scale_coeffs.m

This is GMS down upper converter and down converter in simulink. you may understand the structure in

%*************************************************************************
% function scale_coeffs(h,num_phases)
% This function scales a set of filter coefficients h so that the filter
% will never overflow.  The maximum output of a FIR is the sum of the
% absolute values of the coefficients of any polyphase.  That is because
% the "worst case" condition occurs when the input data is -1 for each
% negative coeff and +1 for each positive coeff.  This function finds
% the maximum value of the sum of the absolute values of the coeffs of
% any phase and returns the scaled coeffs.
% Arguments:
%   h:          unscaled coeffs
%   num_phases: Number of polyphases.  For example, a filter that 
%               interpolates by three has three polyphases.  For
%               decimation filters, use num_phases=1.
%
% Kevin Neilson, Xilinx, April 2007
%*************************************************************************
function [ scaled_h gain_h ] = scale_coeffs(h,num_phases);
for k=1:num_phases
  max_phase_val = sum(abs(h([k:num_phases:end])));
end
max_val = max(max_phase_val);
scaled_h = h./max_val;
gain_h = 1/max_val;
disp(['Scaling coeffs by ',num2str(gain_h)])