Reducing beat error - why does it matter?

Discussion in 'Watch Repair' started by Steve Barnes, Mar 4, 2020.

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

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    After reading many discussions and watching numerous videos on how to reduce beat error, I am told that there is no relationship between beat error and time-keeping - i.e. watches can be significantly 'out of beat' but still keep good time.

    I have searched (here and elsewhere) for views on why it is considered so important to reduce beat error to as close to zero as possible. A target of at most 0.1ms is often recommended.

    If possible, a brief (or even not so brief) explanation of why it matters to reduce beat error would be greatly appreciated.

    Regards
     
  2. Smudgy

    Smudgy Registered User

    May 20, 2003
    2,872
    23
    38
    Country Flag:
    Region Flag:
    One reason for reducing beat error is so that the watch is reliably self starting. When the beat is off even by small amounts it will sometimes hang when wound from a non-running state. The watch can be started by giving the watch a sharp twist in the plane of the balance wheel, but it is very much preferred the the watch start itself.

    There are undoubtedly more reasons, but to my knowledge the self-starting is the most prevalent and what comes to mind
     
  3. Skutt50

    Skutt50 Registered User

    Mar 14, 2008
    3,554
    125
    63
    Male
    Gothenburg
    Country Flag:
    I want to remember that the amplitude increases when the beat error goes down.
     
  4. praezis

    praezis Registered User

    Feb 11, 2008
    461
    26
    28
    Germany
    Country Flag:
    Region Flag:
    That is new to me. Can you explain the reason?

    Main reason why beat error gets so much attention:
    - Chinese TMs display it
    - it is the value, that can be adjusted most easily by amateurs.

    In fact beat error (already that name frightens!) is the least important value of all, even considerable errors level themselves out with every 2nd beat and are noticed on timing machines only.

    Frank
     
  5. DeweyC

    DeweyC Registered User
    NAWCC Member

    Feb 5, 2007
    2,124
    767
    113
    Watchmaker
    Baltimore
    Country Flag:
    OK. This has all been researched by the Swiss of course. According to Tony Simonin, there is no impact on amplitude or rate below .1ms beat error.

    The reason beat error matters, and why it has to be measured and not visually set, is that in the lever escapement the balance swing is actually unequal in the two vibrations. This is due to such things as the excursion is resisted in one direction by balance spring contraction and assisted in the other direction by balance spring expansion.

    For a complete analysis of the natr5ual escapement errors, see Jendritski Watch Adjustment.
     
  6. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    But why does this matter?
     
  7. Al J

    Al J Registered User

    Jul 21, 2009
    594
    69
    28
    Most factory tolerances are not that tight. The most common maximum I see for beat error is given as 0.8 ms maximum.

    Keeping the beat error low can help in particular when a movement has a high enough amplitude to potentially cause rebanking. For example I have an Omega Cal. 321 I'm working on right now that had quite high amplitude, and this caliber will rebank in the 310 to 315 degrees of amplitude area when the beat error is low. Keep in mind amplitude is a value that is displayed on most timing machines as an average of the amplitude in each direction. When the beat error is higher, the greater amplitude in one direction can cause the impulse pin to travel far enough around to contact the outside of the fork horn, and this initiates the cycle of rebanking.

    There is no adjustable stud carrier on this watch, so the collet has to be turned to adjust beat error. Right after service and first measurements on the timing machine showed a higher beat error - around 3 ms, and rebanking at 305 degrees of measured amplitude. In this scenario the average amplitude wasn't enough to cause rebanking, but the amplitude in one direction was. Simply correcting the beat error was enough to stop the movement from rebanking.

    Cheers, Al
     
  8. pocketsrforwatches

    pocketsrforwatches Registered User
    NAWCC Member

    Feb 13, 2008
    274
    10
    18
    Retired from IBM, now a Watchmaker
    Ticonderoga, NY
    Country Flag:
    Region Flag:
    #8 pocketsrforwatches, Mar 5, 2020
    Last edited: Mar 5, 2020
    "I have searched (here and elsewhere) for views on why it is considered so important to reduce beat error to as close to zero as possible. A target of at most 0.1ms is often recommended."

    Are you sure it isn't below 1.0ms?

    0.1ms is near perfect beat.
     
  9. DeweyC

    DeweyC Registered User
    NAWCC Member

    Feb 5, 2007
    2,124
    767
    113
    Watchmaker
    Baltimore
    Country Flag:
    Oops. Thanks. decimal should not be there. I just think in decimals because I make sure beat error is always under 1ms (at least that is the story I am sticking with)

    ALJ reported a good example.

    Why does it matter? It impacts total amplitude and rate stability. This is all explained in Jendritski under natural escapement errors. I also cover natural escapement errors here: https://www.historictimekeepers.com/documents/Watch%20Adjustment.pdf

    If you have access to the Gibri book from 1904 he covers it as well. I suspect this work is the foundation for modern horological theory.

    For completeness, natural errors in the lever escapement yield a net loss in rate. These are all explained by high school physics and are not rocket science. The winding and unwinding of the balance spring is but one such natural error; as is the draw of the pallets that help with secure lock. Impulse after the dead point (the impulse jewel at rest in a properly set up escapement) will result in a loss as well.

