classes and logically got a huge error.

Frank

I wondered if these classes might become further apart over long periods of production, or standard specified limits in constant use ?

[VWatchie]

On 2/7/2025 at 4:17 PM, nickelsilver said:

Locks look great, if anything a little light for such a watch. Rotate the balance by hand and check drop lock on every tooth, on both stones. After drop lock rotate the balance a few degrees more and push the fork so the horn contacts the roller jewel, the escape tooth should remain locked and not go on the impulse face.

The problem seems to have already been solved, so I hope you guys don’t mind me continuing the thread—but this test procedure really caught my attention. Thanks for describing it!  

It took some time and hands-on experimenting with a large movement (UT 6498) to fully grasp the process. Even so, I’m not entirely sure I got everything right. I’d like to try describing it in my own words—please let me know if I’ve misunderstood anything!  

First, I had to look up (or refresh my memory on) 'drop lock.' From what I now understand, this is the moment when the heel of the escape wheel tooth first contacts the locking face of the pallet stone. As the balance continues to rotate a few more degrees in the same direction, the tooth moves slightly further down the locking face. If the rotation continues even further, it will eventually reach total lock (run to banking).  

I assume the reason for advancing the balance 'a few degrees more' is to ensure the tooth ends up somewhere between drop lock and total lock—but still closer to drop lock.  

At this point, we push the fork in the opposite direction of its natural travel on its way toward total lock. If this causes the impulse face of the pallet stone to move onto the impulse face of the escape tooth, then something is wrong and needs adjustment. Instead, the tooth should remain locked, as you mentioned.  

Does this interpretation align with what you meant?

Edited February 9 by VWatchie

[Knebo]

6 hours ago, praezis said:

So the factory will measure hairspring (elastic moment) and balance wheels (moment of inertia) separate, sort them into classes and combine matching ones.

Interesting insight to the manufacturing process. 

 

6 hours ago, praezis said:

Balance wheels and hairsprings cannot be manufactured to <0.07% tolerance (=1min/day)!

1min / day would be ok. I'm seeing over 40min / day. 

 

6 hours ago, Neverenoughwatches said:

wondered if these classes might become further apart over long periods of production, or standard specified limits in constant use ?

That would be an explanation. I don't know exactly, but the two movements may be a decade apart. 

 

ANYWAY, I'll find out:

1. I'll put in a balance complete from another 2414 that I have. If that solves it, we know for sure. 

2. I also still have the matching hairspring to this balance wheel. It needs quite a bit of bending back into shape. Well, more practice to be gained! 

[nickelsilver]

16 hours ago, VWatchie said:

 

First, I had to look up (or refresh my memory on) 'drop lock.' From what I now understand, this is the moment when the heel of the escape wheel tooth first contacts the locking face of the pallet stone. As the balance continues to rotate a few more degrees in the same direction, the tooth moves slightly further down the locking face. If the rotation continues even further, it will eventually reach total lock (run to banking).  

I assume the reason for advancing the balance 'a few degrees more' is to ensure the tooth ends up somewhere between drop lock and total lock—but still closer to drop lock.  

At this point, we push the fork in the opposite direction of its natural travel on its way toward total lock. If this causes the impulse face of the pallet stone to move onto the impulse face of the escape tooth, then something is wrong and needs adjustment. Instead, the tooth should remain locked, as you mentioned.  

Does this interpretation align with what you meant?

That's right. You might see in old texts mention of "corner test"; this is checking safe lock at the moment of drop lock (and that the roller is free to continue its course), but the roller jewel is still pretty much "in" the fork slot. It's also something to check, but doing the fork horn test covers it in 99.999% of cases. The fork horn clearance should be slightly larger than the guard pin clearance, so if it's good the other is too (but you still have to check to be sure guard pin clearance is smaller).

 

I think the most comprehensive escapement set-up description I've done is here:

 

This is a nice description of the "corner test" from the Joseph Bulova School of Watchmaking course, page 250

 

 

Here's an image showing the guard pin taking over the safety action. It's a smaller clearance for two reasons: the guard pin has less contact (friction) than the fork horns, and, depending on the horn shape, if the guard pin clearance is larger than the horn clearance, the roller jewel can hit the outside of the horn, possibly stopping the watch.

Edited February 10 by nickelsilver

[nevenbekriev]

As here the conversation is about matching hairsprings/balances in russian watches, I have some information that will help to understand what is happening.

The supplier of material for hairsprings is only one for USSR/Russia watch factories and they are not able to reproduce it with small enough tolerances. The batches of coils with the profiled wire (may be it is called ribbon or tape?) every time have slightly different thickness and other parameters. The hairsprings always are produced from the wire with same shape - length, diameter, number of coils and angle between beginning/end. Thus, when the wire parameters change with the next batch, the 'strength' of the new hairsprings changes too. To solve this problem, the balance wheels are turned with different moment of inertia to meat the different hairsprings strength for every new batch. Further, within every batch, the hairspring manufactured are tested and separated to 20 groups of strength, the balances again separated in 20 groups of moment of inertia and every group of hairsprings is matched to related group of balances. As the balance wheels are easier to manufacture with desired and predictable parameters, and the hairsprings is harder to make with desired strength, they first make the hairsprings with whatever strength will be and then make balance wheels to match the hairsprings.