Seiko movement regulation and fixing positional accuracy

[JustBlueFish]

Hi guys, first post here. I had a question about regulating a Seiko movement (4r or 6r series), or I suppose any movement.

I've no problem adjusting the beat error and rate of the movement, but how do I deal with the positional accuracy? I have a movement that is showing good amplitude in all positions but there is 20-30 seconds difference in rate in depending on position, is that easily fixable? It would seem that if you're +20 in dial up and -10 in another position, and +10 in yet another position, the odds of getting the watch keeping accurate time constantly is futile. 

Thanks for the info and for the forums. Pleasure to be here.

[Melt]

Imagine where a watch will spend most of it's life and regulate for that. So usually left wrist crown down and bedside table dial up. Best compromise of these two positions should give good results. 

[JustBlueFish]

5 minutes ago, Melt said:

Imagine where a watch will spend most of it's life and regulate for that. So usually left wrist crown down and bedside table dial up. Best compromise of these two positions should give good results. 

That's effectively what I've done in the past, I'm just wondering if there's a better way. Can the positional variance be fixed so that all positions are basically giving the same rate, then adjust down to as close to zero as possible?

[JohnR725]

Positional errors are interesting timing machine versus on the wrist?  So the timing machine is an interesting device that shows the instantaneous rate at the time the watch is on the machine. Based on that we make a prediction of how the watch is going to do over the next 24 hours weeks months whatever. Real test is what does the watch you on your wrist over 24 hours? A lot of the positional errors will cancel out as you move the watch in various positions.

Then multiposition timekeeping takes time to do which is why typically they don't do it. For instance Omega for their non-chronometer grade watches noncertified watches three positions the same as what Seiko has. It's only when you go into their chronometer grade do they time in five positions and then their master chronometer watch it's six positions. Which is why Seiko only times in three positions it takes time and effort to do six position timing and it really isn't necessary.

So to understand where we're starting from I've attached snippets out of Seiko tech sheets. They don't appear that all of the 4R series is identical which is why have two of them and one 6R plus just because a 8r. So this gives you an idea of what Seiko thinks the watch can do.

So a subject in itself you're looking for something called dynamic poising. This assumes that the watches in good running condition and that the balance has been statically poise and everything is reasonably close to being perfect as this is not a magical fix for a watch that has problems. Then I partially did the search for you I have two links below but do a search on this group the discussion has come up more than one so there are other examples this will just get you started.

For removing weight on a balance wheel like this a carbide circuit board drill on the underside of the balance wheel will work. This way don't leave visible marks on top of the balance wheel. The etachron regulator system allows for a lot of variation so as the balance wheel gets lighter up to a point you can regulate that out.  Then you want to take notes when you're doing this is easy to get lost.

https://www.watchrepairtalk.com/topic/1222-dynamic-poising/

http://adjustingvintagewatches.com/category/dynamic-posiing/

[jdm]

I do not recommend trying poising the balance on Seiko mov.ts, especially for beginners. That is a way too difficult and advanced task, and results are not guaranteed, meaning that is a lot easier to do worse than better. Beside 7S, 4R and 6R are all mass produced mov.t s not intended for this type of adjustment, which makes better sense on a balance with Breguet hairspring and microweights on the rim.

Assuming that the hairspring has absolutely perfect geometry, simply turn slightly the Etachron regulator (there is a tool for that but reportedly it's the wrong size for Seiko) in either direction and repeatedly regulate and test dial up and crown down. A variation of 15 s/d (usually crown down is slower) is already good and fully within specifications.

Consider that conventionally the natural positional error caused by laws of physics is 10 s/d. You can somehow trick around these laws, but not beat them. And, as mentioned, the last test is on the wrist in at least a week of normal use.

[JustBlueFish]

Thanks for all the info I appreciate it.

[dhc4ever]

Well my timegrapher arrived today and I've been fiddling around with it.

I have a 2 month old Seiko SRP773 turtle reissue  with a 4R36 movement which was running a small amount slow from the factory.

It also had a 3 ms beat error which I got to 0 with a bit of adjustment and the time to 4 secs a day fast, thats with the face up.

I now know why the turbulion and carosel were invented, face up, face down, crown down, crown horizontal ALL read different times, as much as 30 secs, beat error however seems to max out at 2 ms.

So my question to those that actually know what they are doing is which position do you normally regulate a watch to and how do you regulate it to multiple positions?

Or do you get it the best you can face up and then hope the positional movements null each other out to a large degree?

Cheers

 

Pete

[Tmuir]

Unless you are a master watchmaker and putting a lot of time and effort into the watch, or it is a very high end watch you will never get it perfect in all positions.

You need to check it in all positions and try to adjust it evenly the best you can in all positions.

If you have a 30 second variation between dial up and dial down adjust it so its say 15 seconds fast dial up and 15 seconds slow dial down. If its your watch you are adjusting think about how you wear it. If you sit at a desk and use a computer all day it will spend most of it time in dial up position, so you may decide to adjust it so dial up has less error than dial down.

