Phase Noise Plot to Equivalent C/N ratio conversion?

[Paul]

I'd like to convert a phase noise plot (in dBc/Hz @ offsets) to
an
equivalent Carrier-to-Noise ratio (C/N), over a specific processing
bandwidth.

Yes, these NGs are a long shot for getting info on this sort of
thing,
but it can't hurt to ask.

 

[Jean-Christophe]

      I'd like to convert a phase noise plot (in dBc/Hz @ offsets) to
an
equivalent Carrier-to-Noise ratio (C/N), over a specific processing
bandwidth.

If I got it right : you have a ( phase_noise vs frequency )
and you want to convert this into a ( dBc vs frequency ) ?

What's "dBc" against mW ( and/or mV ) rms ?

 

[[email protected]]

      I'd like to convert a phase noise plot (in dBc/Hz @ offsets) to
an
equivalent Carrier-to-Noise ratio (C/N), over a specific processing
bandwidth.

      Yes, these NGs are a long shot for getting info on this sort of
thing,
but it can't hurt to ask.

Hi Paul,

First undo the "dB" operation, so you get "Noise power density (W/
Hz)" over "carrier power" ratios (W) versus frequency (don’t convert
to voltage ratios).

Integrate versus frequency over the required frequency span, and you
have "noise power" over "Carrier power" ratio. This equals 1/(C/N)
ratio. Make a good judgment on the skirts adjacent to the carrier to
avoid that carrier power is assumed noise power.

Best regards,

Wim
PA3DJS
www.tetech.nl
PM is the one shown without a, b and c.

 

[Paul]

Hi Paul,

First undo the "dB" operation, so you get "Noise power density (W/
Hz)"  over "carrier power" ratios (W) versus frequency (don’t convert
to voltage ratios).

Integrate versus frequency over the required frequency span, and you
have "noise power" over "Carrier power" ratio. This equals 1/(C/N)
ratio. Make a good judgment on the skirts adjacent to the carrier to
avoid that carrier power is assumed noise power.

Best regards,

Wim
PA3DJSwww.tetech.nl
PM is the one shown without a, b and c.

Ok, this is exactly what i did. However, it's debated where
exactly, is the cutoff frequency between the carrier and the noise.

 

[Paul]

It's actually really simple--the CNR is

CNR = 1/(2 (Delta phi)**2)

Cheers

Phil Hobbs

How do you define Delta phi? The change in phase?

How would you get this from a phase noise plot?

Sounds like you are talking out of your ass!

 

[[email protected]]

      Ok, this is exactly what i did.  However, it's debated where
exactly, is the cutoff frequency between the carrier and the noise.

Hello Paul,

I assumed you did measurements with a spectrum analyzer (just scalar
measurements, no vector based stuff). For determining "delta phi" you
need a vector analyzer (or at least a quadrature down conversion
setup).

You might change the RBW setting and look how the spectrum changes.
Below a certain RBW setting, it will not change. In that case the
skirts can be because of the signal's phase/frequency noise, or phase/
frequency noise from the analyzer.

When you have a known stable source, you might be able the measure the
frequency/phase noise contribution from the analyzer. An indication of
the analyzer's noise is the response at zero Hz.

I don't know your application, but noise very close to the carrier
might be suppressed by your application (digital frequency tracking
loop or just block length of the digital demodulation process?). Is
it possible for you to derive the CNR from the output at some stage
in you digital processing? Of course in that case the receiver front-
end must be good.

Best regards,

Wim
PA3DJS
www.tetech.nl
The real address is the one without a, b and c.

 

[Paul]

Hello Paul,

I assumed you did measurements with a spectrum analyzer (just scalar
measurements, no vector based stuff). For determining "delta phi" you
need a vector analyzer (or at least a quadrature down conversion
setup).

You might change the RBW setting and look how the spectrum changes.
Below a certain RBW setting, it will not change. In that case the
skirts can be because of the signal's phase/frequency noise, or phase/
frequency noise from the analyzer.

When you have a known stable source, you might be able the measure the
frequency/phase noise contribution from the analyzer. An indication of
the analyzer's noise is the response at zero Hz.

We measured this with a dedicated Agilent phase noise
analyzer. It contributes very little to the phase noise in most
cases.

I don't know your application, but noise very close to the carrier
might be suppressed by your application (digital frequency tracking
loop or just block length of the digital demodulation process?).  Is
it  possible for you to derive the CNR from the output at some stage
in you digital processing? Of course in that case the receiver front-
end must be good.

Ok, you were on the right track by first converting the dBc/
Hz
to a power ratio, and then integrating by adding these powers
together. I have a basic MATLAB script that does this.

The real question is: where does the Carrier end, and the
Noise begin, when looking at a phase noise plot? I'm sure it's
dependant on your DSP processing, as some people have said
that it's the phase noise outside of your tightest FIR filter that
should be integrated, to get the total Noise.