Citizen HAQ compensation algorithm insights
Like Tree19Likes

Thread: Citizen HAQ compensation algorithm insights

Page 1 of 4 1 2 3 4 LastLast
Results 1 to 10 of 40
  1. #1
    Member
    Join Date
    Feb 2019
    Posts
    98

    Citizen HAQ compensation algorithm insights

    I finally managed to write a program capable of tracking quartz watch accuracy with close to millisecond resolution. NTP time is the source and second hand movement of watch is the input.

    Name:  A640.png
Views: 137
Size:  33.6 KB
    Graph shows results after first two days. Black circles are measurements. There seems to be a saw-tooth pattern in AQ4030's time keeping. The movement clocks with an error of about 11 SPY, but every ~24 hours corrects the error somewhat to be closer to 6.8 SPY on average.
    tomchicago and Nokie like this.

  2. #2
    Member Hans Moleman's Avatar
    Join Date
    Sep 2007
    Location
    New Zealand
    Posts
    1,604

    Re: Citizen HAQ compensation algorithm insights

    Interesting.

    I do value NTP, but not to the millisecond. As far as I remember, you can trust it to 10 ms.
    The best time would be Sunday morning when the network is at its most predictable.

    And I did some measurements on a The Citizen, and as I remember, it baffled me at the time.

    Turns out that it periodically varied the quartz frequency.

    That does not explain why there is a 25 ms span in your measurements.

  3. #3
    Member
    Join Date
    Feb 2019
    Posts
    98

    Re: Citizen HAQ compensation algorithm insights

    Interesting A660 posts, thanks for giving the link. Will read.

    Regarding NTP, with ping delay correction standard deviation between NTP measurements is rather small 1.6 ms.

    25ms of vertical spread is hand tracking variability. Each vertical set *actually* is about 30 closely packed points. Each average should be within a few ms from truth.

  4. Remove Advertisements
    WatchUSeek.com
    Advertisements
     

  5. #4
    Member
    Join Date
    May 2009
    Location
    USA
    Posts
    753

    Re: Citizen HAQ compensation algorithm insights

    This is very interesting thanks. If I read it right you are suggesting that that movement runs most of the day with some level of precision but then once a day it receives a bigger correction if needed. Such a scheme would probably save on power and might possibly explain the variation people notice taking sporadic measurements especially over short intervals.

  6. #5
    Member Nokie's Avatar
    Join Date
    Jul 2011
    Location
    Marin County, CA
    Posts
    27,060

    Re: Citizen HAQ compensation algorithm insights

    Very cool.
    "Either he's dead or my watch has stopped"
    Groucho Marx

    "The only reason for time is so that everything does not happen at once..."
    Albert Einstein

  7. #6
    Member
    Join Date
    Feb 2015
    Location
    NJ
    Posts
    3,088

    Re: Citizen HAQ compensation algorithm insights

    I'm guessing the 25ms distribution in your results is the inhibition. Typically the movement is set to run fast, than inhibition occurs to bring the rate down. For example Omega will run fast for 59s of a minute, than 10ms of inhibition is applied at the 60th second. Not sure how Citizen handles inhibition, but it would explain the distribution of your results.

    Doubt your taking your readings at 86F or wrist temperature, so the saw tooth pattern might be thermocompensation in action.

    The saw tooth pattern is definately not a method to conserve energy, that's the job of asservissement to extend battery life. Most quartz watches even cheaper Fossil's employ inhibition and asservissement.

    As far as TC never really seen how often or how much any watch employs. But assuming the watch is meant to be worn, so measurements taken at 86F would answer the question if this is TC. If you still see the saw tooth pattern its not TC.

  8. #7
    Member
    Join Date
    Feb 2019
    Posts
    98

    Re: Citizen HAQ compensation algorithm insights

    Quote Originally Posted by wbird View Post
    Doubt your taking your readings at 86F or wrist temperature, so the saw tooth pattern might be thermocompensation in action.
    Measurements were done at 31 oC on wrist. So temp was actually pretty constant.

    The saw tooth must be TC: regular quartz does not have this pattern.

  9. #8
    Member
    Join Date
    Feb 2015
    Location
    NJ
    Posts
    3,088

    Re: Citizen HAQ compensation algorithm insights

    Nice work OP. I didn't catch the title at the top of your graph. My bad. The only way to confirm if it is indeed TC, would be if you can influence the size or direction of the saw tooth pattern. I honestly haven't a clue if it speeds up or slows down at room temp, but if it slows down, the saw tooth pattern would have be in the opposite direction if it is TC.

    Just throwing that out there since it is the easiest test to run, assuming your house temp is around 21C.

  10. #9
    Member
    Join Date
    Feb 2019
    Posts
    98

    Re: Citizen HAQ compensation algorithm insights

    I don't promise these data are very accurate. But a quick test at room temp vs wrist looks like this:
    Name:  A643.png
Views: 91
Size:  26.1 KB
    dicioccio likes this.

  11. #10
    Member Hans Moleman's Avatar
    Join Date
    Sep 2007
    Location
    New Zealand
    Posts
    1,604

    Re: Citizen HAQ compensation algorithm insights

    I hate to draw conclusions on measurements that vary so much.

    In the 22 degree graph, how can you draw the blue line to the top of a bunch of measurements?

    How are you picking up the hand movements?

    Averaging is a great tool to filter random noise. But you have to be sure that the measurements are random. That there is no pattern.

Page 1 of 4 1 2 3 4 LastLast

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
    •