The Designer's Guide Community
Forum
Welcome, Guest. Please Login or Register. Please follow the Forum guidelines.
Mar 28th, 2024, 11:48am
Pages: 1
Send Topic Print
Phase noise measurement - fundamental question (Read 7409 times)
sandman
Community Member
***
Offline



Posts: 73

Phase noise measurement - fundamental question
Jun 17th, 2008, 8:46am
 
Hi,

Pls forgive my ignorance of this topic, but I'm new to this area of oscillators and am just picking up some fundamentals.

I noticed from the definition of phase noise that it is the ratio of the power in one sideband, referred to the input carrier frequency on a per hertz of bandwidth spectral density basis,
to the TOTAL SIGNAL POWER (I assume this is integrated power - or what does this mean ?), at Fourier frequency difference f from the carrier. It is a normalized frequency domain measure of phase fluctuation sidebands, expressed in decibels relative to the carrier per hertz.

But when we measure it, it's just the difference (in dBm) between the peak power of the signal (i.e. power at carrier frequency) and the power at the offset frequency. I couldn't figure where the TOTAL SIGNAL POWER fits into this measurement. Am I missing some normalisation factor between the definition and the measurement process or is there a very basic error in my understanding ? If so, what ?

I would greatly appreciate if anyone could describe (in whatever details possible) how the spectrum analyser measures the phase noise.

Thanks in advance ! Cheers.



Back to top
 
 
View Profile   IP Logged
ywguo
Community Fellow
*****
Offline



Posts: 943
Shanghai, PRC
Re: Phase noise measurement - fundamental question
Reply #1 - Jun 17th, 2008, 6:33pm
 
Hi Sandman,

That problem had been confusing me for a long time when I began to study phase noise.

Obviously you notice that the definition for phase noise is different from what we do when measuring it. Then which one is right? The answer is they are both correct. If measure the phase noise accroding to the definition strictly, what is the limit for the integral of the signal power? Infinity? But it's difficult to integrate the signal power to infinity in engineering. Right? So we measure the phase noise in an approximation method. That's the ratio of the power at the offset frequency  to the peak power at carrier frequency in decibel carrier (dBc).  

I hope this is helpful to you.


Yawei
Back to top
 
 
View Profile   IP Logged
sandman
Community Member
***
Offline



Posts: 73

Re: Phase noise measurement - fundamental question
Reply #2 - Jun 18th, 2008, 1:55am
 
Hi Yawei,

Thanks for clearing it up !

But, building upon the approximation for measurement, do we assume that most of the signal power is concentrated at the carrier frequency ? If so, then wouldn't this assumption be incorrect for an oscillator with a very poor Q factor ? (say I have an oscillator with a center frequency several GHz and -3dB bandwidth in the tens of MHz)

If not, what is the approximation being made ? Or what are the assumptions in making these approximations ?

If you or anyone could share how the spectrum analyser actually measures phase noise, that would be great ! I've looked into all sorts of literature, but I can't seem to find a detailed explanation. (every company just posts advertisements for training sessions on phase noise measurement)

Thanks in advance !
Cheers.
Back to top
 
 
View Profile   IP Logged
ywguo
Community Fellow
*****
Offline



Posts: 943
Shanghai, PRC
Re: Phase noise measurement - fundamental question
Reply #3 - Jun 18th, 2008, 11:58pm
 
There is an introduction to spectrum analyzer.

http://www.tutorialsweb.com/rf-measurements/spectrum-analyzer.htm


BTW, the method used in engineering is feasible because the carrier is often a single tone. I haven't seen any oscillator output that is of such a poor Q that results a 3 dB bandwidth in tens of MHz.


Yawei
Back to top
 
 
View Profile   IP Logged
sandman
Community Member
***
Offline



Posts: 73

Re: Phase noise measurement - fundamental question
Reply #4 - Jun 19th, 2008, 4:35am
 
Thanks for the link and your comment.

I still feel that the assumption would yield inaccurate measurement results for an oscillator of low Q. I've come across a few oscillators that operate in the 10s of GHz and have bandwidths that are about 10-30MHz. Would you have any comments on how the machine would measure PN in such a case ?

Cheers.
Back to top
 
 
View Profile   IP Logged
ywguo
Community Fellow
*****
Offline



Posts: 943
Shanghai, PRC
Re: Phase noise measurement - fundamental question
Reply #5 - Jun 19th, 2008, 6:56pm
 
Hi Sandman,

What is the resolution bandwidth (RBW) do you set for the measurement?


Yawei
Back to top
 
 
View Profile   IP Logged
David Lee
Junior Member
**
Offline



Posts: 29

Re: Phase noise measurement - fundamental question
Reply #6 - Jun 20th, 2008, 12:35am
 
Yes, the standard textbook technique will be inaccurate for low-Q (and high-Q) phase noise measurment.

Phase noise is measured like this.
1. Measure the spectrum at f0+fm.
2. Integrate the spectrum from f0/2 to (3/2) f0. Assume the output waveform is sinusoidal, or make it so by filtering.
3. Take the ratio and convert to dBc/Hz.

In practice, when using a spectrum analyzer, step 1 is done by measuring the spectrum at f0+fm using a narrow resolution bandwidth, and step 2 is done by measuring the spectrum at f0 using a much wider resolution bandwidth.
Back to top
 
 

- David
View Profile   IP Logged
sandman
Community Member
***
Offline



Posts: 73

Re: Phase noise measurement - fundamental question
Reply #7 - Jun 20th, 2008, 3:41am
 
Thanks, David and Yawei for your inputs.

I thought I should mention the context of my question, just to put things into perspective.

I have an oscillator that runs at nearly 15GHz and has a -3dB bandwidth of about 30 MHz. It's not a traditional LC or CMOS oscillator and which is why I'm trying to characterize it's phase noise. The carrier power is low (@ -40dBm) and is above the noise floor of the measurement setup by say 30 dB (which I agree is very noisy!).

I'm using the direct measurement method to measure the PN profile of this free running oscillator (using a R&S FSU26 with a software PN add-on kit available from the same company) , which at the moment has a fluctuation in it's central frequency to the tune of a say 1-5MHz. I assume that because of this fluctuation the machine measures frequency noise and not just the intrinsic phase noise of my oscillator. I still don't understand this fully and it would be great if you or anyone would have some inputs or comments on this and what steps I would need to take to measure this noisy, very low Q, very low power oscillator.

There's one thing though - the machine fails to track the oscillator signal even if I increase the error tolerance to large percentages. So, I manually specify the level and frequency of the carrier for the PN measurement. I'm willing to live with some percentage of error in this measurement assuming it is - say, below 10-15%.

Yawei, at present I'm trying two settings for the resolution bandwidth. In one case I allow the machine to set it's on RBW and in another I set it myself to be as large as possible, manually.
In the first setting (automatic) I get an output which has steps (probably due to the changing RBW and the inability of the machine to track the signal peak) in the PN profile. In the second I get a smooth response, but I'm sure the measurement is accurate.

Any inputs are welcome !
Back to top
 
 
View Profile   IP Logged
Pages: 1
Send Topic Print
Copyright 2002-2024 Designer’s Guide Consulting, Inc. Designer’s Guide® is a registered trademark of Designer’s Guide Consulting, Inc. All rights reserved. Send comments or questions to editor@designers-guide.org. Consider submitting a paper or model.