the quest to get a mechanical 400 day to stay within a minute or so a year.

Discussion in '400-Day & Atmos' started by victor miranda, Jan 19, 2017.

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

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Your take is not significantly different from mine Victor. I am closing down on it but...… it is like Will of the Wisp. Very elusive. I reset my test torsions today. Victim has done a month or more and gained 7mins this period of 40 days. I have reset to BST and wound it with 8 clicks. (1.5 / week). I want to see how it does by the time change to GMT at end of month. {20 days}. No changes of regulation, just the wind up. Haller has gained 19m over 6 weeks. These gains all came late say through the last 20 days. Jum/7 and Kundo2 - well they are just a couple of idiots! However a wind of 8 clicks and reset today.
    BTW - it is my belief and understanding NOW that balances, pendulums, (all types) are isochronous by nature over a limited arc. UNTIL that is you want to perpetuate it with impulses. Then it goes awry. It will maintain a DIFFERENT isochronous period ONLY IF the impulse energy is exactly consistent. Now we need to discuss what portion of the total energy is the impulse to assume? A large shove, a small shove, and average shove. We know that a mainspring varies. We know a big difference comes from a small remontoire frequently rewound. A weaker but less variable shove!
    What do we know about pendulum/balance arc length and weights when it comes to consistency? Think on - I think we do know!

    If my Metamec 8 day mantle clock restoration continues to run by tomorrow am I will start my electronic gismo project. All other clocks & watches are satisfactory! So little time.
    Rgds, BJG
     
  2. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    ok we have Identified the problem...

    too many variables.

    forumers,

    from all I can tell, any problems we may be looking to find are
    less that 30 sec per day.
    and that error is hard to see with most 400 day clocks.

    and most weather events like high or low pressure
    tend to be changed in less than a few hours and only last a day most times.
    so being able to tell such shifts seems a priority...
    lubrication seems to be a linear thing --> the clock will gain time no matter what

    jeweled or no oil bearings may be required.

    but the escape needs oil ....

    ok at this point I think it may be possible to set an ordinary 400 day to go
    for a year and get close to a minute year buuuut we may find ourselves
    hoping for a biig storm to slow or speed the clock to make the goal.

    is setting a clock in a vacuum box 'ordinary?'

    at this point I want to test power driving and the effect on the pendulum and regulation....
    and I am going to build an oil bath.
    victor
     
    Berry Greene likes this.
  3. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    An oil bath! Is that different from bath oil?
    Before I get it to that discussion let me say this:- We have some points to clear up.
    ISOCHRONOUS = A constant period or interval. One year / one month / one week / one day / one hour / one second
    Under that definition alone we can just stop the clock. I think we might save a lot of words if we were to say something about eccentricity. Get me?
    I was never thinking of an average performance. Would that count?
    Temperature: All to do with Youngs Modulus and the torsion spring. Holovar have answered that. Yet there are other effects and we know the pendulum stability is one of those. So is the lubrication. Gotta be visco-static. What about the mechanism? Where does the temperature alter it most?
    Air density - Surely must have some bearing on the pendulum? It's moving quite fast through the air at its periphery with a design that is principally aesthetic rather than aero-dynamic - except for the disc type. What don't we make a disc type as in Atmos clocks?
    OK I have wound my test clocks. All have had 6 to 8 clicks which covers a 6 week operation period. "Victim" looks good again - spot on time! Yet it is early days. I need at least a week to get any drift. Haller is the same. Jum/7 is good too and even NoKunDo is on time today. Let's be patient. The main-spring power is the very first problem IMHO - ahead even of temperature, given we use a Holovar suspension. It is also a given that the clocks are on a very firm base and levelled. There can be no draughts, no sudden changes of temperature, and no vibrations even at sub-audio level.
    Do I have all those things in place? To the best of my ability and belief yes I do. However, I do not have them all in identical conditions - as in on the same shelf. I am looking into that. So is my wife. She ain't keen on my chosen sites you see. This could become a domestic. A time driven domestic with more isochronism about it than any clock could ever muster!
    Leave it with me. I'm working on it see.
    Oh lordy is that the time? Gotta go - it's time for my oil bath. :<))
    Best regards, BerryG
     
  4. MartinM

    MartinM Registered User

    Jun 24, 2011
    2,963
    79
    48
    Male
    Medical Insurance Systems Analyst
    El Dorado, CA
    Country Flag:
    Region Flag:
    Out here in California, we just had that annoying time of the year when all my clocks were doing fine and then the temps drop a bit and they all start gaining. It's comforting to see them all gain at the same rate. and I'd love to find that sweet spot where they are perfect in the spring and fall, But I know I never will.
     
