Related

Simple (but awkward) hack to boost Fenix 6 GPS accuracy (and proof of flawed design) - UPDATED

knixen
over 3 years ago

UPDATES BELOW.
As you might have noticed the GPS accuracy on the Fenix 6 is under-performing quite a bit compared to other watches (like the Forerunner 935 etc). What you usually see is that the track is more or less systematically off by a couple of meters in mild conditions and 20-30m in harder conditions, sometimes even more. Remember this is compared to other GPS-watches. Mild conditions are open sky, clear weather etc. Hard conditions is urban environments with high buildings and a lot of turning or running a dense forest. 

However I have noticed that accuracy will increase drastically if you turn watch so it’s pointing upwards to the sky, instead of the usual position at the side.  Like this

So I tried to put the watch on the top of the wrist bone, instead of the flat side, where it would be pointing upwards the sky all the time while running. To prevent it fall off to the side I had to make a “dummy holder” to hold it in steady in this position. Se pics 

The holder piece is made of foam from pipe-isolation. 

It looks awkward but it actually works, the accuracy IS boosted, and it’s on par (or even better) than previous models.

I made several test runs to compare GPS-performance and there is a significant difference. All the systematic errors are basically gone.

Blue track = Sky-facing position on wrist
Red Track = Standard postion on wrist
Green track = True track (Drawn by hand)

Full test run here. http://www.mygpsfiles.com/app/#hdTlx7iz

Device is running the latest version
GPS version 4.10
Software version 9.0
GPS/Glonass setting 1-sec recording

Having the watch this way breaks the optical reading of course, but if you use a HR-band it doesn't matter.

I think it behave this way more or less because of a design flaw by Garmin and it can probably not be fixed by software. What's bothers me a bit is that Garmin could have fixed this easy by integrating the GPS-antenna in armband, or by fixing the antenna position somehow.

UPDATES:
Why did I write this post? 
I want to show what I think is the number one culprit to the bad position performance the vast majority seem to experience with the Fenix 6: Bad antenna design. Could also be in combination with new GPS-chipset. After year of updates, I don't think this is gonna be resolved by a software update.

Why do you care about plotted tracks to be correct,  it's only on the screen? 
See below.

Why don't you use Stryd if you want accurate recordings?
In fact I DO use Stryd already. And it works great, especially for speedworks. But does not help when using the watch as a navigation tool, which I often do when I am running longer distances/trails/Ultras. And bad position performance DO hurt when navigating on intertangled/complex trails. Besides that not everybody want to pay up another $220 when they already have paid a $600 watch. Also wIthout a Stryd, GPS-position IS the main source of information. If it not good it will also affect pace, distance (and to some extent elevation). As you can see there is a deviation of 200 m on my 5km test runs. Another reason for good position accuracy is hitting Segments - Its harder with bad position accuracy, because you could be "outside" the segments, even if you are not. 

What to you mean by GPS-accuracy?
- I mean GNSS position accuracy while running.

Did you try running with spacer, but at the regular postion? 
- Yes, it did not do any difference. Se this image (purple track)

Have been trying running at different times at the day?
-Yes, didn't do any significant difference.

Have tried with just GPS mode (Not using GLONASS, GALILEO)?
-Yes, didn't do any significant difference.

Don't you know that GNSS is never 100% accurate? 
-I'm not comparing to 100% accuracy, I am comparing to other watches that friends are using and that I have been using earlier..

Are you a garmin hater/fanboy?
-I am more a garmin fanboy. I had been using Garmin Forerunner 305, Forerunner 405cx, Forerunner 620, Forerunner 735, Forerunner 935, and been using Fenix 6 for about one year now.  I love the fenix 6 but the position accuracy sucks COMPARED to other watches.

  • 0 over 3 years ago in reply to knixen

    From my POV it's not generally a design flaw, since the traces are very good for cycling and other activities. But I'll call it rather suboptimal for running. And it depends on the user what he does expect and what of this is technically feasible.

    As the discussions always shows there are two different types of users
    1) who get angrily if the trace from a $700 watch isn't pinpoint as can be seen in this thread

    forums.garmin.com/.../fenix-6-pro-gps-problem-accuracy

    and as DCR stated in his review "There’s likely no topic that stirs as much discussion and passion as GPS accuracy. A watch could fall apart and give you dire electrical shocks while doing so, but if it shows you on the wrong side of the road? Oh hell no, bring on the fury of the internet!"

    2) who not care about a small lateral drift of the trace and are happy with the many other functions of the watch

    I'm almost type 2). But I've wondered why my instant pace oscillates strong and I've tried to find a solution. Furthermore the total distance of my route was ~4% shorter than really measured if I wear the watch normal on the wrist. This is mainly because of cutted corners in my route. An accurate instant pace is especially important for long distance running. If you overpace at the start then you are a candidate for "hitting the wall" near the finish. So I've coded some moving average paces with CIQ and I generally have some interest in the field of GNSS functionality. These are my motives.

