Struggling with 1917 watch

graemeW · January 26

15 minutes ago, nevenbekriev said:

Which of the jewels is chipped? Is it in this movement?

Something about the hairspring. The main thing for it is to not touch anything but the regulator pins, including not touching itself. If this is fulfilled, then it may be tilted, not centered, cupped, whatever, even not centered between regulator pins, the amplitude will not suffer. Only one thing - the not centered spring will apply side force to the balance staff and thus increase friction, but this side force is so small that friction created and amplitude decreasing due it will be hard to register. The spring must be centered, flat, parallel and so on, as this is the best way to ensure it will not touch and more - it will look the best possible

This picture I posted is the chip. It's small though and doesn't seem to effect the hole. No marks on the pivot. I read a little since and I gather this is probably fine. I can't do anything about it anyway.

It's back together, doing about 220 amplitude dial up and dial down. Doing about 150-160 in any of the crown positions. It's been running all day, I'll see if it makes it to 24hrs and see what the amplitude is doing then too.

Ill go back through the escapement and hairspring soon but the bulb broke in my microscope so movement work is currently on hold.

Edited January 26 by graemeW

graemeW · January 27

I have it a full wind and left it running. It started at 220° amplitude.

25 hours later it's still running, still keeping good time, amplitude has dropped to 193.

I don't think that's too bad, am I wrong?

I mean it's all still too low, but I expected it to have either stopped, or be limping along.

9 turns to fully wind it again. Ill check at some point how many turns it takes from empty, then I can tell how much power it lost over 25 hours.

Checked it again fully wound, amplitude is up to 230. Oddly though, it was running 9 seconds a day slow after 25 hours, but fully wound it's now 130 seconds a day slow. Why is it slower now it's fully wound?

mbwatch · January 27

1 hour ago, graemeW said:

Why is it slower now it's fully wound?

At low amplitude, watches run faster. So your -130s at 230° is your watch running at its best amplitude. As it runs down and the amplitude drops closer to 180° the rate will speed up.

That said, a delta of 120s is pretty excessive for a movement of any quality. Once amplitude falls below 150 the rates will really start racing and that's the point at which you would normally see such a wide change.

Large rate swings with amplitude changes are also indicative of poise errors on the balance (heavy spots), but you won't see that yet if you are still testing this watch in a horizontal position.

Edited January 27 by mbwatch

graemeW · January 27

11 minutes ago, mbwatch said:

At low amplitude, watches run faster. So your -130s at 230° is your watch running at its best amplitude. As it runs down and the amplitude drops closer to 180° the rate will speed up.

That said, a delta of 120s is pretty excessive for a movement of any quality. Once amplitude falls below 150 the rates will really start racing and that's the point at which you would normally see such a wide change.

Large rate swings with amplitude changes are also indicative of poise errors on the balance (heavy spots), but you won't see that yet if you are still testing this watch in a horizontal position.

Thanks, interesting.

Ill keep checking everything I can.

nevenbekriev · January 27

It is possible that the hairspring is pressed to one of the regulator pins and when amplitude is higher, it gets free between pins. Another thing - it is normal to leave the movement after cleaning/oiling to work for some time (2-3 days, week) and the rate then adjusted finally, as the rate changes a little, then stabilize.

Yes, this balance for sure will need poising. The position errors are huge.

The chipped stone seems to not be serious issue.

graemeW · January 27

1 hour ago, nevenbekriev said:

It is possible that the hairspring is pressed to one of the regulator pins and when amplitude is higher, it gets free between pins. Another thing - it is normal to leave the movement after cleaning/oiling to work for some time (2-3 days, week) and the rate then adjusted finally, as the rate changes a little, then stabilize.

Yes, this balance for sure will need poising. The position errors are huge.

The chipped stone seems to not be serious issue.

Amplitude had risen to 250. Beat error is today is lower than yesterday too. 1.something yesterday, 0.5 earlier and 0.3 now.

I will check the balance poise when I go through the escapement. I have to take the hairspring off anyway.

I will rig up some kind of makeshift posing tool, probably using blades. I have some ridiculously accurate levels to set it up. I should get some idea. Whether or not I could correct any issues is another thing entirely!

I am starting to believe there is hope for this movement though, I was looking faith yesterday.

Edit...

It's getting better. Regulated it to something like good time, beat error 0.4, amplitude in the 260s. Same dial up and dial down.

In the pendant positions, amplitude drops to the 175-190 range, beat error goes up to 1.3-1.5 2 pendant positions are +112 seconds ish, 2 are -112 seconds ish.

