applying a transfer function to the phase noise plot

Dear matcha,

matcha said:

I want to apply a noise transfer function to the phase noise plot obtained by pss+pnoise and integrate it to obtain a filtered Rj.
Can these be done with a calculator?

If you are attempting to exclusively use the ViVA Calculator, to my knowledge, there are no standard frequency domain based noise transfer functions nor any frequency-domain based filters provided in ViVA for use with its Calculator. However, there is nothing to prevent one from writing a custom SKILL function to implement a frequency domain based filter for use with the pnoise output waveform. There will be some processing involved as you must create a waveform to represent the filter as well as evaluate it at the specific offset frequencies of your phase noise data. Units conversion will need to be included as the phase noise data is usually in dBc/Hz. An input for the frequency of your phase noise data will be required if you desire that your integration result be in the time domain.

An alternative is to create an ocean script to do the filtering and integration. Once again, the transfer function will need to be synthesized.

At least two other options exist that I will at least mention if either happens to be of interest.

I do not know the details of the noise transfer function you are interested in using. I know there are some rather complex compliant transfer functions such as those for the higher rate PCIe standards. However, many consist of a single order highpass and/or lowpass corner frequency. If your transfer function falls into the latter category, I have two tools that may be of interest.

1. The first is a Microsoft Excel based workbook into which you insert your phase noise data as a text file and choose a lowpass and highpass corner frequency as well as the nominal frequency of your phase noise data and it will display the phase noise data as a function of frequency and compute the filtered and unfiltered jitter. However, this is rather limited as if your phase noise data includes deterministic components, the random jitter estimate it provides will not be accurate. The workbook uses an integration algorithm that is designed to very accurately integrate a phase noise characteristic even with a limited number of data points per decade of offset frequency. Let me know if this is of interest.

2. The second is a UNIX command line program that accepts your phase noise data as a text file and will accurately synthesize a model of your phase noise, perform spur and deterministic noise source identification and removal, and provide an accurate estimate of random and deterministic jitter components using either brickwall or first-order lowpass and highpass corner frequencies. Results also include the slope(s) of your phase noise data to easily allow identification of flicker noise frequency regions.

Documentation of the program and its operation may be found at:

www.icloud.com/iclouddrive/050gMF6-uU8Bk6izwACyhamfQ#phase%5Fnoise%5Fanalysis%5Ftool%5F020223v1p2

Shawn

Dear matcha,

matcha said:

I want to apply a noise transfer function to the phase noise plot obtained by pss+pnoise and integrate it to obtain a filtered Rj.
Can these be done with a calculator?

If you are attempting to exclusively use the ViVA Calculator, to my knowledge, there are no standard frequency domain based noise transfer functions nor any frequency-domain based filters provided in ViVA for use with its Calculator. However, there is nothing to prevent one from writing a custom SKILL function to implement a frequency domain based filter for use with the pnoise output waveform. There will be some processing involved as you must create a waveform to represent the filter as well as evaluate it at the specific offset frequencies of your phase noise data. Units conversion will need to be included as the phase noise data is usually in dBc/Hz. An input for the frequency of your phase noise data will be required if you desire that your integration result be in the time domain.

An alternative is to create an ocean script to do the filtering and integration. Once again, the transfer function will need to be synthesized.

At least two other options exist that I will at least mention if either happens to be of interest.

I do not know the details of the noise transfer function you are interested in using. I know there are some rather complex compliant transfer functions such as those for the higher rate PCIe standards. However, many consist of a single order highpass and/or lowpass corner frequency. If your transfer function falls into the latter category, I have two tools that may be of interest.

1. The first is a Microsoft Excel based workbook into which you insert your phase noise data as a text file and choose a lowpass and highpass corner frequency as well as the nominal frequency of your phase noise data and it will display the phase noise data as a function of frequency and compute the filtered and unfiltered jitter. However, this is rather limited as if your phase noise data includes deterministic components, the random jitter estimate it provides will not be accurate. The workbook uses an integration algorithm that is designed to very accurately integrate a phase noise characteristic even with a limited number of data points per decade of offset frequency. Let me know if this is of interest.

2. The second is a UNIX command line program that accepts your phase noise data as a text file and will accurately synthesize a model of your phase noise, perform spur and deterministic noise source identification and removal, and provide an accurate estimate of random and deterministic jitter components using either brickwall or first-order lowpass and highpass corner frequencies. Results also include the slope(s) of your phase noise data to easily allow identification of flicker noise frequency regions.

Documentation of the program and its operation may be found at:

www.icloud.com/iclouddrive/050gMF6-uU8Bk6izwACyhamfQ#phase%5Fnoise%5Fanalysis%5Ftool%5F020223v1p2

Shawn

Dear Shawn,

Thank you very much for your reply. As you explained in #1, I am using a similar Excel file to calculate an unfiltered jitter. but I was wondering if there was another way. Thank you for the document. I will read it.

Regards,
matcha

Dear matcha,

matcha said:

I am using a similar Excel file to calculate an unfiltered jitter. 

Great! Please do use caution when integrating a phase noise characteristic with fewer than about 4 points per offset frequency decade. Using a trapezoidal or other low order polynomial based integration algorithms will result in significant errors in their integration estimates. I have a workbook at:

1drv.ms/x/s!AnM-GsAEZPoSrRdbJx3QfcGhKYuf?e=adjHGc

that includes an integration algorithm designed to provide an accurate estimate of the integrated phase noise characteristic with as few as a single point per frequency decade. (Hence, when when you perform a pnoise analysis following a pss analysis, you need not include many points per frequency decade - which can significantly reduce your simulation time.) The workbook also allows for a first-order highpass and lowpass filter as a noise transfer function. The resulting data is provided in graphical and tabular formats. The instructions are provided in the first tab labeled "Variables and Instructions".

Shawn