GPS malfunction of fluke?

[xendula]

Endomondo showed me zigzagging around my trail yesterday, including running on water, and it showed almost 3 times the distance I actually ran, see pics attached.

Factors to consider:
- it was below freezing
- GPS signal on that trail is always a bit tricky, and even on my Gear S used to go in and out
- I do not have a SIM card in my watch, and left my phone in the car
- I was streaming music over BT during my run

I have done all of the above before with no problem. Has anyone else seen this happen before?

I wonder if the GPS is the part that presumably can malfunction in "extreme conditions"?

 

[smatofu]

I got the same results with GhostRacer (amazing app!!!) yesterday. Maybe there was a solar storm...
1. I will be taking the watch off my wrist for a longer time before running.
2. I will wear it loose. The tighter the band is, the weaker GPS reception.

 

[xendula]

Interesting that you had the same results yesterday and yesterday only!

2. I will wear it loose. The tighter the band is, the weaker GPS reception.

Hmm, I could try that, but not sure it will be secure enough not to fall off if it bounces around. I have to use the second hole because of my wrist size.
Today, it did not act crazy, but it did count an extra km vs. what Nike gives me for the same trail on my Gear S and my phone. Upon close inspection of the differences, it could be that Nike measures a straighter run than the trails actually allow. Or maybe not? Very confused... See pic: Endomondo on LG watch on the left, Nike plus on the right (two different days, but same trail).

 

[foxbat121]

Use GPS to measure distance for slower speed like that is bound to be inaccurate due to the inherent inaccuracy of GPS in civilian GPS receivers. Nike+, from my experience on Gear S, uses cloud server, instead of phone or watch app, to calculate the best estimate of routes, then calculate the distance. It requires much more computation resources to do this.

A typical fitness tracker will always fall back to use step counter to estimate the distance instead of GPS in this situation.

The GPS in phones and watches can reach 5 meters (or 15 feet) accuracy at best case. Based on my limited test using the Ware GPS app, it can only reach 10+ meters accuracy for me. Then, to save battery power, these GPS receivers also only samples the GPS position around once per second. So, if your speed is not fast enough, there is no way to calculate how far you have traveled accurately.

 

[xendula]

So would you say, then, that between the two pics you see above, the Nike+ tracking is more accurate because Nike uses cloud servers?

 

[foxbat121]

If you have to use the GPS to measure your track distance, yes, Nike+ does a better job because it filters out those GPS jitteriness due to limited accuracy by calculating a simulated route that best matches your actual route in overall scale. The raw GPS data in this case is quite useless for such direct measurement unless you move at much faster speed, e.g. bike, motocycle or car, anything that moves a few times of accuracy bubble per second. Unfortunately, such calculation is quite resource heavy and complex.

The GPS data is the same. But calculations differ a lot. On Gear S, Nike+ doesn't even have to wait for any GPS lock at all. It uploads raw GPS data to the server to let the server calculate your GPS position data.

 

[smatofu]

Nothing is correct in the post below. GPS in watches could and can estimate distance and speed reasonably well for years without any Internet connection to any servers. Depends on weather, trees, buildings, GPS reception... the error is about +/- 1%.
Maps cannot be used for running distance because people run on sidewalks (not in center of roads), through parks, make shortcuts etc.
In addition, constant Internet transmission and reception would kill any watch battery in minutes.

If you have to use the GPS to measure your track distance, yes, Nike+ does a better job because it filters out those GPS jitteriness due to limited accuracy by calculating a simulated route that best matches your actual route in overall scale. The raw GPS data in this case is quite useless for such direct measurement unless you move at much faster speed, e.g. bike, motocycle or car, anything that moves a few times of accuracy bubble per second. Unfortunately, such calculation is quite resource heavy and complex.

The GPS data is the same. But calculations differ a lot. On Gear S, Nike+ doesn't even have to wait for any GPS lock at all. It uploads raw GPS data to the server to let the server calculate your GPS position data.

 

[foxbat121]

Nothing is correct in the post below. GPS in watches could and can estimate distance and speed reasonably well for years without any Internet connection to any servers. Depends on weather, trees, buildings, GPS reception... the error is about +/- 1%.
Maps cannot be used for running distance because people run on sidewalks (not in center of roads), through parks, make shortcuts etc.
In addition, constant Internet transmission and reception would kill any watch battery in minutes.