    If I've a simple opportunity to enhance my device, I try it and look what happens. I always run with HRM strap and get my running dynamics. Of course, I can buy a stryde with additional cost of >$200 to enhance my instant pace. But additional devices create also additional problems regarding connectivity, maintenance, ...

    Cycling traces are always more accurate because mostly the device is mounted on the stem or weared on the wrist and the watch face points automatically to the sky. And in addition because the speed is higher than the position error. Therefore they have no issues.

  • 0 over 3 years ago in reply to J117

    I decided to wear the watch upside down on my wrist so that 12 o clock point upwards and lo and behold tracks did improve noticeably, I'm now convinced the antenna position is at the top of the watch.  I'm going to continue to wear the watch 'upside down' as see how it goes. 

  • 0 over 3 years ago in reply to knixen

    Don`t bother to convince this guy! Lost cause! There are plenty of posts already about GPS track shifting, which normalizes if you wear it on the inner side of the wrist or as you also notice on the wrist bone pointing the sky. It is either design flaw or production flaw (maybe a whole batch assembled not appropriately or with defect parts idk). To be honest it would be interesting some of these Garmin advocates to share their tracks while running in dense forest or around tall building, wearing the watch on outside of the wrist. I`d be surprised to see an accurate track.

  • 0 over 3 years ago in reply to 7466925

    Same here. I was surprise today on how accurate it was while going between 10 stores buildings and trees. Perfect tracing on mine Near the (1). 
    Before it used to be a very messy part of my tracing and now it is OK.

  • 0 over 3 years ago in reply to subra

    Instant pace oscilates due to GPS issues. If GPS worked well, it'd draw good tracks and instant pace would oscilate less.

  • 0 over 3 years ago

    This isn’t a debate on whether people should do what the OP did to improve accuracy (personally, I wouldn’t do that myself) but maybe on how to improve accuracy.

    If I was concerned about accurate readings (which I’m not because for the most part it’s fine unless the signal is affected by obstacles), then I’d get an idea of what the best satellite configuration would be for my area at the time that I require good GPS readings. Depending on my area, I have to change it based on the available satellites. Generally, GPS/GLONASS works well but I have to adjust that from time to time.

  • 0 over 3 years ago

    Maybe this problem is specific to people while running or doing activities in which the arms are raised at 90 degrees or so to the torso, which points the 12 o'clock position towards the ground.

    As a walker/hiker/cyclist I have good performance from my 6X Pro Solar, on par with my Polar V800 and Wahoo ELEMNT. I cannot run due to knee injury of 3 months and counting or I would test this in running mode, but I cannot complain about my watch as I use it today on the latest firmware. Earlier firmware is a different story.

  • 0 over 3 years ago in reply to knixen
    knixen said:
    Unfortunately that is not true, just take a look in these forums and you will see numerous users with exact same issues. 

    What you see are numerous posts from the same people replying to the new threads that keep getting created, which is not the same as numerous individual instances of the same problem.

    I applaud the posters who spend the time proving their theories, I really do but I am one of those 

    C.sco said:
    people it works perfectly fine while wearing it normal

    I am also one of those people who has tracks from other watches over the same course going back many years to the FR305 that show little or no difference between the track drawn on the map or the distance recorded over the years. That's not to say I don't get bad tracks - I've got a 'beautiful' 945 track from this past weekend which I posted the other day. However, this is something I've never seen before and haven't seen again. Of course I'd be up in arms is this was an every day occurrence. But it isn't. I rarely get tracks that are unacceptable. Interestingly, the distance was still good - this is a 5km measured parkrun course.

    As is often the case, one size/solution does not fit all. One solution for one person does not mean there is a significant design flaw.

    I remain unconvinced that Garmin, or indeed any other manufacturer would design a GPS watch without considering that it could be used on either arm. Perhaps there's the elegant solution - design lefthand and righthand specific watches, and perhaps one for the ambidextrous amongst us. :-)

  • 0 over 3 years ago in reply to VTrunner

    What's the rationale? I can't see how the choice of wrist matters to the watch antenna or other functions.

  • 0 over 3 years ago

    I get much better tracking when riding my bike than when running.  The watch face is pointed to the sky when I ride the bike, and my arm is not moving, of course.  I use the HRM-Tri belt and Stryd pod when running.  Next run, I'm going to wear my bike helmet with the watch strapped to the top and see what happens.  Sounds silly, but that's an easy way to get it pointed up full-time without building anything.