Measuring roll angle using accelerometer

[dan7]

I'm using one axis of accelerometer to measure roll angle of a RC plane. But, it's reading the acceleration at the same time. Anyone can help me filtering this out or is there any algorithm to solve this? I'm programming PIC microcontroller using PicBasic Pro.

I know 2-axis accelerometer would give a better accuracy. For now I'm trying out using only one axis. Any help would be greatly appreciated.

 

[keith1200rs]

Presumably you have mounted it vertically and in line with the aircraft fuselage? However, when the plane is banked I would have thought the centripetal force would combine with the gravitational force meaning you won't get the bank angle.
It should be immune to acceleration forces I would have thought, but I don't really see how it would give roll angle, but I guess I am missing something.

Keith.

 

[dan7]

I mount it horizontally..meaning to say it's lying on a surface and in line with the aircraft fuselage. This is what I'm referring to:

**broken link removed**

The output voltage from the X axis accelerometer relates to the tilt angle which in my case is the roll angle. Centripetal force combines with the gravitational force to give us a tilt angle. Mind explaining why it's immune to the acceleration?

 

[keith1200rs]

I would have thought that the acceleration due to the aircraft's engine would be in an axis which the accelerometer wouldn't be sensitive to? It would still be sensitive to vertical movement of the aircraft though - maybe that is what you mean when you talk about acceleration - in the vertical direction?

Keith.

 

[FvM]

Following a usual axis definition https://en.wikipedia.org/wiki/Flight_dynamics

Inertial axes (or body axes) - based about aircraft CG
X axis - positive forward, through nose of aircraft
Y axis - positive to right of X axis, perpendicular to X axis
Z axis - positive downwards, perpendicular to X-Y plane

a roll sensor would be sensistive along the Y axis and not respond to acceleration in flight direction, but to the lateral acceleration during bank.

In my opinion, the only (principle) way to get the true roll angle independant of lateral accelerations is to add a gyroscope sensor.

 

[dan7]

yes..for my sensor the Y axis is positive forwards, through nose of aircraft and X axis is perpendicular to Y axis. I assume when the aircraft is turning there are 2 forces acting in X and Y direction that's why I'm getting a reading of acceleration in the axis of roll (Y axis, for my case) as well. Acceleration in vertical direction would be detected by Z axis, no?

I have gyroscope together with the accelerometer but I still can't figure out how to program the gyroscope. There's a bias involved and we need to compensate for that. Could you help me with the algorithm?