My Spring Drive accuracy report

[RPF]

Others may disagree but my answer is power. A TC requires a thermistor or a second quartz crystal. The circuitry is in nanowatt territory on the SD. The PR is already 72 hours on a single barrel, way better than most mechanical movements.

[South Pender]

Others may disagree but my answer is power. A TC requires a thermistor or a second quartz crystal. The circuitry is in nanowatt territory on the SD. The PR is already 72 hours on a single barrel, way better than most mechanical movements.

That's a good bet, I think. I started a thread months ago about just this question, and the response was illuminating. The consensus seemed to be that no likely SD buyer would care about this level of accuracy and would be very happy with typical quartz accuracy from what s/he saw as a fundamentally mechanical watch.

[Catalin]

Others may disagree but my answer is power. A TC requires a thermistor or a second quartz crystal. The circuitry is in nanowatt territory on the SD. The PR is already 72 hours on a single barrel, way better than most mechanical movements.

Well, very few will disagree with that (even if some progress WAS done during the last 10-20 years on ultra-low-power and I believe TC might be doable today a decent entry-level around 20 spy )

The other reason is that the SD is not marketed by Seiko as a very high-end quartz but more like the most accurate mechanical (with a corresponding price) - they are basically trying to avoid / delay the fate of the original quartz - but is only a matter of time until the chinese will do something very similar first at 500$ and then at 200$ ... the strong point is that SD really does not need so much the extreme finish of the highest-end mechanicals to achieve its precision so a really honest higher-end chinese will work perfectly as well!

[RPF]

The other reason is that the SD is not marketed by Seiko as a very high-end quartz but more like the most accurate mechanical (with a corresponding price) - they are basically trying to avoid / delay the fate of the original quartz - but is only a matter of time until the chinese will do something very similar first at 500$ and then at 200$ ... the strong point is that SD really does not need so much the extreme finish of the highest-end mechanicals to achieve its precision so a really honest higher-end chinese will work perfectly as well!

I doubt it because it won't sell well. Most people don't know what's a spring drive (or even a mechanical watch). The cost of reverse engineering is also substantial for a product that most people do not care for.

The Chinese will laugh their way to the bank with archaic mechanical movements, including the 2824 which can be copied wholesale with impunity.

I don't anticipate any demand for SD-like movements for years to come when mechanicals are still king of the hill and considered sexy.

The SD was a point of pride to Seiko. IMHO, it's very difficult to make something to that spec, much harder than making a say, 9Sxx mechanical. It'll probably cost more to make outside of Seiko's facilities, due to the absolute quality required. The outstanding patents (around 300) are a big problem too.

Sell a budget version and there'll be no market for it. I suspect it'll be easier to make a smooth sweep quartz with stepper instead.

People don't really care for mechanicals with quartz accuracy.

[South Pender]

Let me borrow a graph from Dwjquest.



At 96 F and 55 F degrees, the non-TC 1538 has approx. zero error and outperforms the TC 9Fxx.

We can't tell anything from readings done at a single temperature, which is what this data is all about. We're seeing good calibration, because the cluster size is small.

As for the watch worn on wrist, we can 1) take it out of the data set and 2) retime it on the winder. At the same time, a control should be established with 1) a TC quartz and 2) a non-TC quartz. This should help us test the hypothesis SD is thermally insensitive. As it is, there is insufficient evidence. Ideally, a plot similar to Dwjquest's will settle the debate conclusively.

And you're confusing performance with spec. +-10s/y is 18x tighter than +-15s/m. These are the specs for the 9Fxx/other TC and various SDs as printed on literature.

Yes, you're right. I was focusing on empirically-determined performance, rather than published specs. I now see (as Catalin pointed out) that the 18:1 came from the ratio of specs. In my opinion, though, empirical evidence is more interesting than specs. It seems plausible to me that Seiko set the SD spec to ± 15 sec./month because they knew that potential buyers weren't interested in HEQ-level accuracy, but were fascinated with the idea of quartz accuracy from what they saw as a basically mechanical movement. Setting the spec to ± 15 sec./month precludes any disappointment and potential returns because Seiko knows that the SDs will greatly outperform this default spec. Thus, I don't feel that published specs are very useful, and that carefully-gathered empirical evidence is much to be preferred. The evidence that ptran provided allows better estimation of the true performance of the SD movement, and use of simple inferential techniques will allow us--with a little more data--to nail down the important aspects of the distribution of SD rate deviations.