    Berry Greene likes this.
  5. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Martin M - it's really good to hear from you on this trace (thread) and also what you have to say. How do you explain it?
    Rgds, BerryG
     
  6. MartinM

    MartinM Registered User

    Jun 24, 2011
    2,963
    79
    48
    Male
    Medical Insurance Systems Analyst
    El Dorado, CA
    Country Flag:
    Region Flag:
    When the temp drops, the elasticity of the suspension spring is reduced, causing it to shorten the period of the pendulum's cycle. As they all have Horolovar springs and are all 8 BPM, they tend to all be affected in a like manner. I have an old Hauck in the bedroom that has a bronze spring and it does the same thing, but with a higher variance.
     
    Berry Greene likes this.
  7. KurtinSA

    KurtinSA Registered User
    NAWCC Member

    Nov 24, 2014
    3,145
    96
    48
    Aerospace Engineer (Ret.)
    San Antonio, TX
    Country Flag:
    Region Flag:
    Martin -

    Are you really seeing that much of a variation in time with the current version of suspension springs? I thought the whole deal with Terwilliger's Ni-Span-C springs was they were relatively flat in terms of modulus change over a wide temperature range of like -50C to 150C. It's hard to see how a drop of 5 or 10 degrees over a short time would register as a change in time regulation, at least enough to really notice.

    Kurt
     
    Berry Greene likes this.
  8. MartinM

    MartinM Registered User

    Jun 24, 2011
    2,963
    79
    48
    Male
    Medical Insurance Systems Analyst
    El Dorado, CA
    Country Flag:
    Region Flag:
    All I can say is that the reality and the claims don't necessarily jibe. When a dozen clocks all have roughly the same change in regulation, I kinda have to go with that.
     
    Berry Greene likes this.
  9. KurtinSA

    KurtinSA Registered User
    NAWCC Member

    Nov 24, 2014
    3,145
    96
    48
    Aerospace Engineer (Ret.)
    San Antonio, TX
    Country Flag:
    Region Flag:
    I thought John might have published some of his testing data. But in the December 2004 issue of the Torsion Times, the thermoelastic coefficient (modulus of elasticity) of a Ni-Span-C suspension spring is shown at ~0 parts per million (ppm) over a range of -45C to +65C. The older Elinvar was +/- 10 ppm for 0C to 100C. Seems pretty darn stable to me.

    Interesting, the coefficient of expansion (the tendency to change shape) is 7.6 ppm/C over 21 to 93C. Steel and bronze higher than that...nothing shown for Elinvar. But the shortening effect of the suspension spring has a lower order effect on the rotation rate of a spring...as we've come to know, the cross sectional area has the most dramatic effect and since Horolovar springs have width tightly controlled, the thickness is what we worry about.

    Kurt
     
    Berry Greene likes this.
  10. Harry Hopkins

    Harry Hopkins Registered User
    NAWCC Member

    Nov 16, 2011
    543
    37
    28
    Retired Instrument Technician
    Mason, IL
    Country Flag:
    Region Flag:
    I was going to do this little test last year and never got around to it but will hopefully have more motivation this year... I have a few 400 day clocks that keep very good time in my temperature controlled home. I also have an unheated but finished room in my home. In about a month here in Illinois that room will be considerably cooler than the rest of the house. I plan to take one of my better time keepers and record how close the time is after 1 week in temp control then put it in the unheated room for the same amount of time and record the same data along with average temps from both weeks. I'm sure someone has done something similar before but I need to see the results myself.
     
