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 >> help:phase noise from VCO Transient sim
https://designers-guide.org/forum/YaBB.pl?num=1196964298

Message started by Sarig on Dec 6^th, 2007, 10:04am

Title: help:phase noise from VCO Transient sim
Post by Sarig on Dec 6^th, 2007, 10:04am

Hello,
I run VCO 5GHz with transient noise simulation.
How do I measure the Phase noise from the sin output ?
Help please.
thanks,
Erez

Title: Re: help:phase noise from VCO Transient sim
Post by jeffyan on Dec 6^th, 2007, 5:48pm

Sarig wrote on Dec 6^th, 2007, 10:04am:

Hello,
I run VCO 5GHz with transient noise simulation.
How do I measure the Phase noise from the sin output ?
Help please.
thanks,
Erez

hi
in general, we use spectre RF to simulate vco's phase noise. for more information,plz refer to spectre user manual.
good luck
jeff

Title: Re: help:phase noise from VCO Transient sim
Post by Sarig on Dec 7^th, 2007, 2:49am

Hi,
Thanks for the answer. Spectre can do PNoise and more but there is no WAY to calculate Phase noise from sin transient output (without using PSS+PNOISE)
Do you know the answer ?
Sarig.

Title: Re: help:phase noise from VCO Transient sim
Post by didac on Dec 7^th, 2007, 3:05am

Hi,
Take a look at:http://eesof.tm.agilent.com/pdf/jitter_phasenoise.pdf, from what I remember from classes at school there are all aproximations but I suppose that it's better than nothing.
Hope it helps,

Title: Re: help:phase noise from VCO Transient sim
Post by Eugene on Dec 7^th, 2007, 8:06am

When you say "transient noise simulation", what exactly do you mean? Does your circuit include sources that inject noise as a function of time? If so, did you create the models yourself or were they part of the tool? If they were part of the tool, I would imagine the tool has some means of analyzing noisey simulations. Most basic circuit simulators like SPICE and Spectre have no inherent noise sources in their transient analyses; with those simulators you would only be looking at noise from the numerical errors in the transient simulation.

Anyway, assuming you are not using an RF tool that will perform small signal noise analysis about a periodic steady state solution, and assuming you indeed have time-dependent noise sources in your circuit, you still need access to some kind of Fourier analysis. What simulator are you using?

Title: Re: help:phase noise from VCO Transient sim
Post by Sarig on Dec 7^th, 2007, 9:27am

Hello all,
Thank you for your comments.
I will explain the issue again for this is not an Easy thing:
1) I want to measure the Phase noise of VCO. I don't want to use PSS+PNOISE simulation which I know very very well.
2) I have a transistor level VCO which I simulate using Spectre: I am doing Transient simuation and include noise option. when doing so the simulation is running transient and include each device noise (flicker+thermal).
3) Now: I have an output which is a Sin wave at 5GHz, this output has a noise in it.
4) So: what to do with this output in order to measure the phase noise ?
5) I want to bring the sin output to MATLAB (I can geneate an excel vector of samples in Spectre), what to do now in MATLAB ?
6)I feel I need some down convert (using for example a verilogA model of Mixer) in my transient simulation and take the mixer output to the matlab else I will not be able to create a vector only of the noise because I will need to much samples in the vector (let say 10pS sample period for a 5GHz signal for 10uS simulation (I want noise upto 10KHz from carry).

So bottom line: simple question. having a Sin wave which you know has noise ---> How to get the Phase noise of it ?

Regards,
Sarig

Title: Re: help:phase noise from VCO Transient sim
Post by David Lee on Dec 7^th, 2007, 9:24pm

For a basic phase noise measurement, do the following:

1. Make periods vs time measurement
2. Calculate: phases = (2 * Pi / mean_period) * cumsum( periods - mean_period )
3. Compute power spectral density of "phases", which gives you "phase noise".

This is basically Eq (2) of my paper: "Analysis of jitter in phase-locked loops"
http://ieeexplore.ieee.org/Xplore/login.jsp?url=/iel5/82/26444/01177400.pdf?arnumber=1177400
Also, have a look at: www.designers-guide.org/Analysis/PLLnoise+jitter.pdf

Title: Re: help:phase noise from VCO Transient sim
Post by Sarig on Dec 8^th, 2007, 2:09am

David,
thanks for the answer.
1. "Phases" has no frequency components in it. Its the accumulated phase which is the Phase jitter of my VCO, so I don't understand what is the meaning of PSD on "phases" (can you explain?)
2. My data is at 5GHz frequency and in order to bring it to Matlab for lets say 1mS (I want Phase noise from 1kHz - 1MHz) I will need enormous excel file.
3. Can I take my VCO output, pass it via Mixer, filter the output at lets say 0.5GHz and do PSD on the result?
It is known that because the signal is sin and the noise is very low the total noise will be the phase noise so simply do PSD (Please comment)

Regards,
Erez

Title: Re: help:phase noise from VCO Transient sim
Post by David Lee on Dec 8^th, 2007, 10:45am

Sarig wrote on Dec 8^th, 2007, 2:09am:

1. "Phases" has no frequency components in it. Its the accumulated phase which is the Phase jitter of my VCO, so I don't understand what is the meaning of PSD on "phases" (can you explain?)
2. My data is at 5GHz frequency and in order to bring it to Matlab for lets say 1mS (I want Phase noise from 1kHz - 1MHz) I will need enormous excel file.
3. Can I take my VCO output, pass it via Mixer, filter the output at lets say 0.5GHz and do PSD on the result?
It is known that because the signal is sin and the noise is very low the total noise will be the phase noise so simply do PSD (Please comment)

Regards,
Erez

Hi Erez,

1. "phases" is a vector of accumulated phase jitter (in radian), for 1st, 2nd, 3rd, 4th, .. n-th cycle. Look up "psd" in your favourite post-processing tool. This function divides your data into multiple segments of "nfft" data points each, do a FFT on each segment, and average the power. The result is a periodogram estimate of the power spectral density in unit of radian^2/Hz.
2. I would be more concerned with your simulator accuracy and your run time. One reason I suggest making a periods vs time measurement is to trim down the raw data size to ease data exporting/importing.
3. Some phase noise measurement instructments use this method. I call this the Mixer method, and this too works fine on several frac-N and int-N PLLs that I compared with lab measurement.

Title: Re: help:phase noise from VCO Transient sim
Post by Sarig on Dec 8^th, 2007, 11:13am

Thanks David,
I will think more about the issue and be happy to update you soon in this site.
Regards,
Erez Sarig