[time-nuts] Re: Types of noise (was: Phase Station 53100A Questions)

[Joseph Gwinn]

On Tue, 15 Feb 2022 03:30:31 -0500, time-nuts-request at lists.febo.com 
wrote:
time-nuts Digest, Vol 214, Issue 17

> Date: Mon, 14 Feb 2022 22:10:10 +0100
> From: Attila Kinali <attila at kinali.ch>
> Subject: [time-nuts] Re: Types of noise (was: Phase Station 53100A
> 	Questions)
> To: Discussion of precise time and frequency measurement
> 	<time-nuts at lists.febo.com>
> Message-ID: <20220214221010.7d39f07da972b13bc6900de7 at kinali.ch>
> Content-Type: text/plain; charset=UTF-8
> 
> On Sun, 13 Feb 2022 17:31:11 -0500
> Joseph Gwinn <joegwinn at comcast.net> wrote:
> 
>> A better word than multiplicative is parametric, the varying 
>> parameters being path loss and path group delay.  This is as seen at 
>> the phase noise test set.
> 
> That is one way how noise can enter a system. Not the only one though.
> And once it is in the system, non-linear elements can increase it. 
> Look at the derivations in [1] (and to a lesser extend [2]), which
> shows how the non-linearity of an element leads to up and down conversion
> of noise and thus turns additive noise into multiplicative noise.
> Unfortunately, whatever measurement system you use, will have non-linear
> elements in it. Starting from the input amplifier and going to the
> sampling system. So this kind of noise amplification is unavoidable.
> (Yes, I know that I am blowing my own horn, but these are the only
> publications that I am aware of that explain this phenomena at all.
> Yes, I find this odd too.)

Yes, and I read these with great interest when they came out.


>>> Amplitude and phase noise are looking at noise from two different
>>> perspective. One is how large the variation of the peak of a sine
>>> wave is, the other is how much the zero crossing varies in time.
>>> Note that all natural noise sources will be both amplitude and
>>> phase noise. 
>> 
>> Hmm.  One case I'm interested in is where the path attenuation varies 
>> according to a random telegraph waveform, due to for instance a loose 
>> connector or cracked center conductor rattling under heavy 
>> vibration.  
> 
> That's an interesting noise model. And one that is oddly specific.
> Did you see that in some application? If yes, could you explain a bit more?

When integrating large systems, people have a lot of trouble with 
loose and/or broken coax connectors, especially under vibration, or 
when a cable is moved for some unrelated reason.  I'm looking for a 
simple way to detect from the outside if this is happening, without 
disassembling everything (and introducing added problems).  Nor is 
the problem necessarily obvious to the eye, even if it is all 
disassembled.

As discussed herein, I suspect that looking at AM (Amplitude 
Modulation) PN (Phase Noise) and comparing it with PM (Phase 
Modulation) PN will flag such problems.

I've also been reading the 200-page 53100A User Manual (dated 3 March 
2021), and it turns out that the measurement hardware does collect 
amplitude difference data, one assumes in a manner paralleling the 
collection of phase difference data (page 38 therein).  The 
strip-chart format is available for phase difference data, and an 
amplitude strip chart could allow one to see rattling directly.  This 
graph type is not currently implemented, but at least the data may be 
available in the TIM file.

 
>> In this, the electrical length does not change.
> 
> I would suggest to be careful with this assumption. 1ps is equivalent
> to 140µm to 200µm of distance in a cable/connector. And with phase
> noise, we are looking at effects that have the equivalent time-deviation
> of a few fs or a few 100nm of distance. Even if the contacts do not 
> move much,
> they still move. And that will have all kinds of effects (including change
> of capacitive coupling and thus impedance and thus reflection).
> I would suggest you do some back of the envelope calculation to see
> whether that movement would have some measurable effect on your system
> or not.

Yes, I think that the cracked center conductor, a common fault, is 
most simply modeled as a tiny series capacitor (perhaps coaxial) that 
may be shorted (when no break), with 50ohm lines on either side, so 
there may be very little length change.

But my instinct is that the path loss change effect will generally be 
far larger, so we may still have a large excess of AM over PM when 
something is rattling.  When I'm in the lab, I'll test the issue 
directly, equipment permitting.


>>  While 
>> the source of the carrier whose PN is being measured will have some 
>> mixture of AM and PM characteristic of that source, the residual 
>> (added) PN will be characteristic of the transit damage encountered 
>> between source and PN test set.  So wouldn't this randomly varying 
>> attenuation yield mostly residual AM PN and little residual PM PN?  
> 
> What is the PN here? Phase noise? If so, then there is no AM PN.

Well, Rubiola uses AM PN and PM PN, and TimeLab says Phase Noise and 
AM Noise, and others use other names.  Plain PN was always a 
misnomer, as there was always an AM component, even though PM 
generally dominated.  Except in laser systems, where AM PN is also 
known as RIN (Relative Intensity Noise), and PM PN was basically 
unmeasurable.  I like the AM PN and PM PN nomenclature because it is 
precise and symmetric, from DC to daylight.  


> Amplitude modulation modulates the amplitude. The zero crossing
> does not change. Thus there is no phase noise. Phase noise comes 
> from phase modulation (or equivalently time modulation).

Yes.

 
> Sure, there are elements that turn AM into PM and vice versa, but
> that is a different topic and I don't want to complicate the
> discussion more than necessary. Though with the on/off modulation
> you have, that might be something you need to look into.

I was talking about the effect of AGC on AM PN in the 
rattling-connection case.  My point was that AGC didn't help there.


> I am not quite sure what kind of system you are trying to measure
> and what you hope to see there. So I'm a bit careful in suggesting
> things until know better what your setup is.

See example at beginning, from radar integration.  My focus is 
transit damage, independent of source PN properties and issues.


Joe Gwinn


> 
> [1] "A Physical Sine-to-Square Converter Noise Model"
> http://people.mpi-inf.mpg.de/~adogan/pubs/IFCS2018_comparator_noise.pdf
> 
> [2] "A Fresh Look at the Design of Low Jitter Hard Limiters"
> http://people.mpi-inf.mpg.de/~adogan/pubs/IFCS2019_collins_isf.pdf
> -- 
> The driving force behind research is the question: "Why?"
> There are things we don't understand and things we always 
> wonder about. And that's why we do research.
> 		-- Kobayashi Makoto

Proceeded by something to the effect of "That's odd", not Eureka.