Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race
-
-
@silentvoyager Favored younger child receives momentary abuse from parent. Abandoned older sibling has mixed feelings. Any Spartans in that race?
-
@Fenr1r There was just one Spartan Sport out of about 100 activities I looked at, and it showed a short distance - something like 14.2 miles if I remember correctly. I forgot to mention it.
-
@silentvoyager can I can the race GPX ?
-
@Dimitrios-Kanellopoulos I’ll share it once I get home from work. The day is just starting for me.
I think the reason it is short is improvement in Sony GPS firmware that removed wobbling combined with the smart recording intervals.
Trails do naturally wobble a lot, especially in my area where trails are rarely straight. GPS algorithms that are optimized for roads tend to cut through those wavy trails. On top of that Suunto has an algorithm where it tends to store points only about every 10 meters or sometimes even further apart. I can see how that would produce better tracks on roads, but on trails true one second recording would be more accurate.
In the past the wobbling kind of compensated for the smart recording so at the end the distance on trails was surprisingly good despite tracks being not very accurate.
There should be an option per sport, whether to use smart recording or record (and update the distance) every second.
-
@silentvoyager Totally agreed you. Option in each sport mode to activate update of GPS coordinates every second is a MUST!
-
@isazi said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
Never used Strava FlyBy before, but looks very cool to extract statistics from public events.
I’ve used it a lot but never thought of it as as such a data source. This stuff is pure gold :]
-
@margusl
it looks interesting… but what you don’t know is- watch settings
- firmware
- settings and firmware of competitor watches
… by this table, Fenix 3 finisher 4:10 nailed it
-
@margusl It is never occurred to me to use the developer tools console to print the data like this. I was opening them one by one. I’ll play with this more tonight. It should be possible to write a chrome extension that will automate extracting data like this.
Looking at this table I realize that slower runners would probably have a bias towards longer distance. When someone stops their GPS positions tend to wander around adding a bit of distance.
-
@Dimitrios-Kanellopoulos Here is the GPX file:
-
@TELE-HO said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
by this table, Fenix 3 finisher 4:10 nailed it
Keep in mind that in road marathons most runners end up running more than the actual marathon distance due to not following ideal tangents, waving around other runners, etc. Road race distances are measured following most straight possible path through every corner, etc.
-
@silentvoyager
…yes I know… but people like to see (edit typo) 42.195km on their watches when they run a marathon, no matter how many other runners they have been circling and how many water stations they used…
it is very hard to proof which watch is the most accurate as we don’t know some parameters.
I like to have an accurate watch, too.
The same run I did on Sunday and before that on Saturday last week showed 4m less ascent and 50meters more distance.
Could be anything from not exactly the same start/end or satellite reception (as I lost satellite reception a week ago I think this was it…) or I went around hikers in a different way…
What I want to say is that a proper statistical analysis is very difficult due to too many variables that we don’t know.
Some records are clearly off precision if you check them, no question. I hope we can help Suunto somehow to improve GPS distance and Baro ascents
edit: this is it by the way: https://beta.quantified-self.io/user/jBm0qOhihUMykVYD8HBJMshQrJc2/event/LZ8RKDysG9Csh4i367AG
edit 2: the rout planning showed 7.06km and 256m ascent…
-
@silentvoyager the distance only from GPS based points confirms your distance.
I would like to run the same analysis on the GPX file of the competition (eg route provided)
You can actually do that your self if you like so.
GPX files do not have distance summaries, so QS uses the vincety algo to calculate the distance out of track points.
For FIT files (dont rememember about TCX ) the distance is read from the file summary
-
@Dimitrios-Kanellopoulos I don’t know if I can get a GPX or FIT file for any other runners. Strava allows me to create a route from someone’s activity but apparently it now snaps the route to trails and that changes the distance.
-
@silentvoyager Good one. You are right it aint easy. I was more wondering about the realroute aka the declared distance.
