Related

GPS: Completely unacceptable

Dennis.stevens1982
over 4 years ago

Hey all,

I have posted my problems with the Fenix 6X Pro in German Forums here and here. Maybe I get some new advices here.

First things first:
FW V5.00
GPS V2.7
3D-Speed: OFF
3D-Distance: OFF
Data recording:Every second

When I used the Fenix for the first time, it ran on FW V4.30.0 and had acceptable performances in measuring distances. On a ~12.5km run it measured for example between 12.01 and 12.1 km. Which was in accuracy comparable to the Forerunner 245 which I have used before- and really OK for me.

My focus in running are the pace and the heart-rate. Which are the most important parameters in addition to my own body sensation. If I run 12km in 5:25 with a HR of approx 160 I am fine. If I feel great after the run, the training was perfect for me.

The pace is a function of time and distance - and now on the Fenix the distance is a huge problem since update to FW 5.00.

On 15th December 2019 (my Fenix received V5.00 the day before) the GPS accuracy became really bad. But each time I follow advices like waiting 25 minutes until the Fenix has updated her satellite-almanach or updating to GPS V2.7, in the first session after such updates the Fenix delivers accurate distances. But each following training it gets worse.

Examples:
1. updating the satellite-allmanach: Distance 12.06km, time 1:05:15, pace 5:25min/km
2. next training (still with GPS V2.6): Distance 11.78km, time 1:08:45, pace 5:50min/km
3. After Update to GPS 2.7: Distance 12.06km, time: 1:07:07h, Pace: 5:34 min/km
4: Next training 11.87km in 1:09:50 h, pace 5:53min/km
5: today 11.37 km, time 1:07:25h, pace 5:56min/km

The distance I run is always the same: 12.5km

Training 1 and 3 proof (to me) that after an update the Fenix works perfectly.

Training 3 and 5 reveal that it is not possible to rely on the pace. Since accuracy (690m deviation) is far beyond acceptable, two trainings in the same distance, with a deviation of 18 seconds are calculated to have a deviation in pace of 22 seconds. And such is not acceptable. 22 seconds in pace can make the difference between an easy aerobe training and an hard training. VO2Max depends on the pace and the heart-rate. In Training 3 the VO2Max was calculated with 49. In training 5 with 47. But the diference in heart rate was only 4 beats.

Important: I do not care for calculated Vo2Max. I just wanted to show how the GPS-accuracy has impact on many other parameters in the device.

Based on my experience, my best solution would be to downgrade to FW V4.30.0. But I was told that this is not possible since I have to downgrade sensor-fw and other software, too.

Can anybody help me with that problem? At the momment sport is just frustrating. And the reason for that is a sportwatch. Unbelievable.

kind regards!

Dennis

Top Replies

All Replies

  • 0 over 4 years ago in reply to Maurycy

    Yes, that is a good track. And it would look even better if the view was straight down. Images from satellites do have a small angle to it (look at the shadows of the buildings).

  • 0 over 4 years ago in reply to Simone.01

    Will do Blush I have 2 round tracks in the city. I’m not sure I can just walk there and run on them, but will try. If not, I will just pick a block (no tall buildings) and do couple of loops.

  • 0 over 4 years ago in reply to Maurycy

    Under those conditions probably every device would deliver good tracks! You see differences in deep woods, hills, cities, edged courses and slow activities.

  • 0 over 4 years ago

    Kidnapped my wife's F6 last night to go for a run on the trails with another F6 and a Marq Athlete. The results you can see. The closest agreement was between the Marq and an F6 on the left wrist. The shortest distance was with another F6 on the right wrist. The others in the groups had distance around 7 km. Given the terrain - steep twisty tracks often with overhead cover and deep gullies, I'd say this was acceptable. Easy to see the tracks are closer together in the more open and straighter sections and a bit spread in the covered and twister sections. I've been running these trails for the past 10 years or so and can assure you that I can produce similar tracks going back to the 310XT and forward.

