Thermocompensated Citizen movements (Cal.0330G and Cal.E510G) tested on the Witschi QT6000

[ppaulusz]

...Unfortunately I do not have a SpringDrive but the person who has the 1998 model The Citizen (Cal.A660H) also has a Seiko SpringDrive and he lives in the same city as me, by the way he is known as "Mechanikus" in our forum so I will try to organise a get together with him and subject some of his fine watches to the "Witschi-test".

Well, Mechanikus visited me today with his fine timepieces and we tested them all on my Witschi at room temperature of 23 degrees of Celsius:

The Citizen (Cal.A610H):
- inhibition period: 10 seconds
- accuracy of the stepping motor (by the magnetic sensor): -0.01 s/d
- accuracy of the quartz oscillator (by the acoustic sensor):
1st second: +7.54 s/d
2nd second: +7.50 s/d
3rd second: +7.54 s/d
4th second: +7.25 s/d
5th second: +6.17 s/d
6th second: +6.09 s/d
7th second: +6.14 s/d
8th second: +6.35 s/d
9th second: +6.19 s/d
10th second: +6.19 s/d
then it started again: the results might have changed a little bit here and there but the trend remained the same for the next lot of 10 seconds then the following lot of 10... and so on... and that is very similar to the test results of my thermocompensated Citizen Exceed watches: continuous inhibition (digital count suppression) combined with the continuous adjustment of the frequency of the quartz oscillator(!).

Citizen Crystron 4 Mega (Cal.7370D):
- inhibition period: none
- accuracy of the stepping motor (by the magnetic sensor): -0.03 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): could not be measured
(note: the Witschi QT6000 is designed to measure 32kHz quartz oscillators only by its acoustic sensor)

Longines Flagship VHP Perpetual Calendar (L.546.2 = ETA 252.611):
- inhibition period: 8 minutes (480 seconds)
- accuracy of the stepping motor (by the magnetic sensor): -0.00 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): between +2.70s/d and +2.90 s/d

Seiko Perpetual Calendar (Cal.8F32):
- inhibition period: 10 seconds
- accuracy of the stepping motor (by the magnetic sensor): +0.23 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): could not be measured
(note: the Witschi QT6000 is designed to measure 32kHz quartz oscillators only by its acoustic sensor)

Junghans Radio-controlled analog quartz:
- inhibition period: none
- accuracy of the stepping motor (by the magnetic sensor): -0.08 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): -0.08 s/d

Seiko Spring-Drive:
Mechanikus observed an accuracy of around +10 s/month over a longer period. The Witschi failed to get proper reading from the watch though we tried many times. Well, actually once we got a result of +0.34 s/d for the stepping motor by the magnetic sensor using 4 minutes (480 seconds) measuring period but we could not repeat it so we can't officially accept it though that result is very similar to Mechanikus' earlier observation.

Casio ProTrek Solar Atomic (my radio-controlled digital multifunction watch):
- inhibition period: none
- accuracy of the quartz oscillator (by the acoustic sensor): +0.50 s/d

[Hans Moleman]

The Citizen (Cal.A610H):
- inhibition period: 10 seconds
- accuracy of the stepping motor (by the magnetic sensor): -0.01 s/d
- accuracy of the quartz oscillator (by the acoustic sensor):
1st second: +7.54 s/d
2nd second: +7.50 s/d
3rd second: +7.54 s/d
4th second: +7.25 s/d
5th second: +6.17 s/d
6th second: +6.09 s/d
7th second: +6.14 s/d
8th second: +6.35 s/d
9th second: +6.19 s/d
10th second: +6.19 s/d
then it started again: the results might have changed a little bit here and there but the trend remained the same for the next lot of 10 seconds then the following lot of 10... and so on... and that is very similar to the test results of my thermocompensated Citizen Exceed watches: continuous inhibition (digital count suppression) combined with the continuous adjustment of the frequency of the quartz oscillator(!).