Nope. The inaccuracy of GPS signal is deliberate. Only the military grade device has the hardware to decode the accurate part of the signal. You are confused with GPS used for driving where the car moved fast enough to ignore the inaccuracy and the GPS mapping software typically snap your position to the road nearby. No so when you use it for walking or running.

If you use any GPS status app like the ones mentioned, you will see you typically have a drift speed of around 3 miles per hour even when you standing still.

 

[smatofu]

Foxbat121,
GPS sport watches are available on the marker for 5 or more years. Millions have been sold. I have had a few of them (Garmin, Motorola, Microsoft, TomTom). Based on my experience, the distance error over 10km is about +/- 100m (+/- 1%). All those watches work standalone, no Internet connection to any server. Anyway, as I wrote before, a server with a map makes little sense for running as runners don't follow roads like cars do.

[If you use any GPS status app like the ones mentioned, you will see you typically have a drift speed of around 3 miles per hour even when you standing still.] You can stand still with that drifting GPS for hours and the position shown by the GPS will be roughly the same. You position will not change by 6 miles after 2 hours

 

[foxbat121]

Most GPS watches use step count, not GPS data to calculate distance. The GPS is only used for position tracking. Besides, Nike+ doesn't necessary need a internet connection. It collects data and uploads it when it connects.

 

[xendula]

Quick update on my end. I wore a Suunto Ambit (3?) on the other wrist yesterday, then compared the distance and maps with Endomondo on LG watch on the same day (left pics below), and the same trails tracked with Nike+ on my phone on a different day (middle pic).
All the way to the right is a different part of my run where I ran in a neighborhood in a straight line.
While not perfect, the Suunto shows me mostly running on the actual road and trail.

Distances tracked:
Suunto: 6.36 km
Urbane with Endomondo: 7.59 km
Phone with Nike +: 6.33 km

Is this along the line with your experience? Has anyone tracked runs with the k8 or Sony SW3, and how were the results? I am thinking of just buying a second watch to wear only on runs. I already sold my Gear S. I hear the S2 is less reliable than the OG Gear S as well, so maybe putting the GPS antenna in the (bendable) band is not such a grand idea.

Foxbat, I hear you on military vs. non-military, but the least I expect is what my phone gives me. I see that even the Suunto has me running next to the trail/road at times, but for my use, that's good enough.

 

[smatofu]

There is nothing military vs non-military at the play here. The problem could be band stretching, bending, or contact with skin. Maybe this is the reason for the recall. I will try to put a piece of fabric/foam between my wrist and the watch band.

BTW, the watch GPS works perfectly well when it is off the wrist!

 

[xendula]

I will try to put a piece of fabric/foam between my wrist and the watch band.

BTW, the watch GPS works perfectly well when it is off the wrist!

I tried carrying the phone in a pocket yesterday, and it indeed tracked as well as my phone.

Today, I had the band closed to about the middle hole, and hiked the watch up on my arm. This got me the best results so far, on the first half (3.5 km), then zigzagging for the second half, which was the exact same trail. I was really sweaty underneath the watch by the end of my run, so I wonder if sweat may affect it. I wore it next to a Suunto, and during the first half, the LG only added about 10% distance as compared to the Suunto, then on the second half it went crazy.

Next test will be wearing it on top of one or two wrist-sweatbands. This will allow me to not have to tighten the band all the way to the second hole, and will keep sweat away from the band.

 

[smatofu]

Xendula,
Sometimes I get results identical to yours. Today, for the 1st time, I was wearing a wrist-sweatband under the watch, and I got good tracking. Yes, sweat or skin contact could be the problem.

 

[xendula]

Re: GPS malfunction or fluke?

Xendula,
Sometimes I get results identical to yours. Today, for the 1st time, I was wearing a wrist-sweatband under the watch, and I got good tracking. Yes, sweat or skin contact could be the problem.

