How to calculate periodic jitter from a spectrum of a signal 1

[WestRace]

I am trying to find a function that can calculate the periodic jitter given a power spectrum such as captured from a spectrum analyzer.

For example, I have a 1.25GHz clock that is being FM modulated by another signal running at 1MHz. I can capture the power spectrum of the 1.25GHz on a spectrum analyzer. From the power spectrum, how do I calculate how much the 1.25GHz is being modulated in term of mUI or ps?

Is there an equation I can use to calculate?

 

[dbengtson]

Hi Westrace:

Jitter is integrated phase noise, so if you are able to capture a phase noise plot of the clock in terms of dBc/Hz, then you can numerically integrate the phase noise to get jitter. I've attached a spreadsheet I grabbed from RF Globalnet as a starting point.

Dave

 

[WestRace]

Thanks for the tips. I am not sure if that is what I am looking for.
I knew how to integrate the noise power spectrum to obtain the phase noise in either mUI or ps-rms.
What I am looking for is kind of a special case I think where you got a carrier being FM modulate by a another signal at a much lower frequency.

 

[sreid]

This is just a general thought on what you want to do. If a sine wave carrier [Fc] is FM modulated with a sine wave [Fm], then the sidebands are at Fc+/-Fm, Fc+/-2Fm, etc. The ampliture of these [infinate number] sidebands are mathematically known. So if you can collect spectrial data over a integral number of cycles [2.4 GHz/1 MHz = 2400] so there is no leakage and the FFT bins at 1 MHZ intervals on either side of 2.4 GHz, then any amplitudes above or below the theoritical amplitudes represents random noise.

I'm certain this concept would be fraught with practical difficulities.

Can you measure the jitter of the carrier, the modulating signal and the modulator [with no input signals] and combine them to get the nmber you want?

 

[biff44]

FM modulation is described by a besel functions, which are a little messy.

However...if it is just little FM deviation(ie. small angle approximations apply), then the sidebands at +/- 1 Mhz from the carrier in dBc are approximately:

sideband dbc = 20 LOG (delta f/2Fm)

where delta f is the frequency deviation, and 2Fm in this case is 2 * 1 Mhz.

So, if you can see how many dBc down the sideband tone is, you can calculate the frequency deviation in Hz to the carrier at 1.25 Ghz. That will yield the pk time jitter.

Rich

http://www.MaguffinMicrowave.com

Microwave and wireless design consulting

 

[zappedagain]

biff44,

I like that! Where did you find that equation?

Z

 

[biff44]

Well, funny you should ask! It is an equation that I have used for years. But when I went to google to search it up--nothing! I fear all the old, fundamental, papers written by the great ones are slipping out of our "internet only" engineering society! A shame.

I do keep a couple of big file cabinets of important paper print outs, and found it there:
Review of the Specification and Measurement of Short-Term Stability, by Robert B. Shields

The copy I had did not say what publication it was from, but it was probably an IEEE or IRE paper.

http://www.MaguffinMicrowave.com

Microwave and wireless design consulting