So it's improved in general, is consistent, but just doesn't like being on its side.

Edited January 27 by graemeW

mbwatch · January 27

I would say you're doing very well with this movement. You might even just skip to dynamic poising, even though you don't have a hardware timegrapher. You'll get better overall results than statically poising with a blade tool. And the +112 / -112 spread indicates very clearly there's a poise issue.

https://adjustingvintagewatches.com/category/dynamic-poising/

Knebo · January 27

That's now a respectable amplitude at least dial up/down. The drop to horizontal is concerning, though.

It's difficult to compare your timegrapher traces to those I'm used to. But it looks a bit wavy now. Something a bit out of round?

I'm not experienced in poising, but ai think you shouldn't waste your time building a static poising tool, but rather apply dynamic poising techniques.

PS: haha, looks like @mbwatch@mbwatch and I were typing the same thing at the same time

graemeW · January 27

36 minutes ago, mbwatch said:

I would say you're doing very well with this movement. You might even just skip to dynamic poising, even though you don't have a hardware timegrapher. You'll get better overall results than statically poising with a blade tool. And the +112 / -112 spread indicates very clearly there's a poise issue.

https://adjustingvintagewatches.com/category/dynamic-poising/

35 minutes ago, Knebo said:

That's now a respectable amplitude at least dial up/down. The drop to horizontal is concerning, though.

It's difficult to compare your timegrapher traces to those I'm used to. But it looks a bit wavy now. Something a bit out of round?

I'm not experienced in poising, but ai think you shouldn't waste your time building a static poising tool, but rather apply dynamic poising techniques.

PS: haha, looks like @mbwatch@mbwatch and I were typing the same thing at the same time

Thanks for the encouragement, I wouldn't have got this far without the support of this forum!

I will read up on dynamic poising.

The balance wheel is a tiny bit wobbly, that might be why the traces are like they are, they vary in how straight they are though, sometimes they are better, sometimes worse. I put it down to the app but it might not be.

I am still concerned that the guide pin seems to brush the roller though. I am still not 100% happy with the hairspring.

I will do all the escapement checks, make 100% sure the hairpsring isn't pushing the balance off again and make sure the terminal curve is spot on.

Then, I assume, dynamic balance last.

It's been more of a challenge than I expected to be honest, but EVERYONE said don't start on antique watches. 99% of people said start on new Chinese the other 1% said start on anything post 1950s. But what can you do. I'm glad it's being a challenge, I'm glad it's forcing me to learn a lot, and learn quickly. I started this hobby as a challenge, not because I thought, or wanted, it to be easy.

Even reading the dynamic poising link. It seems like a slow process! But, it also looks quite a lot of fun. Ill definitely do it that way.

Wjat I don't understand is the mean and standard deviation numbers. The link tells you what they are, and then gives examples of numbers, but I don't understand how he is reaching the values from the numbers he gives. It really doesn't explain it well at all.

Often with these kinds of maths things, I over think it. I assume it's more complicated than it is and struggle to see the obvious. O was great at maths, but HOPELESS at algebra!

I do however understand what he is trying to achieve and how he is achieving it.

mbwatch · January 27

40 minutes ago, graemeW said:

Wjat I don't understand is the mean and standard deviation numbers. The link tells you what they are, and then gives examples of numbers, but I don't understand how he is reaching the values from the numbers he gives. It really doesn't explain it well at all.

Honestly I would not worry much about that. The long and short of dynamic poising is you let the wind out the watch and then wind it only up to a very low amplitude, <150°. At low amplitude, check all the vertical positions to find one that races ahead much faster than the others. In that position, the timing screw directly below the balance staff (the one "hanging at the bottom") is heaviest and needs to have its weight reduced or the one opposite to it made heavier. Repeat until the rate variation between pendant up/down/left/right positions is much smaller and there isn't one position which runs away much faster at low amplitude. You will have then achieved a very excellent poise.

Don't think about standard deviations, just look for a position that is way faster than the others at low amp.

graemeW · January 27

3 minutes ago, mbwatch said:

Honestly I would not worry much about that. The long and short of dynamic poising is you let the wind out the watch and then wind it only up to a very low amplitude, <150°. At low amplitude, check all the vertical positions to find one that races ahead much faster than the others. In that position, the timing screw directly below the balance staff (the one "hanging at the bottom") is heaviest and needs to have its weight reduced or the one opposite to it made heavier. Repeat until the rate variation between pendant up/down/left/right positions is much smaller and there isn't one position which runs away much faster at low amplitude. You will have then achieved a very excellent poise.

