Aug 16, 2009, 02:49 AM
Joined Oct 2007
517 Posts
Quote:
 Originally Posted by RoyLB 50ms should be fine. I'm already behind my own data analysis. Brij sent me gyro data (non-flying) from his board a few days ago and I haven't had time to look at it yet.
Hey guys, here's some data collected from my AeroQuad (the flying brick one) at a sample rate of 50ms. I took data under the different conditions that RoyLB requested. The filenames (and comments within the Excel files) state what the test condition was.

I've been meaning to state this whenever we had talked about programmign special conditions when handling the different flight angles when calculating them with either a complementary filter, atan2 or Kalman filter. If you look at the recorded flight angles (with RoyLB's marvelous complementary filter) you'll notice that for a typical bank (roll + yaw) can be made with less than 12 degrees, and to make the quad move forward you typically just need to pitch forward around 5-10 degrees.

So in practicality it would be rare to need to handle flight angles (in my opinion) > 45 degrees. If you are at those angles you better be doing intentional acrobatics, or your quad is probably doing something you didn't intend.

Also, I was flying over grass, so some of the starting angles for the data may not show close to zero for both axes (probably +/- 2 degrees). The stationary data was done on cement.

Not sure if this is a thing that needs to be fixed in the complementary filter, but if I land hard and look at the ending reported angles, sometimes it'll read around 5 degrees afterwards (but I did land on grass... still sounds a bit big). I can post that data too if interested.

RoyLB, I did my best to climb in altitude in a level manner, but it gets kinda hard when the pilot is viewing the quad at about a 70 degree angle as it's going up.

Thanks for looking at this data guys, hopefully it'll put to rest the stated concerns. I've learned leaps and bounds about control theory, quad construction, basic microcontroller programming and R/C electronics (ESC and motor selection). I think we are all very passionate about what we do, so let's all continue to discuss and agree to disagree in a professional manner as you guys typically do.

It was dangerous for me to be gone so long... I didn't realize how time consuming it was to read all these posts... sigh, now to catch up reading them with y'all.