[SI-LIST] Comments on fluorescent light noise
- From: "Mike Williams" <mike.williams@xxxxxxxxxxxxxxxxxx>
- To: <si-list@xxxxxxxxxxxxx>
- Date: Thu, 27 Sep 2007 13:35:00 -0400
Comments below:
> To: 'Mike Williams'
> Subject: FW: Digest V7 #294
>
> I distinctly remember you talking about this effect many years ago...
>
> Msg: #12 in digest
> Date: Tue, 11 Sep 2007 23:29:49 -0700
> From: Doug Smith <doug@xxxxxxxxxx>
> Subject: [SI-LIST] Fluorescent light noise
>
> Hi Everyone,
>
> For years the electronic fluorescent light over the table I
> do many of
> my experiments has been putting "fuzz" in some scope
> measurements. So,
> I decided to measure what was going on. This month's Technical Tidbit
> on my site is:
>
> Abstract: Electronic fluorescent lights are capable of generating
> significant EMI. The nature of the EMI resulting from internal
> switching used by such a light is presented. The amplitudes measured
> are significant and could affect high impedance analog circuits.
>
> The results are surprising in the amplitude observed at 48 kHz. Go to
> http://emcesd.com . The picture of the scope probe and fluorescent
> light is the link to the article.
>
> Doug
> --
As there is every chance the best practices in use today have gone past my
personal point of reference, let me define what that point is. Several
thousand years ago, I worked principally in what I characterized as "clock
and timing engineering". To me, that had to do with collecting, creating and
enforcing various best practices of managing the tolerancing on widely
distributed high-precision clocks and in further instilling these practices,
as well as rooting out self-defeating beliefs/behaviors/practices in the
teams and organizations that applied them in their work.
My own style in these encounters was to employ a flavor of "informed
design". This often meant for IC manufacturer (ICM) sorts of clients working
with them to provide more usable measurement data to their consumers. And
for systems design (SD) sorts of clients, working to show productive ways of
addressing the mandatory gap between the Utopian data the ICMs knew how to
create and the more situation-specific data the SDs needed. One of the
larger flies in the ointment was that neither side of the equation had very
well fleshed out belief systems about the realities of high-precision
measurement. As a long-time supplier of high-precision time-interval
measurement tools, and as a formerly active consultant that had to deal with
"we still can't do it" sorts of engagements, I can say that we continue to
list The Operator as the single largest source of error in any of the sorts
of measurements that would get onto our/my screen. That, is... THE WAY the
operator uses the gear/tools to measure properties of waveform timing
results in an error that, for the most stable signals being measured by the
best possible combinations of tools, dominates the results he/she gets. The
error can be orders of magnitude larger than the parameter being measured.
Once I had come up with a way to measure cycle-cycle, moment-moment timing
variation, asymptotic non-operator/environmental components of measurement
error shifted from the high hundreds of psec to ones of psec. It became
important as a matter of competitiveness to our clients/customers to try and
get down to the asymptotic capabilities of the systems we were supplying.
Initially, more than a third of our installations were into screen rooms. If
you go back to a then widely circulated reference design I did for the
P5(?)... long time ago... there was less than no room for realistic jitter
in them... and consequently... even less room for measurement error on that
jitter. It was at that time (mid-90's) that I became extremely interested in
the environmental and operator impacts on the quality of the measurements
results. At the time, I enjoyed a close and active relationship with
several divisions of HP that made various instruments pertinent to my field
of work and it was a simple matter to not only examine the phenomena but to
speak with the designers who created the instruments I was interested in to
go beyond what the data sheets said.
That process resulted in a progressively refined set of Measurement Cultural
Practices that became part of the value added for the systems we sold then.
They were included in the training at system delivery, and I think more
impactfully, I employed them myself when, after the training I used those
same principles along with an ultimately well-practiced eye to conduct site
audits. Not every customer had the same uptake on the seriousness of the
matter, and that seemed principally driven by the precision to which they
aspired their devices or systems operate.
So... as to the nature of some of the problems that were uncovered in that
work and addressed by the Measurement Cultural Practices.. Certainly the
neon lighting you found in your posting was a part of it, but there were
literally dozens of others as well. My own taxonomy broke them down into
environmental (electrical [radiated and conducted] and otherwise) and
operator practices. The power environment of the lab was probably the chief
worst offender and my own experience from several dozens of site audits
would place typical electrical/radiated impact as one might find with the
lighting in the lower tier. Power conditioners were the absolute enemy of
high-precision measurement. Counter-intuitive but that's what I found at
several sites. Said another way, I never encountered a power conditioners
that didn't hose up the measurement precision. Maybe they're better today.