Don't think about standard deviations, just look for a position that is way faster than the others at low amp.

That was my plan.

I understood when he was saying about keeping the horizontal timing in mind when deciding whether to add or remove weight.

As a complete beginner with limited tools (well, limited watch tools, I have a 30m square workshop FULL of tools and machines) I'm not sure how I will remove weight, not in the neatest way. Bit of a bodge to just go at the screw heads. Various lightening methods are shown in my book, I'll pick the neatest that I can achieve. I have 2 lathes, but they are WAY too big for watch work!

As for adding weight, I've no idea how I'd do that. Timing washers apparently. I assume I can order those from cousins. Shame as I placed an order about an hour ago!

I assume this will merge with my last post.

Seems timing washers are hard to get. Looks like I will only be able to do this by removing weight.

I assume that means I run the risk of getting all the pendant positions the same, buy the pendant rate being different from the horizontal rate.

I'm not bothered what the overall rate is, I can 're pin the hair spring I assume.

mbwatch · January 27

36 minutes ago, graemeW said:

Seems timing washers are hard to get. Looks like I will only be able to do this by removing weight.

Yes - they come up at auction, but pocket watch washers are less common than wristwatch ("bracelet" timing washers).

The "good way" to remove weight is using a balance screw undercutter tool, which is like a set of tiny circular files that grind metal off the thread side of balance screw heads. The idea being it does not leave visible marks. Those tools are not expensive but have to be bought on ebay.

The easier way is to just take a fine file to the bottom edge of the heavy screw. Removing just a miniscule amount of metal will cause meaningful poise changes. But that method leaves a file mark on the screw, and that goes against the modern watchmaker ethos of leaving no evidence behind of your repairs (except good performance). Some frown on this method.

graemeW · January 27

3 minutes ago, mbwatch said:

Yes - they come up at auction, but pocket watch washers are less common than wristwatch ("bracelet" timing washers).

The "good way" to remove weight is using a balance screw undercutter tool, which is like a set of tiny circular files that grind metal off the thread side of balance screw heads. The idea being it does not leave visible marks. Those tools are not expensive but have to be bought on ebay.

The easier way is to just take a fine file to the bottom edge of the heavy screw. Removing just a miniscule amount of metal will cause meaningful poise changes. But that method leaves a file mark on the screw, and that goes against the modern watchmaker ethos of leaving no evidence behind of your repairs (except good performance). Some frown on this method.

I'll check and see how it looks at the moment. If they have been filed before, I won't mind doing it again.

I can cut the slots a little deeper too I guess, bevel the edges, all kinds of ways.

I don't see leaving well made marks from adjustment as an issue to be honest.

Knebo · January 27

1 hour ago, graemeW said:

Wjat I don't understand is the mean and standard deviation numbers

You're definitely over-thinking it.

The mean is just the normal average of the rates (e.g. 5 positions: -5, 0, +5, +10, +15. Sum it up: -5 + 0 + 5 + 10 + 15 = 25. Divide by number of positions: 25/5=5 is the mean rate).

Standard deviation is the average deviation (always as a positive number) from the mean: -5 deviates by 10 from the mean of +5; 0 deviates by 5 from the mean of +5;.......; +15 deviates by 10 from the mean of +5. So, the average of those deviations is (10+5+0+5+10)/5=6. That's the standard deviation.

graemeW · January 27

3 minutes ago, Knebo said:

You're definitely over-thinking it.

The mean is just the normal average of the rates (e.g. 5 positions: -5, 0, +5, +10, +15. Sum it up: -5 + 0 + 5 + 10 + 15 = 25. Divide by number of positions: 25/5=5 is the mean rate).

Standard deviation is the average deviation (always as a positive number) from the mean: -5 deviates by 10 from the mean of +5; 0 deviates by 5 from the mean of +5;.......; +15 deviates by 10 from the mean of +5. So, the average of those deviations is (10+5+0+5+10)/5=6. That's the standard deviation.

Thanks, I understand that.

I just couldn't quote grasp what the person in the link was saying.

RichardHarris123 · January 27

43 minutes ago, Knebo said:

You're definitely over-thinking it.

The mean is just the normal average of the rates (e.g. 5 positions: -5, 0, +5, +10, +15. Sum it up: -5 + 0 + 5 + 10 + 15 = 25. Divide by number of positions: 25/5=5 is the mean rate).

