Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

Thread: Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

Results 1 to 5 of 5
  1. #1
    Member
    Join Date
    Jan 2009
    Posts
    3,103

    Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

    Seiko Twin Quartz models have been the first affordable Thermo-Compensated (TC) quartz models back around 1978 when first appeared. Even today some very nice (and some very well preserved) models can be found, but after so many years from production (more than 25 currently) the accuracy on those might not automatically be the same as the one initially guaranteed - which was 20, 10 or even 5 seconds/year!

    Some Twin Quartz models however have two trimmers on the back, and those can certainly be adjusted back to under 20s/y with the right knowledge (plus some minor tools and good hands/eyes).

    First a few words on how those models work - the 9923 service manual has a few words on that matter and a small capture of the relevant part can be seen below:



    (alternative picture at http://i7.photobucket.com/albums/y27...compmethod.jpg

    and manuals at http://caranfil.org/timing/seiko_9923A.pdf and http://www.watchuseek.com/seikocitizen/SeikoTechnicalManuals/9923A.pdf )

    and based on that I will describe the way I believe that theory is implemented in 9923 and how that caliber can be both calibrated to a convenient value but also adjusted for temperature (with some limits).

    Basically since the very first quartz wristwatch the implementation used a number of dividers to scale-down from the high frequency of the quartz (8-32 kHz range) to the one impulse / second which we see on the seconds-hand. Those dividers have later advanced to a counter - so in a basic form you have a very, very simple circuit that 'counts down' from 32768 to 0 - and on a 15-bit unsigned counter 32768 is actually the same as 0, so you count from 0 to 0 and the cycle is repeated. A small step forward would be to not count precisely from 32768 (or 0) but instead from some other value (that would be automatically be hardware-reloaded in the counter after reaching 0), and once the circuit has that you can digitally calibrate the watch with that value from which you start counting. Things are more complex than that since one step out of 32768 is a lot too coarse, and actual 'Digital Frequency Adjustment' for quartz watches is implemented over a longer amount of time - at least 8-10 seconds (which gives you the 10 second gate on almost all modern timing machines) with a counter which is practically at least 18-20 bits long.

    Now what the first graph is trying to tell is that when you have two quartz crystals each with it's own counter you will most likely not see the same numbers, and (depending on temperature) one counter will reach zero most likely before the other. However if the two crystals have a similar response to temperature the difference (with sign) can be a decent way to digitally measure temperature - as seen in that graph you have a curve ax^2 +bx + c = 0 and a second curve ax^2 + dx + e = 0, and when subtracting those you get (b-d)x + (c-e) = 0 , which is a simple linear relation. In practical terms it means that you have a main quartz and a secondary quartz, and you count down on the main to zero, and at that point you have a certain value on the secondary counter - and for instance if that value is 5 you have a certain temperature, if 3 another temperature, if 0 another, if -3 another, if -5 another and so on ... Of course that practically not even the 18-bit counter (over 8 seconds) will not show enough difference and instead a longer interval (and counter) needs to be used to see such clear/precise results as I described above - and probably something in the range of 2 minutes to 8 minutes is most likely used.

    Of course that just digitally measuring temperature is not going to automatically solve all problems - and here the second graph comes into play - basically from the value of the difference you can determine a correction value, which will then be added (or more likely counted-down separately as an inhibition delay). The manual even mentions something about squaring that difference to directly calculate the correction, but given the state of ultra-low-power silicon in 1978 it might have been a more practical approach to just have a precomputed table somewhere in a small fixed memory (identical for all watches).

    Now - how does all this information relate to re-adjusting a twin-quartz ? Well, at least in the 9923 caliber there are two trimmers - one marked M which changes the main quartz, and one S for the secondary quartz. Adjusting the Main trimmer is going to obviously correct the main timing - and that should anyways be the very first step taken, clearly aiming for the smallest possible error. That can be a little more tricky than it sounds, since the trimmer is VERY sensitive and it is far from easy to determine the rate that you have - in theory a good timer device with a 10-second mode would be a huge help - however note that the 9923 manual suggests doing the all timer measurements with the crown out on the second click and on a very specific Seiko timer!!! Since not everybody has that timer - the alternative is to use longer-time measurement techniques (together with atomic internet time, like the stopwatch method or ideally the video method). I was using the video method over intervals around one week (sometimes smaller when only coarser adjustments were needed) - and in the first stage I was able to get the main timing to a status where around room temperature the watch had an error of less than 3 seconds/year ! (it might be ideal to aim for the range -1 to 0 s/y at 24 Celsius, but that only if you really aim under 5s/y overall )

