Team
I have been using my phone as a backup when my Stratos watch would crash and after few rides I ended up using the records from Strava app on my Android phone as it measures longer distance ( ~ 0.5km more ) and higher altitude ( ~ 100m more ) than my Stratos watch running latest Firmware 2.3.10.5 or Hybrid ROM 2.3.
It seems to me that latest Firmware either Stock or Custom are using same GPS sensor driver which is simply not measuring distance correctly.
I`m now testing older FW called "STRATOSFied version Rel 11: Full install Version 2.0.27.0" based on recommendation found somewhere on the internets , will report back my test results. ( so far it looks close to what I get on the phone )
If you have similar issue and figured out the way or found the golden FW which measures relatively accurately distance and Altitude please share.
waiting for results!
OK attached is a picture of the result I went for a ride and used 4 devices to measure distance.
Here is the result
Stratos ( FW 2.0.27 ) = 35.71km , 179m elev. gain
Garmin 405 ( yes still alive ) = 35.95km, no elevation
Nexus 5X + Strava app = 36.05km, 197m elevation
Pixel 2 XL + Strava app = 36.80km, 197m elevation
Not sure what GPS chip is used in Pixar 2 XL but it seems to be the answer to the accuracy. I assume Strava would not log fake kilometers/distance.
More than 1km difference in distance seems to be quite a lot.
I will do the same route with latest FW 2.3 and we can compare if any difference.
Turn off the watch for at least 15s to reset sensors, then it will report elevation accurately. Also, you may want to disable 3D data, if it is enabled.
Related
Hi all,
So I wanted to start this topic to share with you a problem with GPS altitude accuracy I've been having since changing custom ROMs and BB to 407.
I know the GPS altitude is not extremely accurate anyways but within 30m it should be fine - which is acceptable by my standards and for my needs. Besides being inaccurate, the altitude reading should be stable in any case.
I use the GPS for when flying - with a clear view of the sky and usually with no clouds at all(or just a bit of cumulus). On the 407BB and with a Longhorn ROM I had instances where the reading was 1000m one second and 700m the next!!
Here is the history of the problem:
- 1 year of use 360 stock Rom&BB + XSsoar - flight track-logs no problem at all: altitude accurate to 30m and no extreme "jumps"
- Installed 407 BB and Longhorn 2.9 - certain track-logs with 100-300m "jumps"(errors) during certian readings. Horizontal accuracy was not too bad but it was better on the stock.
-Installed DSC a week ago - I had no time to test rigorously but with a small walk in the city(not really representative) I still had some(few) problems. Although I cannot complain since the altitude reading in cities is usually buggy with this.
-Now I installed back a 360BB and stock Rom - the accuracy seems to be a bit better - BUT still did not test in the open air - will do one of these days and post the result.
I am attaching a flight track log(KML format for GE) for you to see the problem yourselves. Besides I have attached a track-log of THE SAME FLIGHT done by a Garmin 60CSX gps(my backup GPS for flying) for you to compare the extreme discrepancies.
Code:
Dell Streak 407BB Longhorn ROM TRACKLOG (white):
juventusventus (dot) org (slash) temp (slash) Dell407BBLonghornXCSOAR.kml
Control LOG file with Garmin(this is what my previous(360BB+stock) log files used to look like as well):
juventusventus (dot) org (slash) temp (slash) GARMINcontroltracklog.kml
Hopefully som1 can help and diagnose the problem since I'd rather have the custom ROMS on coz they are much better at everything else other than the GPS accuracy.
A little late in reply but try this
http://www.tabletroms.com/forums/showwiki.php?title=DellStreakFAQ:GPS-FIX
I and all you guys may met that problems many times before, and after doing many test I can see the cause of the problem. This time I show you one case I have just done it few hours ago:
- First I'm using iphone (but it not effect to the test)
- Amazfit app, change to us verson (simple just change the region to my country), so it will sync the activity to Strava.
- Watch Rom: us official 1.3.3a (the result is the same with dev PACEfied rom)
- Do the ride test around, with many turn.
- Sync the activity to Strava, go to strava export TCX file.
- Export TCX file directly from the watch. (How to)
****Okay, now compare:
Strava activity link: https://www.strava.com/activities/1068275813
TCX file export from strava and watch (see attachment files)
How to compare: import TCX files to this size: http://www.gpsvisualizer.com/map_input?form=google
The result you can see at pictures attachment.
****Conclusion:
- The original GPS file from the watch is very good, no drifting at all.