Standard deviation is the average deviation (always as a positive number) from the mean: -5 deviates by 10 from the mean of +5; 0 deviates by 5 from the mean of +5;.......; +15 deviates by 10 from the mean of +5. So, the average of those deviations is (10+5+0+5+10)/5=6. That's the standard deviation.

I get the standard deviation to be just over 7s.

graemeW · January 27

5 minutes ago, RichardHarris123 said:

I get the standard deviation to be just over 7s.

Oh god, don't confuse me!

Can you explain your workings?

Knebo · January 27

6 minutes ago, RichardHarris123 said:

I get the standard deviation to be just over 7s.

Yes, the true formula is that you'd square each deviation, add them up, divide by the number of positions, then take the square root. It's just above 7 then.

In my example:

10^2 + 5^2 + 0^2 + 5^2 + 10^2 = 250

Divide by 5 positions = 50

Square root of 50 = 7.07...

tomh207 · January 27

52 minutes ago, Knebo said:

Yes, the true formula is that you'd square each deviation, add them up, divide by the number of positions, then take the square root. It's just above 7 then.

In my example:

10^2 + 5^2 + 0^2 + 5^2 + 10^2 = 250

Divide by 5 positions = 50

Square root of 50 = 7.07...

It’s getting complimacated around here now with all this mafematiks, makes my head hurt

Tom

nevenbekriev · January 28

OK, it is good to understand the maths, but this is of no big use when real poising needs to be done.

The others in the thread give good advices, but I will try to summarize all shortly.

You don't need to fill all this tables with the rate data in 8 positions. Finding the heavy spot is easy - by slowly rotating the movement together with the microphone, You find the exact position where the rate is fastest while amplitude is about 150. You don't care if it is crown up or half-norhteast or whatever. Then You stop the balance, bring it to neutral position and the heavy spot is down, under the balance staff. If You are afraid to forget where the heavy is, then put a dot with marker. Of course, You have already seen how big the rate differences are, so with the experience gained, You will be able to suggest how mush material You have to remove or add. In beginning You will make small changes and checks. Knowing the rate in horizontals, You decide to remove or add material.

Now, If You deal with some Patek or Vasheron with golden screws, this is another business. Here You have simple and not expensive movement, so filing the screws is not a crime. But better choice is to drill holes in the heads with half smaller drill bit than the screw. No timing washers - use timing wire. Simply untighten the screw, make a wind of thin copper wire under it as the beginning and end are on the bottom side, then tighten the screw and cut the excess with office knife.

Static poising helps to shorten the process when the differences are big. No poising scale - use tweezers. Of course, You can skip static poising.

Have in mind that when differences are not big, the dynamic poising can be done by bending internal curve of the hairspring only, thus moving the center of mass of the spring relatively balance wheel .

Knebo · January 28

1 hour ago, nevenbekriev said:

OK, it is good to understand the maths, but this is of no big use when real poising needs to be done.

The others in the thread give good advices, but I will try to summarize all shortly.

You don't need to fill all this tables with the rate data in 8 positions. Finding the heavy spot is easy - by slowly rotating the movement together with the microphone, You find the exact position where the rate is fastest while amplitude is about 150. You don't care if it is crown up or half-norhteast or whatever. Then You stop the balance, bring it to neutral position and the heavy spot is down, under the balance staff. If You are afraid to forget where the heavy is, then put a dot with marker. Of course, You have already seen how big the rate differences are, so with the experience gained, You will be able to suggest how mush material You have to remove or add. In beginning You will make small changes and checks. Knowing the rate in horizontals, You decide to remove or add material.

Now, If You deal with some Patek or Vasheron with golden screws, this is another business. Here You have simple and not expensive movement, so filing the screws is not a crime. But better choice is to drill holes in the heads with half smaller drill bit than the screw. No timing washers - use timing wire. Simply untighten the screw, make a wind of thin copper wire under it as the beginning and end are on the bottom side, then tighten the screw and cut the excess with office knife.

Static poising helps to shorten the process when the differences are big. No poising scale - use tweezers. Of course, You can skip static poising.

Have in mind that when differences are not big, the dynamic poising can be done by bending internal curve of the hairspring only, thus moving the center of mass of the spring relatively balance wheel .

Good summary

Klassiker · January 28

1 hour ago, nevenbekriev said:

OK, it is good to understand the maths, but this is of no big use when real poising needs to be done.

