Variation in rate

[Bonzer]

I've noticed that if I measure the rate of watches over a longer period of time, there can some considerable variations in rate. The attached screen grab shows an example of a one watch measured over 25 minutes. You can see that the maximum deviation is circa 12 seconds/day (-2 to -14).  I have worse watches!
In an ideal world, there would be a consistent rate.

My question: What could be considered a maximum acceptable variation in rate before coming to the conclusion that there is a problem with the movement?

[jdm]

12 hours ago, Bonzer said:

My question: What could be considered a maximum acceptable variation in rate before coming to the conclusion that there is a problem with the movement?

Similar questions to, how long is a string? Begin  considering that the recommended service for mechanical watches is 3 - 5 years. How many of your have been serviced that recently? How about the bit erratic one that your showing above? In 2021 and actually since the quartz revolution only the richest and most dedicated watch enthusiast duly oblige the service interval, that is set not only to prevent damage, but also to ensure the best possible timekeeping performance. That was important back when the average person only owned one watch and exact time was not universal. Now, that is not not even slightly important anymore.

In other words, even 2 minutes a day with variances in between can be acceptable for an an owner unwilling or unable to spend 150 - 200 Euros after a mechanical watch. But another, which is fixated about precision and buys top watches costing thousands, will expect his to be accurate all the time won't give rest to the watchmaker until they are. 

[Bonzer]

 

9 minutes ago, jdm said:

how long is a string?

Yes, I realize my question is a bit vague.

In the example above, the watch (after a service) is averaging a rate of -6 seconds per day. However, although the rate looks steady on a timegraph measuring for 20 seconds or so, when measured over minutes there are some quite wide variations even though it still averages at -6 s/d.

I suppose another way of putting it: For a watch that has just been serviced, would we expect there to be a solid rate with no real variation? Does the 'erratic' nature of the above graph indicate that there is a problem?

[jdm]

13 minutes ago, Bonzer said:

 

For a watch that has just been serviced,

"Just" as in 5 minutes ago, or 24 and more hours earlier? The latter is the minimum time one need to wait before final assesment and regulation.

13 minutes ago, Bonzer said:

would we expect there to be a solid rate with no real variation? Does the 'erratic' nature of the above graph indicate that there is a problem?

In your graph the rate is probaly plotted as instantaneous measurement with very close vertical bars spaced 5 seconds. That is too small an interval to judge the rate of a mechanical mechanical watch even in a single position.

Barely visible on the left is an amplitude value of 265 deg, which is good and the first indicator of general health.

[Bonzer]

3 minutes ago, jdm said:

just" as in 5 minutes ago, or 24 and more hours earlier?

As in 24 hrs before.

The bars are at 5 minutes intervals in the image shown, so it shows instantaneous measurements taken every 30 seconds.
 

[eccentric59]

Over a 24 hour period are these performance deviations regular? For example, if every 15 minutes there is a marked change in accuracy that seems to be repeating it could indicate that there is a piece of dirt or damaged tooth, jewel or even a spot where a second hand might touch enough to throw off the reading.

I would also ask about the software. Does it report the average rate between intervals or is it extrapolating an immediate rate? The problem might actually be in the reporting and not the movement.

[Bonzer]

53 minutes ago, eccentric59 said:

Over a 24 hour period are these performance deviations regular?

No, they seem pretty random

 

53 minutes ago, eccentric59 said:

I would also ask about the software. Does it report the average rate between intervals or is it extrapolating an immediate rate?

Each point plotted is an immediate rate measurement but it also reports an overall average measurement (-6s/d). When viewing the rate in 'normal' timegraph mode, you can see the rate change, both in this software and in TG. I wonder if the vibration pickup is a contributory factor to a mis-calculation of the rate.

[eccentric59]

How isolated is the pickup? Even normal speech can effect the readings. Footsteps through the floor can also add beats to a reading

[JohnR725]

9 hours ago, Bonzer said:

In an ideal world, there would be a consistent rate.

the above statement is a fantasy and cannot be achieved by mechanical watches.

Simplistic for mechanical watches is amplitude affects timekeeping.

Where do we get amplitude fluctuations from? Normally you would think of the mainspring fully wound until fully unwound there will be variations. But what about an automatic watch if it's on your wrist all day long and providing your active and when it's off your wrist it's on a auto winder then we should have consistent power.

so with an automatic watch we would have consistent power we should have consistent rate but? I'm going to assume that this watch is been recently serviced and is in good running order because if it hasn't been serviced then pour lubrication sticky lubrication dirt whatever will make all of this much worse. the problem now is the gear train from the mainspring barrel to the escapement including the escape wheel we have things the revolving. If the wheels are not perfect there out of round the manufacturing tolerances were sloppy. The same with opinion how the wheel opinion would put together. The pivots how rounded they are a non-jeweled watch for instance the holes could be ovoid shape a whole variety of things will cause the gear train the produce pulsating power. those fluctuations in power will cause fluctuation amplitude which will cause timekeeping fluctuations.

to understand the timing variations in the gear train I snipped out an image showing a gear train.

9 hours ago, Bonzer said:

averaging a rate of -6 seconds per day.

where did the six seconds per day come from?

than the problem with the average rate? Is this a wristwatch or something that remains in a stationary position? If it's a wristwatch and the numbers came from a timing machine I would assume then you time the watch in multiple positions and did the math and came up with the average?

