Rolex service.

[Lost soul]

Variomatic your thread has arrived

[Eleven]

The effect has worn off, it's losing again this morning.

[tickious]

Many variables affect timekeeping. For example temperature. Heat makes makes the hairspring longer and so the watch will run slower, cold the opposite. And positional errors, so besides what you get upto whilst wearing it, the position you leave it in at night can make a difference.

[ExplorerII]

Could you give the watch a full 30 turns on the crown, and ensure that the watch that you are benchmarking it against is accurate to the second. DO NOT wear it for 24 hours and report back your findings please?

[Eleven]

ExplorerII said:

Could you give the watch a full 30 turns on the crown, and ensure that the watch that you are benchmarking it against is accurate to the second. DO NOT wear it for 24 hours and report back your findings please?

Happy to do that, probably after the weekend.

I don't benchmark one watch against the other in that sense, always against timeanddate.com. But I am running two other chronometers at the same time to see how they perform as a "benchmark".

[tickious]

Eleven said:

ExplorerII said:

Could you give the watch a full 30 turns on the crown, and ensure that the watch that you are benchmarking it against is accurate to the second. DO NOT wear it for 24 hours and report back your findings please?

Happy to do that, probably after the weekend.

I don't benchmark one watch against the other in that sense, always against timeanddate.com. But I am running two other chronometers at the same time to see how they perform as a "benchmark".

Talking clocks handy too 123

[PJ S]

tickious said:

Many variables affect timekeeping. For example temperature. Heat makes makes the hairspring longer and so the watch will run slower, cold the opposite. And positional errors, so besides what you get upto whilst wearing it, the position you leave it in at night can make a difference.

Whilst accurate, it's also slightly misleading – a COSC certified movement is tested in 5 positions and 3 temperatures, over 15 days, so as to reduce variances such as you describe.
If the OP's watch is not performing within -4 to +6 secs/day, then it needs regulated or something isn't working 100%, and is causing the wayward timekeeping.

I'd suggest the OP speaks with Duncan @ Genesis Watchmaking – he may be able to sort it, and bill Rolex – rather than the OP having to deal only with Rolex directly for a 3rd time.

[Eleven]

PJ S said:

tickious said:

Many variables affect timekeeping. For example temperature. Heat makes makes the hairspring longer and so the watch will run slower, cold the opposite. And positional errors, so besides what you get upto whilst wearing it, the position you leave it in at night can make a difference.

Whilst accurate, it's also slightly misleading – a COSC certified movement is tested in 5 positions and 3 temperatures, over 15 days, so as to reduce variances such as you describe.
If the OP's watch is not performing within -4 to +6 secs/day, then it needs regulated or something isn't working 100%, and is causing the wayward timekeeping.

I'd suggest the OP speaks with Duncan @ Genesis Watchmaking – he may be able to sort it, and bill Rolex – rather than the OP having to deal only with Rolex directly for a 3rd time.

I am hoping that this time it genuinely is just a matter of regulation, however if it is I am reluctant to send it back to Rolex if it is again going to be away weeks and then come back either damaged, still not keeping time or both.

I've sent watches off to other manufacturers in the past for service or regulation and had no problems at all in terms of condition or performance. Though these have tended to be watches with modified ETA movements - are they inherently easier to get right perhaps?

[GC8]

tickious said:

Eleven said:

ExplorerII said:

Could you give the watch a full 30 turns on the crown, and ensure that the watch that you are benchmarking it against is accurate to the second. DO NOT wear it for 24 hours and report back your findings please?

Happy to do that, probably after the weekend.

I don't benchmark one watch against the other in that sense, always against timeanddate.com. But I am running two other chronometers at the same time to see how they perform as a "benchmark".

Talking clocks handy too 123

Timeis is better, but I would suggest a GPS time app for your smartphone, or referring to a GPS receiver which displays UTC, GPS week and week second.

