1 second Regulator- state of play

Discussion in 'Clock Construction' started by Phil Burman, Nov 30, 2016.

  1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
  1. jhe.1973

    jhe.1973 Registered User
    NAWCC Business

    Feb 12, 2011
    484
    5
    18
    Arizona, USA
    Country Flag:
    Region Flag:
    #51 jhe.1973, Jan 2, 2018
    Last edited: Jan 2, 2018
    Hi Phil,

    This echos my experience. Although I had close to 20 years prior machining experience by the time I built my prototype regulator, there were a few perfectly good shop methods that I had to UN-learn because they would not give me the precision I was trying to achieve.

    Just a suggestion regarding taking notes.

    Would you consider posting all the lessons you learn, even the ones you now have to back up through? This might be of help to all of us here who appreciate the depth with which you are going into your project. After all, there really aren't that many of us who value the degree to which you are studying your efforts. This thread of yours could become your notepad and our reference.

    I don't know just what your background is but I think you would make a great researcher! Thanks so much for your time and dedication.

    I am exactly the same way! One of my favorite prayers is," God grant me patience, but give it to me now". Someday if it ever happens I'll let you know how it works out!

    :whistle:
     
  2. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    308
    1
    18
    Cape Town, South Africa
    Country Flag:
    I too am most interested in what you find. I have made a 1 second regulator. I naively thought that having a heavy and solidly suspended pendulum would make it regular and the clock would keep regular time. Only with experience have I come to see how the train imperfections can get through to the pendulum. My clock has a gravity escape which ensures that the impulse to the pendulum is unaffected by the train imperfections. Only the locking force on the escape varies and that the pendulum has to overcome - but it is just a matter of sliding the locking pallet sideways from under the escape wheel tooth. Yet I can see that that small alteration in force manifests itself as erratic time keeping.

    In one of your earlier posts you suspected that the hourly fluctuation was caused by an imperfection in the hour wheel. Have you been able to confirm that? The 60 X 12 = 720:1 reduction in force from the hour arbor to seconds arbor means that imperfection is much disguised yet it manifests itself. We are really dealing with incredibly sensitive & precise instruments when we are chasing 1 second a week. I am very interested in anything you learn by bitter experience.
     
  3. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    I am interested in a gravity escapement for the reason you describe. Do you have any details, photos and microset charts that you are able to post.

    The variation showed a period slightly longer than 1 hour and was concluded to be a result of the rooms’ thermostat control which, due to the insulating properties of the all wooden case, was nominally 0.1 degrees C. However the attached recently downloaded chart shows a clear 2 hour cycle, with the heating turned off, the second wheel and its' associated pinion has a 2 hour per rev cycle. I don’t worry too much about these cyclical errors as they are on their own self-cancelling over any time frame that matters. The clearest picture of clock performance with Microset is the gain/loss chart, or with GPS the drift. The short term error can be all over the place but the deviation from the true time is very small. For the attached chart the drift from true time when corrected for offset never exceeds 0.02 seconds over nearly 2 days.

    Also the clock can be seen to run faster as the amplitude increases which is the opposite what would be expected with respect to circular error. I’m not sure what is going on but it may be due to temperature changes. For example in and uncompensated pendulum if the temperature increases then the rate will slow due to the increase in pendulum length however with the same value of energy input the amplitude will also increase in order to maintain the same magnitude for the potential energy at the top of the swing. Add the complication of the impact of relatively rapid transient temperature changes on a well-compensated pendulum and you really get into the long grass.

    It is my current conclusion that the main obstacle to my further progress in accuracy is the different rates of temperature equalisation of the different pendulum components during a relatively large and rapid temperature change. I believe there are three practical alternatives that have the potential to overcome this obstacle. 1) Use a pendulum rod with a zero or close to zero coefficient of expansion, such as quartz or super invar, thereby side stepping the problem by eliminating the need for any significant temperature compensation. (super invar is very expensive and quartz is relatively fragile). 2) Where ever possible change the geometry and design of the significant pendulum components such that where possible they all have similar rates of temperature equalisation. 3) Optimise the case design in order to reduce the impact of external ambient temperature transients on the pendulum components.

