Seiko Twin Quartz 9983 Auxillary Oscillator tuning

[steve1811uk]

Hi, I recently received a Seiko twin quartz 9983 watch from my uncle that was in a very bad way. I believe that the auxiliary oscillator and main oscillators have both been adjusted badly in a futile attempt to make it run. Does anyone have a data sheet or information that shows how to adjust the auxiliary oscillator frequency? The data sheet for 9923 that is similar just says do not touch the auxiliary oscillator setting. Thank you.

Edited January 4 by steve1811uk
Images added

[JohnR725]

Usually for the specialties Seiko watches there is no technical information commonly available.

1 hour ago, steve1811uk said:

I believe that the auxiliary oscillator and main oscillators have both been adjusted badly in a futile attempt to make it run

Usually with vintage quartz watches like vintage mechanical watches they need to be properly serviced and lubricated. Versus playing with the oscillators and thinking that somehow will magically fix the problem.

There is a tiny but a information available on regulating at least the watch that you the PDF for like this

If somebody went and played with the other tremor which people love to play with things I'm guessing all it's really going to do is screw up your timekeeping the watch still should run so does the watch run at all? It's not running at all it probably has other conditions and not timing issues.

[steve1811uk]

Hi John, Thanks for the input. The watch does run despite all the abuse it has received before it came to me. It has lots of broken bits though and had been well over oiled. Battery cover was missing and the movement had filled with fluff. Pretty sure I can get it close as I have an electronics background and access to an oscilloscope. I'm going to adjust the main oscillator first so that it peaks at exactly 32768 Hz as shown by the graph. It should run at that frequency at a certain temperature and then raising or lowering the temperature should cause the frequency to fall (at no temperature should it go higher than 32768 Hz). Then I will adjust the aux oscillator so that it gives 32768 Hz when the main oscillator is running at 32768 Hz (again as indicated by the graph). That should only happen at a certain temperature. A hairdryer will be used to change crystal temperatures up and down. I will post any progress.

Edited January 4 by steve1811uk
Added extra

[steve1811uk]

Watch has been adjusted as described above using an oscilloscope with frequency measurement function. This is not the most accurate of methods but the best I could do with equipment to hand. Watch is on the wrist now and I will see how it performs. I do expect that it will need a minor tweak of the main oscillator trimmer as 0.005 Hz translates to 5 seconds per year and I could only measure to 0.1 Hz.

[steve1811uk]

Well, the adjustments although theoretically correct didn't work. Watch running at +2.9 SPD on wrist. I suspect that the oscilloscope probes were loading the oscillator circuits and skewing the frequency readings, also lack of precision on the frequency measurement from the scope. Not given up though. I had a photo showing the trimmer positions as I received it so have gone back to those and will make small adjustments. Currently +0.5 SPD on wrist and +2.5 SPD when cold (approx 5 degrees on the window ledge with window open).

[LittleWatchShop]

16 minutes ago, steve1811uk said:

Well, the adjustments although theoretically correct didn't work. Watch running at +2.9 SPD on wrist. I suspect that the oscilloscope probes were loading the oscillator circuits and skewing the frequency readings, also lack of precision on the frequency measurement from the scope. Not given up though. I had a photo showing the trimmer positions as I received it so have gone back to those and will make small adjustments. Currently +0.5 SPD on wrist and +2.5 SPD when cold (approx 5 degrees on the window ledge with window open).

Interesting architecture.

Measuring accurately using an oscilloscope is a challenge, if not impossible.  You are correct about the load.  Additional capacitance will pull the resonant frequency a small amount.  They used to make testers that measured the frequency of a quartz watch without adding a physical load.  I have a couple, but I don't think they work.  Moreover, with two oscillators...it would not work.

I would give up on the temperature compensation and simply adjust the frequency of one of the oscillators by wearing for a month, adjust, wearing for a month, adjust, etc.

I have to say that Seiko was clever by half.  The human body is a regulated thermal mass.  While the watch is external, to the body, it will still benefit from being in close proximity to the body.

I applaud you tenacity to do this the right way.  Often the project itself is more fun that the end result.

I did some research...these were supposed to achieve 5 seconds per year accuracy.  This was in the 80's.  That is indeed pretty damn good.  Remember that you have to reset your watch twice a hear, so if 5 seconds error is the goal, all you need is 10 seconds per hear.

Here is an interesting article from the NIST.

[steve1811uk]

Thanks for the input. Unfortunately I have a little issue that is thwarting my attempts at adjustment. Occasionally the seconds hand will move forward only half way between markers and then on the next pulse it will move back (taking 2 seconds to basically stay where it is). This is happening very occasionally when cold and more frequently when the date is in the process of changing over. I suspect poor battery, date jumper spring pressing too strong or issue with train wheel teeth, possibly a combination of those. Possibly a weak rotor, does that happen at all, lost magnetism over the years? Will need to fix that first.

18 minutes ago, steve1811uk said:

Thanks for the input. Unfortunately I have a little issue that is thwarting my attempts at adjustment. Occasionally the seconds hand will move forward only half way between markers and then on the next pulse it will move back (taking 2 seconds to basically stay where it is). This is happening very occasionally when cold and more frequently when the date is in the process of changing over. I suspect poor battery, date jumper spring pressing too strong or issue with train wheel teeth, possibly a combination of those. Possibly a weak rotor, does that happen at all, lost magnetism over the years? Will need to fix that first.