    Berry Greene likes this.
  11. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Good stuff from Kurt, Martin and Harry. Oh yes & Victor of course! Thanks chaps. I'll be interested to hear the results of all experimentations - even those previously performed by others all down the years. Rgds, BerryG
     
  12. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    OK difficult post coming up. No-one really likes to be seen as a fool but I'm going to risk it as I am really rather keen to learn. I don't mind a bit if someone edits this or finds a better form of words. I want to speak about 400 days and what it really seems to mean. I have seen 30 hour clocks; 8 day clocks; one month clocks and 400 day clocks.
    Lets ignore the escapement for now. We are asking springs of varying strength (that is thickness & elasticity) to power the hands of a clock for a range of 30hrs to 9600hours from about 7 to 12 half winds. The mainspring barrel size tells us much doesn't it? They are pretty closely matched in many specimens of these clocks..
    I want you now to consider the main task and the variation in gearing to achieve that task. Then I want to speak about friction. Have I got your ear? From a barrel that will be turning some 4 to 6 times between winds we are going to turn that minute arbor through 30 to 9600 times!! Something like a step up of 5 in one case to 1600 times in the other. Check me out! How am I wrong?
    Imagine the friction multiplication. We need to go from good enough pivots to superb low loss purpose made specialities. Yet by and large we just don't do we? The same old steel & brass applies. No wonder we need everything to be clean. The tiniest extra resistance will not only slow us down - it will reduce the power at the escapement.
    Did you forget that bit? Haven't we already said that the variation in spring pressure between fully wound & let down is so wide that it certainly will affect the rate.
    Come along now - shoot me down. Where & why am I wrong in this? I expect to be very wrong because gentlemen & ladies, I do not expect a 400 day to work at all! It is a preposterous suggestion!
    Rgds, BerryG
     
  13. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    the difference and answer is the amount of power it takes to run the pendulum.

    Along with that consideration you have to find a way to limit power from a spring and the desire to get to the gearing that lets
    8 turns on a mainspring turn the hands thru 400 days.

    the mainspring is sized to run the gearing, hands, expected frictions, and a pendulum.
    this is true for any pendulum clock I have met.
    it is not that one can't build a swinger that will run for 400 days.
    it is that the spring is large (and scary when wound) and the bearing surfaces will get large
    and the extra wheel I think will be needed will get large
    and all those large parts will add to cost.
    the biggest problem being that there is little doubt it will cause the regulation to fail without
    some sort of power regulation mechanism.

    the torsion clock is supposedly free of that need.

    if you look at an atmos, it is a big (I am guessing a bit here...) 15 jewel watch.
    it happens to run for a month if not wound.
    I have read the clock is accurate the entire month. ( am still saving to get one)

    we have some evidence that a 400 day mainspring can overpower the escape.
    (and not enough power will stop the clock) so yeah one can state
    a 400 day clock will struggle for regulation from the effects of the mainspring.

    what we do not have is clear evidence of that power curve in the regulation.

    well.... we reasons to think it is there and have some reason to puzzle over what
    causes the other regulation problems we see.

    victor
     
    Berry Greene likes this.
  14. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Hi Victor et al
    I didn't mention the 1200 day clock but they are also out there. Noticeably bigger mainspring but nonetheless..... a brave offering.

    Victor, I am picking in particular at your sentence in the middle:- "the torsion clock is supposedly free of that need" {Meaning any power regulation..?}

    I was drawn to two separate pieces I found on-line. They represent the doubts I suffer because of what I perceive as either a difference of opinion or a misunderstanding. I feel a need to resolve this as part of the quest. I wonder if I am free to quote these. Or if I need to acknowledge the origin?
    First I'll paraphrase then I will email you privately with the precise source links.. I do not want to be accused of plagiarism. But I want it on here because I would appreciate any feedback on the subject.

    First of all I want to set this up. Tell me why should a torsion pendulum be any different in essence to a balance wheel? I mean in respect of its time behaviour. Its isochonism if that's the right word? Also the choice of the horizontal Hermle floating balance for better accuracy that must be kept horizontal? {Like a torsion}
    Here are the quotes:-
    1) "The amplitude of the balance wheel (in the horizontal position) beat should be between 270 & 315 degrees."
    2) "Low amplitude will affect the rate - speeding up on shorter swings...……"
    3) "Note the early use of a cone shaped fusee barrel which is essentially a variable ratio gearbox that evens out the surge of strength a rough steel spring has when fully wound....."
    4) "fusee always been widely used in pocket watches for better accuracy"
    5) "When a clock with a balance wheel runs fast it means there may be a problem with the main spring losing power, counter intuitive as that may sound."

