From: hplabs!pyramid!utzoo!henry@ucbvax.Berkeley.EDU
Date: Mon, 22 Dec 86 18:45:20 pst
To: pyramid!CSL.SRI.COM!RISKS
Subject: Plug-compatible plugs

> Someone discovered by accident that the IBM monochrome display adapter will
> accept a Token Ring connector cable...
> - Why couldn't they have made the token ring connector a different kind than
> the monochrome display connector? Did (or should) the hardware design process
> include any analysis of its consequences in such conjunctions, given known
> human tendencies?

It does in other areas.  In avionics design, it is normally mandatory that
no two functionally-different plugs be physically identical.  This is
usually achieved by keying systems rather than by a vast inventory of
slightly-different connectors, although there are quite a variety used.

The crucial difference is that avionics systems are, to some degree, designed
around the assumption of imperfect maintenance.  The military in particular
has to contend with complex systems maintained by ill-trained technicians
subject to many distractions (e.g. gas masks, bombs falling nearby, etc.).
Unfortunately, the healthy paranoia that this induces in designers doesn't
seem to be present in the computer business.

Computer systems have been designed around the assumption of perfect
maintenance for quite a while, actually.  The cables used to connect most
disks and tapes to their controllers are physically but not logically
symmetrical, with no keying.  At least a 180-degree rotation from one end
to the other isn't generally destructive, the stuff just doesn't work!
Still worse are symmetrical female connectors which plug onto rows of pins
protruding from boards:  not only is it possible to get the connector on
the wrong way, but it is also possible to get it misaligned with the pins,
so that some pins stick past, rather than into, the connector.  The grid of
pins is regular and symmetrical -- they are normally on the 0.1-inch square
grid that is standard for all manner of electronic components -- and there
often is no housing around them to constrain the plug to fit in only one
place.  Slightly fattening the plug to prevent pins sticking past it would
solve this, but nobody seems to bother.  Even some prefabricated sockets
which *do* have outer plastic shells are roomy enough that a narrow plug
can go in misaligned by one row of pins.  (I speak from experience.)  The
D connectors used since time immemorial for RS232 lines, and increasingly
common for all manner of things on personal computers, at least lack these
flaws.

There is no great mystery about why this stupidity occurs:  it's cheap, and
nobody can be bothered improving it.  The offending connectors are available
from a wide variety of competitive sources, and are available in "mass-
terminated" forms that can simply be clamped onto flat cable without the
expensive and largely manual operation of soldering individual wires into
the connector.  A grid of pins sticking up from the board is cheaper than
a prefabricated connector.  It's cheaper to put the pins on the standard
grid than on a special one that would interfere with improper connections,
and cheaper to buy female connectors that have all holes present rather
than having one blocked off for keying.  And so forth.  Often it's possible
to get at least some degree of protection if one tries -- keyed mass-
terminated connectors do exist, for example -- but all too often suppliers
don't bother.  Even something as simple as making one socket male and the
other female offers at least slight protection against wrong hookups.

				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,decvax,pyramid}!utzoo!henry

Newsgroups: comp.risks
X-issue: 12.15
Date: Thu, 22 Aug 91 01:26:20 EDT
From: henry@zoo.toronto.edu
Subject: Re: Bell V22 Osprey crash -- assembly error

>From the Aug 5 issue of Aviation Week:

	The Navy has found an assembly error caused the fifth
	Bell-Boeing V-22 full-scale development aircraft to crash
	June 11 on its first flight...  Reversed polarity on a
	gyro-type device that provided inputs to the flight control
	system was blamed.  The assembly problem was difficult to
	detect, but it was verified as the cause in a flight
	simulator and isolated to the one aircraft...  V-22 aircraft
	should resume flying soon.

Tsk.  While this doesn't seem to have been a computer problem per se,
it does make one wonder about a design that could be mis-assembled
like that.  The military usually tries to avoid this; somebody goofed.

(To digress slightly... one of the most impressive cases of design-for-
correct-assembly I've ever seen was the inside of the Canon CX print engine
used in the HP LaserJet and other first-generation small laser printers.
We service our own LaserJets, and we've had to dig fairly deep at times.
It's complicated and messy and has a lot of connectors... no two of which
are alike.  I don't mean just little keying pins that are easily forced
or overlooked; no two of those connectors are the same *size* even.  And
this is in a unit manufactured by the millions at rock-bottom prices.)

                                         Henry Spencer at U of Toronto Zoology
                                          henry@zoo.toronto.edu   utzoo!henry

   [Also commented on by Bob Rahe <CES00661@udelvm.bitnet> and 
   Tim_Diebert.PARC@xerox.com.  PGN]

Newsgroups: comp.risks
X-issue: 12.42
Date: Sat, 28 Sep 91 18:44:45 EDT
From: henry@zoo.toronto.edu
Subject: Re: V-22 Osprey (Wodehouse, RISKS-12.41)

>consider the case in which the triple sensors are not "reverse-wired" but
 cross-wired (e.g. sensor 2 is connected to input 1 & vs). In this case, with
 "all good" everything is fine. If 3 fails all is ok. However if 1 or 2 fails,
 the other is reported failed, voted out...

Things can get even more interesting if there is more than one set of
wires to the sensors, e.g. for feedback control of some kind.  The second
Saturn V test launch had a double engine failure in the second stage that
was traced to such a problem:  one engine did indeed develop problems,
but the "shut down" command from the control computer went to the neighboring
engine instead.
                   Henry Spencer at U of Toronto Zoology          utzoo!henry