Please notice than when you instantiate it, Filtercoeff~ only outputs coefficients at the rate of the signal vector size to preserve cpu. To have more precision, you have to send it a ‘resamp (int)’ message to select a new rate.
Thus, with bigger vector sizes than the filter changes you are driving, filtercoeff~ will be less accurate than directly sending Max events every vector to the biquad~ (because the coefficients as list msg are getting smoothed during one vector, while the output of Filtercoeff~ will have 'stairs' of the size of the 'resamp' ) … filtercoeff~ will only become more accurate when you send it a “resamp 1″ (or a small number...)
This standard behaviour (when not using the resamp functionality) may seem paradoxical since you usually use filtercoeff~ when you want more precision.
(please list things that you believe to be errors or omissions from the existing refpage)