I now took the watch on two runs and wore it on top of a Banjee Wrist wallet (if you don't know of them yet, you may want to check those out, especially the large version):

1. The first run was tracked almost perfectly with one small fluke in one bend of the trail (the one zoomed in above).

• The second run was tracked perfectly. I had a Banjee underneath and one on top of the watch, which alowed me to close the clasp at a more outside hole, and thus wear it more loosely. It tracked 7.09 km on a 6.6 km run, which I can live with. With the beyond awesome app SyncMyTracks, recommended by mdw1995, I can export/import from a ton of services. Once I imported the run into RunKeeper, it automatically fixed the distance! The distance fix is also something mdw1995 discovered, so, really, I'm just reaping the benefits of other people's experimentation

This seems to confirm that the watch tends to track more accurately when not touching the skin, and when worn more loosely. Being that this watch has a HR sensor on the underside, this would be a big enough embarassment for LG to stop global rollout. I really, really hope that's the one issue that "may malfunction". I can live with that, since I never did anything with the HR data that my Gear S recorded anyway.

It could be that antennas in the bands are just not a good idea. The Gear S2 has similar tracking issues.

 

[miroul]

I thought it would be a good idea to reopen this thread in order to discuss GPS accuracy with the new release of the watch. I bought one 2 weeks ago and started training (running) 4 times a week, doing intervals and long runs. Now, the first time i tried the watch, it was with cell functions off and GPS on. I got really bad results,on a map, you can see my position zig zagging like crazy and while i was running the watch was pumping out km laps in unreal time !

The next day, i tried it for a long run using the cell functions and the GPS on. The GPS connection was acquired a lot faster (I'm guessing due to A-GPS) and my run appeared to be recording normally. When i got home and looked at the map, i saw much better results but still, not as good as you could obtain from your phone or from a GPS training watch. There is still some zig zagging showing up, although a lot less than during my first try. My actual pace was probably about 20 seconds better per km than i would normally have with my Garmin Fenix 3. Which is innacurate enough for me to be worried...

Anyways, i'd like to hear some new feedback from those users that do use the watch for some serious running.

Thanks

Max

 

[nrseife]

I typically turn on location, remove the watch from my wrist (get match faster lock that way; cellular, etc off), run WearGps (to see Nr of satellites it found), put it back on my wrist, and once Ghostracer has a lock start my run. I did not notice any issues with the quality yet (judging from map view in ghostracer after my runs). are you sure you see enough satellites (bare min is 6)?

 

[json405]

I do the same as nrseife. I run Wear GPS right before I run. I haven't ran for a couple of weeks due to injury but when I last tracked it worked fine.

Has anyone tried changing the GPS options in developer mode to high accuracy instead of device only? Not sure if that makes a difference. I just found that option a couple of days ago.

 

[mdw1995]

I've been tracking runs since November and I have updated to the new build. Tracking has been pretty good lately. I had some problems with GPS not recording after a while with Ghostracer but since the update, and since placing a sweat band under the watch, these issues have not returned so far. I ran 7 miles today and it appears it may have missed one or 2 satellite points but I think I was covered by trees during that time. Even with that, when compared to onthegomap.com it was within 0.1 miles. I can live with that. It doesn't track perfectly to the road the whole time, but I haven't had zig zags. I always have cellular on paired to the phone in the car. , received 7 texts dictated to me by audio feedback today, so I can receive forwarded calls and texts. I connect my bluetooth headset, start Google Play music, shuffle my playlist, start Wear GPS to get a signal lock(wait for accuracy to get close to 10 meters or less on the first screen), then start Ghostracer in always on mode. I usually get a lock within 30 seconds without removing the watch. Ran an hour and 6 minutes with 33% battery drop. It probably isn't as accurate as the dedicated GPS fitness trackers, but with everything it does, I'm happy with it.

 

[xendula]

My GPS issues have gone away since I started wearing a wrist wallet underneath the watch, and have it fastened a bit looser than at the second hole I would usually need. I think that my small wrist size (the bend in the antenna) along with it touching my skin was affecting the GPS signal when I first started this thread.

When I get close to my trails, I do this:

1. I shut off BT on my cell
2. Turn on Play Music
3. Optional: turn on cellular (makes no difference in tracking for me but uses up about 50% power per hour or more)
4. Take the watch off and hold it by its buckle
5. Start Wear GPS and wait for a good lock on satellites (may shake the watch a bit in a bad signal area)
6. Put the watch on over a wristwallet
7. Start Endomondo