So we just live with the fact that the Fenix 6 Pro's GPS has issues?

silentvoyager said:

Here is the same trail that I ran very fast downhill (red line), then turned around and walked it slow uphill (blue line). The blue line actually follows the trail quite well (if you look at an OSM based map rather than Google map. But look at the red part of the track! First, notice how it turned into the downhill part way earlier. Can you explain that with just GPS accuracy? That is some magic that Garmin is doing to make it work at slow walking speeds.

On thought is sampling rate - I know they say one per second, but if the watch is lagging for some reason (say CIQ fields, or some other background process) there may be a bottleneck between the GPS hardware and the writing to the watch (perhaps dodgy memory?).. So at speed, the logging whilst appearing to be one per second is stuttering....? That may explain the position stuttering in time but orientation (from compass and accelerometers) being more real time?

Don't know if that's possible, or even relevant, but there's something weird there speed related then - and perhaps it is some computational bottleneck? Just thinking out loud as to possible causes there. And if there is a possible underlying hardware discrepancy/incompatibility with some watches, it may explain why some of us get much better data than others, and I'll certainly say that track example you posted is.. in the true scientific vernacular....utterly crap.

silentvoyager said:

When I am running an ultra, my Fenix 6X is reasonably accurate too. For example, earlier this year it measured 101.6 miles on a 102 mile course, which is very good! But that was, for the most, part, a very slow slog with plenty of walking.

Isn't that the dictionary definition of an ultra :).... The only one I've ever done with any resemblance of speed was a flat 5K x 10 looped one. Never again, loops bore the crap out of me.

ejdzsin said:

Basically my statement about the watch's inability to record accurate data has not been debunked, but confirmed (even tho that probably the same people who say worksforme

I think the idea there is not to undermine your difficulties, which are plain, but to underlie the fact for some of us GPS isn't problematic. Instant pace, well that's a given unfortunately but I work around that with smoothed pace. So I wouldn't really take "worksforme" as an attack on you (and I'm certainly not,  saying that), but just saying that I don't see such blatant GPS issues - and for you they are. Like Silent Voyager, some of those tracks are frankly shocking.

I'm curious to see if there is a way to explain it - whether its pure software, or the possibility of hardware issues to certain batches. I have a 6X sapphire, and for most of the time GPS is fine. I have some problematic areas, but on the whole I can look at them and sort of expect it. You have issues in urban and rural areas, which sort of hints that the problem is more general for you than for others.

I hope you fix it.

jose.cboliveira said:

What I don't understand is why Garmin doesn't give a lock on the road option on recording only on map.This would solve a lot of problems, not all like swimming but for a good percentage of users it will be enough.

I think the Suunto 9 Peak does this, and it also fixes a lot of pace issues apparently (so reviews say). For road running this option would be pretty good, and for trail running a "lock to planned route" may provide also another improvement for some.

Ɛpsilon said:

there's something weird there speed related

I don't consider it weird at all.

Consider 6min/km is ~ 2.75m/s then each data point will have at least 2.75m separation even without allowing for GPS error. - the faster you run the bigger the distance between successive points. A minimum of three data points is needed to get any determination of direction, possibly more if traveling around a low radius curve. Think about hurtling down tight switchbacks that you get on a single track for instance. In those situations the switchbacks might have minimal separation making it hard for the watch to determine direction, never mind pace and distance. Hence, going down a set of tight switchbacks at speed, even without any overhead cover, is very likely to produce a different track compared to walking up the same switchbacks much more slowly.

At risk of getting panned (once again) while there are undoubtedly some problems with some watches, a lot of the time it's people's expectations that are at odds with the capabilities of the technology and simple physics. 

Dr Phil said:

Consider 6min/km is ~ 2.75m/s then each data point will have at least 2.75m separation even without allowing for GPS error. - the faster you run the bigger the distance between successive points. A minimum of three data points is needed to get any determination of direction, possibly more if traveling around a low radius curve. Think about hurtling down tight switchbacks that you get on a single track for instance. In those situations the switchbacks might have minimal separation making it hard for the watch to determine direction, never mind pace and distance. Hence, going down a set of tight switchbacks at speed , even without any overhead cover, is very likely to produce a different track compared to walking up the same switchbacks much more slowly

So you think that is a good explanation why it made that left turn into the downhill (at the top of the map) 25 meters early? At that point I was running steady without changing direction. I was running 5:30/km at that time according to the watch, so that means it turned 8 seconds too soon - that is 8 extra data points that the track could have continued straight before turning. What is an explanation for that?

From there it seems it is hard for the track to recover. You can kind if see it tries to gradually move towards actual positions by incrementally applying vector of speed to each position, but since it has started so far off, it is hard to recover.

And regarding going through switchback turns, it is not exactly the way you described. Obviously I can't just go through a tight turn at that speed and have to slow down before a turn, then accelerate again.

I agree that this is as challenging for a GPS watch as it gets, but I think in case of Garmin Fenix this is a self-inflicted wound, at least partially. I've noticed many times before that it seems to be doing some sort of position buffering where the current position trails behind by as much as 10 seconds. I don't know why Garmin does that, but probably to improve track smoothness at slow speeds. I think the track would look more wobbly without that. Compared Fenix tracks to Suunto 9, for example, Fenix tracks for sure look more smooth. Suunto 9 tracks look more jagged in comparison. But Suunto 9 ends up being much closer to the truth. And I don't even mention Ambit 3, which excels in an environment like this. But that watch has an external GPS antenna, and nobody wants that.

I just realized that any Garmin watch is much better than my phone. Vivoactive 3, fr745, fr935, fr935, 6x pro, 6s is included in the test. Even Amazfit Gtr2 is better than my OnePlus 5t.

As a mountain biker, I have found the GPS to be excellent for my purposes.

I wonder if the speed has anything to do with it. As in going faster leads to a better track. So mountain biking produces a better track than running, and running produces a better track than walking or hiking.

I suggest this because most of the time people report GPS problems, it seems to be with running or walking rather than faster activities.

In my opinion it has more to do with the watch (antenna) position, while on a bike you hold your hand so the watch is facing upwards. 

After beeing frustrated for a while I now accepted there will be no software patch for this. As i mostly use the watch for running it was hard to accept. Wearing the watch on the inner wrist helps a lot, but feels akward.

Niterider4 said:

As a mountain biker, I have found the GPS to be excellent for my purposes.

I wonder if the speed has anything to do with it. As in going faster leads to a better track.

This definitely seems to be a correct observation. I read also several people saying that for the bike, GPS is ok. (Some even say that _running_ in bike mode produces better results, which would be funny.)

Speed is for sure an important factor. But actually I was thinking about another one, which might be even more important in the case of Fenix 6 - a watch, which is suspected to have a poor GPS antenna design: when riding a bike, you usually have your wrists turned in the direction of the sky. The wrists have much-much less, continuous turning/rotating/oscillating movement. I wonder