UPS/SPS testing re: AC protective ground (and RS-232)

[Thad P Floryan]

[apologies if this appears twice; PORTAL has a propensity to crash and/or hang
 precisely at the moment I post an article thus giving no "success" indication]

adams at ucunix.san.uc.EDU (James Warner Adams)
in <9010282317.AA08682 at ucunix.san.uc.edu> writes in response to my comments
about RS-232 pin 1 grounding:

	``Pin 7 should NEVER be directly connected to pin 1.  This is a hallmark
	of poor design.  ... ''

I've never felt "comfortable" about that practice either, but I took my cue
from AT&T documents.  Been selling from 10 to 50 of (my product) every month
since early 1983, mostly to phone companies and US Gov't agencies.  And the
product easily qualified FCC and VDE certification because I used the
techniques in "Digital Design for Interference Specification" (by The Keenan
Corp.)   The product now uses (among other chips) the MC68681 dual-UART, but is
so clean electrically the plastic case (PacTec) did not require NiCr paint or
other shielding.  And pin 1 is internally connected to pin 7 on that product.
You figure.  Quien sabe?  :-)

	``[... other good material deleted here ...]

	Anecdotal evidence aside, good practice dictates that
	a single ground reference point be used when cabling.''

VERY TRUE!  I learned that early in the game while designing low-noise
microwave pre-amps for "government" applications.

	``When installing shielded cables, the shield is generally connected
	BOTH to the hood AND to pin 1.  There should not be an independent pin
	1 wire running inside the shield.  This defeats the whole purpose of
	the shield and invites interference.''

I'll take your word on that (above), but it was my impression a "shield" is an
electrostatic barrier whereas the "pin 1 ground" is simply a protective drain.

	``Thad, did it ever occur to you that perhaps your "site" is one REASON
	for the awful interference you mention?''

Yep, and the reason for my noisy site (8 computers, 6 modems, and misc. other
stuff) is that I literally had to remove the RF shielding from one computer
after an upgrade ... the shield would have electrically shorted out the new
daughterboards I installed; this problem should soon be cured since yet another
modification will permit the vertical clearance I need for re-installation of
the shields.  I've been harping on that manufacturer for over a year now and I
should have new daughterboards "Real Soon Now."

Fortunately, the site is located in a sparsely settled area and I've verified
non-interference with neighbors' TVs, radios, pagers, cellular phones, and
other equipment; beyond 30' laterally or vertically (yes, I climbed my TV tower
as part of the test) the RFI is subdued.

For what it's worth, the "offending" system is 68020-based.  Encircling the
"site" are StarLAN, Ethernet, and RS-232 "networks" stretching out 150', and
the RS-232 operates just fine at 38,400 baud into two systems and 19,200 baud
into all the others.

	``Remember, data processing and communications equipment can
	radiate RFI as well as being affected by it.''

True.  And it's NOT my intention to make light of anyone's studied comments
regarding shielding and grounding.  I hope these recent discussions have
stimulated thought and triggered re-evaluation of everyone's site(s).

Thad Floryan [ thad at cup.portal.com (OR) ..!sun!portal!cup.portal.com!thad ]