[Variomatic]

Lost soul said:

Variomatic your thread has arrived

Eleven][... said:

yesterday I decided to reset the watch because it was nearing 40 seconds slow. Since then it has gained time! Not hugely, but now it's a second fast.

[...] what intrigues me is why stopping the watch and restarting it has made it go from a loss to a gain. Anyone technical care to venture an answer?

The simple answer is lot of things - some related to the "stopping and starting" and some coincidental. Never rule out coincidence!

Setting the time will affect the balance amplitude, and it may not stabilise again for quite a while. If the watch has a bad isochronal error (variation of timekeeping wih amplitude) then that can cause a few seconds gain or loss in a matter of an hour or so after setting, which then settles back to its "normal" rate. If you're checking after a day then that initial gain or loss appears to be part of its normal rate. It'd be unusual for that to make an 8 sec / day difference (it would have to have gained about 8 seconds after setting) but not impossible.

If the watch has a bad positional error then it's usually hopeless to try and work out its rate in use because it will vary (possibly by a lot) depending on what you're doing that day. It might keep great time through the week while you're going to the office doing the same-old-same-old, gain badly on a Saturday when you're on your yacht, then loose time in church on Sunday. That's why COSC testing involves steady positions for fixed lengths of time. If this is what's affecting yours then it'd be coincidence that it varied after you set the time, but it's still a fault for a newly serviced chronometer.

Web based GMT resources are also not 100% reliable thanks to latency - you're actually better off using a good local quartz clock or watch that you know the timekeeping of. Choose your "standard" piece and set it against a web source. Then check it after a month or so and work out its daily rate. That way any internet related inaccuracies are averaged out over a month and become unimportant for any practical matters. Or invest in a cheap r/c digital alarm clock - less than a tenner if you hunt and they'll be within a second at all times.

If you want to get some idea of where the problem may lie you can test reasonably accurately yourself:

Wind it fully and set the time. Leave it sitting dial up and undisturbed. Check it after 4 hours and note any difference in sec / day. Check it again after 12 hours, and again after 24. Compare the difference between the "top", "middle" and "bottom" parts of the wind. You should find that the first two are pretty similar but the last might tail off a bit in either direction.

If there's a big difference (more than a couple of seconds a day) between the "top" and "middle" then it's liable to be noticeable in use - you'll reach the middle of the power range after a couple of hours of watching TV, sitting at a desk, or overnight. In daily use you'll probably never reach the low power range but it's useful to have an idea of how it affects your particular watch.

Now wind it fully again and leave it on its side "crown down". Don't reset it for this - just remember to allow for the error from the first day Take a 24 hour reading and compare it with the full 24 hour reading from the first day. Any more than about 6 - 8 sec / day difference between these two suggests a positional error that's on the high side for a chronometer.

A full COSC test basically follows that pattern but with rates averaged over several days in each position, and with carefully controlled temperatures. They also don't take partial daily readings because isochronal errors will simply cause a fail (they increase positional errors because the balance amplitude is naturally different in each position) and the COSC test isn't designed to diagnose why it fails.


eta: This is a lot easier with a timing machine btw, becase you can take spot readings of daily rate at various levels of wind and position instead of waiting for hours and doing the sums I know a few c ollectors who have no intention at all of opening their watches but have bought one of the Chinese timegrapher units off ebay and now wouldn't live without it!

Edited by Variomatic on Saturday 5th July 12:45

[tickious]

https://play.google.com/store/apps/details?id=com....

[Eleven]

Variomatic said:

The simple answer is lot of things - some related to the "stopping and starting" and some coincidental. Never rule out coincidence!

Setting the time will affect the balance amplitude, and it may not stabilise again for quite a while. If the watch has a bad isochronal error (variation of timekeeping wih amplitude) then that can cause a few seconds gain or loss in a matter of an hour or so after setting, which then settles back to its "normal" rate. If you're checking after a day then that initial gain or loss appears to be part of its normal rate. It'd be unusual for that to make an 8 sec / day difference (it would have to have gained about 8 seconds after setting) but not impossible.

