13 Jun
2006
13 Jun
'06
10:07 a.m.
At 08:08 AM 6/13/06 -0700, you wrote:
On 6/13/2006 at 12:22 PM Jules Richardson wrote:
That's not quite the way that the chain printers that I worked on worked.
The ODEC printers feed the chain from the left and the characters where in
ASCII order. The data for the entire line to be printed was feed into a
buffer and whenever a character on the chain rotated to a position where it
matched the character to be printed then the hammer for that position
fired. The result was the the hammer firing was very erratic and the
printer made a hell of a noise. But it could print an entire line in one
pass of a character set even if you printed nothing but the sme character
in every print position. Of course if you printed somehting like "GFEDCBA"
then you got a lot of matches and a lot of hammers firing at the same
instant! Situations like that were another thing that made these printers
so noisy, there was no rythum at all to the way that the hammers fired.
Each chain usually had 2 or 3 complete sets of characters. Worst case for
one of these printers would be where you printed the same character in
every position.
FWIW print speed was VERY different depending on the width of the paper,
order of characters on the printed line, the maximum number of occurances
of the same character in a line and a host of other factors.
Joe
If not,
Topic came up on a local newsgroup - anyone know
what the typical speed
of a
chain printer was/is? I can't find any hard figures online - all the info
about chain printers seems to just say they were "high speed" and that's
it.
It depended entirely on what was being printed. There were worst-and-best
case patterns. Consider the way a chain printer worked.
You have a spinning chain, a wide ribbon, the paper and (usually about 132)
hammers all making a sandwich. When a character in the chain is at the
right position, the hammer at that position fires and bangs the paper
against the ribbon backed by the type slug. Now, if the pattern being
printed matches the order of characters on the chain (or the non-matching
characters are spaces), you're all done and the paper advances. then the paper has to wait until the proper characters
come around to their
respective positions.
Clearly, the number of different characters in the chain is going to affect
how many times they can be repeated in the chain; a chain containing
lower-case characters in addition to uppercase is going to, on the average,
print more slowly than a chain with just uppercase. Similarly, the
manufacturer may have decided that some characters are so rarely used that
they merit only one or two occurrences, while more frequently-used
characters merit more frequent repetitions. Ordering of the characters on
the chain also matters; placing a "U" next to a "Q" is a good idea.
Most manufacturers give a number based on the average printing speed of a
printer and 1500 LPM was about par for the high-speed models with "easy"
lines going considerably faster.. But with the wrong type chain/train and
pathological data, speed can drop as low as 60 LPM or so. Other factors
could reduce printing speed, such as carriage control--when we printed
documentation, we bolded characters by overstriking, essentially forcing
the printer to print the same line 2 or 3 times.
If you think about it, the chain/train printer is a perfect example of
early parallel processing, although it was rarely stated as such.
Cheers,
Chuck