    HOWEVER - just as the main quartz was out of the original timing, the secondary one has also drifted over that long time - with the immediate result that the entire temperature compensation was seriously out of range - and close to 60 seconds/year difference from room temperature to the warmer one corresponding to always-wearing the watch. And here is where the second trimmer has it's role - it is possible to adjust it so that the overall compensation is back to good values!!! However the adjustment of the second timer is FAR more difficult than the main one - an incredibly small change can result in a VERY, VERY large rate change - since basically you are 'balancing' things on a very-very small needle



    The other thing to remember is that the main trimmer has a more normal "slower anticlockwise and faster clockwise" range of results for all temperatures, while the secondary trimmer will shift the equilibrium-point in a rather symmetrical way - most often rotating it in one direction (counterclockwise in my case) will make the watch slower at room temperature but also at the same time most likely faster on wrist temperature!!!

    So my approach was to have first a good rate at room temperature with only the Main trimmer - I got like +1 or +2 s/y - then to measure the rate warm - at that point it was -60 s/y - then from only the Secondary trimmer I aimed to a warm rate closer to zero - initially I overshoot it and from -60 it went ballistic to +175, but in a number of steps I got that smaller and smaller (to -35). The next step was to check if the room rate was also changed - IT WAS, just as I said, while the warm rate went up the room rate went down (to -30), so extra adjustment on the Main trimmer was again needed - I overshoot that again, but now it seems I am in a range where the DELTA (room to warm) is under 20 s/y and I am aiming for something under 10 s/y, and also a room-rate of under 10s/y !!!

    The other thing that I believe might take place is that any trimmer change might 'settle' over at least 1-2 weeks, so in general it might be worth checking any change for at least 1 week warm and one week room-temperature (ideally 2 and 2) before going for another trimmer adjustment!
    Last edited by Catalin; May 30th, 2010 at 22:13.

  2. #2
    Member T. Wong's Avatar
    Join Date
    Apr 2006
    Location
    Japan
    Posts
    5,770

    Re: Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

    thanks for the explanation, Catalin. I plan to ship my Omega 1441 to MOLEMAN in New Zealand in the near future for his trial to test and regulate the watch....now I have a general idea how he might be putzing with it! :)) ...the Champagne dial version....

    Last edited by T. Wong; May 30th, 2010 at 22:42.

  3. #3
    Member
    Join Date
    Jan 2009
    Posts
    3,103

    Re: Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

    Quote Originally Posted by T. Wong View Post
    thanks for the explanation, Catalin. I plan to ship my Omega 1441 to MOLEMAN in New Zealand in the near future for his trial to test and regulate the watch....now I have a general idea how he might be putzing with it! :)) ...the Champagne dial version....
    ...
    On the Omega I believe things are a lot simpler - there is only the 'main adjustment', and even that one is digital. It will be however VERY interesting to have a look on how the correction curve looks after all those years - there is already one graph (for another watch of the same caliber) posted in this forum, and having a second one might provide the extra statistical information on the matter ...

  4. #4
    Member T. Wong's Avatar
    Join Date
    Apr 2006
    Location
    Japan
    Posts
    5,770

    Re: Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

    Quote Originally Posted by Catalin View Post
    On the Omega I believe things are a lot simpler - there is only the 'main adjustment', and even that one is digital. It will be however VERY interesting to have a look on how the correction curve looks after all those years - there is already one graph (for another watch of the same caliber) posted in this forum, and having a second one might provide the extra statistical information on the matter ...
    Well, I'll leave the data recordings up to MOLEMAN and hope he will post his findings for us all. I'm a newbie to the scientific data but enjoy reading about the processes.
    Recently read one of the Stickies on HEQ crystals and was fascinated by the info on crystal shapes (tuning fork etc...) and your often point of view on crystal deterioration over time question.


  5. #5
    Member
    Join Date
    Dec 2008
    Posts
    2,431

    Re: Re-calibrating and re-adjusting caliber 9923 twin-quartz ...

    Thanks for the write-up on TQ adjustment, saved in case I need to adjust my 9256, currently at +15/0 (Room/Warm).

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

    Posting Permissions

    • You may not post new threads
    • You may not post replies
    • You may not post attachments
    • You may not edit your posts
    •