Costas Loop oscillates in ModelSim while locks in MATLAB!

here is copy:

%NCO bit true model
clear all; close all;
%example NCO parameters
n = 20000;                               %number of test samples
A = 2^15 - 1;                            %max amplitude, amplitude resolution 16 bit signed
M = 2^32;                                %NCO accumulator phase resolution
inc = round(M*.001);                     %NCO accumulator phase increment 
k = 2^12;                                %lut phase resolution (lut size)
lsb = log2(M) - log2(k);                 %LSBs discarded when addressing
lut = round(A*exp(j*2*pi*(0:k-1)/k));    %lut, one cycle sine data
ptr = 0; 
addr = 0;
for i = 1:n     
   y(i) = lut(addr+1);                   %add 1 for matlab LUT    
   ptr = mod(ptr + inc, M);              %phase accumulator(ramp)    
   addr = round(ptr/2^lsb);              %discard LSBs    
   addr(addr >= k) = addr - k;           %check address overflow
end

No I don't want to revert to matlab... as you said, modelsim has proper loop and delays. I am gonna focus on it. I was just making a comparison in behaviour.... But is my analysis right?

That is good path since modelling in matlab for feedback loops needs to be dead accurate match of rtl. Yes your analysis is right if you mean your matlab model which I haven't seen its final format. The nco statement has control over phase just like rtl and this phase is used to control frequency by different jumps over sin/cos functions (luts) so it will have very high floating resolution. but you need to write as y = exp(j*2*pi*linspace(0+offset:N-1+offset,N)*.1); in order to step phase in fractions.

Come on Mr Kazem! The model that we spent days discussing it! We finally said yuppie remember?! Anyway here its is:

% Siraj Muhammad
% 25/3/2015
% BPSK Demodulator
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
load RRC.mat
fc = 0.05000001;
phase_offset = pi/7;
N = length(rf_signal);
bb = zeros(1, N);
bb_f = zeros(1, N);
I_f = zeros(1, N);
Q_f = zeros(1, N);
I_r = zeros(1, N);
Q_r = zeros(1, N);
error = zeros(1, N);
PhErr = zeros(1, N);
error_integral = zeros(1, N);
% Loop Filter Coefficients
% BW = 200;  % Hz
% loop_theta = 2*pi*BW;
% C1 = 4*(loop_theta)^2/(1+sqrt(2)*loop_theta+loop_theta^2);
% C2 = 2*sqrt(2)*loop_theta/(1+sqrt(2)*loop_theta+loop_theta^2);
C1 = 2^2;
C2 = 1/2^8;
for i = 2:N
    % Downconverting to Baseband
    bb(i) = rf_signal(i).*exp(j*2*pi*fc*i+j*phase_offset).*exp(-j*PhErr(i-1));
    % Filtering
    bb_f = filter(RRC, bb(1:i));
    I_f = real(bb_f);
    Q_f = imag(bb_f);
    
    % Error
    error(i) = I_f(i).*Q_f(i);
    
    % Loop Filter
    error_integral(i) = error(i).*C1 + error_integral(i-1);
    PhErr(i) = error(i).*C2 + error_integral(i);
end
figure; subplot 321; plot(real(bb)); title('I Channel')
subplot 322; plot(imag(bb)); title('Q Channel')
subplot 323; plot(I_f); title('I Channel Filter')
subplot 324; plot(Q_f); title('Q Channel Filter')
subplot 325; plot(error); title('Phase Error');
subplot 326; plot(PhErr); title('Loop Filter');

The above code should prove the concept.

I suggest you try:

I am doing some tests... I will do yours as well and let you know.

Hello Mr Kazem,

when I suggested testing the tx/rx chain with the feedback loop deactivated I wanted a sanity check of your chain. Ideally you should have testbench for every module by reading reference outputs from matlab into testbench and check for bit true comparison. That is what we do in the industry. Moreover you need to scale the outputs of each module to some practical resolution e.g 16 bits signed yet I see your dynamic range too high.