D. I. Y. Watch Timing Machine.

[guidovelasquez]

13 minutes ago, CWRNH said:

What is the common?

It is my empirical and partly technical opinion working with sound. But above 40 dB it is very good. Between 45 and 50 is almost perfect. More than 50 would be the ideal.
The WOS screen is also not used as a reference, as it shows very little noise and the peaks are very high. Furthermore, WOS does not apply any gain. What I interpret that determines that the signal is very good.

[CWRNH]

1 minute ago, guidovelasquez said:

It is my empirical and partly technical opinion working with sound. But above 40 dB it is very good. Between 45 and 50 is almost perfect. More than 50 would be the ideal.
The WOS screen is also not used as a reference, as it shows very little noise and the peaks are very high. Furthermore, WOS does not apply any gain. What I interpret that determines that the signal is very good.

So far the best I have had is 55

[CWRNH]

[guidovelasquez]

4 minutes ago, CWRNH said:

the best I have had is 55

It seems to me an excellent result. In my case I do not exceed 52 dB. But that is enough.

[guidovelasquez]

7 minutes ago, CWRNH said:

the best I have had is 55

I suggest you do a test with eTimer, disabling the filter. With this you can see what is the real value of the noise signal coming from the mic + pream.

[CWRNH]

2 minutes ago, guidovelasquez said:

It seems to me an excellent result. In my case I do not exceed 52 dB. But that is enough.

Here is WOS with a Val7750 movement, SNR is not as good, it is at 30.  Probably because the movement is so quiet and inside its heavy case.

[CWRNH]

2 minutes ago, guidovelasquez said:

I suggest you do a test with eTimer, disabling the filter. With this you can see what is the real value of the noise signal coming from the mic + pream.

Was your 52 results with or without filtering?

[guidovelasquez]

1 minute ago, CWRNH said:

without filtering?

 

[CWRNH]

6 hours ago, guidovelasquez said:

 

Looks like you are getting similar results to what I am.

[CWRNH]

7 hours ago, guidovelasquez said:

 

ETA2892 no filter

 

[guidovelasquez]

7 hours ago, CWRNH said:

ETA2892 no filter

 

The eTimer program indicates in red when the signal / noise has a poor ratio. 41.8 dB es good.

[guidovelasquez]

8 hours ago, CWRNH said:

Looks like you are getting similar results to what I am.

[xyzzy]

18 hours ago, CWRNH said:

What Spectrum analyzer are you using?

I used audacity for those pictures.  But I have also added ability to generate spectrograms into tg-timer, https://github.com/xyzzy42/tg/tree/trentpi/python

Looking at sound again, it seem very clear that something is ringing inside the movement, but this happens only on every other beat.  Maybe pallet fork is vibrating after each beat, but only in one direction, due to some difference in the pallet jewels?  Or escape wheel vibrates depending on which pallet jewel is locking it?

Edited April 30, 2021 by xyzzy

[CWRNH]

19 hours ago, xyzzy said:

I used audacity for those pictures.  But I have also added ability to generate spectrograms into tg-timer, https://github.com/xyzzy42/tg/tree/trentpi/python

Looking at sound again, it seem very clear that something is ringing inside the movement, but this happens only on every other beat.  Maybe pallet fork is vibrating after each beat, but only in one direction, due to some difference in the pallet jewels?  Or escape wheel vibrates depending on which pallet jewel is locking it?

I downloaded the source code...  No idea how to use it!  LOL  Any guidance would be greatly appreciated...

[guidovelasquez]

3 hours ago, CWRNH said:

No idea how

Well, if you are interested in investing a few hours here goes:
Source code is like the electronic diagrams of an electronics project. You must run them. But you can modify it according to your goals. Thus, from the source code, you will produce your own version of the program or, if you do not modify anything in the source code, produce the program in someone else's version.
Necistaras as in my case.
1.- Atom to rewrite the source code.
2.- MSYS2 to "compile" the program.
The guidance on how to proceed is in this link

https://github.com/vacaboja/tg

and here I put a clipping of where you should look.

[CWRNH]

4 hours ago, guidovelasquez said:

Well, if you are interested in investing a few hours here goes:
Source code is like the electronic diagrams of an electronics project. You must run them. But you can modify it according to your goals. Thus, from the source code, you will produce your own version of the program or, if you do not modify anything in the source code, produce the program in someone else's version.
Necistaras as in my case.
1.- Atom to rewrite the source code.
2.- MSYS2 to "compile" the program.
The guidance on how to proceed is in this link

https://github.com/vacaboja/tg

and here I put a clipping of where you should look.

thank you

[CWRNH]

On 4/30/2021 at 1:45 PM, guidovelasquez said:

with a tiny wristwatch movement and the filter on i get well over 50, without the filter i get low to mid 40s. 