These were my measurements on the A660



And the hours I spend on why there were two lots of sequences!
Thank you.
There is a fast and a slow period!
Phew! It all comes out in the wash for the one who waits.

[ppaulusz]

These were my measurements on the A660



And the hours I spend on why there were two lots of sequences!
Thank you.
There is a fast and a slow period!
Phew! It all comes out in the wash for the one who waits.

Well, Hans, while it's good news that finally we solved the secret of the The Citizen (and the rest of modern day high-accuracy Citizen movements) - it is somehow sad that there is not much else left to "hunt" for right now...

[Hans Moleman]

Well, Hans, while it's good news that finally we solved the secret of the The Citizen (and the rest of modern day high-accuracy Citizen movements) - it is somehow sad that there is not much else left to "hunt" for right now...

You're a right romantic!
And you're right. We hunt and are disappointed after a catch because we don't need to hunt anymore. And if we never caught anything we'd never hunt either. That's humans for ye!

But not everything is explained yet. We don't know why it is done that way. What advantage did Citizen seek by doing that?

[Catalin]

Well, Mechanikus visited me today with his fine timepieces and we tested them all on my Witschi at room temperature of 23 degrees of Celsius:

The Citizen (Cal.A610H):
- inhibition period: 10 seconds
- accuracy of the stepping motor (by the magnetic sensor): -0.01 s/d
- accuracy of the quartz oscillator (by the acoustic sensor):
1st second: +7.54 s/d
2nd second: +7.50 s/d
3rd second: +7.54 s/d
4th second: +7.25 s/d
5th second: +6.17 s/d
6th second: +6.09 s/d
7th second: +6.14 s/d
8th second: +6.35 s/d
9th second: +6.19 s/d
10th second: +6.19 s/d
then it started again: the results might have changed a little bit here and there but the trend remained the same for the next lot of 10 seconds then the following lot of 10... and so on... and that is very similar to the test results of my thermocompensated Citizen Exceed watches: continuous inhibition (digital count suppression) combined with the continuous adjustment of the frequency of the quartz oscillator(!).

Citizen Crystron 4 Mega (Cal.7370D):
- inhibition period: none
- accuracy of the stepping motor (by the magnetic sensor): -0.03 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): could not be measured
(note: the Witschi QT6000 is designed to measure 32kHz quartz oscillators only by its acoustic sensor)

Longines Flagship VHP Perpetual Calendar (L.546.2 = ETA 252.611):
- inhibition period: 8 minutes (480 seconds)
- accuracy of the stepping motor (by the magnetic sensor): -0.00 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): between +2.70s/d and +2.90 s/d

Seiko Perpetual Calendar (Cal.8F32):
- inhibition period: 10 seconds
- accuracy of the stepping motor (by the magnetic sensor): +0.23 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): could not be measured
(note: the Witschi QT6000 is designed to measure 32kHz quartz oscillators only by its acoustic sensor)

Junghans Radio-controlled analog quartz:
- inhibition period: none
- accuracy of the stepping motor (by the magnetic sensor): -0.08 s/d
- accuracy of the quartz oscillator (by the acoustic sensor): -0.08 s/d

Seiko Spring-Drive:
Mechanikus observed an accuracy of around +10 s/month over a longer period. The Witschi failed to get proper reading from the watch though we tried many times. Well, actually once we got a result of +0.34 s/d for the stepping motor by the magnetic sensor using 4 minutes (480 seconds) measuring period but we could not repeat it so we can't officially accept it though that result is very similar to Mechanikus' earlier observation.

Casio ProTrek Solar Atomic (my radio-controlled digital multifunction watch):
- inhibition period: none
- accuracy of the quartz oscillator (by the acoustic sensor): +0.50 s/d

I would say that the measurements on The Citizen confirm that all TC calibers from Citizen in the last 25+ years are derived from the old 27xx calibers - with two observations - on that 27xx manual the 'cycle' is suggested as being around 2 seconds (but the wording seems unclear) while on both A610 and E510 your measurements suggest a cycle around 10 seconds; still there might be other differences from one to another since the temperature curves that we have seen posted around are not so similar!

