Accelerometer N00b questions

Hey guys - I'm one week into playing with MetaWear and would love some help with explanations of a few questions as someone who is totally new to accelerometers...
  1. For scale, does -2/+2 mean less sensitive movement measurements than -8/+8?
  2. For sample frequency does 50 in the MetaWear app mean 50 hz as in 50 samples of accelerometer data per second?
  3. What is the High Pass Filter, when is it appropriate, and is it available via API or just in MetaWear app?
  4. What is the Low Noise Mode, when is it appropriate, and is it available via API or just in MetaWear app?
  5. What is the Auto Sleep, when is it appropriate, and is it available via API or just in MetaWear app?
  6. What is the benefit of using Root Mean Square data available via the API?

Comments

  • Quick answers:
    1. For scale, does -2/+2 mean less sensitive movement measurements than -8/+8? This scale means that the accelerometer is either measuring a max/min value of 2Gs, 4Gs, or 8Gs. Which is a unit of force, see more information here: http://en.wikipedia.org/wiki/Accelerometer. When you are measuring 8Gs instead of 2Gs it's not the sensitivity that changes but the resolution and the max/min amount of G force the accelerometer can detect.
    2. For sample frequency does 50 in the MetaWear app mean 50 hz as in 50 samples of accelerometer data per second? Yes that is correct.
    3. What is the High Pass Filter, when is it appropriate, and is it available via API or just in MetaWear app? We have a built in high pass filter as part of the API and we give an example of it in the MetaWear App. The most common use for this filter is to remove gravity from the reading. Here is more information about it: http://en.wikipedia.org/wiki/High-pass_filter
    4. What is the Low Noise Mode, when is it appropriate, and is it available via API or just in MetaWear app? The low noise mode is a sample averagerer. This means we are reducing noise by constantly averaging samples. It is part of the API and the MetaWear App serves as an example of it.
    5. What is the Auto Sleep, when is it appropriate, and is it available via API or just in MetaWear app? Auto sleep can be used to reduce power when there is not much motion, it causes the accelerometer to automatically go to sleep or slow itself down (use less power) when not in use.
    6. What is the benefit of using Root Mean Square data available via the API? A lot of algorithms use the RMS for calculations, for example our step counter for our fitness App uses RMS data.
  • Thank you Laura, y'all have been amazing!!!  

    Maybe a part of my problem is understanding how Logging and the other functions in the MetaWear app work.  I have been trying to track my motion during sleep and last night set frequency to 200 but when I emailed myself my log this morning the results seem to indicate otherwise (frequency 2?  pasted below). 
    Do the settings I select apply to both Streaming and Logging or no?  
    For the High Pass Filter/Low Noise Mode/Auto Sleep features, do the settings I select for each only apply if I switch the parent on?  
    Sorry I just have been getting strange results from Logging my sleep and trying to understand what is going on.  

    Scale = 0
    Freq = 2
    HPF On = 0
    HPF Cutoff = 0
    LowNoise On = 0
    Active Power Scheme = 0
    Auto Sleep On = 0
    SleepFreq = 2
    Sleep Power Scheme = 0

  • Correct, the switch must be set to on to enable the High Pass Filter/Low Noise Mode/Auto Sleep.  Note that you must change these settings before you begin streaming or logging (they won't take effect until you stop and restart the log or stream).

    As for the emailed log, those numbers listed are the raw enum values.  This was used just for internal enabling, but I'll update the app to make the log easier to understood.

    Quick note on logging raw accelerometer data, the MetaWear only has space for about 10,000 entries, and a single X, Y, and Z sample takes 2 entries, so you can only "Log" 10,000 (entries) / 2 (entries per sample) / 200 (samples per sec) = 25 second of raw data at 200Hz.  I would suggest you instead just stream data at 100Hz.


  • Thanks Stephen.  That answers my questions.  I was trying to track a whole night of sleep and only seeing 25 seconds in the log LOL.  Thought it looked weird... 
This discussion has been closed.