Timegrapher question - what part of the signal is used for beat error, rate error, amplitude ?

[mikepilk]

Just assembled an old AS 984 and put it on the timegrapher. I fitted a new balance staff and had to tweek the hairspring, so put it in beat by eye - usually I can get it close enough.

The pics are of the movement at rest, with a pallet tooth sat on the input jewel -

 

As you can see, the pallet is pretty central in the banking pins, and lines up with escape/pallet/balance/impulse jewels.  Also the movement fires-up as soon as I wind it, no shaking required. So I was confident I was close.

But the timegrapher shows :

I can't believe I'm over 5ms out - I calc that as something like 25° error.  So I assume it's an error with the timegrapher signal. 

If the beat error is incorrect, is the same part of the signal used for rate error and amplitude ?

I have a PDF of a Witschi training course which says that the first noise (the impulse pin striking the forks) is used for rate deviation and beat error, and the third noise (a tooth hitting the locking pallet stone) is used for amplitude.

Do all timegraphers used the same method - what can I believe ?

 

Edited August 11, 2022 by mikepilk

[praezis]

14 hours ago, mikepilk said:

Do all timegraphers used the same method - what can I believe ?

Yes. 
1) Rate and beat error are derived from the distance between tic - toc - tic ... noises
2) Amplitude is derived from the length of one tic noise.

For 1) using the 1st pulse of the beat noise is optimal, as it is the most precise one. But also another pulse, e.g. the higher 3rd pulse of each beat noise can be used. In this case, no amplitude measurement will be possible.

Btw: your graph tells different: looks like a very small beat error!

Frank

[mikepilk]

46 minutes ago, praezis said:

Btw: your graph tells different: looks like a very small beat error!

That's exactly why I'm posting Frank !  The trace, and visual set-up show small beat error, but the display shows 5.5ms.

If you get a very high beat error (6-8ms?) one trace can wrap-round and overlay the other, so it looks like a small error.  But, because it looks to be close visually, I'm sure this is not the case. 

[clockboy]

I have noticed this as well. Despite being confident by visually inspection when on the timegrapher it is out of beat. I don’t know if the watch when running has a slight lag between the horns of the balance folk and impulse jewel. Maybe!!

[mikepilk]

4 minutes ago, clockboy said:

I have noticed this as well. Despite being confident by visually inspection when on the timegrapher it is out of beat. I don’t know if the watch when running has a slight lag between the horns of the balance folk and impulse jewel. Maybe!!

The fact that it starts ticking as soon as I start to wind suggests that it's close to being in beat (?)

The maximum balance speed (i.e. when it hits the pallet) is given by Vmax = (2π x Max Amplitude)/Period 

ref : https://www.vintagewatchstraps.com/shm.php

Therefore 1 ms error equates to 4.2° angular error of the balance. (A number worth remembering if adjusting beat error)

If the error was 5.5 ms, the balance would be 23° out of alignment. As can be seen from the top pic, this is clearly not the case. 

QED - for some reason timegrapher displaying wrong value 

 

[JohnR725]

Ideally for timing machine diagnostics it be nice to give us more than one picture. A minimum would be dial-up and then dial down With a wristwatch crown down. It really helps for diagnostics to see more than one position.

Then timing machines can have a variety of things that will cause issues. The signal they pick up has to be nice and clean. As soon as the signal is not Kleenex or sounds extra stuff then the timing machine can get confused. Or super low amplitudes can cause a confusion where the machine will read something in the middle of the signal and perceive that the amplitude is really high when it's not. So for instance your amplitude looks nice on the machine how does it visually look?

Then the graphical display has an interesting rollover error affect. We've discussed this somewhere in the message board lines will get farther and farther apart as the watches more and more out of beat and if it reaches the edge of the screen though actually roll over and the lines would come back together egg. I don't know where the exact point of time that occurs it's greater than 4 ms. So this is where usually the numeric display is always right.

Than have your banking pins been moved.  If the banking pins art were there supposed to be that will cause issues with beat.

 

[nickelsilver]

The line on the timing machine looks good, and the position of the fork at rest looks good, and the position of the escape tooth on the pallet at rest looks great, so yes, you are in beat!

 

I don't know if there's a gain adjustment on that model, but likely playing around with it, or squeezing the microphone tighter on the movement, or slipping a piece of paper between the movement and the pickup prong of the holder (or something) will clean up the signal so the displayed beat error matches the line.

