WatchUSeek Watch Forums banner

421 - 440 of 448 Posts

·
Registered
Joined
·
273 Posts
Hello Klaus. Thanks You for reply. I test a Citizen 8110A and i found 39deg lift angel for this movement somewhere. I will try 52.

Edit: Yeah You're right - with 52deg lift angel I have something about 225-230deg of amplitude.
my fault, you were absolutely right. I missed the CTZ8110A with its "strange" liftangle. There is even a liftangle given of 36° (watchguy.co.uk).
"Strange" because I don't understand (by now). Maybe in a few years I'll comment on this once I got it :-d:-d
Nevertheless the escapement is in need of an oil change.
 

·
Registered
Joined
·
1 Posts
Firstly, let me say thank you, contrate_wheel, for developing and releasing this software under the GPL. I was about to buy a Timegrapher 1000, but then found Tg, which is definitely the way to go. I've used Tg quite a bit now, and think it's great! I'm most grateful for this.

I have some improvement suggestions:

1. Tg displays a near-instantaneous value for a watch's current rate. The current rate is useful when regulating. However, once the rate has been set and looks to be satisfactory, I like to run some longer tests, in different positions, to check whether I really do have the rate right, or it was just a short-term blip. Sometimes, the rate of a watch can wander slightly over a period of several minutes, and then swing back again. In this case, having an instantaneous rate measurement isn't useful on its own. A longer, averaged value is also needed. (Whilst perfectionists might only be happy with a dead straight line on the graph, I'm a beginner, and am happy with a watch that just keeps good time overall.)

So I would find it very useful if Tg, in addition to what it already displays, also displayed an averaged value of the rate. This would be a very simple calculation based on the first and last measurements of the trace, taken either from the start time, or the last time the "Clear" button were pressed.

I know a date-time display is on the to-do list. To further facilitate averaging, an elapsed time display (from program start or "Clear" event) would also be useful.

The alternative to this suggestion is leaving the watch running for a day or so, and comparing it manually against an atomic clock time reference. This works fine, but it would be far more convenient if Tg would do this itself over a shorter, user-selectable measurement period. The user would "select" the measurement period by starting Tg or clicking "Clear", then waiting the chosen length of time.

I considered whether averages of the beat error and amplitude would be useful too. Such might be nice, but I don't think they lend themselves to a simple start- and end-point calculation. Certainly, a rate averaged over the measurement period is most needed. I don't know how anyone else here feels about this, but I regard an averaged rate display as an essential feature.

2. For the beat error, Tg displays an absolute value in milliseconds. When adjusting for a zero value one can easily overshoot slightly, whereupon this absolute value is not very helpful, because one doesn't know which side of zero the watch is set. This might be in line with what other timing machines do, but I would find Tg more useful if it displayed a signed number instead. That would make adjustment easier, because one would know which way to adjust when close to the target.

I suppose such a change would be trivial code-wise.

3. I have a suggestion for atomic clock calibration that wouldn't involve NTP, and require no Tg mods at all: write a separate, command-driven program that users run once only, to compute Tg's calibration offset by referring to the computer's sound card and an atomic clock.

This program would prompt the user for just three inputs: how many hours they want the calibration period to last, a "Start" prompt and a "Stop" prompt. For example, the user might want to run this for twenty-four hours, so they would enter 24, then wait for their atomic clock reference to reach a convenient time, and press Enter (Start). Next day, slightly in advance, they would check the atomic clock, and at the right moment press Enter again (Stop). The program would use the reference time interval between these two inputs, combined with sound card timer readings, to compute and display Tg's signed calibration offset. Again, a very simple calculation.

This method would depend on the user's ability to press Enter at an exact moment in time. By choosing a sufficiently long calibration period, accuracy could be comparable to the stability of the sound card's time reference. It would eliminate the need for a quartz-mechanical watch for calibration, and the uncertainty of knowing whether it applied rate compensation. It would also eliminate the uncertainty of a user adding their own bias to compensate for the drift of their cheap quartz watch (should it be added or subtracted?).