- After syncing to Strava, GPS route change alot, drifting happens.
- So the issue here is the step "syncing to Strava"
****Solution:
I have no solution, because I am not the engineer from Huami, I hope people from Huami/Xiaomi see this post and will fix this problem. I have read many feedback from Amazon, Facebook group that they have to give back the watch because the drifting problem. Maybe Strava is not fair to change something when syncing (my thinking). The watch is really good but this problem made many people disappointed.
It is "easy" to spot what happens.
while the Amazfit saves the coordinates with 15 decimal places Strava uses only 6 ( and adds a 0 ).
If we take the first entry
Amazfit:
<Time>2017-07-05T04:35:20Z</Time>
<Position>
<LatitudeDegrees>10.391833305358887</LatitudeDegrees>
<LongitudeDegrees>107.10086822509766</LongitudeDegrees>
</Position>
</Trackpoint>
Strava
<Time>2017-07-05T04:35:20Z</Time>
<Position>
<LatitudeDegrees>10.3918330</LatitudeDegrees>
<LongitudeDegrees>107.1008680</LongitudeDegrees>
</Position>
You can see that there is a difference, but a small one ( ~13cm )
If you then go to
Amazfit:
<Time>2017-07-05T04:40:57Z</Time>
<Position>
<LatitudeDegrees>10.385744094848633</LatitudeDegrees>
<LongitudeDegrees>107.11280822753906</LongitudeDegrees>
</Position>
Strava
<Time>2017-07-05T04:40:57Z</Time>
<Position>
<LatitudeDegrees>10.3857360</LatitudeDegrees>
<LongitudeDegrees>107.1127090</LongitudeDegrees>
</Position>
Here the difference is ~10.8 meters.
so you want to tell me that the watch is very precise and Strava is castrating down the results ?!?!?!!
I have installed the python scripts and extracted the data to GPX files.
One of them I uploaded to Strava and then re downloaded it again.
And the downloaded one was with Single precision.
But it was only off on a few points in a 12.5 Km run.
So it is a combination of Strava storing everything as a single ( not double ) .
But even there the data from the Amazfit was drifting, but it was a run through a forest.
If it was only for the precision it wouldn't really matter that much ( maximum of ~0.5 meters ).
So it seems that
1 - The Amazfit is drifting, and
2 - the upload to Strava sometimes does a strange "correction" to the data.
Thanks a lot Trần :good: this has been reported to the team and we are looking into it.
1immortal said:
so you want to tell me that the watch is very precise and Strava is castrating down the results ?!?!?!!
Click to expand...
Click to collapse
Just my thinking, the problem happened when syncing to Strava, and we got the bad final result.
fzelle said:
I have installed the python scripts and extracted the data to GPX files.
One of them I uploaded to Strava and then re downloaded it again.
And the downloaded one was with Single precision.
But it was only off on a few points in a 12.5 Km run.
So it is a combination of Strava storing everything as a single ( not double ) .
But even there the data from the Amazfit was drifting, but it was a run through a forest.
If it was only for the precision it wouldn't really matter that much ( maximum of ~0.5 meters ).
So it seems that
1 - The Amazfit is drifting, and
2 - the upload to Strava sometimes does a strange "correction" to the data.
Click to expand...
Click to collapse
Sorry for my bad, I dont follow your words (Single precision and Double), are you using your record or my files? Maybe it better with picture, thanks.
netzkobold said:
Thanks a lot Trần :good: this has been reported to the team and we are looking into it.
Click to expand...
Click to collapse
Thank you so much for your responding! :good:
Single/double Precision is a programming/computing term.
Compared to Double they have a smaller range and therefore are easier to store, but not as accurate.
1.2.27 CN
[bold]- Update the GPS firmware
- Optimize the drift of GPS
[/bold]
- New voice control meter air purifier function (such as: open the living room of the air purifier, the air purifier set to automatic mode)
- Repair of some cases caused by the speed or speed display abnormal situation
@netzkobold:
Can we expect this for the pace ( US ) as well?
Has anyone reported this solution to the Huami- team and got a positive feedback to solve the problem in the next update?
mumfordfan said:
Has anyone reported this solution to the Huami- team and got a positive feedback to solve the problem in the next update?
Click to expand...
Click to collapse
If netzkobold said that it was reported to the team, trust me, Huami are working on it.
I also have this problem. The drift is sometimes 20+ meters from the road so some Strava segments are not captured. I have downloaded the raw data from the watch (gpx file via python script) and uploaded that manually to Strava and it is perfect. Definitely something is happening when you sync to Strava via the Amazfit app. I can upload my files if anyone is interested.
