[Elecraft] Mods et al

Sat Nov 9 18:22:00 2002

It is a hard, hard  fact of life, and no doubt a corollary of 
Murphy's Laws, that:

  Important design features or errors of any complex system -hardware 
and / or software - will only be understood after it has gone into 
production.

Hence there are two, and only two, possibilities:

1] the board [or software] looks pristine and un-hacked - no dangling 
blue wires or cut traces - because the vendor couldn't or  wouldn't 
be bothered to  devise   changes to address the errors or missing 
features  and write up the Engineering Change Orders [ECOs] to 
implement them , OR

2] the board or software  has "blue wires"  including cut traces, 
tacked-on components, etc or the software equivalent ["patches" etc] 
, meaning that the product is as good as humans can make it.

The high cost of the third conceivable possibility - a new board 
design or its software counterpart - a new release  for each error - 
rules it out. This  means that the ECOs have to be collected over a 
long period of time - years to forever  - before they can be 
superseded   by a new board [Rev.B] or  software release  containing 
all the fixes, plus expensive new documentation .

Kevin mentions the IBM 360 - the  "Three-Sickly" to its detractors. 
In fact, the entire floating point arithmetic unit of the 360 Model 
91 supercomputer had to be hacked in the field,  as the basic 
algorithms used to do floating point arithmetic were WRONG. Prof. V. 
Kahan of UC Berkeley made his reputation in part by coming up with 
correct fixes to the algorithms in a big hurry. Had he failed, the 
Model 91 might have had to be withdrawn from the market, at about $10 
Million 1968 dollars a pop, plus lawsuit settlements of course.

the other eric

Message: 2
Date: Sat, 09 Nov 2002 16:55:22 +1100
From: "Kevin B. G. Luxford" <[email protected]>
To: Don Wilhelm <[email protected]>
Cc: Vic Rosenthal <[email protected]>, [email protected]
Subject: Re: [Elecraft] Mods  et al

This was a technique used by IBM on their 1800 process control computers
and I guess their 360 range as well.  Every month there would be "sales
changes" or "engineering changes" applied somewhere in the computer.
  From memory, four layer mother boards were used.  Some of the
connections to ICs were via traces within the boards, others were by
wire wrap.  Some changes involved re-routing wirewrap, others involved
additional components being added, others involved removing all wire
wrap from a matrix pin, fitting a tool over the pin and using it to cut
one or more layers below.  Layers that were cut and had to be
re-installed were generally done by adding a wire wrap connection
between two matrix pins that were on either side of the cut trace.

Normal wirewrap used yellow wire.  "Personality" wiring that determined
something unique to this computer were done in purple.  IBM boards
generally looked "professional" from the component side, but a bit of a
rat's nest from the wire wrap side.  As a customer I did not care how
the wiring looked.  I was just grateful that we were dealing with a
company that sent out "engineering change" kits all over the world when
a problem was discovered somewhere, the problem researched and a fix
found.  We had 140 men's limbs and lives hanging on the end of this
computer and regarding the "look" of the circuit boards, well frankly,
my dear, I didn't give a damn.

73
Kevin
VK3DAP

-- 
Eric G Manning  P.Eng, ISP, F.IEEE
Nortel / NewMIC Professor of Network Performance &
New Media Innovation Centre Scientist
University of Victoria

"I hear that if you play the NT 4.0 CD backwards, you get a Satanic message!"
  "That's nothing. If you play it forward, it installs NT 4.0!"

--- StripMime Report -- processed MIME parts ---
multipart/alternative
  text/plain (text body -- kept)
  text/html
The reason this message is shown is because the post was in HTML
or had an attachment. Attachments are not allowed.
Please post in Plain-Text only.---