Not a watchmakers bench - Mechanical Tuning Fork Escapement
Like Tree5Likes

Thread: Not a watchmakers bench - Mechanical Tuning Fork Escapement

Page 1 of 2 1 2 LastLast
Results 1 to 10 of 14
  1. #1
    Member
    Join Date
    Oct 2011
    Posts
    230

    Not a watchmakers bench - Mechanical Tuning Fork Escapement

    Hi All!

    As some of you may know from my previous posts, I'm working on developing an all mechanical tuning fork escapement.

    This was my 'watchmakers bench' last night after working on the escapement. All the equipment is to measure precision of the system. Laser strobing and acoustic frequency/spectrum measurement showed it was locking to about 0.001Hz. This is equivalent approximately to (86400s*0.001Hz = 86.4Hz) a day at 300Hz = 0.288S.

    Earlier in the evening I calibrated an electromechanical Tuning fork to better than 1S day using the same equipment - my escapement was closely matching for stability. Note that, due to how the electro-mechanical Tuning Fork clock is made, it is hard to calibrate more accurately without long term tests.

    My next investments are a GPS modulated 10MHz reference and a bench frequency counter linked to a computer so I can measure and improve long term stability. This is so I can measure performance changes after modifications. I need a much more precise set-up :)

    Here are some of my reference frequency/times based on a 300Hz resonator:

    0.001Hz * 86400 = 86.4/300 = 0.288S/Day

    0.0001Hz * 86400 = 8.64/300 = 0.0288S/Day

    0.00001Hz * 86400 = 0.864/300 = 0.00288S/Day


    Most accurate pendulum clock in free air is within 1S in 100 days ie within 0.01S/day

    (See Guiness Book of Records)
    https://www.youtube.com/watch?v=sQAY-zFkVyQ
    Note - see time point 1.46 for measurement screens

    Most accurate electro-mechanical clock is within within 0.0002S/day

    (Dual pendulum, phase locked in vacuum)
    https://en.wikipedia.org/wiki/Shortt-Synchronome_clock


    Regards to all - Tom
    Attached Images Attached Images

    Watchugeek and Christoph Kemp like this.

  2. #2
    Member
    Join Date
    Feb 2016
    Posts
    117

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    Super cool Tom.

    There is some amazing talent on this board.

  3. #3
    Member
    Join Date
    Oct 2011
    Posts
    230

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    @Watchugeek

    For me - it's mainly persistence ;)

    Tom

  4. #4
    Member dom_'s Avatar
    Join Date
    Jan 2012
    Location
    Mid-Sussex, UK
    Posts
    741

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    Have you done any tests at different temperatures?

  5. #5
    Member
    Join Date
    Oct 2011
    Posts
    230

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    @dom_

    At the moment the system uses an f300Hz resonator from a ESA 9162 movement. And so will be mainly limited by the temperature characteristics of this device (How it is mounted can also effect its fundamental frequency a small amount as seen below).

    I am in the process of developing a long term measurement system. Hence the need for computerized data acquisition and quite extreme precision. At this point I will be able to measure variables like temperature and air pressure and look at exactly how stable the system is. At the moment the system holds frequency over minutes and hours of running under the drive weight - which is not that critical.

    My most recent success is good quality measurement of the TF signal (as well as ensuring mechanical stability using a high precision strobe), without any interference to the system. It may be also possible to run the system in a vacuum to attain very high precision.

    I'll be starting long term measurements over the coming weeks and months when I have acquired all the equipment. The pictures below show the actual sign wave from the TF and the frequency of the system running. There is some phase noise (jitter), which effects measurement but is not enough to stop the scope triggering and the frequency (sign wave period) is solid. Note at this running frequency a clock using the escapement would be a true -10S a day +/- error due to instabilities like temperature etc. I have not yet tried calibrating for true 300Hz as I'm in the process of improving the whole system for long term measurements which at the moment is my main aim to enable me to answer your questions and others with real statistical precision.

    Regards - Tom
    Attached Images Attached Images




  6. #6
    Member dom_'s Avatar
    Join Date
    Jan 2012
    Location
    Mid-Sussex, UK
    Posts
    741

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    Awesome, keep us updated as you can.

