[iim7v7im7]

Hi,

I just received the Feb 2011 issue of WatchTime magazine which contained a review of the Chronoswiss Sirius titled "Sirius Scrutiny" by Martina Richter. There were some surprising issues with the timekeeping of the watch that they reviewed. This article serves as background to my question.

For those of you who do not get the magazine, I will quote the relevant portion:

" The movements rate results were less than ideal. Our electronic timing machine determined that the watch gained an average of 3.8 seconds per day with a fully wound mainspring. That's entrely acceptable, although the difference in rate between the various positions was 13.2 seconds. Fully wound, the Sirius ran fastest (+10.7 seconds) in the dial-up position and slowest (-2.5 seconds) in the crown-left position. After running 24 hours, it showed an average deviation of 12.7 seconds, with deviations of 21.6 seconds among the individual positions and a maximum gain of 24.3 seconds in the dial up position.

Worn on the wrist and wound once per day, the Sirius yielded better results: it gained between nine to 13 seconds. When we left the watch overnight with the dial facing up, it ran extremely fast, and the next day showed a gain of 18 to 24 seconds. These results confirmed those found on the timing machine: only when fully wound does the watch perform with acceptable rate values. When the tension in the mainspring declines (as it periodically will in a hand-sound watch), the Sirius runs fast, and especially so if it is taken off the wrist and left lying face up. This corresponds to the dial-up position on the timing machine, where the Sirius showed its greatest rate deviation of all positions. Wearing a watch on one's wrist compensaates for the greatest deviations (as do the measurements in five positions on the timing machine)and yields an average daily rate of 12.5 seconds after several weeks of wearing."
​

I am presently considering adding a Regulateur 24 (CH 1123) to my collection. It has a Chronoswiss Caliber C.112 which is similar to the Caliber C.111 used in the Sirius that was reviewed; hence my concern. They both utilize re-manufactured Marvin 700 bases. It was my understanding that Chronoswiss has been using these Calibers since 1993 (Orea Hand-Wound and now Sirius and Regulateur 24), so this is a Calber that the company should have ample experience with (e.g. it is not new). These timing results frankly surprised me. No mechanical watch that I have owned performs anywhere near these rates.

I would appreciate hearing from other Chronoswiss owners to get your perspective on this.

Thanks,

Bob

 

[rationaltime]

Hello Bob,

I have no experience with ChronoSwiss. However, your question raises some points
which could apply to other watch movements.

Here are the timing specifications for the ETA (Valjoux) 6498-2 hand wind movement
Top grade. This movement is offered only in Top and COSC grades. I guess it is similar
to the watch you are considering. I used the numbers you posted for the Sirius to
annotate the table.

While the numbers as presented do not show excellent timing performance, they do
come close to meeting the Top grade timing specifications. We are accustomed to
seeing better performance, and that illustrates an important point. The data you
posted represents a sample of one. Really, we would like to know the spread of timing
measurements made on the movements as manufactured. Manufacturers of a wide
variety of products seem reluctant to supply the spread information. Possibly many
of manufacturers do not measure or track the actual performance. Have you heard
of a "go/no gauge"? That type of gauge is often used in a manufacturing process
with no statistical performance monitoring. Often the specifications are created to
reflect the performance limits measured on a sample of product. The bottom line for
the customer is they may receive product performing near the specification limits,
and occasionally outside the limits. If the customer wants better performance they
may need to special order to (and pay extra for) tighter specifications.

You mention using the historical Marvin movement. Perhaps the movements are
modified for the off center hour shaft, but using a historical movement perhaps
implies no design changes are being made to improve the performance of the
escapement.

Thanks,
rationaltime

 

[iim7v7im7]

Hello Bob,

I have no experience with ChronoSwiss. However, your question raises some points
which could apply to other watch movements.

Here are the timing specifications for the ETA (Valjoux) 6498-2 hand wind movement
Top grade. This movement is offered only in Top and COSC grades. I guess it is similar
to the watch you are considering. I used the numbers you posted for the Sirius to
annotate the table.