    Some rather interesting questions emerge from this don't they? BTW - what is not mentioned here is the temperature compensation needed on balance wheels.
    Why? Which component do we need to compensate?
    BjG
     
  15. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    ... the balance wheel is designed to work with reasonable accuracy at angle relative to gravity.

    a torsion pendulum requires gravity to work.
    the regulation is dependent on the gravity.
    similar and not the same.

    right now I can state that most of my clocks do not have 'free' swinging
    in that the fork is often held against the pin of the verge.

    "the torsion clock is supposedly free of that need" {Meaning any power regulation..?} -- yes

    I've read a lot of things... am figuring out how to test them.

    victor
     
    Berry Greene likes this.
  16. MartinM

    MartinM Registered User

    Jun 24, 2011
    2,963
    79
    48
    Male
    Medical Insurance Systems Analyst
    El Dorado, CA
    Country Flag:
    Region Flag:
    In the grand scheme of things, there's not a ton of difference in the total number of beats a clock makes on a full wind whether it be 30 hours, 8 days, 30 days or 400 days. The outliers would include the 1000 day and others with significantly more (or less) total turns of the mainspring.
     
    Berry Greene likes this.
  17. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    #367 Berry Greene, Oct 18, 2018
    Last edited: Oct 18, 2018
    OK Victor I surrender! I told you I was daft!
    Gravity...………………….. Ah yes the missing link. Gravity. Mmm!
    I am chastened. My torsion clock won't work right on the moon but my balance watch does...…. Mmm!
    Ask an American - they should know!
    I know the weight changes everything in a torsion. Yes I already know that. I'm just daft.
    I'm lost!
    End of.
    BjG
     
  18. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Yes Martin - I saw a 1000 day (is it) torsion clock yesterday. A Schatz & Sohne and it has a larger barrel. How it copes with the diversity in mainspring output is mysterious to me. Does it?
    I take your point about the beats Martin - I was saying let's ignore the escape because of all the energy needed just to turn the hands. In 1000 days that minute hand will turn 24000 times. I just find that amazing. Especially when you compare it to a 30 hour clock where the spring turns it only 720 times.
    I know the springs can be longer & thicker. I suppose they can have a different modulus (or whatever it should be called). I still find it amazes me because in overall size it has no comparable increase.
    Rgds
    BerryG
     
  19. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Do you know what a sheep looks like? It looks like I do. I don't have to report this - but I will if you're friendly - my first attempt to build an electronic optical sensor, (for counting torsion pendulum rotations), has ended in failure! I just cannot get the sensor to work. OK I admit that it did come out of my junk - er … spares box but ...….A Honeywell HOA2001 it should be the bees-knees with its built in voltage regulator and Schmitt trigger operation. I'm not done as I have a couple of similar (but not the same) sensors. They are the very much more basic GE H22B1 and the Omron EESB5. I will be trying these - only time prevents me - NO it's NOT my competence. I am committed - or I should be - and that's what counts isn't it?
    Regards, BerryG
     
  20. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    the problem with barnstormers is that you have to fly them where people can see them.
    there are a lot of things near the ground.

    Hi Berry,
    I an unhappy to admit that I do not know any of those devices.
    I have little doubt you will get it working.

    in some related news from what I have seen over the past week,
    the pendulum can vary over 22 degrees of arc in a day....
    not sure what effect that has on regulation however you need to keep it in mind for sensor placement.
    I am up late tonight/morning...

    victor
     
    Berry Greene likes this.
  21. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    #371 Berry Greene, Oct 21, 2018
    Last edited: Oct 21, 2018
    Victor
    I had some interesting links to send you by my surf history is gone. Save to say they were containing statements the facts of which we need to get clear about.
    Such as " torsion pendulum rotation degrees directly affects the rate...…. The business of horizontal balance wheels as in platforms or the swinging balance by Hermle. If it didn't make a difference why do it? I have several platforms mostly in carriage clocks of the 8 day variety. There is also one platform by EDS (who were they?) inside an 8 day onyx mantle clock of unknown manufacture. These clocks are noticeably accurate through their week. The Hermle is spot on. So I want to ask - Is there an advantage to horizontal operation? All the torsion clocks are...…… er…. horizontal.

