GPS Accuracy Issues

Straight comparisons to AW4 and Garmin FR935, the Instict has a distinctive "wobble" and also has cut distance pretty significantly (along with significant variation off actual path). Has anyone noticed the same? DCRainmaker seemed to notice the same as well.

(The only tracks I have so far start/end at my home so I won't be posting the tracks, but I'll try to get something else this weekend.)
  • Hi. I do want to note that I am very happy wth the Instinct. The watch is very lightweight and the option of loading routes for navigation really comes in handy. Also the battery is very good. I can truly use it for day to day stuff, without having to care too much about battery/annoyingly big screen or anything. This distance issue is just a detail in my opinion, but it is little bit inconvenient for me only because I am training for a speedy marathon. For me it is somewhat valueable to know if I am running 4:00 min/km or 4:07 /km, which is roughly the error I now encounter. In any other aspect than this detail (which I think must be possible to solve somehow) I am very happy with the Instinct. Please realize that there is much more to this watch than being perhaps not precisely accurate on the scale of 20 meters per traveled km...

    Anyways, I just ran again with both the FR235 and the instinct, both on the same arm. Now both on GPS+GLONASS, same settings. The Instinct still comes short by ~230 m over ~8.5 km. I will soon upload the runs again. Don't know what to test more. I do see a difference in measured altitude. The Instinct measures an average altitude of ~11 m, whereas the FR 235 measures altitude at ~3 m. The runs are flat and definitely not at 10 m altitude (Amsterdam is not more than few meters above sea level)).I am not sure if this could be related to the distance problem... If so, the FR235 does not have a barometer, which is a difference. I could set the barometer off in the next testrun. Yes, I will do that (next weekend).

    cheers

  • Looking at the raw GPS data. There are missing GPS datapoits in the Instinct .fit files. During these missing data periods, the velocity is commonly reduced, or set equal to zero (even though I was running constant pace). This velocity decrease does not seem to be 'absorbed' by a higher velocity in the neighboring data. Quite often, the data-gap is only 1 second of data long. It is almost as if the watch misses these data deliberately. Why would the Instinct miss out on GPS only for one second on a semi-regular basis, seemingly independent on clear sky/buildings/tree-cover and GPS system? I do not see this problem in the data of the FR235. I am not an expert on this, but it looks more and more like a software issue to me.

  • I have exactly the same problem. I run marathons for speed (under 3:00) and knowing my pace is very important, when  difference from 4:00 to 4:10 plays a lot. I did not go that deep as you, but I confirm, that old Garmins do this work much better, compare to Instinct. I tried 

    FR 225, FR35, FR 925, FR 935, Phenix5 - precise pace measurement.

    FR 45, Instinct - have this problem.

    It looks like last 4-th generation has such a problem, but all watches series 2 and 3 are not affected. Hope Garmin engineers read this forum and will make update soon. Another option is to buy *35 model and sell Instinct.

  • Just updated to the last 5.10 version, will check if it helps. But changelog is quite about this issue... 

  • First run with 5.10 firmware. Pace measurement is good even in trees shadow. Tomorrow will check it again.

    GPS set to GPS only (normal mode) and record track every one second.

  • Same as before, pace is incorrect. Going to buy FR935. Instinct is bad model, has a lot of features but lack of GPS quality.

  • Hi hogmemberisold and alex,

    Thanks for your interest in testing it with me, and reporting some results here. I think it is worth it to share this info. 

    I have now done some more runs, and specifically tested with taking time for the GPS to soak. I now start from a bridge, which has more open sky (i usually starting on the street between two rows of buildings). I now intentionally wait some time after the watch gives the 'GPS-ready' signal, doing some stretching etcetera before starting runnning. This is actually beneficial, because I never did take time to stretch, and it feels quite good :-). Anyways, I clearly see now that the track follows my factual run much much better. The distance is also ~150 m longer, and comes quite close to that recorded previously with the FR235 (<100 m offset to the FR235 on a 8.5 km run). 

    I always thought that GPS soak would not be so important, because i thought the watch would anyways connect to more available satellites while running. But it appears to truly help taking some time before pressing start. If you are in an open space, this does not require 20 minutes or so, as reported at some sites. Taking a minute is more than enough to improve the tracking. 

    Perhaps you could try and see if this also workd for you guys? 

    By the way, the earlier reported 1-sec GPS-gaps in the raw data still exist if i put the watch in 1-sec recording mode. But the velocity and distance now appear normal. These gaps appear to not be present in smart mode. cheers

  • Hi, . Please share how you check for the gaps in raw data. 

  • Hi, Alex,

    I load the fit file to goldencheetah. There I see that there are GPS coordinates missing at some seconds. 

    Sometimes, the speed is set to zero at these seconds. In those cases, I can see in the speed-or-pace plot of garmin connect that there is a missing datapoint if I look closely. I can only see this in the speed/pace plot when i do not expand the speed/pace diagram. As soon as I epand the speed/pace plot, the speed/pace is smoothend or something, so I cannot see the missing speed datapoint. However, in many 1-second gaps, the speed is not zero, and I only see that there are 1-second datagaps by the GPS coordinates, for which i need external software (i.e. goldencheetah). cheers

  • Instinct shows very poor accuracy on 2.8 and 5.1 versions of the software. 70% times it shows correct, but all the other it can be +/- one kilometer on 15 km distance. Route on the map can look quite strange: start or finish point can be few kilometers away from real position, route as well can be few hundred meters away from the real. And it doesn't depend on a weather, metrics can be good when it's cloudy and wrong when it's clear and vise versa. It's not possible to find support's email to complain and provide them files from phone or whatever to fix issues. Instinct clock just a wasting of time and money.