    To square the circle, if the watch is out of beat, this means at impulse (when the pallet slot hits the impulse jewel) the balance spring is not at its neutral force. The balance spring is either winding or unwinding depending upon which way the collet is biased. This introduces an excessive natural escapement error.
     
    GeneJockey and gmorse like this.
  10. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    Thank you Mr Clark. I have now read your explanation of 'Natural errors of the lever escapement' in your work, and all is becoming much clearer.

    You write: 'If the balance spring is still winding when it should be unwinding (beat error) after the
    balance has passed the neutral point, it will impact timing.'

    And also, 'A beat error of under 1 millisecond has no measurable impact on timing,' which you recall above.

    I take from this that the suggestion - 'there is no relationship between beat error and time-keeping' - is therefore not true.

    The reason I ask about beat error is that I bought a Timegrapher recently to check several pocket watch movements I have which work - work in the sense that they wind up, the balance wheel continues to oscillate and the hands turn. Also to learn about how to adjust them.

    My confusion about beat error not affecting timing (which I had read elsewhere) came about when I tested my grandfather's pocket watch of 1888 which had been repaired - at some expense! - and which keeps very good time.

    The Timegrapher records a beat rate of 14400, but displays only scattered dots and cuts out before showing the beat error, the amplitude or the rate. As the working range is 0-9.9ms of beat error, I assumed that this watch had an error over 9.9ms.

    I'm now wondering, with regard to 'natural escapement errors, if there is another reason, as the 'tick' sounds are not continuous. There are five ticks, a missed beat, five ticks, a missed beat etc before the Timegrapher cuts out and starts a new test.

    And yet it keeps good time over 24 hours and more.

    I'm not expecting an instant diagnosis but this pattern may suggest something obvious to forum members.

    Regards
     
  11. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    And now it's doing three ticks, missed beat, three ticks, missed beat etc.
     
  12. Chris Radek

    Chris Radek Registered User
    NAWCC Member

    Apr 13, 2014
    572
    190
    43
    Male
    Watchmaker (now accepting new customers)
    Lincoln, NE, USA
    Country Flag:
    It sounds like this particular watch may have an unusual correct rate. What is the watch?

    Although my timegrapher can be set to many rates, it's very bad at auto-selecting the correct one unless it's one of a few of the very common, like 18000 or 21600.

    (It is also not hard to find watches that have correct rates these machines do not support at all.)
     
  13. DeweyC

    DeweyC Registered User
    NAWCC Member

    Feb 5, 2007
    2,124
    767
    113
    Watchmaker
    Baltimore
    Country Flag:
    Many instruments will not work correctly if the beat is way out of range. So that needs to be corrected to get you within the range of your instrument. This is done by visually looking to adjust the pallet fork lever is in between the two banking pins at rest. But unless you are experienced, it is very easy to damage the balance spring.

    I have two instruments. An MU-700 at my bench for 98% of my work. Then there are odd ball watches like very early Americans that had beats of 16200 and such. For this (and other purposes) I use the microset.

    What is the make and model of the watch?
     
  14. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    The watch is an English lever escapement, case (hallmarked 1888) and movement made by I.J.T. Newsome, of Coventry.

    This may be of no significance but it has an 'Improved Safety Wheel, Patented Mar 3rd 1886'.

    Pallet fork lever at rest is in centre of banking pins.

    I'll test a range of beat rates. Higher ones look promising - 18000, for example, as the ticks are continuous but no beat error reading etc.
     
  15. Chris Radek

    Chris Radek Registered User
    NAWCC Member

    Apr 13, 2014
    572
    190
    43
    Male
    Watchmaker (now accepting new customers)
    Lincoln, NE, USA
    Country Flag:
    Then its correct rate may not be supported by the machine. The only way to know the correct rate for sure is to count the teeth on the wheels and pinions (if it has a second hand you only actually need to count a few.)
     
  16. gmorse

    gmorse Registered User
    NAWCC Member

    Jan 7, 2011
    11,281
    1,580
    113
    Male
    Retired from Xerox
    Breamore, Hampshire, UK
    Country Flag:
    Region Flag:
    Hi Steve,

    English levers typically beat at 14,400 or 18,000, but there are exceptions. I'd expect one as late as your Newsome to be 18,000, but most unlikely to be any higher.

    If you want a timing system that allows you to specify any beat rate you like within a very wide range, (from 60 to 1296000 bph), I can strongly recommend the Delph Electronics eTimer package, which runs on a Windows platform and you can download in a trial version for free. The free version is fully featured but won't accept a microphone input in real time; however it will analyse a recorded waveform from something like Audacity, (also free).

    Regards,

    Graham
     
  17. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    Thank you, gentlemen. At 15360 (after settling down), I get a beat error of 1.2, amplitude 153, rate +137s/day. At 15600, I get beat error of 1.2ms, amplitude 165, rate ----s/day. Above or below don't work.

    That is very satisfying, even if I don't yet know what a rate of ----s/day means. But I now have enough new knowledge to keep me occupied for a while.

