Okay carrying on with some timing information...
So a claim was made above about the timekeeping being very good on these watches, with them running to COSC specs. Before I move on to the Seiko’s performance, just a quick comment on that sort of claim. When most people make this claim, they really mean that the average daily rate (as observed on their wrist) falls within the –4 to +6 seconds range that COSC uses. They don’t actually mean it meets all the COSC specs, as those involve testing over extended periods of time, and with the watch subjected to different temperatures, etc. So I always take this claim with a large grain of salt, because many who make it don’t really understand what COSC testing actually involves.
I am not about to test all of the COSC parameters, but what I will do is refer to Seikos service specs, and Omega’s service specs for their watches – this will look at average rate, positional variation, and isochronism, as these are the most important factors that go into everyday timing.
I fully wound the watch (bonus that this Seiko can be hand wound) and performed timing checks over 6 positions on my timing machine:
Closer look at the results:
So there are a number of things going on here, but the key number that I pay attention to for timekeeping is the Delta, which is the difference in rate between the fastest and slowest positions. It is 16.1 seconds, and right above that you can see the average rate of all 6 positions, and that is –6.5 seconds. As with all Seikos I’ve worked on, the balance amplitudes are quite low compared to modern Swiss watches. Now the one thing that caught my eye as I was taking these readings, is that the beat error was out by quite a bit – the 1.8 ms error displayed is only for the last position (crown right) but it varied so much over the course of the timing that I wrote down the beat error for each of the 6 positions:
Dial down = 1.7
Dial up = 5.3
Crown down = 3.2
Crown left = 1.7
Crown up = 2.2
Crown right = 1.8
The beat error should be quite close to zero, and this sort of variation is very unusual, so I was not feeling very good about the quality here. I decided to use a different screen to do a long term test in just one position, so dial up for a 16 minute period, with the machine plotting an average every 4 seconds:
Closer look at the screen:
The bottom graph (green) is the balance amplitude, and the top graph (red) is the rate. As you can see, both are erratic and the amplitude especially so. I can also display this information numerically, so here is that screen:
So for rate on the top, it shows the average being –7.1 seconds, with the slowest recorded rate at –11.2, and the fastest –0.1. The balance amplitude is much worse, with the average at 270, the lowest being 171, and the highest being 360. So is this really running like crap? Well the balance amplitude is the key to the answer to that question – if the high amplitude was really 360, then the movement would be rebanking, and this would cause an extremely fast rate, yet the rate spread, while not great, does not indicate rebanking. This had me a bit puzzled, sine what I was seeing didn’t add up.
So on the last photo above there is a parameter on the screen called gain. This is essentially the sensitivity of the machine in picking up the acoustic signal from the watch. I pretty much always leave it at the default of 2, and rarely change it. There is a blinking LED that visually shows you the ticking, and it seemed to be blinking cleanly, but on the off chance that there was maybe a weak signal, I changed the gain setting from 2, to 3 – max. setting is 4 just for reference. I ran that extended timing test again for 16 minutes, and the results are dramatically different:
As you can see both graphs are much smoother:
And here we see the numeric results:
Interestingly, the spread of the rate is about the same, but the spread of the amplitude is now only 34 degrees instead of nearly 200. So it seems that for whatever reason a higher gain is needed with this watch. I ran the positional testing again with the higher gain setting:
Here you can see that the Delta is now down to 13.6, but the average rate is still slow at –6.3 seconds. In addition the balance amplitudes are now reading more consistently, and the beat error was also much more stable, so it didn’t have the wild swings that I saw earlier.
Just for confirmation purposes, I tried running that 16 minute test in a couple of other positions, and although crown up was pretty good:
Crown right was quite erratic:
Now as much as people likely think that watches left in one position are always rock solid steady in rate and amplitude, that really isn’t the case. Small variations in the roundness or flatness of wheels can cause small oscillations in a graph like this, and I use this graph feature to diagnose what wheel may be the problem by the period of said oscillation, so if it’s damaged I can replace it, or if it's bent I can straighten it. The variations I’m seeing in the crown right position would certainly be cause for concern, and in a service situation I would be finding the cause and fixing it. By the way I did try setting the gain to 4, and it made no real difference to the results, so this is as good as they get.