  7. #7
    Member
    Join Date
    Oct 2011
    Posts
    230

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    @ _Dom .

    Thank you, a few words of encouragement go a long way.

    I tend to spend 99% of my time thinking and only 1% doing! Dave S from this forum very kindly came to visit me and rightly pointed out that I already had a clock. All I had to do was put some hands on it!

    Actually, mainly due to serendipity, I'm very close to a true clock. My prototypes/proof of principles use the resonator from an ESA 9162/f300 and a set of gears from a 70's Jedco electro-mechanical tuning fork clock. Although using the elctro-mechanical TF clock as gears (because they were geared for a higher frequency than a normal clock) I had never really looked at it's frequency. It turns out it's TF frequency is also 300HZ!

    The Jedco clock has a drive wheel, that transfers the vibrations of the TF to the gear train. This wheel has 40 'teeth'. The teeth are coupled in the Jedco clock using a permanent magnet. My 'Escape' wheel has only 10 'teeth' due to the escapement design/scale. So my 'clock' is running at four times normal time.

    So I either live with my 1/4 hour clock at this point (which could be useful for accuracy tests) or modify/build the motion works so the clock indicates a standard hour.

    Modifying the motion works shouldn't (substantially) effect the energy transmission through the going train and an extra mass can easily be added to the pulley if required. Note: I am driving the system from a pulley/weight on the cannon pinion which has been glued to remove the clutch and drive the train. You can just see the plastic pulleys on the left of the movements on the earlier picture of the test set-up (red and black).

    As mentioned I'm concentrating on system measurement but am slowly edging towards a clock with hands. Ideally, I need a patron as I have little/no funding (this is all done from my bedroom workshop) and I will have to restore a few more watches to fund the computer controlled bench precision counter and frequency reference I need. This is where I'm up to!

    I'll try and keep you all updated of any progress if there's any interest. Hopefully I'll buy the bench frequency counter in a couple of weeks and be then able to move on.

    Regards - Tom
    Last edited by TomBombadil; April 9th, 2016 at 19:02.

  8. #8
    Member
    Join Date
    Oct 2011
    Posts
    230

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    Hi All - Here's the latest on the development of a mechanical tuning fork escapement. I now have a bench frequency counter so can do long term precision tests.

    TF900 Series Frequency Counters | Aim-TTi International

    I talk to the counter via USB and calculate long term frequency trends against NTP time stamps. And also measure short term frequency jitter - this has been revealing:

    As the tuning fork acts as a true escapement the forces in the mechanical train are distributed across the system and back to the tuning fork. This results in phase noise in the measured frequency. Although this averages out it can be used to see the rotation rates of the drive wheels as measured from the short term frequency jitter as seen below. This approach limits the time period of short term frequency measurements using frequency alone to several cycles of any jitter noise ie the hour wheel as the noise needs averaging out (There are techniques to eliminate the phase noise - I'll cover that another time).

    Importantly, the tuning fork is staying in resonance and the measured forces superimposing on the fundamental are not stopping resonance. This can be successfully monitored by strobing a laser using a precision frequency generator. So for your own eyes, below is a plot of short term frequency variation showing the frequency/phase noise as a result of the drive chain. The x-axis is in samples made every 10 Seconds.

    Name:  Frequencies.jpg
Views: 128
Size:  24.2 KB

    Long term accuracy - I came across a report of one of the very first mechanical tuning fork escapements made by Koenig. Koenig was using it to create clock to use as a frequency standard.

    eRittenhouse | Tonometers

    It had a mechanical escapement and vibrated at 64Hz. Koenigs clock even included a thermometer and is suggested to achieve about 0.0001Hz accuracy (at 64Hz that's 0.135s a day). Pretty good. This stands as a baseline for any investigations.

