Huge power discrepancy between Garmin Strava (Garmin set to include zeros)

BenSalibra
1 day ago

I have a new power meter (Stages R8100 dual sided) that seems to be causing significant power data discrepancies between Garmin & Strava.

Garmin Data: 229W

Strava Data: 190W

That's a 39W difference. The data is basically useless if it isn't accurate. Trainerroad is following the Strava data and therefore my training data is very wrong. My Garmin is set to include zeros and can only record in 1sec increments.

Has anyone come across this?

File for reference:

https://connect.garmin.com/modern/activity/20862633269

  • 0 1 day ago

    can you send me somehow the FIT file ? (i dont have access to your original FIT file via GC, you can download from there, "save original" under the wheel icon) file.io free upload for example, or similar file sharing. 

    honestly your hr data is not good imho, too many "freezings" for many seconds so something is not ok with this solar's connections (as i know that is one of the most problematic model unfortunately from the x40 series). did you try to delete all the paired sensors, turn the edge completely off , then on, and re-pair ? in advance: paired via ANT+ everything or BT sensor pairing? 

    and if there are hr freezings, then that is a sign that the conn. is interrupted between sensors and Edge, but the HR is holding/filling these gaps with the last known value for many seconds in the FIT file and onscreen also, the power meter is not, that is putting a 0 watt (and cad) into the file and screen immediately, but the Edge is knowing it that is coming from comm issue (it is some kind of inbuilt error correction or what), and calculating internally with the last known value for 2-3secs for sure , in the summary avgs also, but in the FIT file there are 0s of course , and the 3rd party apps like Strava doesnt know about this conn issue and a 0 is a 0 for them and calculating the avg this way (and this is why it is significantly less despites the 0 includes in power setting )

    so we should find out what can cause this conn chip freezings.

  • 0 22 hours ago in reply to Luko79

    You should be able to download it here: 

    https://drive.google.com/file/d/1Kbsmm8HLIFnf7JHlBxg1Bk7gvsyOA6NS/view?usp=share_link

    Now that you say that, I did notice my HR monitor going in and out of connection throughout the race. I hadn't thought much about it. This 840 has always been a complete PITA. Sometimes, when I get done with a ride it beeps at me telling me things are connecting/disconnecting every few seconds. I've done a factory reset a few times but it still misbehaves.

    I will try to unpair everything. I have not done that recently. Everything is connected via ANT+. In this race in particular, it's possible the edge got close enough to the car to connect to my phone at a certain point in the race and perhaps that caused the dropouts? I usually ride with a lot of connections/sensors. PM, HR, Speed, Di2, Phone w/ Livetrack.

    Thanks for the help!

  • 0 16 hours ago in reply to BenSalibra

    Something is impacting the radio reception rate. The power meter is low at around 83%. Anything below 90% power meter will impact the data. The HRM and speed senser are also low. Do you have a camera or something mounted close to the Edge.

    This explains the power difference. The Edge records one second power. It also records accumulated power as transmitted by the power meter. If data is lost due to a low RX rate then the sum of the 1 seconds values does not match the accumulated power values.

    The Edge uses the accumulated power values to compute average power, while Strava is using the sum of the 1 second records. 

  • 0 11 hours ago in reply to aweatherall

    No cameras or anything just the Edge. I didn't even have my phone with me. I unpaired all sensors and repaired them for my ride today. I didn't notice any dropouts on my ride today. There was still a few watts. I wonder if the many other riders with sensors/garmins caused interference at the race.

    Today's ride:

    3:21:41

    Garmin
    185w Avg
    206w NP
    152 BPM Avg

    Strava
    183w Average
    201w Weighted Avg
    152 BPM Avg

    Seems like there were still some dropouts but less frequently. It was also a more steady state effort than the stop start nature of CX. I wonder if that has anything to do with it? Does Stages record power input from partial revolutions only in the accumulated power file and not in the individual? I've read that some crank based meters require a full pedal stroke to record power for any given second.

    drive.google.com/.../view

  • 0 7 hours ago in reply to BenSalibra

    Hi Benjamin, 

    the 2nd file with solo ride is much better, ~95% reception rates with sensors. that is true that if you started in a big bunch and there were bikes with certain type of powermeters (like Rotor ) that can kill the other signals , especially if that single person is near to you all the time (front of you or behind you). yes, that is true also that for CX or MTB rides these crank-based powermeters are not the best choice, the best one is the spider-based ones with IAV algorhytm like Power2Max NG, these partial and not steady turns on crank are very hard task and if the powermeter has AAV calculation method , the pm is measuring much less and more inaccurate data then the reality (but in your case the lost data packages were the main problem here, the question is that the measurement from Stages is enough accurate or not during a CX race is a different story)

    sometimes the different placement of the Edge can cause signal drop also, so if it is on different place during a race than on trainings (race= stem which is alloy , training= front mount) can be a source of the problem (but if it is always on the same place, front mount, then the problem of the bunch what i described above can be the culprit, besides any Edge's sensor pool issue what the re-pairing is /probably/ solved.

    this tiny difference in 2nd file can come from the different moving time (and probably rounding for an integer is +-1W diff)  , there is a small mismatch (the Strava's threshold value for moving/stopping is a bit different than the Garmin's one) between the 2 values and the higher moving time is causing lower avgpower ofc (if you dont stop , these 2 avgpower values must be equal if the signal strength are perfect and there are no too much rx/tx fails between pm and Edge like in the 1st file)