View Single Post
  #13   Spotlight this post!  
Unread 22-02-2010, 17:11
jhersh jhersh is offline
National Instruments
AKA: Joe Hershberger
FRC #2468 (Appreciate)
Team Role: Mentor
 
Join Date: May 2008
Rookie Year: 1997
Location: Austin, TX
Posts: 1,006
jhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond reputejhersh has a reputation beyond repute
Re: Is there an easy way to download all the source code for WPILib C++ v4.3 at once?

Quote:
Originally Posted by mikets View Post
Hmm, shouldn't integration smooth out noise? Noise is supposed to be superimposing on the actual value (i.e. it will be above at one point and below at another). So both averaging and integration should cancel out noise in the long run, shouldn't it?
Yes, integration (and averaging... accumulation is averaging without the divide) smooths out noise. Deadbanding does not, though... what you should prefer to do if you are getting drift is to improve the calibration for your 0G offset. You could possibly implement a dynamic 0G point based on other sensor data, such as all wheels are stopped so set the 0G point to the current value.

-Joe
Reply With Quote