From: Eric Sosman <Eric.Sosman@sun.com>
Newsgroups: alt.folklore.computers
Message-ID: <3B740786.9007F8AF@sun.com>

Joe Morris wrote:
> 
> Eric Sosman <Eric.Sosman@sun.com> writes:
> 
> >    Apropos of nothing much, I did once hear an extraordinarily
> >impressive Big Bang -- more of a Big Crunch, actually -- when
> >somebody managed to feed a peanut butter sandwich into a 2540
> >card reader/punch in full cry.
> 
> OK, I'll bite.  (pun intended)  *How* did someone manage that stunt?

    First, a brief description of the physical layout of the
read end of a 2540, for those who never had the pleasure.  It
was a gravity-fed arrangement with two stages.  At the bottom,
closest to the point where the cards actually entered the
machine, was a smallish area with a sort of jiggler mechanism
to help get the cards lined up nicely before entering the regions
of tighter tolerances.  The number of cards in this stage had to
be kept within limits: too many and the pressure from the cards
on top would keep those on the bottom from responding adequately
to the jiggler, too few and insufficient pressure made it likely
that the cards would buckle rather than feed smoothly (face
down, nine edge first) into the body of the machine.  Acceptable
pressure was obtained if there were between about twenty-five
and maybe two hundred cards in this stage.

    Of course, two hundred cards wouldn't keep a reader busy
for very long, so there had to be a way for the operator to
load considerably more into the machine.  Above the bottom
stage was a steeply sloping metal ramp perhaps two feet
long, on which some thousands of cards could be stacked.
The bottommost of these rested on a barrier equipped with a
pair of rubber belts; when the belts were stationary the
cards just sat there, but when the 2540 detected that there
were too few cards in the lower stage it would spin the belts
for a while and feed another hundred or so cards into the
bottom section.

    You should now have a picture of a lower reservoir with
a few hundred cards being fed one by one into the main body
of the reader, with the belts replenishing the reservoir
from  the upper stack every ten or fifteen seconds.  You may
also be wondering how the last few cards in the stack got
read, since there wouldn't be enough weight above them to
hold them in position.  For this purpose there was a special
card-sized metal weight that you placed atop the deck to
provide the needed pressure; the weight kept the cards under
control and had flanges and springs and stuff so that the
reader's "do I have any more cards?" sensor wouldn't detect
it when it was the only thing left in the lower reservoir.
To start a new deck, you opened up the jiggling gate, fished
the weight out and put it atop the new batch of cards on the
upper ramp, closed the gate, and hit the START button.

    Of course (and now we get to the story), if you were
feeding a lot of cards through it became tiresome to keep
stopping and restarting the reader every time you needed
to retrieve the weight.  But even a moderately experienced
operator knew how to solve that one: you didn't bother with
the weight at all until the very end of the run, you just
kept an eye on the upper ramp and tossed another box' worth
of cards on whenever it looked like it was getting low.

    That's the approach our guy Andy was taking one day when
he was running three or four trays' worth of cards through
the machine: pile up a few thousand cards, omit the weight,
and make sure to throw another few thousand on every now and
then.  It was lunchtime (obviously) and he happened to be
munching on a peanut-butter sandwich when somebody asked him
to lend a hand with something else.  Why he chose to put the
half-eaten sandwich atop the cards on the input ramp I'll
never know, but he did -- presumably he figured that it'd
take several minutes for the reader to work through that
many cards, and he'd be back in plenty of time.

    Well, the job took a bit longer than he'd expected, and
at some point he noticed that the input stack was getting
rather short.  He knew he didn't want to let it get down
too far without the weight in place, so he hurriedly
grabbed some more cards and loaded them onto the ramp --
and since his other task had put him on the "wrong" side
of the machine, he just reached over the ramp from the
back to do this, and didn't observe the fatal sandwich.
Two or three thousand new cards squashed the sandwich
satisfactorily, and a minute or so later all questions of
smooth or crunchy became academic.

-- 
Eric.Sosman@sun.com