FPGA Central

I implemented the matlab simulation model according to hardware
architecture.The impulse response is ok when the impulse length is
short.But when I increase the length of impulse stimulus, I found
increasing spurs. I don't know whether this is caused by the
instability of the filter or some other reasons. Attached is the
matlab code,and at the end of code I added the matlab internal
function mfilt.cicdecim for comparison.


%Filter Stimulus
%data_in =sin((1:2^15)*pi/2^10);
data_in =[1 zeros(1,4*6000-1)];

%Filter Parameters
R= 4;%decimation factor
M=1;%differential delay
N=5;%cascaded numbers of section

%Integrator Section Registers Initialization
intg_reg1(1) = data_in(1);
intg_reg1_d(1) =0;
intg_reg2(1) = 0;
intg_reg2_d(1) =0;
intg_reg3(1) = 0;
intg_reg3_d(1) =0;
intg_reg4(1) = 0;
intg_reg4_d(1) =0;
intg_reg5(1) = 0;
intg_reg5_d(1) =0;



for i=2:length(data_in)

intg_reg1_d(i) = intg_reg1(i-1);
intg_reg1(i) = intg_reg1_d(i) + data_in(i);

intg_reg2_d(i) = intg_reg2(i-1);
intg_reg2(i) = intg_reg2_d(i) + intg_reg1_d(i);

intg_reg3_d(i) = intg_reg3(i-1);
intg_reg3(i) = intg_reg3_d(i) + intg_reg2_d(i);

intg_reg4_d(i) = intg_reg4(i-1);
intg_reg4(i) = intg_reg4_d(i) + intg_reg3_d(i);

intg_reg5_d(i) = intg_reg5(i-1);
intg_reg5(i) = intg_reg5_d(i) + intg_reg4_d(i);

end

% Decimation by R

deci_data=intg_reg5_d(R:R:end);

%Comb Section Registers Initialization
diff_reg1(1) = deci_data(1);
diff_reg1_d(1) = 0;
diff_reg2(1) = 0;
diff_reg2_d(1) =0;
diff_reg3(1) = 0;
diff_reg3_d(1) = 0;
diff_reg4(1) = 0;
diff_reg4_d(1) = 0;
diff_reg5(1) = 0;
diff_reg5_d(1) = 0;
diff_reg6(1) = 0;

for ii=2:length(deci_data)

diff_reg1(ii) = deci_data(ii);
diff_reg1_d(ii) = diff_reg1(ii-1);

diff_reg2(ii) = diff_reg1(ii)-diff_reg1_d(ii);
diff_reg2_d(ii) = diff_reg2(ii-1);

diff_reg3(ii) = diff_reg2(ii)-diff_reg2_d(ii);
diff_reg3_d(ii) = diff_reg3(ii-1);

diff_reg4(ii) = diff_reg3(ii)-diff_reg3_d(ii);
diff_reg4_d(ii) = diff_reg4(ii-1);

diff_reg5(ii) = diff_reg4(ii)-diff_reg4_d(ii);
diff_reg5_d(ii) = diff_reg5(ii-1);

diff_reg6(ii) = diff_reg5(ii)-diff_reg5_d(ii);
end
data_out = diff_reg6;

data_in =[1 zeros(1,4*10000-1)];

hm=mfilt.cicdecim(4,1,5);
cic_ir=double(filter(hm,data_in));

