Hi, do you have reference to the Chebyshev approximation you do?
maybe different approach is to get error for each degree and then do the cutoff
mlazaric