If the watch has a bad positional error then it's usually hopeless to try and work out its rate in use because it will vary (possibly by a lot) depending on what you're doing that day. It might keep great time through the week while you're going to the office doing the same-old-same-old, gain badly on a Saturday when you're on your yacht, then loose time in church on Sunday. That's why COSC testing involves steady positions for fixed lengths of time. If this is what's affecting yours then it'd be coincidence that it varied after you set the time, but it's still a fault for a newly serviced chronometer.

Web based GMT resources are also not 100% reliable thanks to latency - you're actually better off using a good local quartz clock or watch that you know the timekeeping of. Choose your "standard" piece and set it against a web source. Then check it after a month or so and work out its daily rate. That way any internet related inaccuracies are averaged out over a month and become unimportant for any practical matters. Or invest in a cheap r/c digital alarm clock - less than a tenner if you hunt and they'll be within a second at all times.

If you want to get some idea of where the problem may lie you can test reasonably accurately yourself:

Wind it fully and set the time. Leave it sitting dial up and undisturbed. Check it after 4 hours and note any difference in sec / day. Check it again after 12 hours, and again after 24. Compare the difference between the "top", "middle" and "bottom" parts of the wind. You should find that the first two are pretty similar but the last might tail off a bit in either direction.

If there's a big difference (more than a couple of seconds a day) between the "top" and "middle" then it's liable to be noticeable in use - you'll reach the middle of the power range after a couple of hours of watching TV, sitting at a desk, or overnight. In daily use you'll probably never reach the low power range but it's useful to have an idea of how it affects your particular watch.

Now wind it fully again and leave it on its side "crown down". Don't reset it for this - just remember to allow for the error from the first day Take a 24 hour reading and compare it with the full 24 hour reading from the first day. Any more than about 6 - 8 sec / day difference between these two suggests a positional error that's on the high side for a chronometer.

A full COSC test basically follows that pattern but with rates averaged over several days in each position, and with carefully controlled temperatures. They also don't take partial daily readings because isochronal errors will simply cause a fail (they increase positional errors because the balance amplitude is naturally different in each position) and the COSC test isn't designed to diagnose why it fails.


eta: This is a lot easier with a timing machine btw, becase you can take spot readings of daily rate at various levels of wind and position instead of waiting for hours and doing the sums I know a few c ollectors who have no intention at all of opening their watches but have bought one of the Chinese timegrapher units off ebay and now wouldn't live without it!

Edited by Variomatic on Saturday 5th July 12:45

Thanks for a thorough reply Variomatic. I will try what you suggest but am going to let it run on my wrist for a few days. It seems to have settled down to a loss of 1-3 seconds per day. I will see if it settles still further.

The loss has tended to be less than 3, with an 8 and 5 on consecutive days about a week in.

[Eleven]

Eleven said:

Variomatic said:

The simple answer is lot of things - some related to the "stopping and starting" and some coincidental. Never rule out coincidence!

Setting the time will affect the balance amplitude, and it may not stabilise again for quite a while. If the watch has a bad isochronal error (variation of timekeeping wih amplitude) then that can cause a few seconds gain or loss in a matter of an hour or so after setting, which then settles back to its "normal" rate. If you're checking after a day then that initial gain or loss appears to be part of its normal rate. It'd be unusual for that to make an 8 sec / day difference (it would have to have gained about 8 seconds after setting) but not impossible.

If the watch has a bad positional error then it's usually hopeless to try and work out its rate in use because it will vary (possibly by a lot) depending on what you're doing that day. It might keep great time through the week while you're going to the office doing the same-old-same-old, gain badly on a Saturday when you're on your yacht, then loose time in church on Sunday. That's why COSC testing involves steady positions for fixed lengths of time. If this is what's affecting yours then it'd be coincidence that it varied after you set the time, but it's still a fault for a newly serviced chronometer.

