Related
Options

How to fix / improve wrong current pace on fenix 6 pro?

Mattbest
over 4 years ago

I have, as many other users, a terrible experience with current pace on F6pro, often slower from 1 to 2 min/km from the actual one, in every possible gps setting. I also have to specify that I usually get pretty good tracks (accuracy of 5-10m) from the very beginning of the activity.
Both from my experience and what i've read on other threads, the issue comes from using internal accelerometer to smooth out GPS pace.
I have many proofs of this because:
1. happens particularly in low gps signal areas (trees, buildings etc).
2 It affects expecially the first kilometer, then it seems to get a sort of "autocalibration" and current pace accuracy generally improves.
3. if i artificially increase arm movement cadence the current pace gets faster.
4. in cycling all the metrics are accurate and precise....even in difficult gps conditions.

Since there is no option to set the influence of the accelerometer on the metrics, does someone managed to solve this terrible defect of garmin software? here's my reflections and questions:
1. i can use cycling or create another custom activity for running?but what about the related physiological metrics?vo2max etc...?
2. does exists a way to manually calibrate the internal accelerometer or to get a precise stride length?
3. it has no sense to buy a stryde foot pod: i'm only looking for the decent precision that my old Garmin FR35 (130€) had.
4. as customers, there is something we can do to finally get Garmin fix this thing?

Top Replies

All Replies

  • 0 over 4 years ago in reply to silentvoyager
    silentvoyager said:
    One observation that I've made right away is that the distance at every sample progressively lags behind the distance that can be calculated from coordinates.

    I'm not sure if I've understand your calculation correctly. But you can't calculate the total distance of a wobbling trace by simply adding the distances between every lat/lon.

    A simple example:
    You move exactly in north direction 0 deg, but due to the position error the distance vector between the two lat/lons points in north-east direction 45 deg. So your distance of movement in the north direction should be calculated as

    distance of movement = distance between the lat/lons * cos 45 deg

  • 0 over 4 years ago in reply to subra
    subra said:

    A simple example:
    You move exactly in north direction 0 deg, but due to the position error the distance vector between the two lat/lons points in north-east direction 45 deg. So your distance of movement in the north direction should be calculated as

    distance of movement = distance between the lat/lons * cos 45 deg

    This isn't how it works I think. That kind of position filtering happens even before lat/lon positions are produced by the GNSS chip; otherwise the track would be atrocious. In fact you can see the effect of position smoothing in sharp turns. You can see the track almost always cutting through turns.

    Furthermore when I apply distance correction in Strava on a typical road or suburban trail run, the distance hardy changes. Distance correction in Strava, basically, discards the distance reported by Garmin and connects all lat/lon positions. So basically it replaced the device distance with GNSS distance.

    However on a real trail run, GNSS distance ends up being quite a bit shorter than the device distance, usually around 5%. Trails in my area are quite twisty and wavy, and Fenix seems to be applying additional distance smoothing on already smoothed GPS track. In fact even the original GNSS distance is usually too short on trails, and additional smoothing makes it even more short. As I said above, I understand why Garmin is doing that, and I would be OK if it did that in road running activity, but in trail running activity that additional distance smoothing should be turned off. 

  • 0 over 4 years ago in reply to silentvoyager
    silentvoyager said:
    That kind of position filtering happens even before lat/lon positions are produced by the GNSS chip; otherwise the track would be atrocious. In fact you can see the effect of position smoothing in sharp turns. You can see the track almost always cutting through turns.

    From my POV the lat/lon's are the unfiltered and unsmoothed results of the trilateration from the GPS sensor. And if you run a sharp turn, mostly there isn't a position point exactly at the corner. Usually, you see a point before the corner and after the corner which are connected by a straight line, which cuts the corner. It's simply a polygon.

    silentvoyager said:
    Furthermore when I apply distance correction in Strava on a typical road or suburban trail run, the distance hardy changes. Distance correction in Strava, basically, discards the distance reported by Garmin and connects all lat/lon positions. So basically it replaced the device distance with GNSS distance.

    Yes, there is difference because Strava ignores the distance calculation by Garmin. But I think Strava also uses it's own filtering and smoothing algorithm.

    silentvoyager said:
    GNSS distance ends up being quite a bit shorter than the device distance, usually around 5%. Trails in my area are quite twisty and wavy, and Fenix seems to be applying additional distance smoothing on already smoothed GPS track. In fact even the original GNSS distance is usually too short on trails, and additional smoothing makes it even more short.