While the numbers as presented do not show excellent timing performance, they do
come close to meeting the Top grade timing specifications. We are accustomed to
seeing better performance, and that illustrates an important point. The data you
posted represents a sample of one. Really, we would like to know the spread of timing
measurements made on the movements as manufactured. Manufacturers of a wide
variety of products seem reluctant to supply the spread information. Possibly many
of manufacturers do not measure or track the actual performance. Have you heard
of a "go/no gauge"? That type of gauge is often used in a manufacturing process
with no statistical performance monitoring. Often the specifications are created to
reflect the performance limits measured on a sample of product. The bottom line for
the customer is they may receive product performing near the specification limits,
and occasionally outside the limits. If the customer wants better performance they
may need to special order to (and pay extra for) tighter specifications.

You mention using the historical Marvin movement. Perhaps the movements are
modified for the off center hour shaft, but using a historical movement perhaps
implies no design changes are being made to improve the performance of the
escapement.

Thanks,
rationaltime

Rationaltime,

Thanks for your thoughtful response and I appreciate your comments. Having a background in both quality and statistics, I am very familiar with how manufacturing specifications are developed and implemented in practice. Specifications on a high production products like those used on a ETA 6498-2 top grade movement are likely be based upon historic process capability and reflective of minimally > 3 standard deviations (sigma) of assumed variation so there is a very low likelihood of improperly rejecting a "good" movement. So seeing units performing near or beyond specification limits, while possible is a statistically unlikely event, unless there are special causes of variation underpinning the result (meaning not normal variation).

For example, with a middle rate specification of +/-5 s/d on a well controlled manufacturing process, I would expect 66% of the movements produced to be within +/-1.7 s/d, and 95% of the movements to be within +/-3.4 s/d. So a result of +3.8 s/d, while possible, should be an unlikely event. The same would be true with the maximum divergence and isochronism results (even more so for this). This is why in actual practice well regulated mechanical watches perform well within these published limits even on inexpensive standard or elabore' movements.

Unlike the ETA movements that you cite, the Chronoswiss C.11 and C.112 Calibers are likely low production items. Chronoswiss makes approximately 7,000 watches in total per year and the models utilizing these Calibers represent a small fraction of the volume. The 1950's Marvin 700 base is upgraded with a Glucydur, three-sided balance, a Nivarox I flat balance spring, an Incabloc shock-absorber and the operating frequency of the Marvin 700 is increased to 21,600 vph from the original 18,000 vph rate. The pallets, pallet wheel, screws and bridges are all polished and the movement is finished with Cote Geneve grazed polish. Finally Chronoswiss regulates and tests the watches to 5 positions (similar to COSC) so I suppose that given all of this, and the price point on the watch I expected better time keeping than what was observed in the WT test.

The watch tested (A Rose Gold Sirius) is a $9,900 watch. I have not seen a WT timekeeping report before with an average deviation of 12.7 seconds, with deviations of 21.6 seconds among the individual positions and a maximum gain of 24.3 seconds (CH, dial up position) after running 24 hours.

Again, thanks for your perspective, but I hope that you understand both my curiosity, my perspective on how observed results should relate to specifications and my due diligence before investing >$5k into a new watch.


Appreciated,


Bob

 

[rationaltime]

Hello Bob,

I'm not a statistician, but here are a couple of minor points. First, let's remind
readers the ETA specification is just an example, which does not apply directly
to the Sirius.

My reading of the ETA timing specification is the target mean rate is +5 s/day.
I think ETA reflects the preference of most customers that a watch run slightly
fast rather than slightly slow. In other words, better early than late. I don't
know the spread in the numbers, but if we accept the normal distribution
and 3σ spec limits you suggest, then I think the tested rate result of +3.8 s/day
or better should occur for about half of the tested watches. I think the Sirius
showed good performance for that one calculated fully wound rate.

We should also note ETA does not specify timing divergence for a movement
at the 24 hr run point. While we might be unhappy with the measured result,
the Sirius apparently was not tested for for that performance at the factory.