I've got a new suspect. The stuttering always happens between 54 and 2 seconds regardless of how I set the time. I think the seconds hand pinion could be slightly bent and causing the pipe on the seconds hand to rub the inside of the cannon pinion. Does it sound feasible? Or the hand rubbing the crystal as it's a very slim watch. Will report later.

[steve1811uk]

99% certain that the seconds hand was rubbing the sapphire crystal. Breathing on the back of the crystal to produce condensation revealed a circular line as the condensation evaporated. The line was a couple of mm inside the crystal edge running all the way round, just where the tip of the seconds hand is. I am guessing that the hand was pushed just a little bit harder against the glass at the top of the watch due to the battery behind. Currently, the seconds hand does rise up a touch towards the end and is not perfectly flat. Also the pipe is sat a bit higher than I would like due to how tight it is currently. Anyway, the watch is on test overnight with the crystal and bezel removed. Stay tuned.

1 minute ago, steve1811uk said:

99% certain that the seconds hand was rubbing the sapphire crystal. Breathing on the back of the crystal to produce condensation revealed a circular line as the condensation evaporated. The line was a couple of mm inside the crystal edge running all the way round, just where the tip of the seconds hand is. I am guessing that the hand was pushed just a little bit harder against the glass at the top of the watch due to the battery behind. Currently, the seconds hand does rise up a touch towards the end and is not perfectly flat. Also the pipe is sat a bit higher than I would like due to how tight it is currently. Anyway, the watch is on test overnight with the crystal and bezel removed. Stay tuned.

It's a new seconds hand by the way as the original was missing its pipe.

[JohnR725]

5 hours ago, steve1811uk said:

issue with train wheel teeth

One of the things to remember with quartz watches are there very very intolerant to well anything affecting the gear train I remember when the quartz watches came out initially there was an assessment of no lubrication needed because there is no friction. But the early metal on metal bearing watches would have issues. Then people try to oil their watch but that would have issues because that whales are all too heavy. I had one particular Seiko watch that with oil didn't work without oil didn't work and I think it even had jewels it was a small line Seiko watch that they would put insurance case. And then the Swiss came out with oil for quartz watches which is a very light weight oil and the watch worked perfectly. So I do find that the Swiss quartz oil works really well but if a quartz watches so sensitive the lubrication that would also imply anything with the gear train is going to be an issue

then typically with quartz timing machines they use basically a capacitive pickup. So basically they're not putting a load on the oscillator because of course that's a really tiny micro-powered oscillator and everything is going to be a load. Unless you can figure out how to tap into one of the outputs like the stepping motor.

 

[steve1811uk]

Confirmed that it was the tip of the seconds hand rubbing the glass. With crystal and bezel removed It's not missed a tick since 6pm yesterday including the day and date changing over. I must have done a good job of the cleaning and lubricating for it to be able to run 99% of the time as it was rubbing the glass all the way round. I will sort the hands and then work on timing. Currently +0.5 SPD on wrist and very close to that at 15 degrees. I will need to move both oscillators together to both be a bit slower as the temp compensation seems pretty close. Wish me luck.

[steve1811uk]

I just wanted to share an update . I'm getting quite close with the regulation. Latest is +0.1 SPD off wrist (16 degrees) and +0.04 SPD on wrist.

 

Some observations:

A timegrapher set on 3600 bph is a very useful tool even for quartz to get you in the right ballpark before you attempt the fine regulation. Normally you will see a noisy band across the screen but you can tell if that is sloping up or down, an indication of rate.

A hairdryer can be used to increase or lower temperature quickly. It can be used while observing the rate on the timegrapher.

If the temperature compensation is a long way out then you can see huge swings in rate as the temperature changes, as an example, from 0 SPD on wrist to +3 SPD off wrist.

Occasionally the timegrapher noise clears up (no idea why) and you can see how the compensation is working. See image attached (16 degrees) that shows a slightly negative rate that is compensated every minute or so with a small period where the watch runs faster.

I have tried to mathematically model (using a graphing equation website called Desmos and a simple model based on x squared curves) how the rate will react with small adjustments to the two trimmers. I think this will prove useful and it has shown that small tweaks of both trimmers alternately should be needed to achieve the flat response. Basically the model is showing that adjusting the main oscillator will cause an unintended slope in the temperature response. An adjustment to the temperature response should then required to correct and flatten the slope.

Wish me luck

Also one more observation. When you are getting close then the adjustments required are very small. Without a stereo microscope this way of adjusting would not be possible. The adjustments I am making each time are moving the trimmer by approx just one quarter of the width of its screw slot, possibly even less.

[steve1811uk]

Getting close now, -0.04 SPD off wrist and doing an on wrist check now to determine which trimmer or trimmers should be moved to correct.

I worked out how to get the clean trace on the timegrapher. I noticed that the trace cleared up between 10:30 and 12 midnight. It clears up when the train of wheels is being loaded by the date wheel changing over.

[steve1811uk]

Just posting what the timegrapher shows when the watch is made quite warm with a hairdryer. Note that there are periods where the watch is losing timing (sloping down to the right). That's the main oscillator that is running a bit slower than 32,768 Hz due to the higher temperature. Then there are periods of correction where the watch runs faster (slopes up to the right). The overall rate is showing as pretty much flat after a minute adjustment to the aux oscillator trimmer, as shown by the ruler across the peaks. Note that it takes about 10 minutes for the trace to go right across the screen so it's not a quick process. Much quicker than waiting days between adjustments though. First picture is pre adjustment showing a small rate gain estimated to be about 0.1 SPD.