Web based GMT resources are also not 100% reliable thanks to latency - you're actually better off using a good local quartz clock or watch that you know the timekeeping of. Choose your "standard" piece and set it against a web source. Then check it after a month or so and work out its daily rate. That way any internet related inaccuracies are averaged out over a month and become unimportant for any practical matters. Or invest in a cheap r/c digital alarm clock - less than a tenner if you hunt and they'll be within a second at all times.

If you want to get some idea of where the problem may lie you can test reasonably accurately yourself:

Wind it fully and set the time. Leave it sitting dial up and undisturbed. Check it after 4 hours and note any difference in sec / day. Check it again after 12 hours, and again after 24. Compare the difference between the "top", "middle" and "bottom" parts of the wind. You should find that the first two are pretty similar but the last might tail off a bit in either direction.

If there's a big difference (more than a couple of seconds a day) between the "top" and "middle" then it's liable to be noticeable in use - you'll reach the middle of the power range after a couple of hours of watching TV, sitting at a desk, or overnight. In daily use you'll probably never reach the low power range but it's useful to have an idea of how it affects your particular watch.

Now wind it fully again and leave it on its side "crown down". Don't reset it for this - just remember to allow for the error from the first day Take a 24 hour reading and compare it with the full 24 hour reading from the first day. Any more than about 6 - 8 sec / day difference between these two suggests a positional error that's on the high side for a chronometer.

A full COSC test basically follows that pattern but with rates averaged over several days in each position, and with carefully controlled temperatures. They also don't take partial daily readings because isochronal errors will simply cause a fail (they increase positional errors because the balance amplitude is naturally different in each position) and the COSC test isn't designed to diagnose why it fails.


eta: This is a lot easier with a timing machine btw, becase you can take spot readings of daily rate at various levels of wind and position instead of waiting for hours and doing the sums I know a few c ollectors who have no intention at all of opening their watches but have bought one of the Chinese timegrapher units off ebay and now wouldn't live without it!

Edited by Variomatic on Saturday 5th July 12:45

Thanks for a thorough reply Variomatic. I will try what you suggest but am going to let it run on my wrist for a few days. It seems to have settled down to a loss of 1-3 seconds per day. I will see if it settles still further.

The loss has tended to be less than 3, with an 8 and 5 on consecutive days about a week in.

Okay, so I have worn the watch for a while and timekeeping has gone from a loss to a significant gain. I have tested as Variomatic instructed and the results are below.

Since returning from Rolex, for the third time, timekeeping has looked like this (Rolex told me it was running +2):

Date Day Loss / Gain
18/6/14 1 0
19/6/14 2 -1
20/6/14 3 -2
21/6/14 4 -3
22/6/14 5 0
23/6/14 6 -2
24/6/14 7 0
25/6/14 8 -2
26/6/14 9 -3
27/6/14 10 -3
28/6/14 11 -8
29/6/14 12 -5
30/6/15 13 -3
1/7/14 14 -1
2/7/14 15 -3
3/7/14 16 -2
Reset to zero
4/7/14 17 -1
5/7/14 18 -2
6/7/14 19 -3
7/7/14 20 -2
8/7/14 21 -1
9/7/14 22 0
10/7/14 23 -1
11/7/14 24 0
Reset to zero
12/7/14 25 0
13/7/14 26 0
14/7/14 27 0
15/7/14 28 -1
16/7/14 29 +1
17/7/14 30 +1
18/7/14 31 +2
19/7/14 32 +1
20/7/14 33 +2
21/7/14 34 +1
22/7/14 35 +1
23/7/14 36 +2
24/7/14 37 +1
25/7/14 38 +1
Reset to zero
26/7/14 39 +7
27/7/14 40 +7

Variomatic's testing:

Reset to zero to test resting dial up
After 4 hrs +3
After 12 Hrs +6 (+3 +3)
After 24 Hours + 12 (+6 +6)

Wound fully and rested crown down for 24 hrs
After 24 hours +7

So, initially the watch was running slow but within COSC, it then corrected to an acceptable gain within COSC, but stopping and re-starting the watch has made it run much too fast in both dial up and crown down positions.

