Related

970 WHR accuracy - I'm impressed

1339905
4 days ago

Hi all,

I had been pretty disappointed with the 965's WHR response & accuracy, which was one of the main reasons for me to try the 970. Now today I took both on a 5h trail run, with the 965 additionally recording from a Polar H10. I'll try to attach a graph where I plotted HR deltas of both watches in relation to the H10 and hope it'll work, these are the first 2.7h only, because then I made a break and forgot to pause the 965, so recording got out of sync with the 970. 970 is running 6.17, 965 23.20. Avg. HR was around 135, max. at around 160.

I have to say: WELL DONE, GARMIN!

Best,
r.

  • 0 2 days ago

    Todays run. Very similar after 2.5 km.

    More tests to come, other types of run and sports. It will, for example, be interesting to see the measure of OW. The 970 then under the wet suit sleeve, so really good contact.
    The higher HR in the beginning could be explained by bad contact, but i am not sure, It was wet before start. Lets see... Slight smile

  • 0 1 day ago in reply to hakanwidlund

    it is not caused by the bad contact, that is the typical issue of all garmin HRMs - sometimes it is just completely off for a long period...

  • 0 1 day ago in reply to pxl

    Okay, what is the problem then? If "all" HRMs have this problem then it should be a lot of info about it. Do the HRM 600 have the same issue?
    I have been using the Garmin since sept 2008, a number of different bands (also Polar). As I recall, I have not seen this problem before. Maybe some.

    Anyway, my target is to analyse further if the 970 WHR is good enough, Swim (mainly OW), Run and bike. Previous models have been bad or totally off for me. 225, 745XT, 935, 945, 955, 965 and a Fenix.

  • 0 11 hours ago in reply to hakanwidlund

    I mean all garmin wrist sensors, of course. chest straps are fine.

    the problem lays in the data "correction" algorithm - where there is lack of consistency in the data stream (and there is always inconsistency there, flawless data stream does not exists, so it is always a guesswork), the watch tries to figure out what is going on

    the actual issue is that the watch sticks the guesswork too strongly, too long and it does not believe the real data

    garmin tries to fine-tune this algorithm with FW updates, the issue is that if they fix the algorithm here, it will be broken there - so if it was problematic for someone, but good for other, it could be reversed after a FW update

    that's why after every update there are many posts stating that now it is perfect, others saying, it just broke the wrist HR for them

    could it be fixed ever? probably yes, but then the precision of the data will be worse, so it will act badly in reviews / tests

    as a consequence they might decided to keep the algorithm that is very accurate sometimes, and when it is not, well, it is off completely

    that's why fine tuning is impossible, they are trying to catch a ghost - WHR sensors are not accurate by design, so they report inaccurate data, and that's all

    and we could be much better with an algorithm that is 5% off all the time, rather than one is 99% accurate, well most of the time

    I am - of course - just trying to reverse engineer they mindset and process, but so far everything moves into this direction

  • 0 7 hours ago in reply to pxl

    As shown in the graph, the unexpected higher HR in the first 2.5 k was from the chest strap. The rest wrist/chest where basically the same. Looks hopeful, let's see.