MASH ( SD modulator) power spectrum dc term

>> Se=(1/12)/(fs/2);
>> Sq=Se*(2*sin(pi*f/fs))^6;

Click to expand...

On the book that you suggested, i also saw that formula for the ideal curve. But are you sure it is correct?
According to this , the curve is dependant on the sampling frequency, which as far as i understand it shouldn't be.

Why do you use fs=1GHz ?

I can not believe such high frequency clock for Delta-Sigma-Fractional-N Frequency Synthesizer.

Click to expand...

As i wrote above , it shouldn't matter neither on the measured or ideal curve.
On all the formulas, like 2*sin(pi*f/fs), the discrete frequencies are normalized to sampling frequency. So , a change in sampling frequency would also change the discrete frequencies accordingly resulting in a constant ratio f/fs (no matter the fs).

Anyway, i tried with a more realistic sampling frequency ( 40M ) and still don't get near the ideal curve.
Could you also tell me where did you find the formula using " -10*log(binWidth) " when using the fft method ?

Thank you

You are misunderdtanding.
Se and Sq have to be dependent on sampling frequency.

Surely learn quantization noise.
Can you understand a concept of over sampling ?

Unless you can not understand such easy thing, you can never go on next.

NikosTS said:

Could you also tell me where did you find the formula using " -10*log(binWidth) " when using the fft method ?

Click to expand...

Not formula at all.
It is no more than kindergarden level common sense of Fourier analysis.
Surely read Schreier book.

NikosTS said:

On the book that you suggested, i also saw that formula for the ideal curve.
But are you sure it is correct?

Click to expand...

Of course, correct.

NikosTS said:

According to this , the curve is dependant on the sampling frequency,

Click to expand...

It is very natural.
Can you understand a concept of Over Samplng ?

NikosTS said:

which as far as i understand it shouldn't be.

Click to expand...

Simply, only you are misunderstanding.

NikosTS said:

As i wrote above , it shouldn't matter neither on the measured or ideal curve.

Click to expand...

Wrong.
This is also your misunderstanding.

NikosTS said:

On all the formulas, like 2*sin(pi*f/fs), the discrete frequencies are normalized to sampling frequency.

Click to expand...

This is for frequency domain analysis of NTF(Noise Transfer Function).

NikosTS said:

So , a change in sampling frequency would also change the discrete frequencies accordingly resulting in a constant ratio f/fs (no matter the fs).

Click to expand...

Again wrong.
You are doing time domain simulation.

Surely see your

Spectrum have large variation of -110 ~ -60 at f=fs/2.

You have too many many wrong thought and wrong knowledge.

Good luck you can meet someone who have same thought as you.