On Jun 27, 9:39*am, tonyjian...@hotmail.com wrote:
> I implemented the matlab simulation model according to hardware
> architecture.The impulse response is ok when the impulse length is
> short.But when I increase the length of impulse stimulus, I found
> increasing spurs. I don't know whether this is caused by the
> instability of the filter or some other reasons. Attached is the
> matlab code,and at the end of code I added the matlab internal
> function mfilt.cicdecim for comparison.
>
> %Filter Stimulus
> %data_in =sin((1:2^15)*pi/2^10);
> data_in =[1 zeros(1,4*6000-1)];
>
> %Filter Parameters
> R= 4;%decimation factor
> M=1;%differential delay
> N=5;%cascaded numbers of section
>
> %Integrator Section Registers Initialization
> intg_reg1(1) = data_in(1);
> intg_reg1_d(1) =0;
> intg_reg2(1) = 0;
> intg_reg2_d(1) =0;
> intg_reg3(1) = 0;
> intg_reg3_d(1) =0;
> intg_reg4(1) = 0;
> intg_reg4_d(1) =0;
> intg_reg5(1) = 0;
> intg_reg5_d(1) =0;
>
> for i=2:length(data_in)
>
> * * intg_reg1_d(i) = intg_reg1(i-1);
> * * intg_reg1(i) = intg_reg1_d(i) + data_in(i);
>
> * * intg_reg2_d(i) = intg_reg2(i-1);
> * * intg_reg2(i) = intg_reg2_d(i) + intg_reg1_d(i);
>
> * * intg_reg3_d(i) = intg_reg3(i-1);
> * * intg_reg3(i) = intg_reg3_d(i) + intg_reg2_d(i);
>
> * * intg_reg4_d(i) = intg_reg4(i-1);
> * * intg_reg4(i) = intg_reg4_d(i) + intg_reg3_d(i);
>
> * * intg_reg5_d(i) = intg_reg5(i-1);
> * * intg_reg5(i) = intg_reg5_d(i) + intg_reg4_d(i);
>
> end
>
> % Decimation by R
>
> deci_data=intg_reg5_d(R:R:end);
>
> %Comb Section Registers Initialization
> diff_reg1(1) = deci_data(1);
> diff_reg1_d(1) = 0;
> diff_reg2(1) = 0;
> diff_reg2_d(1) =0;
> diff_reg3(1) = 0;
> diff_reg3_d(1) = 0;
> diff_reg4(1) = 0;
> diff_reg4_d(1) = 0;
> diff_reg5(1) = 0;
> diff_reg5_d(1) = 0;
> diff_reg6(1) = 0;
>
> for ii=2:length(deci_data)
>
> * * diff_reg1(ii) = deci_data(ii);
> * * diff_reg1_d(ii) = diff_reg1(ii-1);
>
> * * diff_reg2(ii) = diff_reg1(ii)-diff_reg1_d(ii);
> * * diff_reg2_d(ii) = diff_reg2(ii-1);
>
> * * diff_reg3(ii) = diff_reg2(ii)-diff_reg2_d(ii);
> * * diff_reg3_d(ii) = diff_reg3(ii-1);
>
> * * diff_reg4(ii) = diff_reg3(ii)-diff_reg3_d(ii);
> * * diff_reg4_d(ii) = diff_reg4(ii-1);
>
> * * diff_reg5(ii) = diff_reg4(ii)-diff_reg4_d(ii);
> * * diff_reg5_d(ii) = diff_reg5(ii-1);
>
> * * diff_reg6(ii) = diff_reg5(ii)-diff_reg5_d(ii);
> end
> data_out = diff_reg6;
>
> data_in =[1 zeros(1,4*10000-1)];
>
> hm=mfilt.cicdecim(4,1,5);
> cic_ir=double(filter(hm,data_in));

I got it! In fact double type in matlab is represented with
54(including 1 sign bit) ,so the so-call floating point simulation is
fixed-point theoritically. Thus overflow is inevitable with the
increase of integration length."Wrap-around" is needed according to
Hogenauer's paper in order to avoid overflow error.

Hi,

I found this thread while doing some matlab simulation
on CIC filters and have a question to your matlab comb section:
Is the comb register initialisation part correct?
If I'm looking at the loop after the initialisation I would
expect that the register initialisation should be for
all X=1..6 diff_regX(1)= deci_data(1);

My implementation is different and takes only two length 5 vectors
for the delay taps. I took your implementation and compared the
results and they differ unless I do modify your initialisation part
as above.