The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> Tones in the spectrum of Sigma Delta Modulator
https://designers-guide.org/forum/YaBB.pl?num=1313592609

Message started by jefkat on Aug 17^th, 2011, 7:50am

Title: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Aug 17^th, 2011, 7:50am

Hi,
I am just simulating an SDM (3rd order) in Matlab. I am not adding any kind of non linearities at the moment yet I see there are some tones in the output spectrum. Do I need to add some noise ? I also added a noise source infront of qunatizer (using randn function of matlab) it helps somewhat but tones are not going away..Who's got a clue to what might be causing it. ?
Thanks for your time.

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by vivkr on Aug 18^th, 2011, 7:03am

jefkat wrote on Aug 17^th, 2011, 7:50am:

Hi,
I am just simulating an SDM (3rd order) in Matlab. I am not adding any kind of non linearities at the moment yet I see there are some tones in the output spectrum. Do I need to add some noise ? I also added a noise source infront of qunatizer (using randn function of matlab) it helps somewhat but tones are not going away..Who's got a clue to what might be causing it. ?
Thanks for your time.

Not adding any kind of nonlinearities? What about the quantizer?

Regards
Vivek

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Aug 19^th, 2011, 7:21am

are you working on a continuous time or switched cap DSM ?

I am guessing a switched cap DSM ?

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Aug 19^th, 2011, 9:46am

yes I am in SC- SDM, you think this has to do with the tones I am seeing? Could you please explain how ? . Quantizer is linear too.

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Aug 19^th, 2011, 9:56am

Here is what I see.(attachment)

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Aug 19^th, 2011, 2:57pm

Hi jefkat,

I see you have a 3rd order DSM, with an input tone @5kHz and a 3rd harmonic at ~15kHz.

This is pretty typical if you assume a non-linear model of your quantizer, the maths+explanation is more involved.

At this point, you'll have to do some reading I'm afraid ... some good texts are :-
Understanding DS Data Converters by Schreier and Temes.

good luck !

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Aug 20^th, 2011, 8:38am

Thanks for the reply Harpoon. I appriciate it. I have read the text you referred too. For me the problem is this : Look at the output spectrum of one bit modulators (many JSSC papers) there are no tones.So a quantizer (linear one) is not the source of non linearity (as some one alluded to in the posts above). I am using two bit , (perfectly linear) quantizer. My amplifiers are ideal so are the integrators . So where these tones are getting generated from.....

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Aug 20^th, 2011, 2:26pm

Hi jefkat,

In essence, a quantizer (no matter how perfect it is) is a non-linear block and will always cause non-linearity to the system, but less so for a 2 bit (4 level ?) quantizer I think.

If you read chapter 3.3 of the book, there is an attempt to explain the harmonics you may get in DS ADCs ...

May I ask if you are in academia or is this going to be a product ?

do let us know what you find ... :)

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Aug 21^st, 2011, 6:38am

Thanks again harpoon. Its a paradox :).. As you said 2 bit should be less non linear than 1 bit and since the psd diagrams of 1 bit SDMs donot have such tones I would say quantizer is (or should be) my least of a concern.

Sure I would post the outcome of this riddle. I am in academic world but this SDM is going to be a product .

harpoon wrote on Aug 20^th, 2011, 2:26pm:

Hi jefkat,

In essence, a quantizer (no matter how perfect it is) is a non-linear block and will always cause non-linearity to the system, but less so for a 2 bit (4 level ?) quantizer I think.

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Aug 21^st, 2011, 9:35am

I would expect a 1 bit quantised to produce harmonic tones as well. Are you sure you are not confusing yourself with first order dsm?

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Aug 21^st, 2011, 9:37am

Btw, this is the least of your worries ... You still have mismatch (dem), ota non idealities, settling, signal level ... etc to deal with !

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Aug 22^nd, 2011, 6:27am

harpoon wrote on Aug 21^st, 2011, 9:35am:

I would expect a 1 bit quantised to produce harmonic tones as well. Are you sure you are not confusing yourself with first order dsm?

A 1-V 140- W 88-dB Audio Sigma-Delta Modulator
in 90-nm CMOS Libin Yao, Student Member, IEEE, Michiel S. J. Steyaert, Fellow, IEEE, and Willy Sansen, Fellow, IEEE.....

This is the spectrum of 1 bit 3rd order SDM . Compare it with the above (2 bit SDM) :). Quantizer is not the reason . and the rest of the circuit is IDEAL.

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Aug 22^nd, 2011, 6:44am

Can you try a 1 bit quant in yr sims and let us know what u see?

Thanks ...

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Sep 19^th, 2011, 5:45am

I did the measurement on the chip (for which sim resuts were showing tones) and there were no tones ..There goes the theory of non linearinty in quantizer :)

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Sep 19^th, 2011, 6:30am

Now we are talking ... Have you employed some dem algorithm to handle mismatch ? What harmonics (and associated levels) do you see in measurement ? Also, what is your input signal level and SNR and SFDR achieved?

Just curious to know if we are comparing apples with apples ...

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Sep 19^th, 2011, 7:08am

Input signal is 1.0V differential.
SNDR is 82dB.
An alternative to DEM is used
3rd order non linearity approximately 84dB below carrier...

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by harpoon on Sep 19^th, 2011, 7:37am

may I ask what "alternative DEM" method you used ?

The results you posted are certainly respectable ...

what is the worse SNDR (sample to sample) did you see and what is the maximum bandwidth of your SD-ADC ?

in any case ... well done !

Title: Re: Tones in the spectrum of Sigma Delta Modulator
Post by jefkat on Sep 19^th, 2011, 9:03am

I would publish and let you know the technique :)
I yet have to do measurements to check variations