The fact that the ultrasonic sensor could not pick the MHz signals is not surprising (and probably even the 196 kHz is too high). For me the more interesting thing is that it could not pick the 32 kHz from the SpringDrive - it suggests my original strictly-technical impression that the 32 kHz from a quartz would only be picked if the quartz manufacturer did not provide an extensive 'vibrational separation' when encapsulating the crystal and the watch does not have some extra level of it's own 'vibrational separation' - I am very curious how different would be the results with something like a Sinn UX ! Another thing that is strongly suggested by the results is that the ultrasonic readings are already very close to a minimum detection limit and as such might be very, very sensitive to any kind of 'noise' - not only actual voices around the device but maybe also even the 'internal ticking' - that might explain why the Aerospace was very 'steady' while the Conquest and Constellation were not! There might also be other sources of 'measurement error' which explains why for instance for your Constellation the Witschi was suggesting over 18 s/y while the watch was obviously calibrated for better than that!

I am a little surprised (in a good way) by the results on the Junghans quartz - what model is that ?

And I am not so surprised by the results on the Casio - my own generic measurements on Casio models was suggesting that they are using a very 'raw' 32 kHz quartz and that very often when non-receiving the error is very large and close to 15-20 s/month !

[ppaulusz]

I would say that the measurements on The Citizen confirm that all TC calibers from Citizen in the last 25+ years are derived from the old 27xx calibers - with two observations - on that 27xx manual the 'cycle' is suggested as being around 2 seconds (but the wording seems unclear) while on both A610 and E510 your measurements suggest a cycle around 10 seconds; still there might be other differences from one to another since the temperature curves that we have seen posted around are not so similar!...

I agree with the above.

[ppaulusz]

...I am a little surprised (in a good way) by the results on the Junghans quartz - what model is that ?

And I am not so surprised by the results on the Casio - my own generic measurements on Casio models was suggesting that they are using a very 'raw' 32 kHz quartz and that very often when non-receiving the error is very large and close to 15-20 s/month !

It's a Junghans caliber 615.

[ppaulusz]

...The fact that the ultrasonic sensor could not pick the MHz signals is not surprising (and probably even the 196 kHz is too high). For me the more interesting thing is that it could not pick the 32 kHz from the SpringDrive - it suggests my original strictly-technical impression that the 32 kHz from a quartz would only be picked if the quartz manufacturer did not provide an extensive 'vibrational separation' when encapsulating the crystal and the watch does not have some extra level of it's own 'vibrational separation' - I am very curious how different would be the results with something like a Sinn UX ! Another thing that is strongly suggested by the results is that the ultrasonic readings are already very close to a minimum detection limit and as such might be very, very sensitive to any kind of 'noise' - not only actual voices around the device but maybe also even the 'internal ticking' - that might explain why the Aerospace was very 'steady' while the Conquest and Constellation were not! There might also be other sources of 'measurement error' which explains why for instance for your Constellation the Witschi was suggesting over 18 s/y while the watch was obviously calibrated for better than that!...

My Constellation is calibrated for on-wrist performance (24 hours/7days) and lately I don't use it at all. The test was done at room temperature and yes, there could be that much differences. Same story with my Aerospace as well!

The interior of the Spring-Drive is the busiest in the quartz watch world. There is nothing like it! I'm pretty sure that we could get something out of its 32kHz crystal if we were to open the watch but that was never an option with a movement like the Spring-Drive.

The Witschi is designed for listening to the 32kHz oscillator's noise by the acoustic sensor. The 4.19MHz and 196kHz oscillators could not be recognized by it though the accuracy of those movements still can be measured by the magnetic sensor that receives the signals from the stepping motor. There is a capacitive sensor in my Witschi as well but using it requires opening the watch and using the supplied contact accessories - that is a bit beyond my knowledge and is better suited to the professional watchmakers. The Witschi can do a great deal more than the standard closed-case measurements I've been using it for.

