I have been thinking about this problem and think that I've got a suggestion now, but I'm struggling with the implementation. I'll enclose the suggested patch anyway....and hopefully someone will have an idea how to fix it.
The equations I suggest for four signal amplitudes whose squares add up to 1 comes from a geometric idea. If we have a circle of radius 1/2, and center coordinates x1,y1 within a circle (radius 1/2) centered at the origin, then the squares of the locations of intercepts always add up to 1.
So, I tried to implement this, in a few steps:
limit it to 1 2. divide the radius by 2, and convert back to cartesian coordinates 3. use the following equations for signal amplitudes s1=y1+sqrt( 0.25 - x1^2) s2=sqrt( 0.25 - x1^2) - y1 s3=x1+sqrt( 0.25 - y1^2) s4=sqrt( 0.25 - y1^2) - x1
s1 and s2 correspond to the speakers opposite from one another, same with s3, s4
what a mess....
Chuck