On Thu, 06 May 2004 21:53:11 -0500, in <alt.comp.periphs.mainboard.asus>,
[snip]
It took me a while to find a doc that even remotely addressed
the motherboard grounding issue. This one hints at it:
http://www.casereview.org/wtx/WTX-Spec-11.pdf (pg.33)
*****
5.2.1 Motherboard Grounding
The motherboard needs to tie into the chassis with the lowest
electrical impedance possible. Motherboard mounting features must
provide good ground paths into the adapter plate; this is the
responsibility of the board designer. [snip]
My interpretation of this, is "EMI trumps single point grounding".
Better to have a little extra noise in the sound output, than having
your computer function as a radio transmitter.
[snip]
Well, maybe. But save for the unfortunate (and perhaps unintentional or
misguided?) use of the plural "paths", I see nothing which conflicts with the
ideal that the mobo itself should provide a proper ground plane with a
low-impedance path to *a* ground point. Given that scenario, both EMI/RFI and
ground loops are properly dealt with.
Besides, ground loops can (and often will) lead to a *lot* more than just
"extra noise in the sound output". All manner of flaky, seemingly random,
difficult to diagnose misbehaviors can (and often do) have their roots in
marginal or improper grounding.
5.2.2 I/O Shield
The I/O shield for the motherboard connectors is one of the most
critical aspects of limiting EMI. Poorly grounded connectors will
allow noise to escape on their associated cables. Also, weak
connector grounding can make a motherboard much more susceptible
to functional errors or damage from electrostatic discharge (ESD).
The I/O shield should make secure contact to the metal connector
faces at multiple locations, with spacing between contacts not to
exceed 0.250 inches [6.35mm] at both the bottom and top of each
connector D-shell.
[snip]
Again, this all about shielding, as opposed to grounding per se. I don't mean
to imply that proper shielding should be ignored; but it doesn't trump proper
grounding (in fact, it can't exist without it).
Howard Johnston's web site has an article on single point grounding,
"moats and float" and the like. But, the PC already has a ground
path through the ground wire of the PS, and every connector on the
back of the computer is a secondary (and unacceptable to that theory)
path.
[snip]
Yes and no.
Yes, ideally, the shield in each cable would only be connected to *one* of the
shells (and preferably, all on the same end -- typically the one which
connects to the main system unit).
But the other issue in play here (and part of the reason the previous
paragraph is often rendered moot) is that, at least very often, the "ground"
used by those I/O cables is *not* chassis ground in the first place; instead,
it is a derived ground which, to one degree or another, is permitted to float
WRT to true chassis (or Earth) ground. This works because, for shielding
purposes, that derived ground is the only one relevant to the signal(s) being
"protected". Another reason this is done is because the designers cannot
count on both "their" device and whatever is at the other end of the cable
being connected to the same Earth ground.
Now, *IF* that derived ground is itself "dirty" (i.e., excessively
noise-ridden, WRT Earth ground), then EMI/RFI radiation might rear it's ugly
head; but typically, the currents involved are far too small for this to be a
serious concern. And besides, this concern still applies to situations where
the cable's shield is connected to both shells.
As a consequence, I think the "ground flood" theory is being
invoked, where an attempt is being made to bring everything to a
common ground potential, with the majority of DC ground currents
being carried inside the motherboard itself. As this is a compromise
between analog noise performance (ground loops) and meeting FCC, and
the former has no legal ramifications and the latter does, it is \
easy to see which issue wins.
[snip]
Well, that depends (in part) on your POV. In terms of "legalities", there are
two things you need to keep in mind:
A. As far as the FCC is concerned, limiting EMI/RFI radiation not only
"trumps" such petty concerns as "extra noise in the sound output", it also
trumps any concern about your computer actually working correctly or being in
any way useful. IOW, making serious, even crippling, compromises to other
aspects of the system's design integrity is perfectly acceptable, if it
accomplishes the sole goal the bureaucracy is interested in.
B. As a practical matter, FCC compliance is only a concern to manufacturers of
off-the-shelf electronic products, who are required to submit their prototypes
to rigorous lab testing before they can affix that little "This product
complies with Part 15 <blah> <blah> <blah>" sticker (which in turn is required
to be there on certain classes of products sold to the general public). The
home-brewer or small-time systems integrator simply will not appear on their
RADAR (at least, unless you somehow manage to create such a *huge* RF problem
that one of your neighbors complains about your screwing up his TV reception,
or similar -- highly unlikely, to say the least <~>).
One last comment: While I may disagree (at least in detail) with some of your
contentions, I do appreciate that you are attempting to take a rational and
logical approach to these issues, as opposed to the reductio ad absurdum
generalizations and ad hominem pot shots found in the other f'up to my earlier
article. Thank you for that.
--
Jay T. Blocksom
--------------------------------
Appropriate Technology, Inc.
usenet01[at]appropriate-tech.net
"They that can give up essential liberty to obtain a little temporary
safety deserve neither liberty nor safety."
-- Benjamin Franklin, Historical Review of Pennsylvania, 1759.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
Unsolicited advertising sent to this E-Mail address is expressly prohibited
under USC Title 47, Section 227. Violators are subject to charge of up to
$1,500 per incident or treble actual costs, whichever is greater.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
)