    Here's a rather interesting link. It shows us an old Keinzle with a disc pendulum. Is this shape a better way to go? What do you say? Aerodynamic?
    Kienzle Disc Pendulum 400 Day Clock – ClockInfo.com

    These are the issues that make me want to count the pendulum swings. After that - if it looks to be a possible answer, I would be looking for an alternative impulse method. I see the clockwork to be just driving the hands. The escape just releases the clockwork to step the time forward. It no longer provides impulses. The impulses are applied to the pendulum magnetically in the same way as the electronic balance which predated the arrival of the quartz. These also keep excellent time but they are very fast as in short periods. We are stuck with the existing period. I don't believe the fork has a part to play as it promotes flutter. The verge release is also electric/electronic.
    So that's my thinking which so far doesn't include temperature issues. I think we have to accept that the Horolovar springs have this covered so we're back to the pendulum, its material, it's shape, and maybe some temperature & humidity compensation.
    Regards, BerryG
     
  22. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    #372 victor miranda, Oct 21, 2018
    Last edited: Oct 21, 2018
    I set a pendulum up without the fork.
    (in dead clyde ) and did a little timing
    it takes about 5 to 6 minutes to oscillate from a bit over 360 degrees of arc
    to a little less than 180.
    and after a few attempts I got it started on a steady run without any side swing.

    here is the run of times for each cycle.
    raw data thumb clicks and mark I eyeball
    10.13
    10.37
    10.09
    10.33
    10.37
    10.15
    10.34
    10.26
    10.33
    10.18
    10.25
    10.3
    10.31
    10.35
    10.27
    10.24
    10.32
    10.29
    10.18
    10.29
    10.4
    10.25
    10.31
    10.28
    10.19
    10.37
    10.25
    10.29
    10.29
    10.27
    10.29
    10.28
    10.28
    10.33
    the stopwatch says that took 5:49.43

    I did some tinkering with a spreadsheet
    the "--5 avg time" is the average of the 5 cycles up to that line.
    10.13
    10.37
    10.09
    10.33
    10.37 --5 avg time 10.258
    10.15 --5 avg time 10.262
    10.34 --5 avg time 10.256
    10.26 --5 avg time 10.29
    10.33 --5 avg time 10.29
    10.18 --5 avg time 10.252
    10.25 --5 avg time 10.272
    10.30 --5 avg time 10.264
    10.31 --5 avg time 10.274
    10.35 --5 avg time 10.278
    10.27 --5 avg time 10.296
    10.24 --5 avg time 10.294
    10.32 --5 avg time 10.298
    10.29 --5 avg time 10.294
    10.18 --5 avg time 10.26
    10.29 --5 avg time 10.264
    10.40 --5 avg time 10.296
    10.25 --5 avg time 10.282
    10.31 --5 avg time 10.286
    10.28 --5 avg time 10.306
    10.19 --5 avg time 10.286
    10.37 --5 avg time 10.28
    10.25 --5 avg time 10.28
    10.29 --5 avg time 10.276
    10.29 --5 avg time 10.278
    10.27 --5 avg time 10.294
    10.29 --5 avg time 10.278
    10.28 --5 avg time 10.284
    10.28 --5 avg time 10.282
    10.33 --4 avg time 10.295
    349.43
    STW time 05:49.43
    avg time 10.2773529412

    from all that there is some reason to think more arc takes less time to cycle.
    and the difference is .02 seconds per cycle from 360 to 180

    among the things I noticed is that the slowing or shrinking of the arc
    was not linear in that it seemed to get to 270 in about 2 minutes
    and the next 3 and a half to get to 180.

    so air may slow the bob quicker... effect on time to cycle may be involved.
     
    Berry Greene likes this.
  23. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Wow Victor you have humbled me again. Great stuff. This is precisely what I wanted to chase. You have worked hard on this. Well done sir!

    I can report that I do now have a working opto-sensor. It provides me with a good signal switch from about 0.5V to the VCC level which I hope to set at 4V and run from a slab lithium battery (Laptop PC type). I think this will enable me to compare your results with mine. The setting up for this sensor is likely to be difficult. I will need to attach a lightweight (paper) vane to one of the bobs which is centred in the detector when at rest. I admit there will be a slight air drag.

