Watch compass not working smooth on LG G Watch R

[Lineflyer]

I do now own a LG G Watch R and tested c:geo wear on it. I am however not sure about the exact cause of the problem, so I try to descibe my experience:

• Basically the compass if alway off by 180°. I never had a correct direction up to now.
• If I turn the watch there are some positions which make the arrow move permanently more or less randomly. Turning the watch a bit more makes it go back to the 180° offset.
• It seems a bit as if slow movements would keep the arrow at 180° offset while fast movements might get it unsteady also after the movement is stopped (but that is only a guess).

I checked my watch compass with this app:
https://play.google.com/store/apps/details?id=com.runar.wearcompass

It is working fine as you can see here:
http://youtu.be/40m_54qRALY

Next is a video of the typical error case with c:geo wear (Note: The watch is not moved, good conditions are checked with the bearing app...compass hardware is working fine):
http://youtu.be/ZOySZNEN9f8

If I set c:geo wear to use the smartphone compass everything is OK.

Tell me if I can help by testing any APK snapshots or alike.

Environment:
Samsung Galaxy S4, Android 4.4.2
c:geo wear Version 2.1.1
LG G Watch R 5.0.1 (Android Wear 1.0.5.1627902, Play Services 6.5.90(1610726-534)
Build LWX48P

[TAZ427]

I've been experiencing the same issue with my Asus ZenWatch.

Environment:
Samsung Galaxy S4, Android 4.4.4
c:geo wear Version 2.1.1
ASUS ZenWatch B4A0
SW 5.0.1
Android Wear ver. 1.0.5.1630507

[cjmoran]

If you disable watch compass, does the compass displayed on the watch match the compass displayed by selecting "Navigate > Compass" in c:geo?

Edit: Must have missed that line in your post, you said phone compass appears to work well.

If so, then this is very confusing because all the position calculations are exactly the same regardless of whether they happen on the phone or watch; they are defined in this file: https://github.com/culmor30/cgeo-wear/blob/master/locationUtils/src/main/java/com/javadog/LocationUtils/LocationUtils.java

Perhaps there's some 'offset' variable the system provides which I'm not accounting for? I really don't want to go the route of "offset compatibility lists" for each Android Wear device because that would be very difficult to maintain. Any suggestions?

Also, @Lineflyer, I suspect that there's some trickery being done with that other compass app, unless the G Watch R has a much, much better compass than the Moto 360. It looks almost like it calculates magnetic north and then just uses accelerometer values to simulate turning. I'd be interested to know if that compass "drifts" after leaving it on for a while or moving around.

[Lineflyer]

I will try to get more experience with how the watch compass behaves and also compare to other wear apps and the behavior if the phone compass is used and come back to you in some days.

[TAZ427]

I was looking at the settings and I'm not sure exactly how to disable the
watch compass. I checked again and I believe it's actually offset 90deg
and not 180deg at least for me. That said, I've had problems w/
another caching program on the phone (WhereYouGo) being offset by 90 as
well. All the other caching and compass related ones on the phone have
been spot on.

That said, I believe the sensor's Native orientation is 90 off (I remember
seeing this somewhere in one of the diagnostic programs, but can't seem to
locate it right now), but the OS compensates when passing to programs
through typical APIs. However, the way WhereYouGo pulls it it has the
offset (I switch to map mode on WhereYouGo to do the WhereIGo's)

I'll see if I can dig further. If you can let me know how to disable the
watch compass I'll give your first test a try.

EDIT: Maybe you just mean turn it (compass on watch) off so it's not running it background. This I've done with same results. I'm uploading a video showing my compass right now. I'll edit this to add the link. I also did some off axis rotation and some quick jerks which show that it's not simply using accelerometers after a measurement.

[cjmoran]

@Lineflyer, thanks :)

@TAZ427, I meant that you should try opening the c:geo wear phone app and uncheck the "Use watch compass" preference. That way, the phone's compass sensor is used instead (but the direction to walk is still displayed on your watch).

[TAZ427]

I tested it w/ the watch compass disabled via the c:geo wear phone app. It consistently matched what the compass (from the listing and in compass mode) within c:geo displayed. I tried it in both orientations and the direction continued to match.

Also, here's a link showing my watch tracking of it's internal compass http://youtu.be/Iit6M1pwZgQ

[cjmoran]

Okay, so it looks like there is some issue specific to Wear devices, unless I'm overlooking something. I'll experiment in trying to fix this.

[Lineflyer]

OK, as promised here are some more details:
First of all I made some more test runs using the watch compass on my LG G Watch R and the main problem is still as described initially.

1. The compass still has a stable 180° offset (I tried calibrating, etc, etc.) but is still working normal in other apps.

Here are screenshots with comparison of phone and watch in every orientation:

1. If I continuously rotate the watch there are some (2-3) angles which make the compass "jumpy". If I stop rotating at this point it keeps jumping randomly, if I continue rotation it is back to the 180° offset. Here is another video of this:
   http://youtu.be/89WYoS0KN2Y
2. If I select to use the phone compass the watch arrow is identical to the phone compass. However I managed to reproduce the jumpy behaviour also with this setting, but it only lasted for a few seconds then it got stable again.
   Pure speculation: It could be that both modes (watch or phone compass) have the same bug here, but the compass of the phone delivers a kind of filtered data, which the watch does not.

[Lineflyer]

@culmor30

Please let me know If I can help e.g. by testing development builds, etc.
I was out caching today and tried to use cgeo wear on my LG. It was quite disappointing due to the issues mentioned above.

[cjmoran]

The best I can do for now is mark this as 'help wanted'. I'd love if someone could take a look at the source and let me know if I'm doing something wrong (or potentially fix the issue with a pull request) because I can't replicate this bug on my Moto 360 and I have no idea what's causing it.

I've tried posting on twitter for my area asking if I can borrow someone's G Watch R to test with, but nobody responded (lol).

[Lineflyer]

I can try to organize a Zenwatch for testing. According to @TAZ427 it has the same issue.