Ok will try to be clear;
I have the latest sw 2.3.4 on Stratos 2 installed
I passed my watch in metric using the app watch amazefit. No way to do it on the watch.
When doing a sport (e.g running) the distance is set in mile. Was confused first time.
When saved activitiy, the distance appear in KM in the app ... and also in the watch ..
So seems that it's only during the activity (only tested with running so far).
Anyone else..?
Thx
Change to miles, save, restart, change back to km and restart once more. If you see "mileage" on display, it actually doesn't change with units, read it as "distance".
thx
Works I have now the distance in KM in the watch during activities.
I ve notices a difference between strava and Xiaomi : strato 6km; strava 5,8km
Hi everone,
I am using Amazfit Stratos, Rom version 2.3.4.0 .I have a climbing activity on 1 spetember 2018.
The total mileage shown on watch & Amazfit App is 15.1km, but it only 12.2km (12.6km if consider 3d distance) via Strava or Google Earth after I download the gpx file, it is 23% different. The huge mileage different also happened when climbing mountain with elevation more than 500m, the others activity will have 3% to 6% different (the watch reported higher).
Anyone have this problem? How to solve it?
Is anyone noted that the elevation info always missing in the first few point of the GPX file downloaded?
Probably you had 3D Data on (search for what it means).
lfom said:
Probably you had 3D Data on (search for what it means).
Click to expand...
Click to collapse
Yes, I know what is 3d data, I have set 3d data on for this climbing activity.
But it is 12.2km (without consider 3d) and 12.6km (with consider 3d), calculated by myself and checked using "Google Earth".
Will it because of movement of the hand where I wear the watch?
Sounds like a problem with the altitude measurement. I'm not happy with that too. Opening compass, doubletapping on it and waiting for a GPS fix before starting an activity is helping a bit.
I see several reasons for this:
1) The phone app is smoothing out the altitude data by using digital elevation map data, which at the end comes to different mileages.
2) In the mountains unprecise GPS signal can lead to significant differences, 30m 2D can sum up to 100m or more 3D, a balance between weak signal filtering and loosing signal is not easy to achieve.
3) During my tours I had another GPS running parallel. What made a real difference was, what I wrote under 2) and that my Pace did not exactly recognize pauses, which resulted in several pausing clusters which added to the mileage.
My conclusion:
I'm using the watch but after the activity I:
1) export it to tcx format via the exporter available in this forum.
2) manually delete the first trackpoints which either have no altititude or an unrealistic speed.
3) correct the altitude with external tools if I see it necessary.
Some of this after activity steps are done by the phone app automatically. Since the phone app is not giving me options for this I use other tool to know exactly what is going on.
Edit: The handmovement shouldn't influence outdoor results.
Did you check the altitude graph? It seems that run/walk use GPS for altitude whereas trail run/climb use the barometer. I still think it's related to 3D Data: the higher the altitude change, the higher will be the final difference in distance when compared to common "2D" data.
---------- Post added at 01:39 PM ---------- Previous post was at 01:24 PM ----------
Also, latest international firmware for Stratos is 2.3.6.1, and you are still running 2.3.4.0... Latest Amasfit app version may make a difference too... Auto-pause may make a difference too: either disable or use something other than "stand still".
lfom said:
Auto-pause may make a difference too: either disable or use something other than "stand still".
Click to expand...
Click to collapse
Are there other settings for Auto-pause than "on/off"? If so, where can I find them on the pace?
Profil Climbing and profil Trail Running have only Auto-pause "on/off" on ROM Pacificator WOS 2.6.3.1.
nhedgehog said:
Are there other settings for Auto-pause than "on/off"? If so, where can I find them on the pace?
Profil Climbing and profil Trail Running have only Auto-pause "on/off" on ROM Pacificator WOS 2.6.3.1.
Click to expand...
Click to collapse
Not all activities have the other options, then you can try disabling it and use the button to pause manually. When recording walkings I found that it was too sensitive and it added a lot of bad data points to GPS when enabled.
lfom said:
Not all activities have the other options, then you can try disabling it and use the button to pause manually. When recording walkings I found that it was too sensitive and it added a lot of bad data points to GPS when enabled.
Click to expand...
Click to collapse
I experienced the same phenomena. Thats why I stopped using "Auto-pause" and was wondering now about the possibility of finetuning autopause. Which activities have this, I couldn't find any?