    In Philip Woodwards’ book “Woodward on Time” he makes the statement that 1 second per week for a weight driven pendulum clock is “a piece of cake” and it is this that challenged me to set my target at 1 second per month.

    upload_2018-1-3_22-7-11.png
     
  4. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    308
    1
    18
    Cape Town, South Africa
    Country Flag:
    I have started to write up my gravity escape. It will take quite a few days before I can post. I will do it in a new thread.
     
  5. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    Great, can't wait :D

    Phil
     
  6. Bruce A W

    Bruce A W New Member

    Jan 26, 2018
    1
    0
    1
    Male
    Country Flag:
    Phil, I am just doing some experimenting with building a clock from scratch. I am impressed with your work, have learned a great deal from your entries here, and am frankly somewhat in awe! I have several questions about the your use of involute gears. What pitch diameter/mod are you using? Have you done anything special with the teeth for this project. How accurately do you make the teeth? Do you use pins to measure the pitch diameter or do you just use the depth of cut as your guide? What is the diameter of your arbors? I would appreciate any information you could share.
     
  7. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    #57 Phil Burman, Jan 27, 2018
    Last edited: Jan 27, 2018
    Hello Bruce, thanks for the compliments. The tooth module used in this movement is listed in post 1. Maximum involute size is generally dictated by the acceptable size of the movement, minimum size is dictated by the required durability. Nothing special done regarding tooth profile, I use involute gear cutters (Chinese) some of which are not too accurate with regard to profile.

    I use this page for involute data, the bottom of the second page gives information on standard clearances.

    https://mdmetric.com/tech/modulegeardata.pdf

    The gears transmit very little power and run very slowly so it is much, much better to have more rather than less clearance/backlash, I use the clearance value as per the above link. Ultimately the necessary clearance is dependant of the concentricity of teeth pitch circle diameter relative to the arbor pivot. Measuring with pins is somewhat irrelevant, I use the DRO on my mill to achieve the required depth and the drill the end plates according to calculation, although a depthing tool is the correct way to achieve the desirable degree of clearance, automatically taking care of all the inaccuracies of construction. I have a mid 19th century English drop dial wall clock where the escape pinion teeth are literally worn half way through and the movement rattles like a handful of nails in tin can, yet it keeps extremely good time.

    The arbor diameters range from 8mm to 2.5mm but far more important is the Escape wheel mass together with all pivot diameters. The escape should be of the lowest mass you feel happy with and all pivots as small a diameter as possible. The issues are reduced moment of inertia, so material reduction at the wheel diameter is worth much more than material removed at the arbor diameter. Smaller pivot diameters require less power, halving the pivot diameter halves the power requirement needed to overcome the pivot friction.

    Regards
    Phil

    PS: where are you located.
     
  8. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    A quick update on the current status.

    A novel bob design with external temperature compensation - may be an original design? The advantages are much more stable with regard to "wobble" at the 3 point contact between the temperature compensator and the bob, and also the three temperature compensators react to temperature changes without lag in relation to the pendulum rod. The disadvantages are the aerodynamic drag results in a lower Q value and the increase in overall diameter will require a larger cabinet.

    The other photo shows a four sensor temperature date logger to monitor and record temperatures at different heights in the cabinet. It has been clear for some time that the single temperature sensor of the Microset timer was not showing the whole story of what was happening inside the cabinet with respect to changes in temperature stratification.

    bob with external compensation 2.jpg 4 channel temperature logger 1.jpg
     
  9. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    308
    1
    18
    Cape Town, South Africa
    Country Flag:
    That temperature compensator is really novel. The three point support for the bob set way out there is beautifully stable.
     
  10. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    #60 Phil Burman, Feb 23, 2018
    Last edited: Feb 23, 2018
    Temperature compensation with the new bob and cabinet vertical temperature profile for the same period:


    temperature compensation.JPG cabinet temperature profile.jpg

    Orange plot on the temperature compensation plot is barometric pressure, maximum range is approximately 2.6mbar
     
  11. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    To those that may be interested:

    For my Strasser escapement recent results have shown that a reduce drive weight actually gives a larger pendulum amplitude but with a much greater, random variation in rate.

    drive weight 6.0 3.5 Kg

    rate variation 0.15 2.20 s/day

    ang displace 2.25 2.56 no units

    My conclusion as to why:

    Unlike a deadbeat escapement the Strasser provides impulse to the pendulum prior to the lock not after the lock. Consequently both the escape wheel tooth and the pallet are in motion from the start to the end of the impulse. AS the escape tooth moves forward to make contact with and provide the impulse the pallet impulse face is also in motion toward the escape tooth. The point at which they make contact is a function of the angular acceleration of the pallet, which is fixed by the pendulum length, and the angular acceleration of the escape, which is controlled by the mass of the drive weight, minus the frictional losses in the train.