-
@silentvoyager said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
@Dimitrios-Kanellopoulos I don’t know if I can get a GPX or FIT file for any other runners. Strava allows me to create a route from someone’s activity but apparently it now snaps the route to trails and that changes the distance.
Try this - https://mapstogpx.com/strava/ , it builds GPX from Strava data streams, i.e. while Strava route export is not involved, some filtering and processing might be.
@silentvoyager said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
When someone stops their GPS positions tend to wander around adding a bit of distance.
Close, but not quite there - this is more like an example where slower runners tend to forget themselves, and add some extra distance due being lost
(F5 with longest distance spent about 30mins off route)@TELE-HO said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
@margusl
it looks interesting… but what you don’t know is- watch settings
- firmware
- settings and firmware of competitor watches
… by this table, Fenix 3 finisher 4:10 nailed it
Going too deep into limited data set that might or might not be already altered by someone will no be best use of ones time :] I can just say that there are multiple more significant factors there: rain, tripping wet forest, multiple start groups, neto timing, starting watch when already on the course, sightseeing, getting lost. Though It is possible to spot some trends and check if spread over distance and/or elevation somehow correlates with watch brand/antenna type/GPS chip/etc. Using Strava as a source also poses a huge bias on sample distribution: as a platform it does not appeal to everyone and it’s privacy settings form another filter. So I’d be very cautious to conclude anything out of this, definitely not most/least accurate watch. Consumer trends? Maybe. But knowing that doesn’t make this fiddling any less fun
-
@margusl said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
But knowing that doesn’t make this fiddling any less fun
-
yesterday and today I took short rides and compared on random/different short routes the track accuracies for Switzerland of the different GNSS combination settings…
my takeaway: it is a tiny slightly little bit more accurate with GPS+Galileo but definitely not worth the higher battery consumption.
I did not compare the distance of cours, as all routes are different.… but the activity with GPS+GLONASS has for 100% sure issues with the total ascent…
GPS+Galileo is way more realistic!
I suspect that the baro sensor holes was covered and it caused this pumping effect as I already know from Suunto Core… that’s a huge disadvantage and I hope that I don’t see this frequently. But to be honest, from a design engineer point of view: the location for the pressure sensor is a faulty design
especially when you have to tighten the strap enough to get halfway reasonable OHR readings…living sideways
-
Here is a comparison between Ambit3 Peak, Suunto 9 with one year ago firmware, and Suunto 9 with the current firmware. The routes weren’t exactly the same, but there are some common parts that allow to do some comparisons.
https://quantified-self.io/user/JMRgYAdyBBXBsMbxUHGVRwlKoKq2/event/xT4vJjRS3TPvm8WuF6y2
I’d say that in general the track of Suunto 9 with the current firmware is improved - it stays closer to Ambit3 Peak track, especially on straight segments. But at the same time it has a tendency to cut bends and corners. Here is one example of where it completely cuts through a series of short switchbacks:
Compared to that the previous firmware had much more random wobbling, but I think that was also a reason of why its distance was closer to the real distance on trails - it would cut corners in one place and compensate that with some wobbling in another place. The new firmware doesn’t wobble as much but it still cuts corners, so it ends up being short on trails.
-
@TELE-HO said in Suunto 9 with the recent GPS firmware was among least accurate GPS watches in today's 25K trail race:
yesterday and today I took short rides and compared on random/different short routes the track accuracies for Switzerland of the different GNSS combination settings…
my takeaway: it is a tiny slightly little bit more accurate with GPS+Galileo but definitely not worth the higher battery consumption.
I did not compare the distance of cours, as all routes are different.… but the activity with GPS+GLONASS has for 100% sure issues with the total ascent…
GPS+Galileo is way more realistic!
I suspect that the baro sensor holes was covered and it caused this pumping effect as I already know from Suunto Core… that’s a huge disadvantage and I hope that I don’t see this frequently. But to be honest, from a design engineer point of view: the location for the pressure sensor is a faulty design
especially when you have to tighten the strap enough to get halfway reasonable OHR readings…with GPS+GLONASS seems have better result than others
