Reply With QuoteJust downloaded this.Originally Posted by mihaixp
Seems to focus OK - will see how it goes when I check my '82 tomorrow!
Cool app just checked my new BLNR GMT and its gaining 1.5 seconds per day. Fairly happy with that.
Just downloaded this.
Seems to focus OK - will see how it goes when I check my '82 tomorrow!
Spot on!
Here's a few of mine:
The Smiths, (my grandfathers watch) and the Titan are around 50 years old and, to my knowledge, have never been serviced although it is on my to do list.
Wow! Those old boys are doing really well!!
Have had the android 'Watch Accuracy'app for a couple of days now.
Timed my '82 and it seems to be consistent with my checking against the speaking clock!
Have set up profiles for a couple of other watches, so will see how they get on!
I'm happy with that!
Kello can provide hours of entertainment too, it uses the phone's microphone to 'listen' to the beat then analyses to produce an estimated gain / loss....
Last edited by colin t; 17th May 2014 at 09:55.
The two pictures above make for an interesting comparison.
Here's one of my more accurate ones it seems. I think I should get the app on my phone, it looks awesome.
Pfffft. My HAQ GS was +-0 seconds after 6 months. :D
just downloaded the app. Seems cool.
My Airking, not bad:
I have an Orient Star , THIS one which has been my most accurate so far and it was a while since I checked it so spurred by this very thread I went and followed it's accuracy for the last two weeks. It has been worn around 9-14 hours per day during week-days and left face up on my desk, nothing special. At week-ends it was not worn, just handwound on Saturday nights so it keeps on going. Well, I did remember it was no more than 2-3 sec. fast per day but low and behold, it surprised me even more. After the first full day it was going about 3-4 seconds fast (all measurements against time.is ) and then I did not measure it again until 7 days have past, at which point it was again about 2-3 seconds fast per day. At this point I stopped/hacked it and re-synchronized it and it was worn again using the same routine. Today was another 7 days past since and it's basically on the dot, <1s fast.
Yeap, I remain a big fan or Orient Star, I've went through quite a few watches in the last few years but besides a Stowa that I got of here and which was pretty accurate, can't think of anything in this price range that can achieve such feats of accuracy. And although this is my most accurate, I currently have 6 Orient Star models, another incomingand none is more than 15-20 seconds off per week, in fact most within 10-15 sec. fast per week. I'll bow to those Japanese folks that manage to get these out of the factory at such a good level of accuracy whilst not bragging about it with any certificate that would easily double the price of these.
My week old Stienhart pilot
Just checked my newly acquired 1481 powered vintage Omega Geneve using the Android ap. I timed it at +6secs over 24 hours against the Radio 4 pips. The ap said +6.6secs per day. Once again this ap is proving to be very accurate.
Usint Twixt app on iOs, but am really not convinced by it... some days it says the watch is 30-60 seconds out (it's not), other days it says just a couple of seconds out..
Plus, it seems to use USA time - if I do a pic before lunchtime, it still has the previous days date, and it usually >1m out. After lunchtime it seems fine...
It's given me some daft readings too - one day it said 5.4 minutes per day for a Grand Seiko GMT when the watch was within 9 seconds of the official time and losing an average of around a second a day. Suffice to say the watch is perfectly acceptable for accuracy though it does gain and lose markedly between on the wrist and off the wrist.Usint Twixt app on iOs, but am really not convinced by it... some days it says the watch is 30-60 seconds out (it's not), other days it says just a couple of seconds out..
Plus, it seems to use USA time - if I do a pic before lunchtime, it still has the previous days date, and it usually >1m out. After lunchtime it seems fine...
I don't know how to take a screen print in iOS - if anyone wants to enlighten me I'll attach one.
Press the home & power button quickly and simultaneously (on an iPhone / iPad) to take a screen shot.
It will save to your photos, and can then be emailed / uploaded etc
Not only does the watch have to be running properly, but the measuring system has to be running properly, and calibrated properly. I calibrate my computer based measuring system (watch microphone -> mic in -> analysis software) before each use (with GPS). On average the correction factor is 2-3 seconds. When using a photography based system, which I assume is being used by this app, you have to take into account the amount of time it takes to capture the image and time stamp it along with the accuracy of the phone's clock (i.e., the time stamping). I would be surprised if time to photograph and time stamp is constant or instantaneous, or that the phone's clock doesn't introduce errors. For all that, however, it is a good way of telling whether the watch is running well or not.
Best wishes,
Bob
PS It may be that your phone corrects its time daily via a NTP server, but it is likely to drift during the day. Given the heat in some phones, the drift may be significant.
RLF
Last edited by rfrazier; 26th May 2014 at 11:21.
Thank a for the tip on the screen print I've never found Apple stuff to be intuitive and wish it came with printed instructions! Anyway here's the shot:
It all goes a bit awry on the 13th of May with some of the shots benefitting from a wifi connection and the 5.4min one done off line. Oddly despite the wide variances in daily rate, the cumulative total on the top left seems about right.
This is a watch that performs best when worn everyday; I tend to wear it every other day.
Interesting, an iPod touch was used which with a wifi connection is running one second slow when compared to a G-Shock atomic / multiband 6.
The watch does vary a lot between on the wrist and off the wrist and whether it's fully wound. I just don't see where the daily accuracy figures are coming from unless they are predictions based upon performance so far / within the day.
Well I've taken another shot and in the space of 90 minutes or so the GS has gained a second compared to the iPod but then the iPod has lost a second compared to the G-Shock and we get this:
I'm going to stick with the G-Shock if I want to check the timing. Incidentally my first shot with app was stating accuracy within a second a day on the GS.
