Phase Noise

can we then assume that from 0-F Hz, the phase noise model follows the
1/f^2 slope?
also since the simulink only allow us 2 input parameters ie phase noise
level and freq offset, can we safe assume that this is the breakoff
power/freq for the 1/f slope?


		
This message was sent using the Comp.DSP web interface on
www.DSPRelated.com

please see

http://www.mathworks.com/access/helpdesk/help/toolbox/commblks/ref/phasenoise.html


In the example given, the slope is 1/F.

The phase noise level (density) is -120 dBc/Hz at 100 Hz.

In a way, the use of 2 numbers is redundant.
For convenience, they are probably allowing you to specify both the
density and the frequency at which that density occurs.  In reality,
since the slope is fixed at 1/F, there is only one degree of freedom.
i.e. specifying -120 dBc/Hz at 100 Hz (as in the example)  gives the
exact same results as specifying -130 dBc/Hz at 1kHz.  Check to see if
this is true.

If you want more details, I would look into the shape of the digital
filter they use to shape the noise.  Without this filter, if the random
number generator noise spectral  density is flat with frequency and
applied to an angle modulator as shown, the phase noise density would
be flat with frequency.  Therefore, the digital filter must have a -3
dB per octave response over a wide range of frequencies.

I'm not sure a random number generator has a flat spectral density?
Does it?

(Disclaimer, I'm not a Simulink expert, the above is all based upon
general phase noise theory)

Mark

Mark wrote:
> please see
>
>
http://www.mathworks.com/access/helpdesk/help/toolbox/commblks/ref/phasenoise.html
>
>
> In the example given, the slope is 1/F.
>
> The phase noise level (density) is -120 dBc/Hz at 100 Hz.
>

see also;

http://www.mathworks.com/support/solutions/data/1-1CAZP.html?solution=1-1CAZP

Mark wrote:
> please see
> 
> http://www.mathworks.com/access/helpdesk/help/toolbox/commblks/ref/phasenoise.html
> 
> 
> In the example given, the slope is 1/F.
> 
> The phase noise level (density) is -120 dBc/Hz at 100 Hz.
> 
> In a way, the use of 2 numbers is redundant.

Are you sure? Typical 1/f noise sinks below the Johnson noise at some 
frequency and disappears (glub glub). !/f noise plagued semiconductors 
at first. Efforts to make low-noise audio transistors succeeded only in 
pushing down the turnover frequency. The frequency has been pushed so 
low that it is called popcorn noise in op-amps; an audible Poisson 
distribution. The specified frequency may be that frequency below which 
the noise can be considered "excess".

   ...

> I'm not sure a random number generator has a flat spectral density?
> Does it?

That depends entirely on the generator. It can be either way.

> (Disclaimer, I'm not a Simulink expert, the above is all based upon
> general phase noise theory)

Same here.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Jerry Avins wrote:
> Mark wrote:
> > please see
> >
> >
http://www.mathworks.com/access/helpdesk/help/toolbox/commblks/ref/phasenoi=
se.html
> >
> >
> > In the example given, the slope is 1/F.
> >
> > The phase noise level (density) is -120 dBc/Hz at 100 Hz.
> >
> > In a way, the use of 2 numbers is redundant.
>
> Are you sure? Typical 1/f noise sinks below the Johnson noise at some

> frequency and disappears (glub glub). !/f noise plagued
semiconductors
> at first. Efforts to make low-noise audio transistors succeeded only
in
> pushing down the turnover frequency. The frequency has been pushed so

> low that it is called popcorn noise in op-amps; an audible Poisson
> distribution. The specified frequency may be that frequency below
which
> the noise can be considered "excess".
>
>    ...
>
> > I'm not sure a random number generator has a flat spectral density?
> > Does it?
>
> That depends entirely on the generator. It can be either way.
>
> > (Disclaimer, I'm not a Simulink expert, the above is all based upon
> > general phase noise theory)
>
> Same here.
>
> Jerry
> --
> Engineering is the art of making what you want from things you can
get.
>
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=
=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF=AF



No I'm not "sure" but the Mathworks help seem to indicate that:

http://www.mathworks.com/access/helpdesk/help/toolbox/commblks/ref/phasenoi=
se.html

there is no metion that F is a breakpoint of any kind but I can see how
that would be a logical way to look at it...but I don't think they are
looking at it that way.

I think you specify the density D and the F and the function generaes
phase noise with
density D at offset F.  Since the slope is fixed at 1/F it is somewhat
redundent.

What do you think?


Mark

just did a test on simulink, the results are the same for -60dBc @ 100Hz,
-70dBc @ 1kHz and -90dBc @ 100kHz. so its safe to conclude that the input
parameters for simulink is not the breaking point for the 1/f slope.

anyway can we safely assume that the 1/f slope or the flicker noise is
dominant for phase noise; hence we onli consider that and not the 1/f^2
slope?
		
This message was sent using the Comp.DSP web interface on
www.DSPRelated.com

"elbib00" <elbib00@hotmail.com> wrote in message 
news:DaadnZeb94vm29ffRVn-rg@giganews.com...
> just did a test on simulink, the results are the same for -60dBc @ 100Hz,
> -70dBc @ 1kHz and -90dBc @ 100kHz. so its safe to conclude that the input
> parameters for simulink is not the breaking point for the 1/f slope.
>
> anyway can we safely assume that the 1/f slope or the flicker noise is
> dominant for phase noise; hence we onli consider that and not the 1/f^2
> slope?
>
> This message was sent using the Comp.DSP web interface on
> www.DSPRelated.com

Hi elbib - google on "flicker noise" then decide whether you want to assume 
that it's dominant in your model.  Personally, I don't think it's safe to 
assume anything about phase noise - best to measure it.
Best of luck - Mike