Acknowledged
1 month ago
CIQQA-3335

Bad accelerometer data captured by FR55 after latest update

I develop CIQ apps for tracking racket sports ( https://apps.garmin.com/developer/6a6d6521-4f13-470f-aa51-bc8d4ad0886b/apps ). Core functionality of my apps is capturing and recognizing strokes based on accelerometer signal.

Recently I got couple of reports from Fr55 users that on their devices no strokes are being detected. Looks like latest update introduced some issue with accelerometer data aquisition. I'm pending users to provide me some raw data but I'm wondering if anyone has similiar issue?

SzymonP
Parents
  • 2 days ago

    Another test was done (by @flocsy) using FR55 (FW11.03) and FR965 by recording 30sec activities - he was trying to perform similiar hand movements for both recordings.

    FR965:

    FR55:


    Raw data demonstrates similiar behavior to reported earlier:

    Every 25 samples (1 second) the signal repeats but changes every 100 samples (4sec). FR55 signal, which was recorded using similiar hand movements, demonstrates some similarities so it is not 100% fake - for FR965 you can clearily see 4 hand strokes at second half of recording and for FR55 you can also see that 4 high signal patterns show up. @flocsy suggestion was that the the sensor records data for about a second, and then it sends the same buffer 4 times. Like it needs 3 seconds to reset the sensor because of some buffer overflow issue.

Comment
  • 2 days ago

    Another test was done (by @flocsy) using FR55 (FW11.03) and FR965 by recording 30sec activities - he was trying to perform similiar hand movements for both recordings.

    FR965:

    FR55:


    Raw data demonstrates similiar behavior to reported earlier:

    Every 25 samples (1 second) the signal repeats but changes every 100 samples (4sec). FR55 signal, which was recorded using similiar hand movements, demonstrates some similarities so it is not 100% fake - for FR965 you can clearily see 4 hand strokes at second half of recording and for FR55 you can also see that 4 high signal patterns show up. @flocsy suggestion was that the the sensor records data for about a second, and then it sends the same buffer 4 times. Like it needs 3 seconds to reset the sensor because of some buffer overflow issue.

Children