Keep it Level

[misiu02]

Good Afternoon,

My first post, it looks like there is alot of good information here. I have one quick question though.... I have the majority of the issues solved (theoreticly anyway) for my project. But i dont seem to be able to find good, inexpensive peice of hardware that would low me to gather tilt information.

My project is basicly running off of a small computer and I need something that would allow a constant imput on how far off from nutral its tilting.


Any help would be appreciated.

[zlite]

FMA Co-Pilot (http://fmadirect.com/products.htm?cat=29&nid=7) (IR sensor).

What sounds simple--"Where's down?"--is not, especially in a vehicle moving in three dimensions. Anything involving gyros and accelerometers is very complicated (Google "Kalman Filter"). Absolute frames of reference (IR sensors look at the gradient between sky and earth) is the way to go if you want cheap and simple.

[misiu02]

I hadnt thought about using IR sensors. My initial thought was to use one of these... Phidgets Encoder (http://www.trossenrobotics.com/store/p/3235-Phidget-Encoder.aspx) suspend it somehow and then place weight on one end to keep it pointed "DOWN" (see horrable drawing attached)

When the plane tilts the weight will always point down but the sensor should rotate and I could base the "level" or tilt of the plane off that value. one for for the left right tilt/axis and one for the forward/backwards axis.

Ill try and make a real diagram of what im trying to do when im not using my cell phone for internet access.

[zlite]

Sigh. Here we go again (we should have a sticky link called "Autopilot Physics 101"!).

When the plane tilts the weight will always point down but the sensor should rotate and I could base the "level" or tilt of the plane off that value.

No it won't. If the plane is in a turn, as it probably will be when it's tilted, centrifugal forces will "pull" the weight away from down. You learned all this in high school!

[dmgoedde]

Sigh. Here we go again (we should have a sticky link called "Autopilot Physics 101"!).



No it won't. If the plane is in a turn, as it probably will be when it's tilted, centrifugal forces will "pull" the weight away from down. You learned all this in high school!

In case there is any doubt in what zlite wrote, I concur 100%.

Actually, the only way a weight *might* work is if it were NOT a pendulum but rather was perfectly balanced and massive with a large radius of gyration and on a perfectly frictionless shaft with axis lying on a line between nose and tail. In this case, as the plane banked the weight should NOT rotate within its own reference frame while the plane does rotate, and in principle you could measure this with some encoder and then have an idea of plane's new position. I have never heard of this being done, but it brings to mind a mechanical gyro that I had on a 30 sized helicopter circa 1995.

Dean

[Jack Crossfire]

If you assume roll is proportional to yaw, you can put a yaw gyro between aileron servo & receiver. That should level it when the servo is neutral, possibly eliminating the need for roll sensing.