As for dwjquest's graph, it would seem--from your earlier logic that a watchmaker will set a movement to demonstrate zero deviation at wrist temperature--that Omega considered one of the temperatures of 55°F and 96°F wrist temperature. But neither is that. Wrist temperature is somewhere in the 84°F-86°F range, certainly not 96°F. Furthermore, although the data in dwjquest's graph is from one watch, they suggest far greater rate variability as a function of temperature than ptran's data suggest for the seven SDs. It's true that, with the latter data, we are not examining rate variation of a single watch as a function of temperature change, but the small difference between the one worn watch, on the one hand, and the average of the 6 on the winders, on the other, provides us with information--albeit from a different perspective--about the same phenomenon.

[RPF]

Yes, you're right. I was focusing on empirically-determined performance, rather than published specs. I now see (as Catalin pointed out) that the 18:1 came from the ratio of specs. In my opinion, though, empirical evidence is more interesting than specs. It seems plausible to me that Seiko set the SD spec to ± 15 sec./month because they knew that potential buyers weren't interested in HEQ-level accuracy, but were fascinated with the idea of quartz accuracy from what they saw as a basically mechanical movement. Setting the spec to ± 15 sec./month precludes any disappointment and potential returns because Seiko knows that the SDs will greatly outperform this default spec. Thus, I don't feel that published specs are very useful, and that carefully-gathered empirical evidence is much to be preferred. The evidence that ptran provided allows better estimation of the true performance of the SD movement, and use of simple inferential techniques will allow us--with a little more data--to nail down the important aspects of the distribution of SD rate deviations.

As for dwjquest's graph, it would seem--from your earlier logic that a watchmaker will set a movement to demonstrate zero deviation at wrist temperature--that Omega considered one of the temperatures of 55°F and 96°F wrist temperature. But neither is that. Wrist temperature is somewhere in the 84°F-86°F range, certainly not 96°F. Furthermore, although the data in dwjquest's graph is from one watch, they suggest far greater rate variability as a function of temperature than ptran's data suggest for the seven SDs. It's true that, with the latter data, we are not examining rate variation of a single watch as a function of temperature change, but the small difference between the one worn watch, on the one hand, and the average of the 6 on the winders, on the other, provides us with information--albeit from a different perspective--about the same phenomenon.

Re: Omega. I was saying ptran's data (as well as my own anecdotal evidence from physical SD specimens owned by others, including friends) indicate a pretty good rate on the wrist, which suggests calibration. So far, many report 1-2 s/m accuracy. They worst I heard was around 4-5 s/m. That's very good for a watch spec-ed to +-15s/m.

I didn't say Omega calibrates their watches to wrist temperature. In fact, they don't even tell you the spec-ed accuracy up front. Note however that the Omega 1538 remained within spec for the temperature band. That's all they're required to deliver, for a movement spec-ed to a ~30s/m band within a 30C band. I.E., 1s/m per degree C.

One more thing. Wrist temperature depends on environmental factors. For example, wearing a watch in the death of winter and on a sunny hot day results in different on-wrist temperatures. 84-86F is too tight a band, and must have been specific to certain conditions. But that's a minor point.

Re: SD outperformance. I don't believe that. I trust specs more. It's unreasonable to expect better because the physics of a non-TC tuning fork quartz at 32khz is a fixed parabola. Unless Seiko is employing TC, there is no reason to expect anything other than that, based on what we have uncovered on this forum. We may see tight calibration at a common temperature but that's about it. 1s/m per degree is the limits of physical and human law.

[South Pender]

Re: Omega. I was saying ptran's data (as well as my own anecdotal evidence from physical SD specimens owned by others, including friends) indicate a pretty good rate on the wrist, which suggests calibration. So far, many report 1-2 s/m accuracy. They worst I heard was around 4-5 s/m. That's very good for a watch spec-ed to +-15s/m.

