Watch Accuracy Meter - Timegrapher App

[JAFO]

I have been considering a Weishi timegrapher, but they are all £160 plus on ebay now.
I saw a watch for sale on ebay with a timing report from a mobile phone app.

I downloaded a few from the play store to see how they worked. . The one they used on ebay (toolwatch) required you to login, but there was another one called "Watch Accuracy Meter" (actually, it might even have been our own @Fat-Sam in his Speedmaster for sale listing)

You put your watch dial near your phone mic, press start, and it over a minute, it draws a line showing the accuracy. I was a tracking a phone for 4pm yesterday. The app said +11 seconds, and the actual time was +10. I did notice that the accuracy of that watch declined as the power reserve went down. I found you had to put the watch around he phone so the dial lay flat on the phone.

I've checked all my watches at around lunchtime, and most were what I expected, so I will check the actual results tomorrow. I couldn't get a reading of one watch. I can only think the ticking isn't loud enough.

Screenshot below of a trace.

[Viognier]

I've always questioned mobile apps accuracy. Yours seems close. On another forum (from a watchmaker), he stated the apps using microphones are not terribly accurate. Your experience suggests that app might be more accurate.

[MiniMpi]

I have used this app before, I found it very accurate.
Here is the readout for my C9 SS which is basically spot on with the COSC cert position.

[Robotaz]

I can’t find anything for iPhone that uses a microphone. Anyone?

[ajax87]

I can’t find anything for iPhone that uses a microphone. Anyone?

I couldn’t find this for iPhone either.

[jkbarnes]

How does this app work, exactly?

[tikkathree]

I can’t find anything for iPhone that uses a microphone. Anyone?

I couldn’t find this for iPhone either.

Sounds like it might be available in Android

[JAFO]

How does this app work, exactly?

First, I have an android phone. It doesn't say on any of the apps I downloaded whether they are also available for I-phones. As FMs are not able to find versions of this for I-phones, I guess there aren't any.

I understand time-graphers work by listening to the tiny clicks made by the escapement as it changes direction and goes backwards and forwards, and measures the frequency of those clicks. A watch running at 28,800 should have 8 clicks per second (28,800 per hour/60/60), and at 21,600 should have 6. By measuring the actual time taken, the accuracy of the watch can be evaluated.

These time-graphers for phones use the phones microphone to listen and count the clicks, rather than an external device.

So, doing some maths, a watch that should tick 28,800 times per hour that gains 9 seconds per day is actually ticking at 28803 times per hour. That's 3 too many per hour. 24 hours by 3 extra ticks per hour is 72 ticks per day, which is 9 seconds, as there 8 ticks per second. 3 ticks in 28800 is about 1/100 of a percent wrong, so these errors are just tiny fractions.

I don't know which measurements yield the beat error and amplitude. I presume the beat error is the irregularity of ticks within the tick count, so that even ticks and odd ticks are not quite equally spaced, as they should be.

I was surprised it worked so well, but I guess I shouldn't be. We know mobile phone clocks are virtually completely accurate, and clearly the microphones are pretty sensitive.

[RichM]

It worked surprisingly well for 2 of mine insofar as it showed fast/slow correctly...

... and did show what I already knew that my C65 SH21 LE runs 9 secs fast!!

[jkbarnes]

How does this app work, exactly?

First, I have an android phone. It doesn't say on any of the apps I downloaded whether they are also available for I-phones. As FMs are not able to find versions of this for I-phones, I guess there aren't any.

I understand time-graphers work by listening to the tiny clicks made by the escapement as it changes direction and goes backwards and forwards, and measures the frequency of those clicks. A watch running at 28,800 should have 8 clicks per second (28,800 per hour/60/60), and at 21,600 should have 6. By measuring the actual time taken, the accuracy of the watch can be evaluated.

These time-graphers for phones use the phones microphone to listen and count the clicks, rather than an external device.

So, doing some maths, a watch that should tick 28,800 times per hour that gains 9 seconds per day is actually ticking at 28003 times per hour. That's 3 too many per hour. 24 hours by 3 extra ticks per hour is 72 ticks per day, which is 9 seconds, as there 8 ticks per second. 3 ticks in 28800 is about 1/100 of a percent wrong, so these errors are just tiny fractions.

I don't know which measurements yield the beat error and amplitude. I presume the beat error is the irregularity of ticks within the tick count, so that even ticks and odd ticks are not quite equally spaced, as they should be.

I was surprised it worked so well, but I guess I shouldn't be. We know mobile phone clocks are virtually completely accurate, and clearly the microphones are pretty sensitive.

Thank you. That’s a very clear explanation.

[rkovars]

That first plot isn't great. MiniMpi's is good but he also didn't have the app calculate the beat rate. He entered it manually. I'm guessing that the noise reduction algorithm in the app (they need this to distinguish the ticks from all of the other noise around) has a better time if the beat rate is known.

I would hesitate to use it often as there a magnets all over the iPhone and iPad. Some of them are rare earth magnets and very strong. Probably best to use the microphone in a set of headphones that you can keep the watch away from the phone.

This would be good for a quick spot check but I wouldn't use it for any diagnostics. If all you are after is an accuracy measurement a timing app like WatchTracker or Toolwatch would be all you need.

[JAFO]

My first plot wasn't taken in an ideal position. The phone was actually on charge, and I just wanted to get something I could upload. When I take more care, mine are denser, and more like the plot MiniMpi posted. I take the point you are making about magnets though.

I think on the back of this, I will open up the cases of a few watches and see how easy it looks to tweak the regulation.

[Amor Vincit Omnia]

I have always felt that these machines and apps are best suited to the serious tinkerer.

I tend to set my watch, then wear it, leave it on the winder or leave it in the box for 24 hours, check it again and so on. If, over several days or weeks, a pattern emerges then I have a fair idea of the rate of my watch and whether I should set it a bit slow or a bit fast accordingly.

As I’m not immensely bothered about split-second accuracy that’s fine for me.

[Robotaz]

I have always felt that these machines and apps are best suited to the serious tinkerer.

I tend to set my watch, then wear it, leave it on the winder or leave it in the box for 24 hours, check it again and so on. If, over several days or weeks, a pattern emerges then I have a fair idea of the rate of my watch and whether I should set it a bit slow or a bit fast accordingly.

As I’m not immensely bothered about split-second accuracy that’s fine for me.

It’s good habit twice a year or so to note timekeeping over 24 hours just to have a sense of service needs. This is why I’d like an app. I could check it really just once during the day and not have as much work put into checking, and probably have better results anyway since I’m getting lazier and lazier about taking care of my watches.

[rkovars]

The two apps I mentioned above are handy because you don't have to be religious about checking it at a certain time. It will do all of the math for you. You just need to keep the watch wound. I do a check over a week wearing the same watch checking the time once a day or so (you can also check multiple times a day if you want). You can also have different runs so if you let it run down you can start a new run from where you set it back to 0.