Variomatic, what do you think is going on?

[Lost soul]

Get it sent off to Variomatic for a service , he is the man who can

[Jules360]

I am astonished that 2 seconds a day is so important that you spend 6 full weeks testing the watch. As long as i know it's "about 6:20", that's usually good enough for me.

I can only conclude that you have too much time on your hands (or towards the start of the test period, too little)

[Eleven]

Lost soul said:

Get it sent off to Variomatic for a service , he is the man who can

It's just been serviced by Rolex. I shouldn't really need to pay an independent to service it again quite so quickly.

[Eleven]

Jules360 said:

I am astonished that 2 seconds a day is so important that you spend 6 full weeks testing the watch. As long as i know it's "about 6:20", that's usually good enough for me.

I can only conclude that you have too much time on your hands (or towards the start of the test period, too little)

2 seconds isn't important. If the watch is 2 seconds adrift either way it's fine. But if it is inconsistent or gaining 7 seconds a day as it is currently it is a problem. Two reasons for this; Firstly the watch is a chronometer and should be running at an average of -4 to +6 seconds per day as well as being consistent (especially as it has just been serviced by the manufacturer). Secondly I bought the watch for travelling and it needs to be fairly accurate and consistent. Buying a COSC rated watch and keeping it serviced should ensure this.

It's a question of horses for courses really. I am wearing a quality watch today that isn't a chronometer and it hasn't been recently serviced. I don't much care if it's a minute or so a day inaccurate. Accuracy is not what it's about, so it doesn't matter.

[Variomatic]

It's a tricky one with those figures - they're not within spec, but they're not really bad enough to suggest a particular fault. Apart from the odd -8 and -5 (which could be freak values), it could be regulated back a couple of seconds to leave it just about within COSC spec overall.

The isochronism is excellent with a steady gain acrosss the whole power range but the 5 s/day positional error is a little on the high side. Could be he balance very slighly out of poise, the hairspring slightly off centre, or almost negligible wear on a balance pivot.

If the first 40 days were measured in use rather than static it could be enough to account for most of the variation.

[Eleven]

Variomatic said:

It's a tricky one with those figures - they're not within spec, but they're not really bad enough to suggest a particular fault. Apart from the odd -8 and -5 (which could be freak values), it could be regulated back a couple of seconds to leave it just about within COSC spec overall.

The isochronism is excellent with a steady gain acrosss the whole power range but the 5 s/day positional error is a little on the high side. Could be he balance very slighly out of poise, the hairspring slightly off centre, or almost negligible wear on a balance pivot.

If the first 40 days were measured in use rather than static it could be enough to account for most of the variation.

Yes, the first 40 days were on the wrist. The tests you suggested were obviously static.

When the watch went from a loss to a gain I assumed that it had settled down to the rate that Rolex had set it at, and that would be it. That it has speeded up significantly after restarting has been disappointing.

I put it in the safe "12 down" this morning to test another position but it fell over to "dial down". I checked it this evening and it's gained 9 seconds after 10 hours. I'll check in the morning to see what it's done after 24 hours.

It's looking like the watch is going to be going back to Rolex, again, doesn't it!

A side note. I picked up a K series GMT II a couple of days ago which was serviced at Rolex within a month of the watch in question here. It's been running +10s/24hrs dial up for the past 2 days....

[Grandfondo]

I have a Daytona and a Tag chronograph for everyday and they both keep st time!!!
I bought a Cheap Seiko when the Tag was away for a service a couple of years ago and pulled it out of a drawer after not wearing it for months it was bang on whereas the "quality chronometers" are minutes out after a few days.