By checking it in all positions it can alert you to issues you didn't previously notice, for example if there is massive variation between dial up and dial down in the amplitude this could indicate an issue with the balance, maybe you didn't clean or oil one of the jewels properly or the hairspring is out of true and rubbing, or a pivot is worn.

[jdm]

On 7/12/2018 at 11:57 AM, dhc4ever said:

I now know why the turbulion and carosel were invented, face up, face down, crown down, crown horizontal ALL read different times, as much as 30 secs, beat error however seems to max out at 2 ms.

Very strange that the beat error goes up to 2ms. Should stay under 0.5ms in all positions.

Quote

 

So my question to those that actually know what they are doing is which position do you normally regulate a watch to and how do you regulate it to multiple positions?

The two most important positions are dial up and crown down. On a Seiko like your you can be happy with 10-15 s/d difference. The main way to correct this to turn the Etachron regulator pin one way or the other, say 2 deg at time until find best compromise. All this data and info is in the service sheet. But in the end is the wrist performance that counts

 

On 7/12/2018 at 1:54 PM, Tmuir said:

Unless you are a master watchmaker and putting a lot of time and effort into the watch, or it is a very high end watch you will never get it perfect in all positions.

You need to check it in all positions and try to adjust it evenly the best you can in all positions.

Nobody can completely eliminate the positional error as it's cause by laws of physic. At most, tricks can be used as in accuracy competitions. But it's mostly pointless, and just a luxury game for the mind. Just  like the writing "adjusted in XY postions", was and is manly a selling point. Ask a "master watchmaker" how do they actually do positional adjustment and the most you will get is a smile and some nice sentence like "adjustment starts at mainspring".

[JohnR725]

You were talking about the rate and the beat error you failed to mention what the amplitude was?

Then it would help if you are doing your tests correctly so I've attached an image out of the Seiko manual. So you wind the watch fully up and then you let it stabilize they recommend a minimum of five minutes up to 30 works. Then a couple of other things not mentioned when changing to various positions allow about 30 seconds to stabilize. Watches themselves are interesting in that they keep phenomenal time averaged over time. Instantaneously they fluctuate the timing machine does better if it measures the watch over time. You didn't mention which timing machine you got but if it's Chinese it powers up and defaults to a measuring time of 4 seconds. It's much better if you have a longer time like 20 seconds minimum 30 seconds would work good.

Then the term regulating to multiple positions is an amusing term.  So some things you can't change in other words  a low quality watch you're not going to regulate it to a Rolex timekeeping. But if we assume it's a normal wristwatch in good running condition dial-up and dial down should be almost identical. Then when you go to the crown/Pendant positions gravity is bad for timekeeping. This is where you can end up spending a lot of time poising the balance wheel both statically and dynamically to eliminate the positional errors as best you can. Then once the positional errors are more or less eliminated the watch should keep reasonable time in multiple positions.

 

Seiko 4R35B 4R36A.pdf

[jdm]

2 hours ago, JohnR725 said:

This is where you can end up spending a lot of time poising the balance wheel both statically and dynamically to eliminate the positional errors as best you can.

You can poise the balance all you want but that will have negligible effect on dial/pendant positional error, because as you mentioned that is mainly caused by gravity (of the hairsping), which can't be "poised". What Seiko did to win (and then be booted out) of the Swiss Chronometer competition was to attach a minuscule weight to the hairspring. They were disadvantaged at start because they were (and still are) using a flat HS with regulator, while others others had a free sprung Breguet one.

[JohnR725]

8 hours ago, jdm said:

(of the hairsping), which can't be "poised"

I've seen this before for unknown reasons silly watchmaking schools think they can teach their students to poise hairsprings? I snipped out the relevant section in the image the rest can be found at the link below.

http://raulhorology.com/2012/11/the-mark-of-a-true-watchmaker-hairsprings-part-2-vibrating-the-spring-using-a-luthy-tool/

[jdm]

4 hours ago, JohnR725 said:

I've seen this before for unknown reasons silly watchmaking schools think they can teach their students to poise hairsprings?

Going back to our normal context of "realistic hobbyist watchmaking", let me quote the sentence into which I recognize myself more in the article above:  

"things don’t go to plan it can a real nightmare and very stressful. I have lost count of the number of times I have spent hours at this point with no success"

[JohnR725]

It's not just hobbyists is all of us In watch repair knowing when to let go. The obsession for perfect is very bad for one's health.

 

[dhc4ever]

Well gents thanks for the answers and background.

I have no experience or interest in stuffing around with hair springs or poising the balance.

That would be a disaster in the making at this stage.

I am zeroing in on an acceptable level of accuracy, make adjustments on the timegrapher and then wear the watch for a few days o see whats its doing on the wrist and adjusting accordingly, not as fast a process as I had hoped.

By all means keep the discussion going its interesting even if most of it is way beyond my abilities.

[dhc4ever]

John,

Thanks for that data sheet, very useful.