This suggestion need not replace Tg's existing calibration procedure, but supplement it.

Alternatively, this function could be incorporated into Tg directly somehow, and I leave it to contrate_wheel to figure out something that wouldn't mess up its nice, clean interface.

Note: The code to implement this might need measures to prevent out-swapping during the measurement period, or have an extra input to trigger an in-swap shortly prior to the final input. That would make four inputs in total, but still very simple.

4. I removed Suggestion 4, as I hadn't thought it through properly. It depended on Tg's yet-to-be-implemented date-time display. All I will say is that a calibration-adjusted date-time display based on the sound card's timer would be very useful.

5. This one is just a minor niggle: Tg forgets what the current working directory is between multiple file saves and loads.

If you are saving a number of snapshots multiple levels deep in your folder structure, you have to re-browse to the required folder each time, starting at the top. This can get tedious. It would be nice if Tg had an internal variable to re-supply the current working directory as a default for each file save/load call. Browsing to a different folder and saving/loading should update this internal variable.

The same applies to the "Save All Snapshots" and "Open" options. Could these all pick up and use the current working directory?
 

·
Registered
Joined
·
525 Posts
I stumbled onto TG yesterday, what a great program! I have an old Vibrograph MU700, I'm using the multi-position watch holder with a little circuit between it and the sound card. It works great.

It appearss that if the sound card does anything other than listen to the watch the timing gets disrupted and is not stable. This may be different with different sound cards. It would not be difficult to make a one second 'tick' machine with an accurate and temperature controlled crystal oscillator that could serve as the time standard. It could go in on one channel, the watch on the other. An alternative to a dedicated ticker is a function generator. I don't know how difficult it would be to implement in the code to accommodate an external reference clock, but it seems that it has the potential to make TG very accurate.

Anyway, thanks for all the hard work!
 

·
Registered
Joined
·
72 Posts
For those not having quartz watch to calibrate the app, I would suggest you just buy $5 quartz watch, calibrate it in say 10-20 days interval (how-to: https://www.watchuseek.com/f9/methods-determining-accuracy-watch-382752.html ), keeping them on shelf, and then calibrate the TG app with that watch.

Among quartz watches I've got, there's very cheap Q&Q watch, that I'm tracking for 6 month now (android watch track app WatchCheck) and it has gained less than 5 sec (making it +10 spy on-shelf)...
 

·
Registered
Joined
·
87 Posts
This is really awesome! I was looking for something on the app store (mainly looking at the timegrapher apps) and was trying to install Hairspring (which didn't work). Then I stumbled upon this! I'm at work now, so I haven't had the pleasure of trying it out (the only mic I have is a Bluetooth headphone mic, and it did not work, but I checked the System settings and it wasn't really sensing sound either when I tapped it, so it's the mic's fault).

This would be awesome if we could port it with React-Native and get it on IOS and Android!
 

·
Registered
Joined
·
20 Posts
This is really awesome! I was looking for something on the app store (mainly looking at the timegrapher apps) and was trying to install Hairspring (which didn't work). Then I stumbled upon this! I'm at work now, so I haven't had the pleasure of trying it out (the only mic I have is a Bluetooth headphone mic, and it did not work, but I checked the System settings and it wasn't really sensing sound either when I tapped it, so it's the mic's fault).

This would be awesome if we could port it with React-Native and get it on IOS and Android!
Good Idea. |>:-!
 

·
Registered
Joined
·
3 Posts
Hi guys,
Anyone know what happend with download links on project's page? I used this program some time ago to regulate my 7s26 and it's doing great gaining +3 seconds per day on avarage. Now I wanted to download and use it for my next project and all links disappeared.
 

