The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Measurements >> Phase Noise and Jitter Measurements >> power spectral density simulation of DSM used in fractional PLL
https://designers-guide.org/forum/YaBB.pl?num=1484736874

Message started by lwzunique on Jan 18^th, 2017, 2:54am

Title: power spectral density simulation of DSM used in fractional PLL
Post by lwzunique on Jan 18^th, 2017, 2:54am

I use the dsm model provided by cadence pllMMLib:SDM_MASH_3rd_111_macro. and run trans analysis，and then do psd to the output of the decimal output, range is -3 to 4. but the spectral is wrong?

the psd setting is as in the picture.

the frequency of the input clk is 16.384Mhz.

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by lwzunique on Jan 18^th, 2017, 3:02am

the psd setting is as in the picture.

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by cheap_salary on Jan 18^th, 2017, 3:51am

lwzunique wrote on Jan 18^th, 2017, 2:54am:

I use the dsm model provided by cadence pllMMLib:SDM_MASH_3rd_111_macro.
and run trans analysis,and then do psd to the output of the decimal output, range is -3 to 4.
but the spectral is wrong?

Plot it x-axis as log scale.
And apply Hanning window.

Did you confirm validity of output sequence of "pllMMLib/SDM_MASH_3rd_111_macro" ?

I don't use "pllMMLib/SDM_MASH_3rd_111_macro".
See http://www.designers-guide.org/Forum/YaBB.pl?num=1361756819/3#3

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by lwzunique on Jan 18^th, 2017, 5:55pm

cheap_salary wrote on Jan 18^th, 2017, 3:51am:

lwzunique wrote on Jan 18^th, 2017, 2:54am:

I use the dsm model provided by cadence pllMMLib:SDM_MASH_3rd_111_macro.
and run trans analysis,and then do psd to the output of the decimal output, range is -3 to 4.
but the spectral is wrong?

the output sequence of "pllMMLib/SDM_MASH_3rd_111_macro" is -3 to 4, and its average value is the input fractional data, I think it is valid.

and when do psd to the decimal output, I also made it simple at every delatN number(I mean each decimal number in the psd time range.) hanning widow is also adopted. but the result, the shape of the psd curve not like you showed in this topic:http://www.designers-guide.org/Forum/YaBB.pl?num=1361756819/3#3, and the magnitude of the psd is also not in accordance with that in:http://literature.cdn.keysight.com/litweb/pdf/5989-9471EN.pdf, and have great difference with the result I got with simulink.

in cadence spectre simulation, my input clock's T is 50ns, so begin at 75ns and end at 3276.875u, there will be 65536 simples. I show it in the picture.

my question is:
1.why my curve shape is not like yours?
2.why the magnitude of the curves are so different from each other?
3.when do PSD, we should also apply dB10 to PSD function, right?
4.why should we choose hanning window, and if the window size surpass 1024, the psd result will be wrong, why?

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by lwzunique on Jan 18^th, 2017, 6:41pm

I changed the clock to 1MHz and BIT to 20 as you did in:http://www.designers-guide.org/Forum/YaBB.pl?num=1361756819/3#3
but in my simulation the order is mash111 not mash1111. the result is more like.

another question: let make it question 5: what do the smooth factors mean, it doesn't work when I changed the value?

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by cheap_salary on Jan 18^th, 2017, 8:11pm

lwzunique wrote on Jan 18^th, 2017, 6:41pm:

but in my simulation the order is mash111 not mash1111.

Always you don't see anything carefully.

My examples are MASH-11. MASH-111 and MASH-1111.

Definition of FFT is different between MATLAB and Cadence ADE.

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by lwzunique on Jan 18^th, 2017, 8:49pm

thank you,@cheap_salary, in the:http://www.designers-guide.org/Forum/YaBB.pl?num=1361756819/3#3,how to get the smoothed-spectrum?

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by cheap_salary on Jan 18^th, 2017, 8:59pm

See delta-sigma toolbox of MATLAB.

I ported "circ_smooth()" to Cadence ADE as Skill function.

https://jp.mathworks.com/matlabcentral/fileexchange/19-delta-sigma-toolbox/content/delsig/circ_smooth.m

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by cheap_salary on Jan 19^th, 2017, 4:22am

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

the output sequence of "pllMMLib/SDM_MASH_3rd_111_macro" is -3 to 4,

No. It is no more than range of output codes.
You have not confirmed output sequences at all.

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

and its average value is the input fractional data, I think it is valid.

Maybe, output sequences are correct.

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

and the magnitude of the psd is also not in accordance with that in:http://literature.cdn.keysight.com/litweb/pdf/5989-9471EN.pdf, and have great difference with the result I got with simulink.

Definition of fft() is different between Keysight ADS PostProcessing and MathWorks MATLAB.

Study fft().

Definition of fft() is same between Keysight ADS PostProcessing and Cadence ADE PostProcessing.

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

1.why my curve shape is not like yours?

http://www.designers-guide.org/Forum/Attachments/QQ20170119103715.jpg seems reasonable.
This PSD is for delta_N, that is, frequency, so slope is +60dB/dec=2*order*20dB/dec.
PSD for phase is +40dB/dec=2*(order-1)*20dB/dec.
See http://www.designers-guide.org/Forum/YaBB.pl?num=1444923186/6#6

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

2.why the magnitude of the curves are so different from each other?

Simply your setting for psd() is not appropriate.

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

3.when do PSD, we should also apply dB10 to PSD function, right?

Right.

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

4.why should we choose hanning window,

See Appendix-A of the followings.
https://www.amazon.com/Understanding-Delta-Sigma-Converters-Richard-Schreier/dp/0471465852/ref=la_B000APOGIC_1_1?s=books&ie=UTF8&qid=1484826787&sr=1-1

lwzunique wrote on Jan 18^th, 2017, 5:55pm:

and if the window size surpass 1024, the psd result will be wrong, why?

Study psd().
See http://www.designers-guide.org/Forum/YaBB.pl?num=1405235104

lwzunique wrote on Jan 18^th, 2017, 6:41pm:

another question: let make it question 5: what do the smooth factors mean, it doesn't work when I changed the value?

I can not understand what you want to mean at all.

Title: Re: power spectral density simulation of DSM used in fractional PLL
Post by lwzunique on Jan 19^th, 2017, 4:21pm

when doing psd in cadence spectre calculator, there is an option:smoothing factor.