    As soon as you give any GPS irrespective whether MediaTek, Sony or whatever a 'challenge' corners get cut, recorded tracks wander away from the trails, and distance variations increase. Here we're looking at a max variation of 380m horizontal and 3m vertical between watches in challenging conditions on a route that starts and finishes in the same location.

    I'm not interested in getting into a discussion about which track is better than the other. Any one of these in isolation would be acceptable to me. The differences look bigger when overlaid but the returned distances are within my expectations from a wrist-worn GPS watch.

  • 0 over 4 years ago
    Simone.01 said:
    Sony chipset is famous for cutted bends, also in open air and not in cities.

    And I'll say again - ANY GPS chipset in a watch will cut corners under challenging conditions whether in an urban 'street-canyon' environment or dense overhead cover and/or deep gullies. The issue is not limited to the Sony chipset.

  • 0 over 4 years ago in reply to philipshambrook
    philipshambrook said:
    Given the terrain - steep twisty tracks often with overhead cover I'd say this was acceptable.

    This is my experience as well. Last saturday I did a ≈22km trail together with five others. A quite heavy tree cover, bad weather (both windy, rainy and cloudy).

    I used my F6X, the rest was a F6 Pro, a FR935, an AW and two I don't remember.

    In the end I had 21.72, the F6 Pro got 21.44, the FR935 21.47. The rest between 21.4 and 21.5. I find that very acceptable, we know from sources (it is a marked nature trail) that it is about 22k but no one knows the exact distance. The path changes from year to year due to deforestation and other natural causes.

    The following picture shows the F6X, F6 Pro and FR 935 on Strava Fly By:

    If I zoom in it is still a good precision:

    Unfortunately no scale but straight from top to bottom is about 550m. Black is my F6X, light green F6 Pro and blue is the FR935.

  • 0 over 4 years ago in reply to Dennis.stevens1982

    We’re assuming that the watch is wrong. It could very well be that the 12.1km measured by a bike speedo was wrong. Can we verify this distance?

  • 0 over 4 years ago in reply to Simone.01

    Here You go. Short, but for me great GPS for that conditions.

    Fenix - 2,52 km (2,536 km from GPX) - https://connect.garmin.com/modern/activity/4594352032
    AW - 2,56 km (2,58 from FIT file) - https://connect.garmin.com/modern/activity/4594424763

    1 lap (according to Topo maps) is ~360 m x 7 laps = ~2,520 km so for me - Fantastic.

  • 0 over 4 years ago

    I think often, we misunderstand the limitations of the technology therefore have high expectations of it’s capabilities.

    Consumer GPS typically has an accuracy of +/- 3m on every single sample point under “best case scenarios” - generally hand held devices with external antennas in an open field with clear skies. GPS.gov advises typical accuracy for a smartphone is +/-4.9m. A watch is likely going to be worse.

    How many individual sample points are contained in a 20km run at 1s recording? I’d guess 10,000 to 13,000 sample points, each with an error rate >3m. Sure, algorithms can smooth out those lines but it doesn’t overcome the fact that each individual sample has an inherent error of >3m. That is a limitation of this technology for every single manufacturer.

    How can we overcome this? We either accept that there will be some variance due to the limitations and be happy that the data “is pretty close” or we need to move to higher accuracy equipment.

    We need to understand that accuracy means how close the recorded values are to a known value. We need to understand that precision means the repeatability of the readings. If the device consistently measures slightly short (or long) but the results are within 1-2% of each other, then precision and repeatability is there, so we can therefore use it as a training tool to compare against previous results on the same device.

    If your needs are such that you need extremely accurate results with high precision, then I’d suggest you look into commercial GPS receivers and professional sensors. I’d suggest that only elite athletes looking for the last 1-2% of performance would need this and they’ll likely not fit on your wrist.

    If your concern is just accurate pace readings, invest in a footpod.