So with the gain change, and the crown right stability results notwithstanding, I certainly feel much better about the movement now. But how does it stack up to the claim of running within COSC specs? Clearly the average rate doesn’t meet that criteria as it’s running too slow, but that is easily resolved by a simple adjustment, and really the same is true for the beat error as this watch has a moveable stud carrier.
But what about the positional variation – this is really the foundation for the daily rate specs, because if you have wide variation in positions, you will have to rely on luck for the average rate to fall within specs. So for Omega they allow the Delta to be as much as 12 seconds at full wind, measured over 5 positions. For all COSC watches, the 6th position (crown right) is not tested. If I look at the last set of full wind timing results, it’s 13.6 seconds over 6 positions, but if I take off the last one (crown right) then it reduces to just 10 seconds, so it falls within that spec. Great right? Well technically yes, but if you look at the first two readings, they are dial down and dial up, and those are the slowest and fastest positions.
These should be the same or very close, so to have the entire variation between these two positions indicates that something isn’t right. It could be a lubrication difference between the two sets of balance jewels, it could be that one pivot is flatter that the other, or there is a cleanliness issue, but the starting point for any timing work is that these two should be as close as possible. On a newer watch the timing should be within a couple of seconds – more slack for vintage due to wear issues that can be present. So good, but not as good as it could be certainly.
How about isochronism? Just to clarify, isochronism is a measure of an oscillator’s ability to maintain it’s period, regardless of it’s amplitude. So when the balance amplitude drops, how does this affect the rate? To find out I fully wound the watch again and waited 24 hours, then took those same 6 readings:
Closer look:
So let’s look at that Delta again – it has now gone up to 30.3 seconds, and if I delete the last reading it doesn’t make any difference. The average rate has sped up considerably, and now instead of –6.3 it’s flipped to +7.6. Curiously the difference between dial down and dial up is now less by about 1/2 of what it was. So how does this compare to Omega specs? Well 24 hours after full wind, Omega allows the Delta to increase from the 12 seconds at full wind noted earlier, to 15 seconds, so an increase in variation of just 3 seconds. Here if we again delete that crown right position, the Delta has increased by almost 20 seconds, nearly tripling the Delta, so these results are frankly not good at all.
So the conclusion that can be drawn from this is that certainly this sample would not meet the COSC criteria by a long shot. But how do the numbers here compare to what Seiko requires?
Well that wasn’t as easy to find as I had hoped. I first found a technical guide from 2006 for the 6R15 movement, and for average daily rate it gave tolerances that anything between –15 and +25 seconds per day is within spec. So that is a possible range of 40 seconds, and COSC specs call for –4 to +6 and that is just 10 seconds. Omega specs are from –1 to +6 for their COSC watches, which is only a 7 second range. This guide had no information on positional variation or isochronism.
So this 2006 document didn’t say anything about how many positions are tested or any other information. So just looking at the average rate we had of –13.6, then that falls within their range of –15 to +25. PASS!
However I found a later guide dated 2011, and it had the same spec for gain/loss, but it clarified it a bit by saying that –15 to +25 spec was:
“Daily rate worn on the wrist at temperature-range between 5°C and 35°C)”
It also gave some additional criteria, but it’s worded very differently than the specs I’m used to looking at. It gave 3 testing positions, and gave rate tolerances at full wind for each position separately.
For dial up it said that the tolerance was +/- 10 seconds. For crown left the tolerance was +/- 15 seconds, and for crown down the tolerance was +/- 10 seconds. So if I take the fastest end of the tolerances given at +15, and the slowest end of the tolerances at –15, they allow a Delta of 30 seconds over 3 positions. So at full wind the watch I have is well within that range for Delta, but because the dial up position must be +/- 10 seconds, this watch fails Seiko’s criteria as the dial up rate is –13.2 seconds. FAIL!
So what about isochronism? They apply a general tolerance of an additional 10 seconds variation 24 hours after full wind. So the allowed tolerances for each position now go to -15 to +15, –20 to +20, and –15 to +15 respectively. If I look at the timing checks at 24 hours after full wind, they are all within those specs. PASS!
So depending on what criteria you use, this watch may pass or fail Seiko’s own specs. But in any case it’s a very long ways from being at the level of accuracy that brands like Omega, Tudor, or Rolex require.
Since there is a bit to digest in the timing information, I'll stop here and leave the remainder for another post.
Cheers, Al
