Question About Simulink Simulation of Delta-Sigma Modulators

[naalald]

delta sigma modulator

Hi All,
The attached are the simulink model of a 1st order delta-sigma modulator and its result. But when I compare the output fft and SNR of my model with that of Schreier toolbox for a 1st order modulator with OSR=32 and for the same inputs, I see that the results are different. I have 28.5 dB SND and Schreier toolbox has 41.6 dB SNR.
Could you please help me figure out where is my mistake and why I get this poor SNR?
Thanks a lot.

 

[jiangxb]

simulink delta sigma model

why the threshold of quantizer is 0.5? i think it should be 0.

 

[naalald]

delta sigma modulation simulink

Hi jiangxb,
Thank you very much for your reply. That is because the output levels of the Quantier are 0 and +1 and I have to select +0.5 for the threshold. In fact In Simulink there is no flip-flop that its outputs are -1 and +1. Do you think this does not work?

 

[jiangxb]

sigma delta modulator using simulink

i think the cause is the setup of quantizer threshold. i suggest that setting the common-mode value 0 and using sign function as quantizer.

 

[naalald]

delta sigma model simulink

Thanks for your reply.
But if I use the sign function for the quantizer, how can I simulate the sampling? How can I set the oversampling ratio? There would be no clock in the circuit then!

 

[jiangxb]

delta modulation simulink flip flop

the modules in simulink have option - sampling time, it's the inverse of sampling frequency. set input sinewave frequency of modulator in signal bandwidth. you needn't set oversampling rate which is used in decimation filter only.

 

[naalald]

about simulink

Thanks jiangxb,
You were right. Now that I have changed the quantizer and used a sign function, the result is much better. But as I can't define the clock concept well, the result still has to improve (I defined a function instead of the quantizer as below). By changing the oversampling ration a little, the SNR changes a lot.

The function that I defined for the quantizer:

function y = fcn(u, clk)

if clk
if u<0
y = -1;
else
y = 1;
end
else
y = 0;
end

The attached is the new results which are much better than the previous one.

 

[jiangxb]

delta function in simulink

you needn't set clock because they are discrete-time themselves. set SAMPLE TIME of one module and -1 for others(-1 for inherited).

 

[naalald]

sigma delta simulink model

Thanks.
I used a sign block instead of defining a function and also I eliminated the clock concept and used the sample time of the source block. The results are better now.
Thank you very much for your help!

 

[gaboor]

simulink sample question

Hi All,
I find this topic very interesting as i am trying to build a model for 1st and 2nd order delta sigma modulation. I just started using Simulink and i wonder if anyone can help me.
Thanks