Does one need to normalize nonlinear inequality constraints in fmincon?

Hi Everyone

I am using fmincon with the 'sqp' algorithm to solve an optimal trajectory. I initially tried 'interior-point' to no avail. My input profile/trajectory is discretised to 60 points. Thus, I have 60 variables for my optimization to run.

I am struggling to converge to a good solution and often get this message during my GlobalSearch run:

    Warning: Matrix is close to singular or badly scaled. Results may be inaccurate. RCOND =  3.010150e-19. 
> In C:\MATLAB\R2012a Student\toolbox\optim\optim\private\stopTestSQP.p>stopTestSQP at 204

I have normalised and centered by variables but should I be normalising my constraints as well? eg. in my case my angles must be less than 90degree. Should I normalise and make them less than 1?

Or does this error relate to something else?

Thanks in advance, Amir