Gear that was capable of low 1's of psec repeatability would demonstrate
significant sways in terms of the mean values of certain measured parameters
(not the variation, but the means of the parameters) that tracked perfectly
the change in the air temperature. One of the reasons I was a HUGE Denali
fan when it came out was that "spoke temperature", and I had built into M1
(our scope analysis product family) the ability to track and I think display
that, as well as reporting to the operator on long time-scale sorts of
measurements when the temp would vary by more than a certain amount. The
reason being, the impact it had on the long term means. Actually, to this
day, the time spent in labs getting to the bottom of why the world wasn't
working for them is a lingering ability to nail the temperature swing over
time within a degree or two. Yet another useless ability:)
There were lots of other things that also fed into the errors... Where the
DUT was relative to the instruments (I once had to cross the US *just* to
move the DUT to the other side of the scope merely to get him his numbers).
How the boxes "talked to each other".. for example, was a box more emissive
out a certain surface? Was a box more susceptible through a certain surface?
Did these vulnerabilities point to a better way to orient the instruments
during measurement? Was Channel One more susceptible than Channel Four on
the scope to measurable influence from the CRT. The closest channel always
performed worse on picosecond-level measurements than those that were
farther.
Am I in Japan or the US? Yup... that was the biggest and most consistent
difference I ever encountered, and looking into it carefully showed it was a
larger effect than anyone seemed to be aware. Am I on a lab 40 floors up or
am I close to ground level? Well these last two, I am told by friends at T&M
companies has been addressed in modern equipment though I haven't verified
that for myself. I trust that it has.
This is just a bunch of iceberg tips.. I'll stop the list here and note that
from my own personal work in this area, the number of sources of measurement
"influence" was, when this was more a part of my average day, poorly
understood by both customers and solutions providers.. the need to make psec
level timing measurements just wasn't part of their experience base. When
our sales went too high to work with each customer as we could at the start,
we had to cease including site audits with a system purchase though I did/do
continue to help long-time "good guy" customers who asked for it after that.
My POV as the guy from ASA that was working with all our precision-oriented
clients was that we work in an industry that is more human than technical,
and thus ruled more by political and personal proclivities than the maniacal
pursuit of correctness (The Great Rj/Dj Debacle being more than compelling
evidence of the validity of this perspective, in my own view). We hold, and
when we're in labs and have the opportunity to observe, that a missing
awareness in the vast majority of operators of fundamental measurement
cultural practices and environmentals makes it pretty common for measurement
results to differ significantly from the asymptotic performance of the tools
they are using. They are, in a sense, turning their $150k scope into a $35k
scope.
Given the governing dynamics that drive how engineering (including matters
that are of importance to signal integrity engineers) is being practiced in
the current age.. committees blessing the "idea" that 15 different answers
to a "measurement" are all "right", or the incessant push to do more with
fewer/lesser engineers, etc... I really don't see issues like this becoming
important enough to address because there will never be a critical mass of
people who care, or at some point, will even know they ought to care. I know
our products actually have INCREDIBLE abilities around this subject buried
inside of them to help address many of these issues. Other than in very
rarely encountered scenarios, we don't see THAT flavor of thinking
(engineering excellence) driving customer behavior. So we don't bother to
talk very much about those things anymore. Honestly, the features that
dominate salable T&M product are more things like "ease of use",
productivity, convenience etc. In interactions with our own customers
(obviously very smart guys to our way of thinking:), we rarely see the
engineering excellence features driving their ACTIONS. Sure, you always see
intellectual curiosity about it but their own personal realities in an
industry whose dynamics places many other things above engineering
excellence, it's product content like "one button reporting" or the fact the
tools automatically set up the right measurements for you or the work-flow
speedup stuff or the fact that all the compliance tests are included for
free that drive their ACTUAL behaviors. It's almost ALWAYS the product
engineering features that appeal to the "non-experts".. or overwhelmed
engineers... or the lab manager getting nickel and dime'd by his own
management that creates the bulk of our sales... not the clever neat stuff
that represented more high-concept measurement science matters. Personally,
having been rather passionate about my engineering field at one time, it's
disappointing to a degree that our products are known more for their
contribution to "the masses" than to the science or the early adopters, even
though that has was our focus for a very long time.
Sorry to go beyond commenting on just the lights. I have an insurmountable
difficulty in separating self-defeating industry governing dynamics from
purely technical matters. Goodonya for going there though. Know that there's
many more issues present than just the influence of the lights if you're
going to working with psec-level precision measurements. Back to
work..........
Mike Williams
Pres.
ASA Corp.
www.TheJitterSolution.com
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field
or to administer your membership from a web page, go to:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List technical documents are available at:
http://www.si-list.net
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
- » [SI-LIST] Comments on fluorescent light noise