    Long term frequency testing involves counting the pulses of the tuning fork and average them over known time periods. There are lags in a both the counter end and the computer that have to be considered. To get a good time stamp the computer clock has to be as accurate as possible. One way of doing this is to install the NTP service. This uses the NTP protocol to control/update and monitor the PC's time to national standards. A monitor program can also be installed to show how well correction and control are being achieved. Details can be found in the following link. Note: if you install the NTP service I'd suggest using a separate directory such as C:/NTP rather than the default and it should stop any installation file ownership rights problems etc

    https://www.meinbergglobal.com/english/sw/ntp.htm

    In this way we can get a time reference to a given level of indeterminacy. On my own system, the clock offset averages out at better than 0.002S from UTC using the NTP Service. A beginning and end count can be given a time stamp as they are collected and the the frequency calculated. The following graph shows the tuning fork escapement frequency being calculated over increasing periods ie how the frequency converges, effectively averaging over the periods. Various fixed intervals can be used to measure and sample drifts over different periods etc.

    Name:  Long Term Frequencies.jpg
Views: 129
Size:  6.5 KB

    So next:

    I need to improve the power reserve as is currently only 6 hours max. I'm modifying the drive, so instead of driving from the cannon pinion I'll have a separate drive wheel with a ratchet/chain and extended gears. I'd also like to sort out a direct GPS disciplined reference for both my counter and time stamping. I also need to calibrate/modify the tuning fork to get 300Hz for a real clock. I am missing out details but hopefully giving enough to give an idea of my directions. And hopefully, Ill be able to provide a real time or near real time internet feed with graphing so anyone can monitor how the mechanism is performing.


    Tom
    MNTN likes this.

  9. #9
    Member
    Join Date
    Nov 2008
    Location
    New York
    Posts
    460

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    Your theory has little problem.

    Suppose your projected accuracy range is 1 spd (second per day) error. Think about 1 day has 86400 seconds. So your projected accuracy in 1 second is 1/86400=11ppm. In order to accurately measure 11ppm error, your equipment should around 1ppm in accuracy, stability and repeatability. This is a very high demand.

    Now look at your Owen scope -- I actually have the exact same scope and this one cannot provide 1ppm reference frequency unless you provide external rubidium frequency standard. You still can use tuning fork or scope to calibrate watch. But it is almost impossible to tune watch in one shot to 1 spd because your standard is not reliable.

    Measuring mechanical timing error, especially the measurement of long term stability (>24hr), is still a frontier in horology. Such device could easily cost you half million dollars.
    Last edited by dkbs; May 30th, 2016 at 00:40.

  10. #10
    Member
    Join Date
    Oct 2011
    Posts
    230

    Re: Not a watchmakers bench - Mechanical Tuning Fork Escapement

    @dkbms

    I am using an NTP time server to discipline a computer clock to provide accurate time stamps (not an oscilloscope). The precision required can only be done over long time periods as I have described. For long term measurements I am using the bench counter to count absolute counts, not frequency which has problems as I have described and shown above. These are averaged over long time periods using an NTP reference that disciplines the PC clock. An extremely accurate long term frequency measurement can then be made. If the counter misses no counts then the long term measurement of accuracy increases over time. NTP is based on international standards - look at how it works at the different stratum. I am making use of a system that cost billions. But I am also going to use a GPS disciplined 10MHz reference and time stamp to complement my system this can also be used as external reference to the frequency counter to ensure a very high performance.

    https://en.wikipedia.org/wiki/Network_Time_Protocol

    This will tell you more about GPS disciplined reference and how they are used in metrology calibration laboratories.

    http://tf.nist.gov/general/pdf/2297.pdf

    This is what I am using!!!

    https://en.wikipedia.org/wiki/Networ...e:Usno-amc.jpg

    Tom
    Last edited by TomBombadil; May 30th, 2016 at 01:51.

Page 1 of 2 1 2 LastLast

Thread Information

Users Browsing this Thread

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

    Similar Threads

    1. Help with Mechanical Tuning Fork Escapement
      By TomBombadil in forum Watchmaking
      Replies: 6
      Last Post: January 19th, 2016, 23:32
    2. All Mechanical Tuning Fork Escapement
      By TomBombadil in forum Watchmaking
      Replies: 3
      Last Post: November 2nd, 2015, 01:29
    3. Replies: 20
      Last Post: December 12th, 2013, 11:33

    Posting Permissions

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

    Search Engine Friendly URLs by vBSEO 3.6.0 PL2