Edit:
I found one: Cycling (1km/h steps)
Thats understandable but with the bad HDOP&VDOP filtering the Pace has, configurable Autopause in 0.1km/h steps could help in other activities too.
nhedgehog said:
I experienced the same phenomena. Thats why I stopped using "Auto-pause" and was wondering now about the possibility of finetuning autopause. Which activities have this, I couldn't find any?
Click to expand...
Click to collapse
Run has an settings based on pace, Outdoor cycling has it based on speed. "Common" walk doesn't have it, it's either on or off.
[/COLOR]Also, latest international firmware for Stratos is 2.3.6.1, and you are still running 2.3.4.0... Latest Amasfit app version may make a difference too... Auto-pause may make a difference too: either disable or use something other than "stand still".[/QUOTE]
the firmware now is 2.3.6.1 (2.3.4.0 is when that particular climbing activity on 1 september ),
I always off the 'Auto pause' except the first time i use the watch, because it keep on pausing (because my activity is inside the forest).
https://drive.google.com/open?id=1KSCwCRNmbnLdRgJaI1x9lkoogJN0Qv25
this is the list of all my activity, it is true that '3d data' will affect the total mileage. When I on '3d data' to climb the mountain, the total mileage is 20+% higher, but it is 3 to 5% higher when only the small hill.
However, activity after 29 september , '3d data' is set to 'off', the total mileage still 3 to 5 % higher.
lfom said:
Did you check the altitude graph? It seems that run/walk use GPS for altitude whereas trail run/climb use the barometer. I still think it's related to 3D Data: the higher the altitude change, the higher will be the final difference in distance when compared to common "2D" data.
https://drive.google.com/open?id=10whj9izJoNCdXGQOxWMZ-pvTdYv2fW5e
https://drive.google.com/open?id=1PjBOYg-Op3nOSbuGoq2i7gDrRceBq7On
https://drive.google.com/open?id=1T-pJWxehu0a7TpqhL1-7S3Is7GkLR5eb
https://drive.google.com/open?id=1PXFCi2TMMsf94Kt5GxrWCkF-J_3S8JQp
those are the elevation profile of activity on '1/9 climb 3d-on', '3/11 walk 3d-off', '10/11 run 3d-off' & '17/11 trail run 3d-off', it seem that all altitude record by Amazfit is lower than which corrected by 'Strava' or ' gpsvisualizer.com' or record by handphone.
Any solution to keep the total mileage within 1 to 2%? after off the 'auto pause' & '3d data' ? 3 to 5% is still consider too high.
Click to expand...
Click to collapse
Having a look at your pictures your altitude data is much better than that what my pace produces before this was "corrected" by the phone app.
I guess that means it comes from the phone app.
So that leaves us 3 options:
1)
Those 3 altitude correcting sources you use:
Amazfit
Strava
GPS visualizer
are probably using different altitude maps - meaning you can't compare them and have to decide which one you use. I guess the one from amazfit
has a lower resolution compared to those from Strava&GPS visualizer. You probably can't change the altitude mapsource of those 3 apps - just decide which one you prefer and stick to it.
2)
If I'm using altitude correction I don't rely on either of those apps but use LIDAR maps which have a higher resolution and therefore representing the altitude of the area better:
http://data.opendataportal.at/dataset/dtm-germany
To do that I need the RAW data from the watch, which I get as mentioned via a python script as tcx files directly from the watch. Edit: One could use the gpx file which the watch procides under activities but the program I use for reprocessing the track does not recognize the included heartrate data. Otherwise internal created gpx and external created tcx don't differ much.
My resolution: another app
3)
The third option I see would be relying on the altitude data the watch provides, which I personaly found not reliable at all. I guess that has something to do with the filtering of weak signal from the gps chip, which is done by the gps firmware. To solve this problem we can only hope for some better filtering via GPS firmware by the manufacturer.
Hello,
I recently purchase Amafit Stratos main for sports activity and good GPS tracking.
My decision was made according many good reviews but what a surprise?
After several activities i was very very disappointed by its GPS tracking, im on latest 2.3.7.0. ROM.
For example walking only 6km, it was off bu 200m.
The gps map was showing gps drift and bad road tracking, i was compare it with polar m200 and garmin fenix 5.
When it upload data with Strava non of data didnt match, i mean total distance and every km i run.
I even make factory reset, but results was the same, bad unprecise gps and different data with strava .
So i search again on net for someone like me complaining about bad gps accuracy for stratos but nothing found!
Please help.