[Bonzer]

On 4/30/2021 at 3:15 AM, CWRNH said:

What amp design have you used to achieve this result?

Also, is this still using the clip-on piezo mic?

I've tried to look back at this huge thread, but not really got the full information.

Thanks!

 

[JohnR725]

1 hour ago, Bonzer said:

What amp design have you used to achieve this result?

Also, is this still using the clip-on piezo mic?

I've tried to look back at this huge thread, but not really got the full information.

have you look at the website for the software he has the official amp.

then just because a lot of us had our own designs sprinkled through the current 44 pages of discussion. Which probably does make for interesting and confusing reading

then there's the confusion factor for all of us? The picture above looks nice doesn't it but were missing details as you've asked about. Like a pocket watch will look nicer than a wristwatch. We really all need to buy the exact same watch for running tests so we all of the same thing and we can compare. or we need to convince the Chinese to clone the witschi test module so we all have the exact same thing so pictures like the one posted above would have meaning because we could all try the same thing and see that ours does the same thing.

[xyzzy]

On 5/1/2021 at 8:00 AM, CWRNH said:

I downloaded the source code...  No idea how to use it!  LOL  Any guidance would be greatly appreciated...

If you go to the page and scroll down, there are instructions for building for a number of different operating systems.  I have written more instructions for Linux and MacOS, but do not use Windows, so I can not improve those myself, which is one of the reasons they are probably the worst.  Also because Windows makes these things harder.

I can and do provide installable packages for Linux, where to get those is also described if one scrolls down, but do not know how to do this for Windows or have a Windows computer to do it on.

[guidovelasquez]

17 minutes ago, JohnR725 said:

We really all need to buy the exact same watch for running tests so we all of the same thing and we can compare.

We do not necessarily have to have the same clock, the same amplifier or sensor or microphone.
Reasonable results can be obtained in a number of ways. Also keep in mind that programs or software can handle very poor sounds. Of course the ideal is a good sign.
Here I leave the results and the photos as I obtain them.

rec0504-131406.wav

 

[JohnR725]

20 hours ago, guidovelasquez said:

We do not necessarily have to have the same clock, the same amplifier or sensor or microphone.
Reasonable results can be obtained in a number of ways. Also keep in mind that programs or software can handle very poor sounds. Of course the ideal is a good sign.
Here I leave the results and the photos as I obtain them.

personally I still like the witschi test device but it probably cost an arm and a leg and six fingers and your firstborn.

so I'm intrigued and enthusiastic about your idea but were still stuck with what I'm trying to get at is we need a reference device. It should simulate a pocket watch and a little tiny ladies watch with set parameters of all of us in the entire group put it on our what ever we all would get the exact same results which is hard to do if everyone has everything different.

The problem I've seen when answering questions on the discussion group at least watch repair questions is my watch is doing this well is it really? Are we basing answers on faulty diagnostics. This actually happened in a rather lengthy discussion with a little tiny ladies watch where we just weren't sure where we stood with us faulty diagnostics or faulty watch. We needed a tiny reference to verify that the timing whatever was working.

Or here when people put their circuits together how do they know it's actually working out of they know it's sensitive enough.

I was just thinking of you all have the same watch at least we would all have the same results but you indicate that's not necessary so I'm looking forward to your ideas of solving this problem.

then probably too many variables but I wonder if we could take your wave file for instance play it on a computer perhaps a small speaker because it has a moving coil I wonder if that could simulate a watch? But were still stuck with a standard if your computer is outputting the volume at twice the level of mine were measurement results may look different.

 

[guidovelasquez]

6 hours ago, JohnR725 said:

We needed a tiny reference to verify that the timing whatever was working.

Well, you're right. We would need a very reliable standard to do our tests.
Here I have a proposal. It is not created by me. It comes with Graham Baxter's eTimer software. It is a waveform in wav format.  Is 2min and 40 seg, the sound of a 18000 BPH or 5BPS clock. Executed properly and with any of our timing machine programs, it should give the following results. -1 second of disface per day. 0.0 beat error and 262 degrees of amplitude when the lift angle is set at 52 degrees. Not much but it has helped me to learn.

 

Edited May 5, 2021 by guidovelasquez
delete sound file.

[guidovelasquez]

48 minutes ago, guidovelasquez said:

It is a waveform

 

1461445751_Cal.2min40seg.wav

[LittleWatchShop]

9 hours ago, JohnR725 said:

we need a reference device

Frankly, the reference device is an atomic clock.  So most, if not all, of this thread is OCD.  Interesting...fun...and useful in a vacuum, but at the end of the day, a mechanical movement pales in comparison to digital electronics. So...why bother? 

Fine, I am OK with doing your best, but polishing a turd has no real benefit.  Accept that mechanical watches are not accurate and move on.  Value the beauty of their architecture and reset them once per week.  Rent or download the movie "Castaway" and face reality!