[Catalin]

My Constellation is calibrated for on-wrist performance (24 hours/7days) and lately I don't use it at all. The test was done at room temperature and yes, there could be that much differences. Same story with my Aerospace as well!
...

Hmm, my own Conquest VHP Perpetual certainly has less than 6-8 s/y difference from 'warm' to 'room' and the same range seems to be suggested by the numbers that I have seen around here ... 18 s/y might be at most a battery in the final months or something similar ... or just I said - some unavoidable measurement errors on very short intervals!

[ppaulusz]

Hmm, my own Conquest VHP Perpetual certainly has less than 6-8 s/y difference from 'warm' to 'room' and the same range seems to be suggested by the numbers that I have seen around here ... 18 s/y might be at most a battery in the final months or something similar ... or just I said - some unavoidable measurement errors on very short intervals!

Well, my Aerospace at room temperature runs -0.05 s/d. From the 20th of this month I will start using it again 24hr/7days and see what happens.

By the way, you bought your VHP just a couple of weeks ago so that would be a very short interval indeed.

[Mechanikus]

The test with Witschi at Ppaulusz was very interesting. I have summarised the results and attached the picturres of the measurements taken. You can find it here:

Makszy's view on horology - Accuracy and Development

[Catalin]

The test with Witschi at Ppaulusz was very interesting. I have summarised the results and attached the picturres of the measurements taken. You can find it here:

Makszy's view on horology - Accuracy and Development

Very nice!

However I believe a better English translation for the label on the column with the quartz ultrasonic results would be (instead of 'Inhibition') something like 'raw quartz rate before inhibition' or some variation around those words ...

[ppaulusz]

Very nice!

However I believe a better English translation for the label on the column with the quartz ultrasonic results would be (instead of 'Inhibition') something like 'raw quartz rate before inhibition' or some variation around those words ...

That is right. There were two kinds of measurements:
- Measurement of the raw quartz crystal (that works only with 32kHz movements - in the case of the Spring-Drive the opening of the watchcase might be needed) by the acoustic sensor and that means a state before inhibition (bypassing the inhibition).
- Measurement of the stepping motor (works with all kind of analog quartz watch, except the Spring-Drive) by the magnetic sensor and in case of inhibition it shows the state after the inhibition/correction (the correct inhibition period or multiply of it needs to be applied for the setting of the duration of the measurement).

[Catalin]

However I believe what we see is the 'generic inhibition' that is most likely present in any modern 10$ watch with a Miyota movement - I would actually be very curious what you could see on something like that (Q&Q for instance).
...

Actually do not disregard that at all but instead rephrase it (it seems that even after only 4 hours of sleep after a party there was a part of my brain that was getting things instinctively - the other part that was supposed to explain those in words and ideas seems to have been more affected ) - I believe we see a generic 10 seconds inhibition used for raw rate correction coupled with a independent 2-frequency-variable-fill-factor TC method which seems to be a slightly more modern version of the method used in the older 27xx calibers!

Once you realize that the Witschi ultrasonic numbers have a certain variability/error (normal given the very weak/noisy signal that is measured) and you connect the dots with Moleman's data you see how that is precisely the case!

[ppaulusz]

Actually do not disregard that at all but instead rephrase it (it seems that even after only 4 hours of sleep after a party there was a part of my brain that was getting things instinctively - the other part that was supposed to explain those in words and ideas seems to have been more affected ) - I believe we see a generic 10 seconds inhibition used for raw rate correction coupled with a independent 2-frequency-variable-fill-factor TC method which seems to be a slightly more modern version of the method used in the older 27xx calibers!

Once you realize that the Witschi ultrasonic numbers have a certain variability/error (normal given the very weak/noisy signal that is measured) and you connect the dots with Moleman's data you see how that is precisely the case!