Do you have the iPod on auto time ?
Erm what's that? I think there's a reason I wear mechanical watches; most of this technology either escapes me or provides little interest. I'll check the settings...
I've not used the app, but here is a first guess as to how it works.
1. Time stamped picture (P1) with the hands at a certain position (H1).
2. Lapse of time.
3. Time stamped picture (P2) with the hands at a certain position (H2).
4. Actual time lapse (ATL) is taken to be P2 - P1 (seconds elapsed since the previous time stamped picture).
5. Stack the second picture on the first. Represent the middle of each hand by degrees of an arc.
6. Given the ATL, you would expect the hands of the watch in the second picture to be a particular distance away (in an arc) from the hands of the watch in the first picture, call this the "expected hand position (EHP)".
7. The hands in the second picture are actually a particular distance away. Call this the "actual hand position (AHP).
8. Watch deviation (WD) is (EHP - AHP), which is given in degrees. Translate the degrees into seconds. Call this result the "watch deviation (WD). (This assumes that the watch is no more than about +/- 6 hours, or something like that)
9. Then you calculate the daily deviation by dividing the WD by number of days. So, for example, 5 seconds deviation after 2 days and 3 hours would be 5/2.1250 or 2.3529 s/day. (Of course, most, if not all, of the decimal places in this calculation are not significant.)
If I'm right, especially about the stacking, the accuracy is going to be influenced by the light and quality of the picture.
Best wishes,
Bob
PS I think it can be done without stacking, simply by calculating from the picture a numerical representation of the middle of the hands in an arc..
RLF
Last edited by rfrazier; 26th May 2014 at 15:04.
'Date & time' shows 'set automatically' is on. Given the other chap with the brand new Rollie Explorer II is around 30 secs a day adrift according to the app; I'm inclined to think there's a bug here somewhere.
Would the angle between the hands be affected by the angle that the shot was taken at? Common sense suggests it would though I'd hope the app would compensate especially given that you have to centre the 12 o'clock position and surrounding markers.
Perhaps the shot has to be taken dead straight and it would help if the iPod didn't gain or lose.
Right so let's try and break it (the app not the watch) and take a ridiculous shot at an extreme angle which if stacking is what's going on would create a high variance with the last shot:
It's a deliberately silly shot and you'll notice the problem in aligning the 12 o'clock.
Last edited by AKM; 26th May 2014 at 15:19.
That's actually a really useful experiment, and confirms what my non-technologically savvy mind suspected.Right so let's try and break it (the app not the watch) and take a ridiculous shot at an extreme angle which if stacking is what's going on would create a high variance with the last shot:
It's a deliberately silly shot and you'll notice the problem in aligning the 12 o'clock.
I bet even the tiniest angle could give an error of a few seconds.
I suppose the best bet is to do as many readings as possible - maybe even 2 or 3 a day, over 2-3 weeks, and then see what the average is - hopefully this would filter out a lot of the error?
Try Watch Tracker. It's nice and easy to use since you just tap the screen when your watch shows the appointed time. No taking photos, no problems recognising and aligning the hands. That makes it very quick to add data points (just two taps) and you can track any number of watches simultaneously.
It includes an NTP client, so you will be checking against atomic time regardless of the state of the clock on your device. It's even smart enough to figure out, from your watch's last measured offset and rate, what time to offer you onscreen for the data point. In other words, if it thinks your watch will be 10 seconds fast today, it will show you the time about 15 seconds ahead of the true time and wait for you to tap. Nifty.
I think this is the right way to do a watch tracking app; the other approaches (recognise photo of watch / listen to movement via microphone) are fun and interesting, but probably not as reliable and certainly not as easy to use in practice.
For simple timing, I suspect that you are right. Average reaction time is within 0.25 seconds, I think, so this method is plenty accurate.
Best wishes,
Bob
Last edited by rfrazier; 27th May 2014 at 21:01.
Watch tracker looks good. I'm in!
I have compared my watch to a mobile phone and computer and that made me think my watch was losing/gaining a little. I was wrong, the problem is that mobiles and computers generally have very poor clocks that lose or gain the odd second between the time when they sync with correct time.
This site will tell you correct time and how far your device is away from it http://time.is/
Using that kind of app may be fine to measure a mechanical watch, but for a high accuracy timepiece I suspect that errors in the mobile's clock will mess things up, unless it has just been synced of course.
Timekeeping is not really an issue for me since I have a high quality watch, Grand Seiko quartz.
Yes, so...
for those of us in the Dark Ages (or with Windows phones....), the GMT website (or whatever similar) and a basic little spreadsheet to record it onto is all you need ;)
Rather please with my newest purchase, running an ETA 2836-2 movement:
Start of Test: 10.6.14 0841hrs
End of Test: 18.6.14 1433hrs
Total (h:m:s): 197:52:00
Error (secs): -2
Av error secs/day: -0.243
Av. error secs/month: -7.3
Not been off the wrist more than a handful of hours, and possibly still "settling in", but it can settle at that for all I care !
Last edited by notenoughwrists; 18th June 2014 at 15:34.
PRS 18Q
Can't grumble at that!
-0.02 sec a day. Didn't use the app.
"Bite my shiny metal ass."
- Bender Bending Rodríguez
A slight annoyance with Watch Tracker is it doesn't show the name of the watch on the graphs screen. What is it?
Ha!
Not bad for a watch from 1977:
This is the seiko Sportura World time. My most accurate watch outside atomic.
And its a cheap I compared to the MM200 DD!!
Posting Permissions
