Errors in the Walking Activity

Stavros _Clinica Orl

@isazi … good suggestion; now consider this … the basic metabolic rate is fixed in time (ie tot calories per hour) … so the variation of the error I am seeing should be declining as the activity continues ( 2-2.5 hours). I am seeing this, but I will validate it and post the data.

Egika

Let me stress my initial comment on this: why would a model based on limited input data (phone) produce better results as one that has additional input data (watch)?
It makes sense to compare energy consumption computed by a Suunto and a Garmin watch (and you will find, that they match closely).
Phone’s calculations are more or less guessing here.

my2c

Stavros _Clinica Orl

@egika I agree with your suggestion. The problem here arises from the type of activity (walking not running) and the assumption that the S9P algorithm might extrapolate data in an erroneous manner. Obviously a comparison with another source (Garmin, Polar) would indicate any discrepancies. Unfortunately I have only one watch now and most of the watch comparisons in Internet are done for the running modality.

Stavros _Clinica Orl

@egika and the other senior member of the forum: CAN someone make a test with the S9P and another sports watch for the Walking modality ? Even a 1 KM session would give interesting data. Thanks

Efejota

@stavros-_clinica-orl I have used walking activity often lately. My results are more or less consistent (which may indicate overstimation or not), as I usually repeat the route at a similar pace. I checked the last 5 walking activities and I get a similar ration. I divided nr of steps in the app and distance and the results are: 0,88; 0,80; 0,83 and 0,83. I have a very different result in one case for which my step is clearly longer (up to 1 m.), but it is a different route with a long downhill section. I always wear my watch in the left arm and have no means to evaluate with another gadget (ok, my phone, but I rather trust the S9P over the iPhone data, to be hones).
J

herlas

I’ve heard on the forum step count are spot on the S5 watch, hopefully that can come to S9P via firmware

Stavros _Clinica Orl

@efejota thanks for the input; in my case , after many years of training and walking I know that my inter-walking distance is approximately 80 cm, so for a 10 KM session I would expect numbers above 10000 and not below (ie 9200). This is what I have noticed in multiple occasions. As I said , step count is not important as the distance, but the price tag of the watch dictates that certain errors should not be happening. I will organize another 10 K session this week with the newest firmware update (2.19.42) to check upon steps and calories consumed, and I will post the data.

DMytro

@stavros-_clinica-orl well, what if suunto invests into different directions i.e. navigation and has all ‘non-important’ functions either absent or at the bare minimum, just so that people who buy suunto for its core features have a few toy metrics, ‘nice-to-haves’. And other people, who find such metrics important, buy something else instead.

I don’t say that it’s so or that it’s a good way to go around things, just offering you a possible explanation.

Egika

@dmytro There has been substantial effort to get the step count on left hand in S9P correct. So I would not say that the basics don’t care.

DMytro

@egika then I’m looking forward to the update, where those efforts come to fruition:)
Still wouldn’t matter to me though, but I’m happy for others.

Stavros _Clinica Orl

Dear members of the FORUM, as promised I did some additional testing in the walking modality , using the latest firmware version. The collected data are suggesting major computation flaws in the GPS assisted walling session. The test was composed on estimating various parameters on two 5 KM routes (termed R1 , R2) on flat terrain (to eliminate GPS signal losses). The speed on both sessions was approximately 10 min/KM (slower speeds are more easily controlled). I used the Suunto 9 in two modalities ( one per session) in a simple step calculation or as a GPS assisted activity. The Walkmeter app was running in IOS 15.3 in an Iphone 11 Max pro.

Steps 9SP (R1) 4913 —

Steps 9SP+ GPS (R2) – 4438

Steps Walkmeter (R1, R2) 6076 6100

Calories 9SP 260 ----

Calories 9SP + GPS (R2) ---- 398 (453 TOTAL)

Calories Walkmeter (R1,R2) 278 275

As you can see the GPS activity introduces significant errors in terms of step and calories estimation.

I hope that the administrators can signal the case and forward the data to the appropriate SUUNTO technical team.

Stavros

vcebrian

Hi @stavros-_clinica-orl ,are you wearing the watch on the left or right hand? I had the same problem and I think it’s a problem with the calculation algorithm when you wear the watch in the left hand.

Two examples:

Slow walk of 2,9km:
- Suunto Spartan Sport (right hand): 3312 steps. (87cm/step)
- Suunto 9 Peak (left hand): 2264 steps. (128cm/step) Impossible!!
Fast walk of 2,3 km:
- Suunto Spartan Sport (left hand): 2343 steps. (98 cm/step)
- Suunto 9 Peak (right hand): 2246 steps. (102 cm/step) Both possible!!

I have more trainings behaving in the same way, so please try it out.

Also, if you analize the Cadence graphic in the left hand S9P behaves erratic some times, but in right hand things are OK.

vcebrian

isazi

@vcebrian a fix for left hand is coming.

vcebrian

@isazi Is there any date scheduled?

isazi

@vcebrian can’t say, but usually there’s a software release every quarter.

Stavros _Clinica Orl

I have been asked by the Italian customer service to report my findings on the error issues related to the Walking activities, using the watch on the Left and the Right hand (I am right-handed). I tested the 9SP in 2 ten KM sessions, with the WalkMeter App in GPS mode in my Iphone 11 Pro . In the second one I even used another cheaper Wearable to see differences in the performance. Details on the testing have been provided in my previous post.
I was surprised to see the differences between L and R tests… even in the GPS assisted activities. 9SP shows erroneous estimations in the step count, distances (both wearables overestimated the distance by 80 m) and calories consumption. The Left hand measurements are the worst.
Of course these are samples of the observed problems, numerous repetitions are needed to define the margins of these errors.

image url)

isazi

@stavros-_clinica-orl left hand fix is coming with next update.

Egika

@isazi said in Errors in the Walking Activity:

@stavros-_clinica-orl left hand fix is coming with next update.

Yeah, this has now been posted like 5 times in this thread. From top to bottom.
Seems like nobody listens or cares…
I will stop repeating it.