Let's subject a Cal.27xx to the Witschi test! If it behaves like the already tested current Citizen movements (A610, E510, 0330) then you got it right, if it behaves differently (eg: shows no sign of inhibition) then you got it wrong. It's really that simple, isn't it? Now, it's time to find an affordable Citizen with Cal.27xx...

[Catalin]

Let's subject a Cal.27xx to the Witschi test! If it behaves like the already tested current Citizen movements (A610, E510, 0330) then you got it right, if it behaves differently (eg: shows no sign of inhibition) then you got it wrong. It's really that simple, isn't it? Now, it's time to find an affordable Citizen with Cal.27xx...

Of course that is again wrong - just as in the other thread - you are still not quite understanding the difference from what 'inhibition' does for rate adjustment and what it does in digital thermo-compensation!

[ppaulusz]

Let's subject a Cal.27xx to the Witschi test! If it behaves like the already tested current Citizen movements (A610, E510, 0330) then you got it right, if it behaves differently (eg: shows no sign of inhibition) then you got it wrong. It's really that simple, isn't it? Now, it's time to find an affordable Citizen with Cal.27xx...

Looks like the test of the Cal.27xx won't be needed after all... See post #28 here:
2011 HEQ summary

[ppaulusz]

Of course that is again wrong - just as in the other thread - you are still not quite understanding the difference from what 'inhibition' does for rate adjustment and what it does in digital thermo-compensation!

I had read this thread first then replied here then read the other thread. Obviously I could not have forseen what was happening in the other thread when I replied here.
Post #37 in this thread explains everything about the above pretty well.

[Catalin]

Actually do not disregard that at all but instead rephrase it (it seems that even after only 4 hours of sleep after a party there was a part of my brain that was getting things instinctively - the other part that was supposed to explain those in words and ideas seems to have been more affected ) - I believe we see a generic 10 seconds inhibition used for raw rate correction coupled with a independent 2-frequency-variable-fill-factor TC method which seems to be a slightly more modern version of the method used in the older 27xx calibers!

Once you realize that the Witschi ultrasonic numbers have a certain variability/error (normal given the very weak/noisy signal that is measured) and you connect the dots with Moleman's data you see how that is precisely the case!

Since apparently for many people the 2-frequency-variable-fill-factor TC method described in the old Citizen 27xx manual might be 'not so obvious' I will describe it in very few words - basically the method works by having one quartz in a slightly more complex oscillating circuit which can be 'commuted' between two states - one in which a single capacitor is added (with one resulting frequency) and one in which two capacitors are added (with a slightly different resulting frequency) - hence the 2-frequency part. Now if you take a slightly longer interval and divide it variably into two sub-intervals and during the first sub-interval you only activate the first capacitor and then during the second sub-intervals you activate both capacitors you will generate a resulting frequency which is a linear combination of the two base frequencies - and by changing the fill-factor you can generate a resulting frequency almost anywhere in between the two base frequencies! (and by changing the fill factor depending on the actual temperature you can compensate for the changes in frequency resulted from the quartz dependency on temperature plus eventually some other parts of the oscillating circuit).

In the case of the 'modern' Citizen TC calibers that frequency is fine-tuned to an extreme point but NOT to precisely 32768 Hz but instead to something different and apparently odd-looking - let's take an example like 32770.1 Hz (which is about 5.54 seconds/day fast) and then 'fed' as raw input to potentially any caliber that supports what Citizen calls DFC (Digital Frequency Count - which is just raw rate inhibition) - and for instance in that caliber every 10 seconds a certain number of clocks are skipped - in our case about 21 (the number is specific to each watch but does not vary with the temperature or anything), with the end result of a frequency very-very close to 32768 Hz with a precision in the 5-10 s/y range (the numbers that I used are actual values extrapolated from the E510 numbers on the Witschi test numbers provided above in this thread).