    I have yet another sensor to try (if this one proves in the least effective), which is the reflective type. I would not then need a vane - but just a white mark (Snowpake?) on the bob. However it might be more open to incidental light pollution and the one think I do not want is spurious counts. Photos will follow if I have any kind of success.

    In my previous post I referred to lost links. I have now found one of those and I think it's worth a read. https://clockinfo.com/posts/7952
    Here (in red) is the part that caught my eye It concerns regulation procedure but it also carries the statement in red at the end. The answers are probably "out there" - we just haven't found them yet.
    1. When the pendulum comes to a stop at the end of a rotation, grasp the pendulum in one hand and hold it stationary. Lift it up slightly to take the pressure off the suspension spring;
    2. With your other hand turn the regulating disc a small amount (Clockwise for slower, Counterclockwise for faster);
    3. Release the pendulum, and the clock will start going again.
    This procedure is good because it doesn’t disturb the rotation of the pendulum very much, reducing the timekeeping variations caused by the changing the amount of swing of the pendulum. And it doesn’t disturb the pendulum as much as taking it off would.

    With luck there will be more from me later.
    Best regards & thanks for the info
    BerryG
     
  24. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    #374 Berry Greene, Oct 23, 2018
    Last edited: Oct 23, 2018
    Good morning!
    I have progress to report and some photos. I even have a short video but I cannot post it here directly as in AVI format. They are not too large - being just under the limit of an individual post. I'll see about converting them or do the YouTube upload and link}.

    All that in good time. However, I can post the photos which, I hope, very largely speak for themselves.
    I will just say that this preliminary experiment is on NoKunDo as I do not want to wreck my other experiments and trial - yet!
    Here we go then:- Oh wow it's a big photo. {Well it was until I posted! The system has shrunken it down. I hope it's clear}.
    You can see the PCB with the black sensor. It has an opening not very visible in the picture.
    Electronic pendulum counter  prototype-(2).JPG


    Here you see the lightweight black paper vane that I have attached to the bob suspension strut. It is arranged to pass through the aperture in the sensor with each beat. {It looks as though the camera can see through it! Mmm!
    There will be a change in signal level as the vane passes and interrupts the light path. I get a count of each beat. Electronic pendulum counter  prototype-(9).JPG


    This is actually a frequency counter I made many years ago (when younger!). It also measures periods & will count events. I have set it to count events. There is a divide by 10 feature which will extend the maximum count if I need that. I have not yet worked out the numbers but we all know that if this is a 6 second beat we will get 10 to the minute therefore 600/hr. In a day it's 14,400 beats. The range before roll-over is 99,999,999 which is more than enough surely? {I make it 19 years! Can that be correct?}
    Electronic pendulum counter  prototype-(5).JPG

    Shows the vane passing through the sensor
    Electronic pendulum counter  prototype-(10).JPG

    During this process I see that I might in fact be able to upload the videos directly as they are just inside the limit of a single post. I will try after this is transmitted.

    Regards,
    BerryG
     
  25. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    that looks good to me.
    count on the leading edge? .... (I want to sound like I know what I am doing)

    I have to point out that the clock should do the count.
    you may want to watch the frequency instead to see if the pendulum holds to 6 seconds a beat.

    I plan to try a few things a little later today to test escape and fork interaction...

    victor
     
  26. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Victor YOU do know what you're doing. I, on the other hand, am English!

    Yes that's correct. The HOA2001 Honeywell sensor gives a Lo out when blocked and this rises to a Hi when it clears. My counter advances on a rising edge.
    This is my preferred sensor but I have experimented with 3 types. {2 are the interrupt type and one is the reflective type}. This one is probably the best since it also incorporates a Schmitt trigger device which removes the possibility of spurious bounce at the vane edges. (Which BTW is now foil. No light penetration; thin smooth edge}.

    You must understand that so far this is all experimental prototype work. There have been some false starts and changes to the rig. I now need to tidy up and stabilise the test bed. A very firm perch would be good.

    However, things look promising in the sense that the pendulum rotations are just what you would expect. i.e. 600 beats per hour, or close to that. I am timing to within one beat of GMT but I will be closer than this eventually. Everything so far points to flutter being the main cause of the gains. I daresay that the spring drive will also marginally affect the pendulum time but in relation to the flutter the effect is small. {That's just my initial impression}.

