timing machine

[gerene]

I do have this timegrapher a few years now and I am very happy with it. It really helps regulating a watch. It shows you the number of seconds a day slow or fast, amplitude and beat error together with the traces.
I do understand that a perfect watch would have a single horizontal line in all positions (but never encountered one ).
I was wondering if someone knows what an acceptable beat error would be. Is there an upper limit where it would not pass… Some watches are relatively easy to correct the beat error (e.g. etachron regulator system) whilst others are more involved to put in beat: turning the hairspring collet with the risk of damaging the hairspring. In the latter case I would prefer not to touch it when it is within tolerance.
What do you guys do?
Also what is considered to be a good amplitude and when should corrective action be taken?

thanks,

Jan

[david pierce]

Jan,
In short you want the beat as even and close to one solid line as you can get it. A number of other timing factors rely on this such as lift angle and amplitude. Once the beat is even you need to find the lift angle for the watch you are working on. The lift angle is the angle formed when the roller jewel makes contact with the first pallet fork and and releases contact with the second pallet fork. This angle can be either found from a chart supplied by the manufacturer or calculated with the following procedure:
1) Run the power out of the mainspring.
2) Wind the watch one click at a time and watch the balance wheel oscillate back and forth until the wheel oscillates 360 degrees in each direction. Since each rotation of the balance wheel moves the pallet in one direction, the amplitude is called 180 degrees.
3) Look at the watch timer and adjust the timer until the amplitude reads 180 degrees. You have now calibrated the timing machine to your watch. The lift angle will then show up on the screen.

Lift angles usually fall in around 49-53 degrees but this does not mean that this is the correct angle for the particular watch your are working on. The amplitude is usually around 240 degrees and up. The amplitude is directly proportional to the amount of force required to release the pallet jewels from the escape wheel lock. If it is too low the pallet jewels can stick a little and time inconsistently. If it is too high the roller jewel can roll around too far and smack the back side of the pallet fork.
david

[gerene]

Thank you David for your extensive reply.

I understand that having the beat as close as possible to one line is the ultimate goal. However, getting it there is not always easy. It is not difficult with regulators having a movable stud, but more so when the hairspring collet has to be turned. In the first case the balance stays assembled and the watch ticking on the timegrapher. One can then move the stud untill the lines come together. When the collet has to be moved, one has to make a mental note of the direction and amount of correction to apply, disassemble the balance and move the collet, with the risk of damaging the hairspring in the process.
Therefore I was wondering what would be an acceptable beat error. For instance if it is 2 ms or less it is better left alone.
What are your thoughts about that?

Jan

[david pierce]

Jan,
As I am not a professional watch repair guy and have no intentions of ever becomming one, I personally would fuss with it until it was as perfect as I could get it. A true professional will get it to work with the least amount of time and parts that they can get by with and charge the maximum amount of money to the customer that they can get away with.
Two milliseconds will give you an error of + or – 2.88 seconds/day or 86.4 seconds per month. That certainly sounds reasonable for a mechanical watch but keep in mind that the other timing factors are riding on the beat being even. A digital watch should be around + or – 5 seconds per month.
david

[gerene]

Hello David,

I am not a proffessional watch guy either, nor intend to become one. Just having fun in trying to get these little marvels to work again as good as I can. I too try to get them as perfect as possible and reasonable.
I thougth that the 2ms beat error referred to the difference in time for the oscilation in one direction versus the other. Don’t know if I explained this clear enough and if it is correct. Of course if this is true then the proportional error would depend on the beat rate.
I did not understand your calculation of 2.88 seconsd/day, would you care to explain? I am trying to understand this fully and would appreciate everybody’s opinion and knowledge.

Jan

[david pierce]

Jan,
I took .002 seconds X 60 X 24 = 2.88/day. 2.88/day X 30 days = 86.4/month.
david

[tmac1956]

@david pierce wrote:

Jan,
I took .002 seconds X 60 X 24 = 2.88/day. 2.88/day X 30 days = 86.4/month.
david

Except in a leap year…

Tom

[gerene]

Oh, now I understand what you where referring too. I was confused because I was convinced that the beat rate should be in the equation .

Well, to some it up I think that the beat error should be as small as possible and preferably zero. If the watch is not in beat, the arc of swing will not be even on both sides and amongst other thing, most likely the watch will not start when wound.

Jan

[david pierce]

Jan,
One also must consider the escape wheel and pallet functions that are affected such as impulse, drop, lock, draft and slide.
These functions must have sufficient time to properly take place. If the watch is out of beat then the allowed time is not evenly divided between the entry pallet and the exit pallet. There may be more than enough time for one half and not enough time for the other half. All of the subsequent settings, angles, and positions need to be made around an on beat situation.
david

[gerene]

That is correct David and this was actually the base for the original question. How much time, in function of the beat rate, is used by those functions or formulated otherwise, how much beat error can be tollerated before there is too much adverse effect. :

Jan

[david pierce]

Jan,
I would specualate that the larger watch factories have also asked the same question. They certainly have the financial muscle to get ultra high speed recording equipment and analyze the individual events taking place. I think another less expensive way to find out would be to take a watch, hook it up to the timer, and then intentionally turn the hairspring collar to throw the watch out of beat. Then look at the readings to see how the overall performance is affected. I would think that anything that comes in around 1 minute a month or less would be good. This would put the beat at .00138 or about 1 1/2 milliseconds. If I am wrong go ahead and shoot me but remember, no one else stuck their neck out on this one.
david

[arutha]

Hi Jan,
found this on the NAWCC forum which should give you a good idea.
http://mb.nawcc.org/showthread.php?28397-What-is-an-accepted-Beat-Error-to-be-considered-In-Beat
Paul

[gerene]

Hi David,

I have no intention to shoot anyone and I appreciate that you took the time to think about this and reply. I am just trying to figure out what would be a good and acceptabe value and enjoy exchaning thougths on it. I realize it is not an easy question to answer as is also said in the thread Paul is referring to. But I thought it is an interresting one. Don’t feel offended by anything I said, it was certainly not my intention 😳 and keep in mind that English is not my mother tongue, so I might sound harsher than I intend to.

Paul,

thanks for the link, it indeed gives a good idea on what others (more knowledgeable than me) think about it.

Jan

[david pierce]

Jan,
Far from it. These types of questions inspire research and thought. Thanks for putting the topic on the forum.
david

[arutha]

I have had to re think how I say things in text form as it can be a bit ambiguous in certain situations. We all play very well together on this forum which just about makes me redundant apart from re-sizing the odd picture, deleting duplicate posts and moving the odd post to the correct area of the forum, just how I like it
Paul.

[Author]

Posts