I didn't say Omega calibrates their watches to wrist temperature. In fact, they don't even tell you the spec-ed accuracy up front. Note however that the Omega 1538 remained within spec for the temperature band. That's all they're required to deliver, for a movement spec-ed to a ~30s/m band within a 30C band. I.E., 1s/m per degree C.

One more thing. Wrist temperature depends on environmental factors. For example, wearing a watch in the death of winter and on a sunny hot day results in different on-wrist temperatures. 84-86F is too tight a band, and must have been specific to certain conditions. But that's a minor point.

Re: SD outperformance. I don't believe that. I trust specs more. It's unreasonable to expect better because the physics of a non-TC tuning fork quartz at 32khz is a fixed parabola. Unless Seiko is employing TC, there is no reason to expect anything other than that, based on what we have uncovered on this forum. We may see tight calibration at a common temperature but that's about it. 1s/m per degree is the limits of physical and human law.

Good points. However, I do wonder why Omega would choose to calibrate their quartz movements to zero deviation at temperatures that will never be experienced. I would have thought that Seiko's practice in this respect would be the industry norm. Calibration that would lower the parabola would seem to be the key to much-improved performance at the more usual temperature points--wearing and storing. Your other point about the limits of quartz technology perhaps ignores the fact that there can be variations in the quality of the quartz oscillator (including pre-aging) and, particularly, the way and care with which it's cut. Since SD is Seiko's current pride and joy, and prices are high, is it not possible that they have selected only the best quartz crystals for them, thus improving SD performance beyond standard ± 15 sec./month quartz levels?

[Catalin]

...
The SD was a point of pride to Seiko. IMHO, it's very difficult to make something to that spec, much harder than making a say, 9Sxx mechanical. It'll probably cost more to make outside of Seiko's facilities, due to the absolute quality required. The outstanding patents (around 300) are a big problem too.

Sell a budget version and there'll be no market for it. I suspect it'll be easier to make a smooth sweep quartz with stepper instead.

People don't really care for mechanicals with quartz accuracy.

I disagree on the first two points - it MIGHT have been difficult to make initially when the ultra-low power was indeed a challenge, but today that is no longer the case.

And as for your opinion on 9S being easier to build MECHANICALLY - I can only say one thing - the Seiko marketing is waaaay too effective on you

The truth is that Seiko has chosen for MARKETING reasons to pair the quartz elements with rather high-end movements to (partially) justify the price - HOWEVER the mechanical part could just as well be VASTLY inferior with very similar final results (and in case you wonder why that sounds familiar - this is just like when the inexpensive japanese quartz has partially destroyed the original over-expensive swiss quartz business model in which the quality needed for a very good product was actually inflated for marketing reasons).

Regarding the last ideas - marketing and availability are strong 'weapons' against weak minds - that is the MAIN reason why we have seen the diminishing interest in quartz and the interest in mechanicals - but make no mistake, that has little to do with technical achievements!

[Catalin]

...
Calibration that would lower the parabola would seem to be the key to much-improved performance at the more usual temperature points--wearing and storing. Your other point about the limits of quartz technology perhaps ignores the fact that there can be variations in the quality of the quartz oscillator (including pre-aging) and, particularly, the way and care with which it's cut. Since SD is Seiko's current pride and joy, and prices are high, is it not possible that they have selected only the best quartz crystals for them, thus improving SD performance beyond standard ± 15 sec./month quartz levels?

The process of flattening the parabola would be more like 'adjusting for temperature' and in most designs that I am aware would be a rather VERY tricky procedure.

The much simpler standard calibration just involves shifting pretty much the same parabola up or down.

I really do not see even Seiko using a quartz cut in SD that is much more advanced than what they already have - and they only have (many-many models and sizes) at 32 kHz and basically one model for the 192 kHz - my feeling is that the second was never seriously considered for power-consumption and volume reasons plus also the fact that it might still need far more complex electronics around it to go to 20 spy - but I might be wrong.

[South Pender]

The process of flattening the parabola would be more like 'adjusting for temperature' and in most designs that I am aware would be a rather VERY tricky procedure.

The much simpler standard calibration just involves shifting pretty much the same parabola up or down.

When I wrote "lower the parabola," I meant shifting it (the same parabola) down.