9 hours ago, Bonzer said:

I suppose another way of putting it: For a watch that has just been serviced, would we expect there to be a solid rate with no real variation? Does the 'erratic' nature of the above graph indicate that there is a problem?

 

7 hours ago, eccentric59 said:

I would also ask about the software. Does it report the average rate between intervals or is it extrapolating an immediate rate? The problem might actually be in the reporting and not the movement.

the last two quotes are related? Let's look at this differently let's look at how does a watch company expect us to look at the problem and see how that compares with what you're looking at.

And understand what I'm saying let's look at the 6497 I snipped out a couple of thing is off the manufacturing information sheet. This is where they have all those nifty technical specifications. then you notice the timing specifications there is two columns that's because this particular manufacturing information she has two different grades labeled Standard & Elaboré. you notice for the higher grade watch the time an additional position and require closer tolerances. If it was a chronometer grade watch than typically they would time in six positions.

So you notice the manufacturers really just concerned with does their watch keep time. Then isochronism desire watch keep time from fully wound up until 24 hours later. Then even amplitude that usually is quite a obsession on this group all their concerns the of is whether it has too much when it's fully wound up. Then a minimum 24 hours later and of course the watch has to keep time.

then for the rest of my answer ill probably be I'm not sure if this is a be helpful at all but let's see what happens?

8 hours ago, Bonzer said:

No, they seem pretty random

if you're graphical display is random random is actually a good thing. reason why random is good as long as it's truthful is that it's not indicating a specific problem. If you look at the image to the gear train if one of those gears was bad you would see a fluctuation that would be a problem.

now not sure if the rest of this is not be helpful or not but let's find out. The last several images come from one watch hasn't been serviced in a very long time I don't actually know the service history. I'm guessing that I probably put lubrication on the escape wheel perhaps can't remember it may have just timed it exactly as I found it. Then yes the timing machine did get it right or I manually set it it really is 16,200 beats per hour. Timing machine is the one we haven't workI wish the Chinese would clone it because it's way too expensive from the Swiss. All of our watches are timed and six positions it works nice for diagnostic purposes. I'll skip over the effect it has on the boss who's a perfectionist freak and seeing all these numbers and wanting them to be perfect is definitely an issue.

let's look at the numbers that was one of the complaints I had above with the average of your watch is it really the average in one position or is it the average of multiple positions? So here we can see dial-up and dial down ideally they should be identical. Obviously they're not it's a problem it causes a timing problem. Even worseis usually around a 30° drop from the dial positions to the crown positions and here there is. Plus one of the crown positions is actually better than one of the dial positions that's definitely not good. So watch should be serviced

 that is to images of the graphical display where you can tell the effect of various positions because sometimes it doesn't look that bad and other times it definitely looks undesirable.

Then we get the time plot one of my favorite for troubleshooting purposes. the reason I like the time plot is if I'm seeing variation of the numbers not necessarily in the graphical display but if I start seeing number variations I go the time plot to see if there's a geartrain problem. So there's a geartrain problem that I usually don't worry about the number fluctuations. Watches are beautiful in that they average out all of this. At least they usually average at all out.

So at the time of this time plot the machine thinks the watches 23.2 seconds fast. On the amplitude is easy to see the problems were having a five-minute variation which is usually the pinion meshing with the mainspring barrel. Then another wheel has an issue that's a repeating pattern. But look what that does the timekeeping up above. Amplitude is reasonably linear and up above it is very very jagged.  so we get a very jagged instantaneous timekeeping an average of 23 seconds.

It's where you have to be careful when you're looking at a graphical display what exactly are you looking at. If you look at the numbers obviously this watch needs to be serviced  or should be serviced.

then just because the plots wonderfully time plot something that we know to be really decent. Usually really decent is a Rolex mass-produced and they keep really decent time. What so to time plot could we expect and yes I know I only get one Rolex I should probably do more. But it also shows a problem with time plots and your graphical display of how exactly it gets its numbers and how exactly does it average those numbers. Because of you look at the time plots I have the really jagged amounts depending upon where you sample and whatever you have a really horrible looking display which means what exactly?

 

[praezis]

Hi Bonzer,

seems to me, it is rather a problem of your TM software than of your watch. It does not filter out noise events well enough and so its rate calculation gets distorted.

 Typical is, one peak by a noise event is followed by a peak in the other direction after one averaging period (measuring time or however it is called by your device). In your graph we can see two such events, with a distance between opposite peaks of 2.5 s.

For peace of mind, just select a longer measuring time/ averaging period - peaks will disappear 

Frank

 

[Bonzer]

Well, I always learn a lot from your replies John, thanks for taking the time to explain and illustrate.

13 hours ago, JohnR725 said:

where did the six seconds per day come from?

than the problem with the average rate? Is this a wristwatch or something that remains in a stationary position? If it's a wristwatch and the numbers came from a timing machine I would assume then you time the watch in multiple positions and did the math and came up with the average?

No, this is one position - DD and the machine averaged it. I should have explained that.

19 hours ago, eccentric59 said:

How isolated is the pickup?

It is very poor. This is a home-made pickup. 

 

10 hours ago, praezis said:

seems to me, it is rather a problem of your TM software than of your watch

...or the pickup, as above!