    Every turn of a course is a challenge for the SW, because it has to differentiate between a turn of the course and the position error. The SW needs always some points to decide that it's a turn at the course.

    Anyway, whether lat/lon's are filtered/smoothed or not, mostly the traces in .gpx, .tcx or .fit of all kinds of runs looks wobbling due to the position errors. These artefacts should be eliminated before the calculation of distance or speed. I think the SW estimates the heading/bearing of the run and takes only the distance component in this direction into account. And this always shortens the total distance of a zigzag-polygon.

  • 0 over 3 years ago in reply to silentvoyager

    I bought a heritage garmin foot pod today (actually part of a watch package...) for $25. It is ant+, hopefully it will fix this issue on my 6x Pro Sapphire.

    I'm having the exact same issues you describe. This is insane that we all have this problem. My 920XT never had this issue - I just got tired of the missing features and square shape after 5 years.

    My pace bounced between 7:45 and 10:30 on a 5 mile run yesterday. My mile time splits were 8:38 to 8:42 for all 5 miles. I did not run intervals, but I ran on wooded gravel roads and the pace is just all over the place the entire run. My actual pace was like clockwork....obviously the watch could not keep up on the Pace field...while distance appears to be working fine.

    Using Lap Pace (I hopped to the second screen on the Run Layout), each mile started off crazy (9-10min/mi), but by about 0.4 miles into the "lap"/mile, the lap pace corrected to exactly where I was planning to run (8:38-8:42). I'll use this until the foot pod arrives...if that actually does anything.

    Note: the new Garmin Running Dynamics Pod (at least reviewed by DC Rainmaker) does NOT do pace, so I had to find a discontinued one online. It is absolutely nuts that Garmin has not fixed this issue.

  • 0 over 3 years ago in reply to subra
    subra said:
    From my POV the lat/lon's are the unfiltered and unsmoothed results of the trilateration from the GPS sensor. And if you run a sharp turn, mostly there isn't a position point exactly at the corner. Usually, you see a point before the corner and after the corner which are connected by a straight line, which cuts the corner. It's simply a polygon.

    That's not how it appears to work to me. The track often cuts well inside a corner, with multiple position points cutting through a path that I didn't run. The way it appears to work from looking at the track and a few other observations is that the "current" position on the watch trails behind my actual position by 5-10 seconds when GPS signal isn't strong, and  it tries to make a direct path from that position to where I actually am, which often appears as if the recorded track entered corners way too early. Think about a trailer towed behind a car on a long line - that is how it looks to me a lot of the time. 

    There other examples of the current position trailing behind. For example, sometimes it would tell me I am 100+ ft (30+ m) from a Strava segment as I am passing exactly by its start, then as I run away from that segment and perpendicular to it, the distance to it keeps reducing. I observed that multiple times.

  • 0 over 3 years ago in reply to silentvoyager

    What I don't get: It works very well with the Forerunner models. Can't they use the code? I've never heard of these kind of issues with a FR945, which should have the same sensors inside.

  • 0 over 3 years ago in reply to 8947930
    8947930 said:
    I bought a heritage garmin foot pod today (actually part of a watch package...) for $25. It is ant+, hopefully it will fix this issue on my 6x Pro Sapphire

    I don’t want to be that guy, but I kept the ant+ footpod from my FR645 era and no, it hasn’t solved the unreliable pace issue on my 6s sapphire. Maybe a stryd would be the solution, but here in Italy is hard to find, and too expensive IMHO.
    The only training where I find it to be acceptable are those where I need to keep the average lap pace on a specific value, but that’s the easy case: just a fraction with measured time and distance.

  • 0 over 3 years ago in reply to Labronico

    Thanks for that update. Very unfortunate that the foot pod did not fix the issue!

  • 0 over 3 years ago in reply to 8947930

    I tried this with a run by using an old Garmin pod, but ran into several problems: calibration: how the heck can I calibrate it if the pace varies so widely even across same routes? And I ended up with this below. Until I've had it and walked home. no idea what happened there, but this pace was even worse than anything I've ever seen with GPS.

  • 0 over 3 years ago in reply to 6928399

    Just talked to Garmin Support:

    - Settings / Data Recording: Set to 1 per second instead of Smart recording (1 ever 10 sec)

    - Connect the watch to Garmin Express - download all map updates, as those come with GPS updates as well

    Those are the first troubleshooting steps.