The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> RF Simulators >> Spctre to predicts PSRR for OScillator
https://designers-guide.org/forum/YaBB.pl?num=1108757064

Message started by ic_engr on Feb 18^th, 2005, 12:04pm

Title: Spctre to predicts PSRR for OScillator
Post by ic_engr on Feb 18^th, 2005, 12:04pm

I ran PSS and PXF for my oscillator. I shose sweeptype as absolute with f=1Hz to 10MHz.

After completion of PXF, the PXF form assks me for selecting the Vsource. I selected the VDD source. It gave me a plot from 0 to 10MHz. Is this the frequency of the VDD from DC to 10MHz or is it harmonic of the oscillator.

The PSRR output shows 40dB at 1Hz and goes dwon to 0 at ~1kHz. from 2kHz to 10MHz it stays flat at ~-3dB.

Can some one explain me what all this is.

What does 40dB reflect in terms of change in frequency of fundamental being 4MHz.

Is the plot voltage vs freq or not.

I will appreciate your help on this matter.

Also do I have to select Modulation in the PXF form. Do i have to make AC=1 on the Vsource.

I ensured the freq axis is absin.

Any help will be appreciated.

Thx

ic_engr

Title: Re: Spctre to predicts PSRR for OScillator
Post by Ken Kundert on Feb 18^th, 2005, 1:55pm

With PXF the frequency range that you specify on the analysis is always the frequency range of the observed signal. Thus, since you specified 1Hz to 10MHz absolute, that is what you get. So you are measuring the transfer function from VDD from baseband to baseband.

For PSR for an oscillator, you probably want the transfer function from baseband to the carrier. So will want to use a relative sweep. That would cause the output frequency of your transfer function to range between f₀ + 1Hz to f₀ + 10MHz. When plotting VDD, you will need to choose the +1 sideband (or perhaps it is the -1 sideband, I always forget, use 'spectre -h pxf' for find out for sure). This will result in you computing the transfer function from 1Hz --> 10MHz to f₀ + 1Hz --> f₀ + 10MHz.

If you chose the output as a pair of node and the input as a voltage source (VDD), then the transfer function will be the voltage gain from the source to the output nodes. So a result of 40dB @ 1Hz would imply that a 1Hz signal at VDD is amplified by a factor of 100 before appearing at the output at f₀ + 1Hz.

You do not have to set mag or pacmag on the source.

-Ken

Title: Re: Spctre to predicts PSRR for OScillator
Post by ic_engr on Feb 18^th, 2005, 2:04pm

Thanks Ken,

I thought the output will show the transfer function from VDD at 1Hz to a change in Frequency of the oscillator.

I am interested to find out how change in VDD at various frequencies will change oscillator frequency and not oscillator voltage.

Is there any hope of getting this information from PXF simulation.

Thanks for your time on this

ic_engr

Title: Re: Spctre to predicts PSRR for OScillator
Post by Ken Kundert on Feb 18^th, 2005, 3:27pm

You can use small-signal FM modulation theory to map the transfer function in voltage to a transfer function in frequency.

-Ken

Title: Re: Spctre to predicts PSRR for OScillator
Post by ic_engr on Feb 19^th, 2005, 11:52am

Ken,

Where can I find information on this. Have you any links or lead on this.

Thanks very much

ic_engr

Title: Re: Spectre to predict PSRR for Oscillator
Post by Ken Kundert on Feb 19^th, 2005, 3:56pm

Any book on communication theory, such as Ziemer and Trantor.

-Ken

Title: Re: Spectre to predict PSRR for Oscillator
Post by Ken Kundert on Feb 19^th, 2005, 5:56pm

Take a look at Section 6.1.2 on AM and PM conversion, and in particular equation 64, in http://www.designers-guide.com/Analysis/rf-sim.pdf.

-Ken

Title: Re: Spctre to predicts PSRR for OScillator
Post by ic_engr on Feb 21^st, 2005, 6:42am

Ken,

I did read that section and equation 64. I am a bit confused as to what is Ac, the denominator of eqn 64 ?. Also i sL and U the results from PXF analysis.

Is it possible to show an examle using equation 64. That will be great help.

Thx

ic_engr

Title: Re: Spectre to predict PSRR for Oscillator
Post by Ken Kundert on Feb 21^st, 2005, 11:19pm

A_c, L and U are the lengths of the three phasors. A_c is the amplitude of the carrier, U is the amplitude of the upper sideband, and L is the amplitude of the lower sideband.

A_c is a large signal and so is computed using the PSS analysis, L and U are small-signals and are computed with a PAC analysis.

You also need phi_c. This is the phase of the carrier and is computed during the PSS analysis.

I'll try to see if I can find an example.

-Ken

Title: Re: Spectre to predict PSRR for Oscillator
Post by Ken Kundert on Feb 22^nd, 2005, 7:17am

As for an example, take a look at http://www.designers-guide.com/Analysis/am-pm-conv.pdf.

-Ken