Accelerometer

OK,

I don't think my question (https://www.electro-tech-online.com/threads/processing-accelerometer-data.116380/) was clear enough. So here it is rephrased.

I get 0-255 for X, Y and Z from the Chronos EZ430.

When I line up X and Y and keep the watch static, I get 80 for Z... therefore the value for gravity must be 80.

So I divide these values by 80 which gives me a total of 1 when I add up X, Y and Z.

I have created an windows application with a projection of a 3D object which I intend to represent the rotation of the watch on.

To work out the angles (degrees) for X, Y and Z to do this rotation, I do the following.

x = (Math.Asin(X) / Math.PI * 180)
y = (Math.Asin(Y) / Math.PI * 180)
z = (Math.Asin(Z) / Math.PI * 180)


The 3D object moves almost as I would expect but there are clear bugs in my working out of these angles.

The code for the 3D object is very much proven and I use it in many other bits of code. So what is wrong with the above please?

To determine the effect of gravity, orient any given axis upwards, then turn the device over, and divide the difference in readings by two. For example, if +Z reads 80, and -Z (180 degrees from first position) reads 60, the difference is 20, so 2g causes a Δ reading of 20 counts, so 1g is 10 counts.

MikeMl said:

To determine the effect of gravity, orient any given axis upwards, then turn the device over, and divide the difference in readings by two. For example, if +Z reads 80, and -Z (180 degrees from first position) reads 60, the difference is 20, so 2g causes a Δ reading of 20 counts, so 1g is 10 counts.

Click to expand...

I did this and got 76.5 which seems reasonable. When I rotate the object with the watch now it seems better.



The way the object moves now gives me the impression these sums are the issue. Ummmmm

beakie said:

OK,

To work out the angles (degrees) for X, Y and Z to do this rotation, I do the following.

x = (Math.Asin(X) / Math.PI * 180)
y = (Math.Asin(Y) / Math.PI * 180)
z = (Math.Asin(Z) / Math.PI * 180)

Click to expand...

beakie said:

The way the object moves now gives me the impression these sums are the issue. Ummmmm

Click to expand...

I am convinced that bit is wrong now. This is the graph that I get when I rotate the watch when its laying face side down. No change on X, Y or Z at all!



Clearly I have misunderstood what I need to do to get a rotation vector from the sampled data.

beakie said:

Clearly I have misunderstood what I need to do to get a rotation vector from the sampled data.

Click to expand...

I realise now that because gravity is only acting on one axis, I will only see 1 on that axis and 0 on the rest... so what do I need to do to get the tilt?