[time-nuts] Upper limit on phase noise from two oscillators. (fwd)

Wed Apr 27 22:55:35 EDT 2005

Dr. Kirkby,

[Note that this list is set up to involve everybody in a single reply.]

I found the paper at http://www.g8wrb.org/useful-stuff/Untitled-11.shtml
by searching for "phase noise" at www.g8wrb.org that was listed in your
14 April mail. Apparently the secure html wasn't.

I had no intention of suggesting that you were duplicitous in your
postings. T'were closer to an old man being crotchety about the sudden
increase of postings on this group. How silly, when each posting is worth
reading.

As it turns out, my interest in precision time was fueled by the Shortt
clock at the British Science Museum many years ago. I can still see the
exhibits but the details have gone fuzzy. Since then I have used eBay to
acquire 4 Rubidium, 1 Cesium, countless counters and time code generators,
and a 36" Invar rod. I left MIT in 1960 with a BSME, but my career was in
electronics and computer control. So why not look into the limitations of
physical oscillators like the pendulum? Now a mast rises behind the house
with two HP GPS antennas and a lightning rod, with two Z3801's. There is
a Liebert 2 KVA UPS with eight 12 volt batteries that cost twice the cost
of the Liebert (and demand respect).

But I have other work to do and the project moves slowly, as we all know.

This group appears to be concerned about performance figures such as the
Allan variance to the exclusion of physical clocks and oscillators. Still,
I have learned a lot by listening. Tom van Baak is a fount of information.
John Ackermann put the list together and maintains it, and we are richer
for it. The most amazing specialists drop by to help out.

FWIW, I've done a dozen eBay transactions with rdr-electronics and never
been disappointed. Suzees/cozees in Hawaii was another matter.

Maybe we could take a discussion of physical clocks off the list? I haven't
looked at a newsgroup since AOL started up.

Bill Hawkins

-----Original Message-----
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com]On
Behalf Of Dr. David Kirkby
Sent: Wednesday, April 27, 2005 7:48 PM
To: time-nuts at febo.com
Subject: RE: [time-nuts] Upper limit on phase noise from two
oscillators. (fwd)

On Wed, 27 Apr 2005, Bill Hawkins wrote:

> Dr. Kirkby,
>
> Mike S's query solidified my resolve to get to the bottom of this.
>
> Searching for "phase noise" at your web site brings up a paper

It is not a paper, just a Mathematica notebook, mainly for my own use.

Could you tell me where you found it, since it was put up temporarily for
Wolfram Research, the producers of Mathematica, to sort out an issue of
fonts. I would like to remove it, as it not intended to be made too
public at this point in time, but since you found it, I will explain
*briefly* about it.

However, it has *nothing* to do with why I was asking questions on here.

> that
> concerns the use of lock-in amplifiers to extract 1 microvolt signals
> from 3 volts of noise.

That is an example of what a commerical lock-in, such as the Standford
SRS830  can do.

> The setup is described as a modulated laser
> and a light receiver. Apparently, the laser is modulated with the same
> sine wave as the lock-in (phase detector) amplifier. [The same technique
> is used by the ancient Fluke 207 VLF receiver to pull WWVB out of today's
> man-made noise. The 207 also compensates for offset.]

No, this is not what your old Fluke does.

The laser is modulated at 70MHz. Light passes through human tissue and
undergoes a delay. The delay, and hence how far the light has travelled in
tissue, is measured. Whilst not the subject of my PhD, there is a
description of the process, and an analogue method of measuring this in my
PhD

http://www.medphys.ucl.ac.uk/research/borl/homepages/davek/phd/phd.html

in chapter 1, page 62, section 1.6.3.

The system you found the Mathematica notebook of (I asssume it is that,
please point me to the URL, as I can't find it myself!!) is a digital
implementation of a system that achieves the same as figure 1.16, but
dispenses with all the analogue mixers. It uses direct RF sampling of the
70MHz signal at a frequency well below 70MHz, undersampling techniques and
a digital implementation of a lock-in amplifier via a DSP chip.

That work will be the subject of a paper, as soon as I can get the system
working!! At the minute, I have problems in that the fast/expensive data
acquisition system, that is supposed to be able to sample signals to
200MHz, is far from doing that. It has been returned to the supplier.

So what you found was a Mathematica simulation of what I will eventually
implement on the DSP. The hardware is built, but the software is not.

> It seems that the sensitivity of the receiver would be affected by the
> phase noise level of the oscillator.

Since that project is designed to measure phase differences then phase
noise is semi-important, but since the data will be averaged for a few
seconds, I don't think this is a big deal. It will just use a Marconi
signal generator to produce the 70MHz - nothing too special. The system
shown in my PhD was a bit more critical, as that made measurements every
50ms or so, and had far more stringent requirments.

> Is that your reason for raising the noise level of this group? When the
> full extent of a question is not revealed, it sets off a wave of noisy
> speculation that is not far from shared ignorance, in some cases. Not
> that I haven't learned anything from all this, you understand.

NO, my interest in this group started some time back with a discussion of
quartz control of a pendulum clock. Do a search on Google in the
alt.horology newsgroup and you will find where I am coming from.

e.g.
http://groups.google.co.uk/groups?hl=en&lr=&selm=4250307d%40212.67.96.135
(I did use the email address  Dave (nospam at nowhere.com), but if you check
the IP address, you will see is came from the same as my messages to here
- 213.78.42.15 You will see my thoughts on there about Rubidium sources
and other ways to control the clock. You might find some messages with my
real email addresses, but I tend to not use them more than I have to.

I soon realised that to do this, I needed some way of accurately measuring
frequency - more accurate than any test equipment I have will allow.

I can assure you I have NOT been trying to trick anyone about my interest
in producing a decent quality oscillator. It is purely for home/amateur
use - hence I have been buying the bits from eBay.

The thing you found on the web site about the lockins is commerically
funded work, that funds my salary at the university. That uses an
expensive commerical data acquisition system.

Of course, the idea of using a lock-in to measure phase noise that I
mentioned did come about since I use lockins a lot and was considering
whether using one would allow the phase noise to be measured without
reference to a second high quality oscillator, that is at least 10x better
than the one you are trying to measure. I made it clear there that the
lock-ins were ones I had at work.

Sorry if you felt I was pulling the wool over your eyes about this, but
these are very different interests - one is professional, one is home.

I don't know where SM5BSZ has put the latest Linrad mail archieves, but if
you search on "Detecting phase modulation with undersampling" you
will find questions from me about undersampling and measuring phase with a
DSP. I made it VERY clear there that was for professional use, and NOT a
ham project, just in case someone felt I was trying to get professional
advice from hams who think my intersts are purely amateur.

It is not my intention to fool anyone. Sorry if you did not mean it that
way, but I can understand a certain sense of confusion. But they are
different. I just wanted to put the record straight.

Dr. David Kirkby.
Senior Research Fellow,
Department of Medical Physics and Bioengineering,
Univeristy College London,
Malet Place Engineering Building,
Gower St,
London,
WC1E 6DT
Tel: 020 7679 0273

_______________________________________________
time-nuts mailing list
time-nuts at febo.com
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts