Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
AW5 LTE GPS accuracy problem
- Thread starter rbart
- Start date
- Sort by reaction score
Reposting it here as it is not wanted in the Workoutdoors thread:
Repeated my run from Wednesday with same conditions (LTE on, iPhone disconnected). This time with WOD instead of Apple workouts.
The track is a little bit more wobbly and distance overall more or less the same.
One area is completely off, have a look:
![Bildschirmfoto 2019-10-26 um 10.15.37.png]( "Bildschirmfoto 2019-10-26 um 10.15.37.png")
Yellow is the workout from Apple workouts, blue is WOD from today. The area is (obviously) a small forest. Not too dense tree covers though.
Also, what can be seen in the track, sometimes the track is not connected.
Can be seen here better, interestingly one my way from and back home almost at the same position (no idea what is there exactly, nothing obvious):
![Bildschirmfoto 2019-10-26 um 10.16.17.png]( "Bildschirmfoto 2019-10-26 um 10.16.17.png")
@cfc : I can share the TCX Files with you, if you want.
Repeated my run from Wednesday with same conditions (LTE on, iPhone disconnected). This time with WOD instead of Apple workouts.
The track is a little bit more wobbly and distance overall more or less the same.
One area is completely off, have a look:
Yellow is the workout from Apple workouts, blue is WOD from today. The area is (obviously) a small forest. Not too dense tree covers though.
Also, what can be seen in the track, sometimes the track is not connected.
Can be seen here better, interestingly one my way from and back home almost at the same position (no idea what is there exactly, nothing obvious):
@cfc : I can share the TCX Files with you, if you want.
Great. Would be interesting to see if you do the same trail with the Apple workout app and LTE on.
iOS location APIs make use of several things, the GPS is one thing but in certain modes the location API will use other means of gaining location such as cellular and wifi spot location information. I assume that watchOS is exactly the same in this respect.
(linked one of the references and the macOS API works the same way https://developer.apple.com/documentation/corelocation/cllocationaccuracy).
Accuracy is at the expense of power and if you only need coarse accuracy and the LTE 'has' to be on anyway then it may well use that instead of GPS. The API (above) shows that it's open to the developer to select a specific accuracy level within their call to the location services.
Also note that at any given time GPS accuracy can very quite a lot by virtue of atmospheric conditions and azimuth angle between the satellites, if they are closely grouped then triangulation becomes very narrow and atmospheric canditions, even with augmentation like WAAS or some internet provided augmentation, may vary.
One thing you don't mention is if you are using an app or the inbuilt health apps. I would try it with the inbuilt apps to see if they give you a better result because this may be solely down to the accuracy level that the app requests from the location services.
Did a run with Apple app today on my track at work and what can I say? Track is perfect, in spite of having LTE on. Better than all my Garmin watches ever did and consistent with my recording one week ago. No hiccups, smooth on track, etc.
Starting tomorrow, I will do a series of runs on my home track with all three apps: Apple, WOD, iSmoothrun. All with LTE on and off and will put the results as a comparison to mygpsfiles.
Hope we can nail down the problem.
Starting tomorrow, I will do a series of runs on my home track with all three apps: Apple, WOD, iSmoothrun. All with LTE on and off and will put the results as a comparison to mygpsfiles.
Hope we can nail down the problem.
My first runs with the AW5 (I was coming from a Garmin background) were terrible. My assumption is, that they are related to the motion sensors not being calibrated yet. Also, my first run was without sufficient GPS reception at the start and GPS coming from the left behind iPhone at home.
What also changed is the crown position now (turning the watch upside down) which also significantly improved track accuracy.
To be more systematic, I will do the tests now on the same route with different apps and LTE on and off as described before. My hypotheses for the moment is that LTE is impacting the track accuracy, but only in 3rd party apps. It would be a very big coincidence if only the days where I used the Apple workout app are the days were there are no hiccups and the tracks are brilliant. But we'll see what these six runs will produce as results.
Two samples (again, only a hypotheses). Weather conditions similar, tree cover also.
One from last Friday, with WOD with LTE on:
Today, Apple workouts, also LTE on:
First run of the little sample project on my home route. With Apple workouts and LTE on. No hiccups.
[automerge]1572350858[/automerge]
Most probably, yes. Or a combination of software and hardware.
[automerge]1572350858[/automerge]
Most probably, yes. Or a combination of software and hardware.
Yes, updated on Tuesday.
Interestingly, the area where I had this very big hiccup is the same where WOD also struggles (and Apple workouts is ok).
Tomorrow next run with WOD and LTE on.
I am a software developer and I do not believe in randomness or luck.
That's why I do the tests with different apps and try to keep the conditions as similar as possible.
So far, the Apple workout app did not show any hiccups on the specific area. Call it luck, maybe. How big has the sample size to be that you do not say it is luck?
From about six runs I had the issues ALWAYS at the same place. But we'll see when the samples are collected.
We'll see.
I have done lot of runs with built-in app and it's always the same behavior: big hiccups somewhere when using LTE 90% of times, 0 hiccup with LTE off.
Same behaviour with all apps, but it's not a surprise, they all use the same source to get position from Apple API, the only difference can be in some post processing.
The problem is a mix of hardware/software: maybe when the LTE signal gets stronger because of bad reception corrupts the GPS reception ? Maybe the watch use cellular position instead of GPS or movement sensor in some situation ?
Only Apple can tell, but I am sure they don't care of this issue, there are very few people complaining so ...
I think this bug is hard to reproduce in dev labs ...
Same behaviour with all apps, but it's not a surprise, they all use the same source to get position from Apple API, the only difference can be in some post processing.
The problem is a mix of hardware/software: maybe when the LTE signal gets stronger because of bad reception corrupts the GPS reception ? Maybe the watch use cellular position instead of GPS or movement sensor in some situation ?
Only Apple can tell, but I am sure they don't care of this issue, there are very few people complaining so ...
I think this bug is hard to reproduce in dev labs ...
You are so sure about this. I am not. Could well be that Apple is doing something differently, something that might not be available in the public APIs or whatever.
We'll see in the results of my tests if there is a pattern.
Must be something like this. I'll show you my results and you will see very strange tracks, e.g. doing circles when I was running straight. But you will know this too.
My hope is to have some data which allows a comparison and I will hand over the data to the app developers (when there is something different between Apple workouts and their apps) and maybe they have a better channel to Apple than us as end users.
One more new (or not so new) hypotheses: when people tell that they have zero problems with their AW4 LTE, what is different to AW5 LTE? Yes, the compass. Maybe this f*cks up the tracking in area where there is bad GPS reception, bad LTE and the Apple workout app can somehow compensate it, the 3rd party not? Might be or might not be.
I would love to do all 6 runs on one day to get the results as early as possible, but as my foot is still injured, this is not the best idea ;-). Will do the series (LTE on/off) probably with Stryd app also, the developers seem responsive there.
My little sampling experiment has ended now. Did six runs on my home route now testing three apps: Apple workouts, WorkoutDoors, iSmoothrun. All disconnected from the iPhone and one run with LTE on and one with LTE off.
Distance results from 10.15km to 10.30km so more or less in the same range. 10.30km was the only run that had a hiccup, more on that later. Compared to my Garmin runs on this route, this is about 2% measuring less as the route is about 10.00km, also measured in GPSies.
Well, result is: Apple workouts and WorkoutDoors both produced almost perfect tracks, with or without LTE.
iSmoothRun failed on the run where I had LTE activated and worked perfect on the run without LTE.
This is where iSmoothrun with LTE failed (I followed the path):
![Bildschirmfoto 2019-11-05 um 08.43.49.png]( "Bildschirmfoto 2019-11-05 um 08.43.49.png")
This is also the area where I had major hiccups with WorkoutDoors, but not on this series of runs.
Ok, to sum it up: three runs with LTE off: no problems. Three runs with LTE on: one with problems. Distance in all six runs in reasonable range. Could of course be random, as I have no better explanation why WorkoutDoors worked under the difficult conditions with LTE activated without a hiccup.
What could be the case though is that I received an email right to the time where iSmoothrun hiccuped, but I am not too sure about this anymore. Maybe some CPU shortage?
I will from now on run with WorkoutDoors with activated LTE and see what happens.
Distance results from 10.15km to 10.30km so more or less in the same range. 10.30km was the only run that had a hiccup, more on that later. Compared to my Garmin runs on this route, this is about 2% measuring less as the route is about 10.00km, also measured in GPSies.
Well, result is: Apple workouts and WorkoutDoors both produced almost perfect tracks, with or without LTE.
iSmoothRun failed on the run where I had LTE activated and worked perfect on the run without LTE.
This is where iSmoothrun with LTE failed (I followed the path):
This is also the area where I had major hiccups with WorkoutDoors, but not on this series of runs.
Ok, to sum it up: three runs with LTE off: no problems. Three runs with LTE on: one with problems. Distance in all six runs in reasonable range. Could of course be random, as I have no better explanation why WorkoutDoors worked under the difficult conditions with LTE activated without a hiccup.
What could be the case though is that I received an email right to the time where iSmoothrun hiccuped, but I am not too sure about this anymore. Maybe some CPU shortage?
I will from now on run with WorkoutDoors with activated LTE and see what happens.
I was also wondering if the hiccups are not linked to a specific "data" activity on the watch in bad reception condition: push notification, mail ...
For me it's absolutely not related to the app, I have hiccups in all apps I use (Apple, Workoutdoors, Intervals pro) , and it can also be clean in all apps. But it seems normal for me as all apps use the same Apple workout API.
For me it's absolutely not related to the app, I have hiccups in all apps I use (Apple, Workoutdoors, Intervals pro) , and it can also be clean in all apps. But it seems normal for me as all apps use the same Apple workout API.