    Thank you all.

    Regards
     
  18. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    Apologies, Chris and gmorse, I hadn't seen your posts before I sent above reply. Will come back later (European time).
     
  19. Chris Radek

    Chris Radek Registered User
    NAWCC Member

    Apr 13, 2014
    572
    190
    43
    Male
    Watchmaker (now accepting new customers)
    Lincoln, NE, USA
    Country Flag:
    The correct rate might be 15428.5714, which is a configuration I found recently in an English watch: 4th wheel 60 teeth, escape pinion 7.

    Your beat error reading is probably right, but the amplitude is going to be way wrong for an English lever unless you can determine the lift angle.
     
  20. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    Yes, Chris, it is indeed what you say. FW 60, EP 7, EW 15. And thus the beat figure that you give.

    Well, well, well. So nearer 15360 than 15600. Perhaps that's why there was no reading for the rate on 15600.

    For some reason, I'd thought that beat rates would be integers. But not so.

    I will pursue the question of the amplitude and try to work out the lift angle (without taking the watch apart).

    Graham, thanks for that link. Have downloaded and looked at the Delph timer. At first glance the beat settings are also integers.

    Does this mean that for watches that have beats such as 15428.5714, electronic timers show results for the nearest beat setting available?

    That beat rates may not be integers raises other questions too, but for another time, I think. When I've done more research - and read the whole of 'Watch Adjustment' by Mr Clark. ;-)

    Regards
     
  21. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    Just an update Chris. The earlier test was when wound only a few turns. Fully wound, it is reading for 15360: 0.4ms beat error, 212 amplitude.
     
  22. gmorse

    gmorse Registered User
    NAWCC Member

    Jan 7, 2011
    11,281
    1,580
    113
    Male
    Retired from Xerox
    Breamore, Hampshire, UK
    Country Flag:
    Region Flag:
    Hi Steve,

    No, they are integers in the pull down list, but you can type any value between 60 and 1296000 bph, with a single digit to the right of the decimal point, into the box at top right, (giving a resolution of one beat in ten hours), which will get you as close as you need. It also allows you to specify the lift angle for levers the same way. For cylinders and verges you can set it to clock mode, and it has settings for coaxial and tuning fork escapements as well.

    Regards,

    Graham
     
  23. GeneJockey

    GeneJockey Registered User
    NAWCC Member

    Mar 2, 2012
    1,598
    618
    113
    Male
    Scientist
    San Francisco Peninsula
    Country Flag:
    Region Flag:
    If you have a smart phone, you could take a slow motion video of the watch running for 10 seconds, and observe how many degrees the balance travels from stop to stop. Divide by 2, and there's your amplitude. Then you can go back and adjust lift angle on the machine till it gives the correct amplitude.
     
  24. karlmansson

    karlmansson Registered User

    Apr 20, 2013
    2,612
    115
    63
    Doctor
    Linköping, Sweden
    Country Flag:
    A more direct way to do this is to have the movement in the pickup of the timegrapher, put a dot on the balance rim with a fine tipped marker (of course not a good idea if you don't have a way of cleaning the balance afterwards) and then wind the watch. When the dot appears as a single dot you will have reached 180 deg amplitude. You need to do this in intervals though as the sample time of the timegrapher is usually around 10s. When you are certain that you have a correct reading, you adjust the lift angle on the time grapher until the amplitude reading is 180 deg.

    Regards
    Karl
     
  25. Smudgy

    Smudgy Registered User

    May 20, 2003
    2,872
    23
    38
    Country Flag:
    Region Flag:
    The rate of ---- s/day is your error. If you have a value of +32 s/day that means your watch will gain 32 seconds a day (which is a lot). The error will probably change depending on the habits of the person wearing the watch (more active users will lose more time than sedentary people), and may change with the mainspring unwinding due to isochronal error.
     
  26. gmorse

    gmorse Registered User
    NAWCC Member

    Jan 7, 2011
    11,281
    1,580
    113
    Male
    Retired from Xerox
    Breamore, Hampshire, UK
    Country Flag:
    Region Flag:
    Hi Steve,

    I think if your Timegrapher is showing '----s/day', it either means that it can't calculate a value or the value is outside its range. Probably the former since it couldn't detect the correct beat rate.

    Regards,

    Graham
     
  27. Smudgy

    Smudgy Registered User

    May 20, 2003
    2,872
    23
    38
    Country Flag:
    Region Flag:
    I thought he was just using the dashes to represent whatever random number may be shown there instead of putting in a particular digit.
     
  28. Steve Barnes

    Steve Barnes Registered User

    Apr 16, 2019
    41
    12
    8
    Male
    Roanne, France
    Country Flag:
    When ignorance is NOT bliss! Having discovered how much I don't know about rate, amplitude, beat error, lift angle, beats per hour (not to mention 'natural escapement errors' etc), I now need to go off and do many more timing tests to get to grips with this subject.

    Meanwhile, thank you all for your help as I now have lots of practical info to work with (Karl, I like the dot idea - my felt-tip's at the ready!)

    Regards

    SteveB
     

Share This Page