 

 

[Neverenoughwatches]

3 hours ago, mikepilk said:

 

If you get a very high beat error (6-8ms?) one trace can wrap-round and overlay the other, so it looks like a small error. 

This confused me the first time i experienced it. A wide line spacing from a high beat error bringing the two signals back together. Worth noting for beginners. 

[mikepilk]

46 minutes ago, JohnR725 said:

Ideally for timing machine diagnostics it be nice to give us more than one picture. A minimum would be dial-up and then dial down With a wristwatch crown down. It really helps for diagnostics to see more than one position.

DU and DD look very similar, and the amplitude looks about right to me. 

CR and CD don't look good though - I think it's because the hairspring isn't perfectly vertical between the guide pins and the gap's a bit big. I need to have another go at the hairspring - took some work to get it nearly flat and the terminal curve was all over the place ... and it's as soft as a wet noodle - I distorted it twice just refitting the balance 

I need some tea and a think - it's too hot here for fiddling with watches 

 

[Neverenoughwatches]

16 minutes ago, mikepilk said:

DU and DD look very similar, and the amplitude looks about right to me. 

CR and CD don't look good though - I think it's because the hairspring isn't perfectly vertical between the guide pins and the gap's a bit big. I need to have another go at the hairspring - took some work to get it nearly flat and the terminal curve was all over the place ... and it's as soft as a wet noodle - I distorted it twice just refitting the balance 

I need some tea and a think - it's too hot here for fiddling with watches 

 

A cold beer mike but just one, too many and it wont be just the end curve thats all over the place . I'm having the same problem with a Unitas, got the end curve just nice but a vertical bend in the spring's  height near the stud is doing my head in. I'm becoming tempted to send my lad for some ciders and i haven't drank for 15 years lol

[mikepilk]

If the bends are near the stud, it's often easier to do it with the balance installed.

I recently had to make a vertical bend at the start of the terminal curve and wasn't sure how. I posted it here how-do-i-make-a-vertical-hairspring-bend - as you see, using tweezers pushed in to some soft wood it's surprisingly easy and controllable. It should fix your problem !

I just need to do a twist near the guide pins to get it vertical, I'll try it without removing the balance from the movement. There is also a slight vertical bend, but I might leave that as is - the end of this spring has been tweaked a lot and I fear it may give way soon 

 

[Neverenoughwatches]

9 minutes ago, mikepilk said:

If the bends are near the stud, it's often easier to do it with the balance installed.

I'm  finding it difficult to get in with the balance installed. Whats left is a long sweeping bend thats throwing the opposite end of the h/s up from the stud when installed. I feel like I'm giving it too much stress and then its just laughing at me and returning back to is bent state. Vertical damage is bloody resistant to straightening.  I'm nearing Gerts method of launching it but i will try your idea first . Appreciate that Mike 

[praezis]

@mikepilk
May I add a little adjustment as things are not quite as bad:

Your 5,5 ms above equates an out-of-center error of the balance of 12 degrees.
Still much!

deg = B.E.ms x Ampl x BPH / 23 e5

Frank

[dadistic]

Frank ( @praezis), please excuse my poor math skills

So,  just making sure I'm doing this right,  23 e5 is 23 to the fifth power.

deg = 5.5 X 274 X 18,000 /  23.e+5

12 = 27,126,000 /  23.e+5

23.e+5 = 27,126,000 / 12

2. e+5 = 2,260,500   

You must have rounded

27,126,500 / 2,300,000 = 11.79413 degrees.

Woot!

Thanks!

 

 

[mikepilk]

2 hours ago, praezis said:

@mikepilk
May I add a little adjustment as things are not quite as bad:

Your 5,5 ms above equates an out-of-center error of the balance of 12 degrees.
Still much!

deg = B.E.ms x Ampl x BPH / 23 e5

Frank

My calculation above shows that 5.5ms is 23° out.  Or do I need to halve that ?

The real puzzle is, why it visually looks in line, and the trace shows little separation (0.1ms?), yet it displays 5.5ms?

[praezis]

16 hours ago, mikepilk said:

My calculation above shows that 5.5ms is 23° out.  Or do I need to halve that ?

Obviously yes. My own 1st calculation gave the same result as yours, but I felt, this must be too much.
The manual of a timing machine (Timomat) led me to the correct calculation.

This picture from another forum reminded me of your post. Seems like vertical screen is about 5.5 ms here, so your TG may tell the truth and one line wrapped around once - though it is puzzling.

Frank

Edited August 15, 2022 by praezis