RE: [SI-LIST] : What's your favorite Screwy SI Concept?

About this list Date view Thread view Subject view Author view

From: Chan, Michael (Michael.Chan@compaq.com)
Date: Fri Jan 07 2000 - 13:21:30 PST


My 2 cents is to find out how a "screwy rule" being drawn out. A
rule ( no matter it is good or bad ) has to come out from some initatives
and/or motivatives. I don't think a rule can come out from nowhere. I
believe
a rule based on solid fundamental concepts cannot be wrong by that much and
any derivative from real world observations can be corrected and/or
improved.

I believe building some test structures in order to see whether a rule will
break or not cannot help to try to understand how a rule being drawn up and
whether a rule make sense or not.

For example, take the famous " 20H RULE " as a case. Can anyone, especially
the originator(s) ( if there is some ) of the 20H Rule explains the physical

concepts behind how this rule is being drawn up?

Michael Chan

-----Original Message-----
From: Michael Vrbanac [mailto:vrbanacm@swbell.net]
Sent: Friday, January 07, 2000 2:35 PM
To: si-list@silab.eng.sun.com
Subject: Re: [SI-LIST] : What's your favorite Screwy SI Concept?

DougS,

Very good response. I don't always trust them either. One really
needs to know exactly how the simulator performs and utilizes data
so they know how much faith to put into the results. Otherwise,
its merely a "black box" approach and the user is gambling on the
outcome. The test board is the only way to gain an understanding
of the "real physics" because a test board, while it might not do
what you intended if you didn't plan right, it never "lies" to you
(measurement error notwithstanding.)

For the group, (a commentary and proposal)

I have watched the responses over a few days and what I have
seen has been interesting. Some responses are really humorous.
I haven't seen anybody who I thought was actually "wrong" but
mostly some who might not have it right for every case. I'm not
one to blindly accept those "mysterious global rules" either especially
those without any foundation whatsoever. I don't like "black magic"
stuff so I almost always put such "rules" to the test (without "mercy"
if you know what I mean.) Sometimes, though, it may be prudent
to step back and see if there are times when a "screwy rule" might
make some sense and what prompted its creation..

Since it was mentioned, let's take the 20H rule as an example. It is
very conceivable that many have created test boards and have
indeed seen no favorable results supporting the 20H rule. That is
a good thing. In those situations, it was obvious that it was not necessary
nor helpful. Is it then possible that it still could be of value in another
type of application? Certainly, but only if you understand the 20H rule and
use it to accomplish what it was intended to do. Not every situation
will call for it and in some (perhaps many) cases it would be superfluous.
Yes, I have seen it work but only when it was used properly in a
situation that called for it. So just as one shouldn't blindly trust a
simulator,
one shouldn't blindly use some of those "screwy rules". Even at that, we
don't stop using simulators and we sometimes end up using "screwy
rules" because they are needed. There are other good examples of
this situation but I'll save that for another time.

The problem with "screwy rules" (assuming that they are useful
somewhere) is that many times a very helpful rule for a certain type of
solution is incorrectly applied or rigidly enforced when there really is
no call for it. The first case is just ignorance and that can be cured
easily.
The second is a matter of convenience for those who are "enforcing"
it. The truth of the matter is that SI and EMC along with other product
development tasks must deal with varying conditions and situations.
The sad truth is that our "implementation systems" aren't really flexible
enough to "bend" with the new requirements of every fast moving product
development cycle. So we end up making long lists of design guidelines
that by necessity must include some of those "screwy rules" and it
is totally impossible to justify the use of each one in all cases. If we
all had enough time (don't we wish!), we could filter out all the
unnecessary
rules for a particular project and be done with it. We can all dream
about that. So for practical reasons alone, I honestly don't see the
"screwy rule" problem going away anytime soon unless we do something
constructive about it.

The way I look at it... Maxwell's equations rule... but always
keep your eyes open for new ways to apply them to the situation
at hand. If you don't, someday, you might find yourself using a
"screwy rule" you've fought against (or vice versa) and have to face
your peers with the embarrassing fact that you were dead wrong.

To avoid this, I would propose then we would look at the "screwy rules"
and figure out why they exist and in what situations they would be useful
if any at all. That's a long process but isn't that what this forum is all
about?

Michael E. Vrbanac

Doug Smith wrote:

> Hi Jon,
>
> I would tend to build up a test board to try something like this. I
> learned a long time ago not to trust simulators unless I could build (or
> calculate by hand if possible) a similar but simpler test case and get
> the simulator to agree with calculations or measurement. I build a lot
> of test circuits and do measurements on them. I call that "putting the
> problem on the analog computer." Many of the test boards or other
> apparatus I build eventually end up in published papers or my courses as
> demos.
>
> I have seen cases where simulation output was believable, but not
> accurate because of some problem, either an input problem (garbage in ->
> garbage out) or a problem within the simulator itself.
>
> Given the above, does anyone know of any "hard" data to support the
> simulation?
>
> DougS (as opposed to DougM who posted the original 20H item)
>
> Jon Powell wrote:
> >
> > Doug Wrote:
> > Screwy SI Concept #2: The 20H Rule
> >
> > Why It's Screwy: Highly debatable concept and it's
> > not really an SI concept at all.
> >
> > Why It's Hung Around: Easy to remember
> >
> > Jon Writes:
> > About 6 months ago I did a test to see what was going on with this "20H"
> > rule. I simulated some simple ground planes with simple sources using a
> > 3D simulator. (I am not making any claims, just telling what I saw), the
> > results: the 20H boards quickly put all of their energy into the
> > enclosure. And the regular one retained the signal (a lot longer) Now,
> > perhaps this doesn't sound like I needed a simulation to know this but
> > this is my guess:
> >
> > 1) If you have a good enclosure, (or don't care about EMI off the board)
> > th 20H rule can reduce noise on the board. (And perhaps EMI off the
> > connected cables?)
> > 2) If you don't have an enclosure, and you care about EMI off the board,
> > the 20H rule may not be a good thing.
> >
> > --
> > Jon Powell
> > Director of HSSD Consulting Services
> > Viewlogic Systems, INC.
> > 805 988 8250
> >
> > **** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list, for more help, put HELP.
> > si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
> > ****
>
> --
> --------------------------------------------------------------------
> ___ _ Doug Smith
> \ / ) Manager EMC Development & Test
> ========= Auspex Systems
> _ / \ / \ _ 2300 Central Expressway
> / /\ \ ] / /\ \ Santa Clara, CA 95050-2516
> | q-----( ) | o | Phone/FAX: 408-566-2157/2020
> \ _ / ] \ _ / Email: dsmith@auspex.com
> --------------------------------------------------------------------
>
> **** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list, for more help, put HELP.
> si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
> ****

**** To unsubscribe from si-list: send e-mail to
majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE
si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
****

**** To unsubscribe from si-list: send e-mail to majordomo@silab.eng.sun.com. In the BODY of message put: UNSUBSCRIBE si-list, for more help, put HELP.
si-list archives are accessible at http://www.qsl.net/wb6tpu/si-list
****


About this list Date view Thread view Subject view Author view

This archive was generated by hypermail 2b29 : Thu Apr 20 2000 - 11:34:29 PDT