    I hope in due course to publish my figures but to do so now would be premature. I not only want more time - it might be that I don't need NoKunDo to be a part of the reckoning. It always maintains but we are not best of friends and I cannot take back some of the nasty things I have uttered about it! Revenge is an ever present possibility!

    There are two parts to this.
    First: Does the pendulum swing as expected? Beat as expected. I might need several hours if not days to get confident it matches to passing time.
    Or if there is an error I need to quantify that and make certain it is a consistent error. The smaller these are the longer I need for them to extend into whole beats.
    Second: Does the clockwork measure the expected time? Match the pendulum beats? Or are there missed escapes? Or extra escapes?

    What do I expect? Ahh..... one needs to be careful not to anticipate. I will accumulate the data and then rationalise. Of course I will!

    Best regards,
    BerryG
     
  27. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Not so easy!
    All sorts of complications. Doing my brain in!
    From stray light to mains borne interference!
    Trying to keep on track. What is it we really want to know here?
    Answer: Does the escape mechanism miss or gain any escapes?
    So 600 beats - and hands move an hour - and so on.

    TO DO:
    Improve installation. Shelf;
    Eradicate or suppress any mains borne or indeed RF interference.
    Improve logging methods.

    DONE:
    Different vane material
    Different sensor type
    Different probe connection {1x or 10x}
    Light shield fitted
    Generate look up chart {e.g. I beat = 6s - 50 beats= 5min - 100 beats =10min - 150 beats = 15m - 600beats = 1hour - etc.}

    So some way to go yet. I am anxious but need yet more patience and time.
    Got other jobs on. Not everything in life is a regulator! th?id=OIP.jdBr7gT8fH1tZNh6aAsKcgHaHB&w=218&h=207&c=7&o=5&pid=1.jpg


    Just an update you know. A simple if slow progression towards The Quest.

    Leave it with me.
    Rgds, BerryG
     
  28. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    #378 Berry Greene, Nov 10, 2018
    Last edited: Nov 10, 2018
    Oh gosh the tabulation of my results here is rubbish. I will have to come back but can't just erase the entry. Must write something! I would ask that the moderator removes this entry, So sorry.
     
  29. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    "Publish and be damned". Whoever said that knew a thing or two. These results have been long in mixer. I would welcome anyone that has a working knowledge of the formula's in Excel. Despite that I am now going to publish my first page of results using my electronic pendulum counter applied to my KUNDO 2 400 day unit I call "NoKunDo" on account of its reluctance to keep accurate time. There are periods when it does well and then, usually overnight, there is a slow change almost invariably a gain but not always. Victor was the first I remember to point out the likelihood of "flutter" and it occurred to me that we needed to tie down the number of pendulum rotations (beats to you) more exactly. In pursuing this I constructed an optical sensor which is coupled up to an electronic counter. There are pictures of my very first efforts published previously and although I have added some more decoupling and earthing straps (to exclude RF interference) it remains substantially the same. It is with its partner, the Excel Spreadsheet, that I have long been fighting. Although not perfect in every respect I do now have a working example and some results to publish which I will bring down as a screenshot in the manner actually suggested by Victor Miranda the father of the Quest.

    upload_2018-11-19_8-33-12.png

    In my view it shows the clock dial gets ahead of the beats actually completed. Put another way it needs fewer beats than it should to be where it is. Flutter is the most likely cause. What do you think?

    Please note that although I keep track of GMT on the chart this is not the primary accuracy we are looking for here. I am looking for discrepancies with the beats of the pendulum which has a 180 deg rotation and maintains through thick & thin. The overswing is 1/2" and even. The clock is on a firm perch and away from draughts. The temperature variation does not exceed 2 or 3C deg. Indeed although ahead of the beat the overall clock shows a slight loss against GMT. I assume this arises from a very small discrepancy from the nominal 6 secs/beat? What do you think?