Yes, the standard deviation is a useful concept when talking about bell-curve distributions of a characteristic within a population. The variation of these values is something different.

graemeW · January 28

1 hour ago, nevenbekriev said:

OK, it is good to understand the maths, but this is of no big use when real poising needs to be done.

The others in the thread give good advices, but I will try to summarize all shortly.

You don't need to fill all this tables with the rate data in 8 positions. Finding the heavy spot is easy - by slowly rotating the movement together with the microphone, You find the exact position where the rate is fastest while amplitude is about 150. You don't care if it is crown up or half-norhteast or whatever. Then You stop the balance, bring it to neutral position and the heavy spot is down, under the balance staff. If You are afraid to forget where the heavy is, then put a dot with marker. Of course, You have already seen how big the rate differences are, so with the experience gained, You will be able to suggest how mush material You have to remove or add. In beginning You will make small changes and checks. Knowing the rate in horizontals, You decide to remove or add material.

Now, If You deal with some Patek or Vasheron with golden screws, this is another business. Here You have simple and not expensive movement, so filing the screws is not a crime. But better choice is to drill holes in the heads with half smaller drill bit than the screw. No timing washers - use timing wire. Simply untighten the screw, make a wind of thin copper wire under it as the beginning and end are on the bottom side, then tighten the screw and cut the excess with office knife.

Static poising helps to shorten the process when the differences are big. No poising scale - use tweezers. Of course, You can skip static poising.

Have in mind that when differences are not big, the dynamic poising can be done by bending internal curve of the hairspring only, thus moving the center of mass of the spring relatively balance wheel .

That's made the whole thing feel a lot more approachable. Thanks.

mbwatch · January 28

7 hours ago, nevenbekriev said:

No timing washers - use timing wire.

I think I have used this tip in the past but forgot all about it! I stripped the insulation off a stranded copper electronics wire and used the fine strands to wrap as "timing washers" when I didn't have the right size on hand. I have to do this again in the next week so I'm happy for the reminder before I went digging through my washers again.

graemeW · January 29

I might get some terms wrong, but had chance to look at this more today.

Checked the lock on the pallet jewels. Couldn't quite get my head around the 2° of lock thing, but the pallets seem to lock and release at the right times. The run to the banking is there, but small, I believe that's good. The shake between the guide pin to roller and pallet fork to banking pins is good too.

Only thing I found wrong was the guide pin was a little bent. I discovered that when checking where it points in each direction in relation to the level hole when the pallet jewels release. Straightened that before checking the shake between the guile pin/roller and fork/banking. The shake is the same on both sides. Probably wouldn't have been before so straighter the pin.

I also tweeted the hair spring some more to get it more convincingly between the regulator pins while the collar is central. I think it's pretty good. Collar doesn't move at all when sweeping the regulation range.

The bottom of the balance wheel has seen some serious action!

The screw at 3 ocklock is the heavy point according to the tweezers poise set up.

The pins, and indeed the wheel itself, have had loads taken off from 12.30 to 2.

Also at 9 and 7.

At least that means when I poise it I don't have to worry about how I do it. It's had loads taken off though. Maybe I'll end up adding weight to poise it, I don't know.

But I'll try and do it dynamically once it's back together.

At least I'm happy now I've checked the escapement that there are no issues, at least none big enough to show up to a novice.

Edit, I can't poise it yet. Need to set beat error, see how it runs to make sure it's all ok, and then service it again really.

Edit

Beat error is a 2, I'll adjust that. It's about 200sd slow, but that's not an issue right now. Amplitude is back down to 200 though. It has just gone back together though. It's been apart maybe 10 times since it was cleaned and lubricated. Maybe all the oil just isn't where it should be. I'll leave it for a bit.

Watching the hair spring though, it's moving back and forth with each rotation, hitting the banking pin and the boot (is that what's it's called?) In a rather satisfying manor. This pleases me. I believe that's how it should be.

Edit again, beat error is a pain to set! I think the hairspring collar is too loose. I can gently hold the balance wheel and poke the collar with an oiled and get it to move.

Checked free oscilations and it's back down to 25 seconds.

I think strip, clean, lubricate, go through all the checks as it goes together hopefully for the last time. Then try and poise it.

Struggling with 1917 watch

Recommended Posts

Top Posters In This Topic

Popular Days

Top Posters In This Topic

Popular Days

Popular Posts

nevenbekriev

VWatchie

GPrideaux

Posted Images

Join the conversation

Recently Browsing

Topics

Posts