    Now, if the drive weight is reduced then the escape angular acceleration is also reduced so the pallet face has time to move further toward contact with the escape tooth, therefore the escape tooth meets the pallet face early to give a long duration impulse. However with a reduced drive weight any variability in magnitude of the frictional losses in the train has proportionally a larger influence on the magnitude of the impulse.

    So overall a smaller weight gives a long impulse duration but with large variation in pulse magnitude. Whereas when the drive weight is large the reverse situation exists and I get a relatively short impulse duration with a smaller variation in impulse magnitude. Demonstrated by the two plots. Too larger weight would actually provide insufficient impulse to keep the the clock running - because the escape moved to quickly to the lock position, almost before the impulse pallet "arrived".

    I'm wondering if a deadbeat might be a better choice over the Strasser in that the frictional losses during the lock may be less troublesome that the effects of variation in angular acceleration of the escape wheel with the Strasser?


    3.5kg drive weight
    upload_2018-7-31_1-34-49.png


    6.0 kg drive weight
    upload_2018-7-31_1-35-29.png

    View attachment 487440
     
  12. tok-tokkie

    tok-tokkie Registered User

    Nov 25, 2010
    308
    1
    18
    Cape Town, South Africa
    Country Flag:
    That is a massive difference - particularly if you read the Y scale. The MicroSet certainly gives you a wonderful insight into what is actually happening.
     
  13. John MacArthur

    John MacArthur Registered User
    NAWCC Member

    Feb 13, 2007
    297
    24
    18
    In re-reading this thread, I remembered thinking about the business of co-axial hands vs. "regulator" style. It bears on something I've been trying to solve. In many of the high-grade regulators with separate hands, the minute hand is mounted on a friction sleeve that grips the center arbor behind the dial. There is often a counter-weight attached to this friction sleeve that balances the minute hand as it goes around. Not having this hand balanced must have some affect on the timekeeping, even though it would be cyclical, and in theory should average out. This is not possible to do with co-axial hands, as the minute hand is mounted on the outer end of the cannon pinion, which is within the hour hand sleeve, and isolated by the friction spring from the rest of the train. At any rate, I have not come up with a solution, as I prefer the co-axial arrangement, not for precision, but for general ease of telling time at a glance.
    Thoughts are welcome.
    Johnny
     
  14. Phil Burman

    Phil Burman Registered User

    Mar 8, 2014
    172
    17
    18
    Suffolk England
    Country Flag:
    My thoughts. I think there are possible three solutions.

    1) Balance weight behind dial on the driven wheel down-stream of the friction wheel (see attachment). The simplest implementation might be a very thick wheel, crossed out then with some of the crossings filled with lead to provide the counter balance.

    2) Balance weight in front of dial on the minute hand. This solution would have a tendency to block the reading of the second hand on an astronomical type dial, hindering accurate readings, but may be OK if all hands are coaxial or where a separate seconds dial is relatively small in diameter.

    3) No balance weight on the minute hand but keep the hand as light as possible to minimise variation in rate, possibly with some balance weight assistance on the hand.


    Here’s the arrangement on my regulator.

    minute hand counter weight.JPG regulator hands.JPG regulator overall arrangement.JPG
     
  15. John MacArthur

    John MacArthur Registered User
    NAWCC Member

    Feb 13, 2007
    297
    24
    18
    Thanks Phil -- Your # 1 is what has been on my mind - make the gear on the bottom of the cannon pinion look sort of like a locomotive wheel, with no cutouts on opposite side, and maybe even thickened. Usually the cannon pinion has a pretty small gear on it's bottom end, so that the intermediate gear doesn't have to do all the reduction. I may play around with this design, however. An alternative is to make the minute hand analogous to the second hand, with a counter-weight built on the opposite end, shorter and fatter than the pointed end. None of these seems as elegant as #3, which is the usual option, and I have quite a ways to go before I really have to think about this.

    Johnny
     

Share This Page