    BTW I think my next move, having got the system up & running, will be to transfer to a more reliable clock. However it all takes T.I.M.E. :<))
    BerryG

    Screenshot-Spreadsheet section E1.2.png
     
  30. MartinM

    MartinM Registered User

    Jun 24, 2011
    2,963
    79
    48
    Male
    Medical Insurance Systems Analyst
    El Dorado, CA
    Country Flag:
    Region Flag:
    What are you using for a time reference? I ask because I just made a discovery that solved an issue I thought I was having with my Seth Thomas Regulator #2 for nearly a year.
    I set an atomic clock on top of the regulator and every week, when it was time to wind the clock, I'd consult the atomic clock and make the necessary adjustment to the regulation.
    I FINALLY realized that the atomic clock wasn't phoning home nearly as often as I had imagined it should and whatever it uses for an internal time reference between check-ins is very poor. Once I started using NIST, it is essentially perfect.
     
  31. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Hello Martin. Wow that is a surprise but I wonder now if it should be. There exists an opportunity to use a poor local reference if corrections are to be quite frequent.

    Before I get further into that discussion I need to explain that my investigation is not yet for pursuing absolute sidereal time accuracy. I am looking for an absolute correlation between torsion pendulum beats and the progress of the clock-work. In a word - probably "flutter." I did it because I really thought I could - me with my electronics background etc etc. In truth I have found it much harder than I could ever have imagined. Not least the spreadsheet production!

    Yes I do annotate the precise GMT as given by my atomic regulated clock. {Frankfurt beacon I think}. Back a while now I did check it with the following link and more latterly with my Lap-top PC clock. which is always very close whenever I check it. Try this link: Time.is - exact time, any time zone
    It actually tells me the error in my PC clock (right now is +1.9s) but it is typically very much closer than that.

    As you know the torsion clocks are notorious for their inaccuracy and the reasons fly about all over the place. I am in little doubt that whilst the Horolovar suspensions do overcome changes in the spring modulus with changes of temperature, there are many other reason for inaccuracy. The mainspring must be high up on the list if remontoire working with balances yields such undisputed improvements in isochronous performance. Surely I could offset that by much more frequent winding? I have been trying that monthly.

    Yet it is flutter that is my main focus at the moment. This weird phenomenon that waits in the shadows of night to cheat the escape and advance the progress. My results so far support that probability - but I am not yet convinced that my method is without its own idiosyncrasies. More time and other clocks are needed before I can be sure of that. At least I have a Spreadsheet that now seems to work. It saves those awkward mental calculations or the waiting to catch whole hours for the time checks.

    I just wish I knew more about the Atmos clock performance. How accurate are they? I know they are generally kept wound and run for only short periods when they are not. Perhaps I should post that question on this Forum somewhere more appropriate.

    From the foregoing you will see that I need more help than I can give. I apologise for that Martin. Thanks for that insight.
    Rgds, BerryG
     
  32. mauleg

    mauleg Registered User
    NAWCC Member

    Dec 26, 2012
    848
    136
    43
    Country Flag:
    Mine is the most accurate clock I own; on par with quartz; this accuracy is reflected in the experience of others who own these clocks. Keep in mind, however, that the atmos is a clock on an entirely different level from the majority of torsion clocks. Tolerances and quality on an Atomos are more in line with a fine watch.
     
    Berry Greene likes this.
  33. Berry Greene

    Berry Greene Registered User

    Oct 2, 2017
    405
    23
    18
    Male
    Retired electronics engineer
    Chichester
    Country Flag:
    Region Flag:
    Thank you Mauleg - just what I wanted to know. Most interesting. That removes lot of my doubts about torsion operation. It can be done evidently. I shall research the ATMOS the more now.
    Rgds, BerryG
     
  34. victor miranda

    victor miranda Registered User

    Jan 13, 2017
    319
    30
    28
    computer programming stuff
    Baltimore
    Country Flag:
    Region Flag:
    Forumers,
    When I am frustrated I tend to take the struggle inward.

    what I decided was the cause of variable rates for the pendulum
    was as simple as the fork getting stopped by the escape pin.

    my idea was to allow the pendulum to swing free and give it an impulse.
    I tried it out and I can't report success,
    I included a few photos

    it does work in that the pendulum swigs free.
    it seemed like it lacked power so I added some weight to drive it.
    It stopped faster and occasionally would jam.

    for this idea to work, a way to reduce the friction is required
    and that would solidly undo the 'ordinary' part of my search.
    photos of the swing-free.
    free-swing.JPG free-swing2.JPG
     

Share This Page