My primary point to readers is not all watches show the same time keeping
results. There is a variation in the watches as delivered. While we hope for
the best, this should not be a surprise. It is part of the manufacturing process.
There has been development in mechanical watch movements over the years.
The numbers gradually improve. However, when we are using old technology
progress tends to be slow.

Bob, I think you should study the watches prior to making a purchase of
this magnitude. As we can see by the posts on the forum, members ask
questions about far less expensive watches. We support that. It seems
right to keep searching until you have enough information to make a
purchase decision. Along the way we may all learn some things from the
discussion.

I have one more thought about these watches. I suggest waiting until
after the Basel watch show next month where new models are often
introduced. Perhaps there will be a new watch model for you to consider.

Thanks,
rationaltime

 

[iim7v7im7]

Rationaltime,

Again you make some very "rational" (forgive me, I couldn't resist) remarks. Your point about running fast is also right on target and it is what drives the asymmetry of most time specifications (e.g. COSC being -4, +6 s/d). The average deviation of +3.8 s/d did not bother me whatsoever. Just looking at the 9 watch tests in the last two issues of WT the average deviation ranged from +0.3 to +4.2 s/d. But for example, the watch at +4.2 s/d had a deviation of only 6 sec.

Rate Results: (deviation in s/d when fully wound/after 24 hours)

Dial up: +10.7/+24.3
Dial down: +9.9/+19.1
Crown up: +0.8/+13.8
Crown down: 0.0/+2.7
Crown left: -2.5/+3.4
Greatest Dev: 13.2/21.6
Avg. Dev: +3.8/+12.7
​

It was the deviation between the positions and the performance after 24 hours that just struck me as odd, particularly those after 24 hours on a manual wind 46 hour power reserve watch where isochronism can be regulated.

As far as Chronoswiss's testing of their watches, below is a quote from Chronoswiss's marketing materials:

"Your watch is tested in five different positions for at least 72 hours with a machine that simulates wrist motions. The results of the rate protocol are written into a service checkbook that you receive back with your watch."
​

I suspect that they must rewind the watch every 24 hours (once daily). It is just in actual practice most owners will not wear a watch to bed and it will be place either on a dresser, night table or in a watch case. This is where these types of time keeping deviations will cause problems.

In the end, this won't keep me from buying a watch. I was just curious if there was something inherent in an old, not well distributed Marvin caliber.

Best,

Bob

​

Hello Bob,

I'm not a statistician, but here are a couple of minor points. First, let's remind
readers the ETA specification is just an example, which does not apply directly
to the Sirius.

My reading of the ETA timing specification is the target mean rate is +5 s/day.
I think ETA reflects the preference of most customers that a watch run slightly
fast rather than slightly slow. In other words, better early than late. I don't
know the spread in the numbers, but if we accept the normal distribution
and 3σ spec limits you suggest, then I think the tested rate result of +3.8 s/day
or better should occur for about half of the tested watches. I think the Sirius
showed good performance for that one calculated fully wound rate.

We should also note ETA does not specify timing divergence for a movement
at the 24 hr run point. While we might be unhappy with the measured result,
the Sirius apparently was not tested for for that performance at the factory.

My primary point to readers is not all watches show the same time keeping
results. There is a variation in the watches as delivered. While we hope for
the best, this should not be a surprise. It is part of the manufacturing process.
There has been development in mechanical watch movements over the years.
The numbers gradually improve. However, when we are using old technology
progress tends to be slow.

Bob, I think you should study the watches prior to making a purchase of
this magnitude. As we can see by the posts on the forum, members ask
questions about far less expensive watches. We support that. It seems
right to keep searching until you have enough information to make a
purchase decision. Along the way we may all learn some things from the
discussion.

I have one more thought about these watches. I suggest waiting until
after the Basel watch show next month where new models are often
introduced. Perhaps there will be a new watch model for you to consider.

Thanks,
rationaltime