·
Registered
Joined
·
2,159 Posts

·
Registered
Joined
·
87 Posts
Discussion Starter #432
Hi guys,
Anyone know what happend with download links on project's page? I used this program some time ago to regulate my 7s26 and it's doing great gaining +3 seconds per day on avarage. Now I wanted to download and use it for my next project and all links disappeared.
Thanks a lot for letting me know: I accidentally deleted the files! Admittedly, I got a bit scared when I saw that all links were indeed missing. Fortunately, after reviewing my logs, it turned out that it was simply my fault for inadvertently deleting the wrong folder. Now they should back online...
 

·
Registered
Joined
·
1 Posts
Thanks a lot for letting me know...
@contrate_wheel

Thanks a lot to you for this extremely useful, well thought out and made software!
I have one question: I would like to reduce the threshold used to set the time mark in the main tick for calculating the amplitude. I use a normal microphone which works well due to your superb filtering. However, there is always some echo in the signal, and therefore the time mark is set always too late resulting in a false (small) amplitude calculation.
I found in "algo.c" a "compute_amplitude" proc with the line
double threshold = fmax(.01 * glob_max, 1.4 * max);

Are these the factors setting the threshold amplitude relative to max? Can you please advise me on the correct factor to play with?
Thanks again
 

·
Registered
Joined
·
437 Posts
Guys I've just downloaded version 0.5.0 and am using this for the first time.

I need some help, wondering if there is a user guide anywhere? I can't seem to find things like amplitude.

Thanks!
 

·
Registered
Joined
·
29 Posts
I am confused about how to interpret these results in this particular case.

I captured this snapshot of an Omega Speedmaster cal. 863. It has not been serviced for 5 years, and it is fully wound.
The shape and location of the peaks do not move much when I change the position of the watch. It seems quite stable.


Rate: -5 s/d (-4 to -16 s/d depending on position)

Beat error: 0.2 ms (number looks good at first glance...)
Should I trust this value?
The top trace shows that the software missed the first peak, but it seems very low and there is more than 3 peaks in total?​

Amplitude: 288 deg (number looks good at first glance...)
But the first peak was missed... what is correct here?




(If I could wish one thing, it would be to add a slider/value for the trigger sensitivity for finding the first peak)

Capture.PNG
 

·
Registered
Joined
·
2 Posts
I would trust beat error, because it is calculated from the loudest sound, wich is usually the the locking.
Regarding Amplitude, the peaks displayed are different from theory. This might be an effect of filtering, FFT calculations, or because real lif is different from theory.
Not sure the peak not detected is the first one from escapement
 

·
Registered
Joined
·
2 Posts
I would trust beat error, because it is calculated from the loudest sound, wich is usually the the locking.
Regarding Amplitude, the peaks displayed are different from theory. This might be an effect of filtering, FFT calculations, or because real lif is different from theory.
Not sure the peak not detected is the first one from escapement
 

·
Registered
Joined
·
2 Posts
This looks great. Thanks for your hard work. I am in the process of obtaining the parts to build the amplifier and am having problems finding a switch. I would greatly appreciate it if someone would post a part number or a link to help me find a suitable part. I can't wait to get this project started.
 

·
Registered
Joined
·
34 Posts
Whew just finished reading through all 44 pages of this TG thread as well as 30 pages on Watch O Scope on the WRT forums. The latter has some great discussion and plans for homebrew piezo preamplifiers as well as designs and ideas for timing microphone/watch holder setups. Good stuff. I think there's some good features with that program on long-interval graphs for watching rate, amplitude overnight or longer periods.

I'm a big fan of Paul's blog Adjustingvintagewatches.com, I'd love to see TG with some features or tabs added that help keep track of timing results over 6 positions and the 8 vertical positions for dynamic poising, and also the positional difference and average statistics. Extra bonus if there's an option to keep track over multiple rounds of adjustment on the same movement.

Good work Marcello and contributors, thanks for all your efforts and generosity!
 
421 - 440 of 448 Posts
Top