Ta

[JohnR725]

One of my coworkers likes Seiko movements and is busy casing up his own watches. So I inquired and he had suitable watches for testing. Both of these are brand-new no attempt at regulation these are the way they came. Seiko is interesting in that they have their Seiko branded watches like one of these is a 4R35B but you also have other watches don't have the Seiko name like NH36A which is supposed to be equivalent to Seiko 4R36A. I find the non-Seiko documentation more interesting in that it specifies things a little nicer. Like for instance to wind up the watch requires turning the crown 55 times. In the absence of that you can turn the ratchet wheel eight times to wind the watch up. Then for testing they recommend 10 to 60 minutes after fully winding. The measuring time is 20 seconds with 20 seconds between positions which works out fine I have the machine set the 30 seconds for both. Then for both tech sheets they only recommend three position timing which is dial up, 9 o'clock up and 6 o'clock up. Basically most watches are only tested in three positions it's only when you go to a chronometer grade where your paying way more money is the number of positions they test in go up.

Testing was identical for both watches wound them up by Winding the ratchet wheel. Then nice witschi timing machine at work complete with automatic microphone so the testing is done in six positions.

Then I ran all the tests twice to show you an interesting occurrence of watches versus timing machine. Mechanical watches average their imperfections over time. Timing machine is interesting the graphical display you see instantaneously what the watches doing. Numerically the numbers like in the results below are they average of 30 seconds. Sometimes when you look at the numbers at the time of the test they may not always agree with what you see graphically because they are in average. Then run the test again and depending upon a variety of thing the numbers can change.

[jdm]

1 hour ago, JohnR725 said:

So I inquired and he had suitable watches for testing. Both of these are brand-new no attempt at regulation these are the way they came.

As you found yourself, both brand new Seiko complete watches, and SII spare mov.ts most often require regulation.

Unfortunately many buyers don't realize that they are dealing with mass produced mechanical devices, and when the watch doesn't perform like a quartz one, they are quick to return or complain on forums.

BTW, I have featured this topic.

[dhc4ever]

Well I read the seiko tech sheet John posted.

Waited a few days to see what my watch was doing on the wrist and then put it on the time grapher. It was running 11 sec a day slow.

Face down it was pretty close at -8 , 0 beat error and over 250 deg, face up I cant remember but it was also slow.

I ended up adjusting it to +12 face up, beat error didnt go above .1ms amp stayed above 250 deg.

On the wrist after 3 days its + or - 2 secs.

At this point I call that a success.

Interesting seeing the difference between the timegrapher readings and whats going on on your wrist.

Sort of answers the question "how hard can it be?"

Answer, depends.....

[jdm]

10 hours ago, dhc4ever said:

I ended up adjusting it to +12 face up, beat error didnt go above .1ms amp stayed above 250 deg.
On the wrist after 3 days its + or - 2 secs.

That is because usually pendant down is slower, so they average. For final testing, professionals use an automatic arm, that cycles all positions and re-wind the auto. Pretty cheap from China these days.

[JohnR725]

So for continuing education on timekeeping 24 hours later back on the timing machine again. As the amplitude decreases errors in the timekeeping increase especially positional errors.  Then the weird part the first watch I had on the machine one of the numbers looks really weird so I highlighted that for you. If you look at the graphical display you'll see there was a problem with the waveform? But run the test again the problems gone?

Any time I get really weird timing I take the watch to another machine at work. It does a time plot currently set to 16 minutes. What it does it lets you see the effect of the gear train power transmission fluctuations. Properly engineered watches have relatively even power distribution unless there's a wheel that has a defect. The Seiko watches seem to have a defect perhaps. The perhaps is because these are cheap watches and a defect like this will average out.

 

 

[jdm]

14 hours ago, JohnR725 said:

So for continuing education on timekeeping 24 hours later back on the timing machine again.

As you have demonstrated, 24h w/o winding is pretty much the most a 7S / 4R can take before accuracy degrades greatly (accuracy vs MS power is first form of isochronism desired in watches, that is shared by clocks also). One has to remember that these are based on an automatic-only design, meant to be worn everyday, by the working class let me add - Seiko has a long history making automatic-only movements, before the 4R added manual winding. 

One could check how  better the 6R does with it SPRON mainspring and 50h nominal power reserve. But my machine isn't good as your and there is no way I'll sit collating pictures as you did, thanks for that.

 

 

[saswatch88]

Like the majority of the replies this is a seiko I don’t even think they are timed in even three positions from the factory. And I have on so many seikos Right out of the box they are all over the place most of the time. We can all appreciate you wanting to learn dynamic poising but as stated Is a very difficult task the movement is not a Chronometer So I would think trying to adjust in six positions Will just have you pulling your hair out. Besides the numbers you posted don’t seem all that bad for a seiko. As you move your wrist it will even out. We do have a force of nature here on earth called gravity, its the unseen demon haunting every movement known to exist. Dont fight it just wear it. Hell even the moon phases will affect timing.


Sent from my iPhone using Tapatalk

[Nucejoe]

What kind of variation between FU and FD positions would prompt suspecting a worn pivot? TIA