Newsgroups: comp.sys.ibm.pc
Path: utzoo!utgpu!jarvis.csri.toronto.edu!csri.toronto.edu!kaarel
From: kaarel@csri.toronto.edu (Kaarel Truuvert)
Subject: Perplexing Parallel Printing Problems Puzzle Poster
Message-ID: <1989Jul7.161409.16895@jarvis.csri.toronto.edu>
Organization: University of Toronto, CSRI


An employer for whom I do consulting work is experiencing a mysterious
problem with printing things. I'll first ask a quick question which might
enable some to resolve my problems without reading the rest of this (rather
long) message. Then I'll describe the setup, the problem and finally
my suspicions and questions.

The quick question:
What differences are there with the way a real IBM PC (with the
printer port on a Hercules graphics card) and a Compaq 286 (i don't
know what sort of card the parallel port is on - it's the same one as the
9-pin serial port is on) use their parallel ports? For example,
does a PC (or the Hercules card) "send a stronger signal" than the
286 (or its card)?
Also, do different programs (e.g. the Compaq MS-DOS(3.3) "PRINT" and "COPY"
(to lpt1:) commands) send data differently? How so?

The setup:
A group of various flavours of PCs (a PC, an XT, 3 Compaq 286s, an AT clone)
all have parallel cables leading to three print buffers ("The Logical
Connection") which are daisychained together with simple twisted pair
(as per the manufacturer's specifications). One of these buffers then
sends the output via another parallel cable to a Postscript laser
printer ("QMS PS-820").

The problem:
Printing from the Compaq 286s and the AT clone produces, in some cases,
garbled output. (Printing from the PC and XT always works in any setup,
even those in which a 286 substituted for the PC results in errors.)
After playing around with all sorts of permutations of the
printing setup, the following symptoms can be reported:
1. printing via WordPerfect and the DOS "PRINT" command always works fine.
2. printing via one particular print buffer (not the one directly attached
   to the printer) always works fine.
3. printing via the DOS "COPY" command (to lpt1:) and via "Allways"
   (an add-on to Lotus 1-2-3 enabling Postscript printing) produces
   problems when connected to either of the other two print buffers.
4. The garbled data is almost exactly the same in repeated tests of the
   same PC-cable-buffer-cable-printer configuration, but different if,
   for example, we use another cable, or connect to another print
   buffer.
5. All parallel cables involved are quite long. Connecting a 286 via a short
   parallel cable to any print buffer or directly to the printer results
   in correct output. Connecting a 286 directly to the printer using the
   long cable usually used to send from the buffers to the printer
   results in garbled output (different every time).

My suspicions and questions:
It seems as if it's a problem with signal degradation through the long
parallel cables, but only sometimes. Here's a series of questions and
my guesses to the answers:
Why do the PC and XT (and I suspect the print buffer output port)
work ok? (They send "stronger signals"?)
Why does one of the 3 buffers always work ok? (The buffers are internally
different, one using a chip set that can handle weaker signals? - I did
check the model numbers - it does seem that the working buffer is a
different (later?) version.)
Why do some programs work while others don't? (They handle parallel
output differently? What is the difference?)
Why did this problem crop up now, after months of problem-free operation?
(The hot, humid summer weather has physically affected the cables?
Or maybe the air conditioning? We also replaced the print buffers
recently -- perhaps the old set were all like the one ok one now?
But the setup with the current buffers was working fine for about a
week (I think..))

Any help with solving this mystery would be greatly appreciated.
I apologize for producing such a long article, but in a case like
this, any small detail might be important and hence shoul not be omitted.

Thanks in advance to potential detectives.

Kaarel Truuvert				kaarel@csri.utoronto.ca
Sometimes a student			kaarel@csri.toronto.edu
Sometimes a consultant