#: 19742 S12/OS9/68000 (OSK)
    20-Feb-94  21:40:44
Sb: #19709-#Printer Problems
Fm: Carl Kreider 71076,76
To: Bob van der Poel 76510,2203 (X)

Bob,

Your printer problems may be an artifact of the 'better' parallel chip, the
68230.  It has a pipeline in it.  Instead of interrupting you when it sees an
ack, it transfers the char in the FIFO on the ack and interrupts you to refill
the FIFO.  This means that data will get lost because a cable beyond a certain
length will cause false acks by coupling back the stb signal or data.  In
general, it drops at least every other character.  I have never seen the
problem on an MM/1 (mainly because I don't normally use one) but I really
tripped over it on the VME crate.  It took a couple of days fooling with
drivers and another day with a 'scope and analyzer on the port before I saw
what was going on.  I then tried terminators and drivers and etc but to no
avail.  I finally wrote a driver that fixes the problem.  It even works on a 30
foot cable!  I really doubt that the trouble you and other MM/1 owners have
seen is due to wimpy cable drive.  I am sure the problems are due to that I
have just described.  My new dirver has been used on an MM/1 and did fix the
problem there.  The only problem is that I only put in support for one of the
two ports since that was all that was available on the Force CPU boards.  I
have an MM/1 again for awhile and I intend to get it running on the MM/1 on
both sides of the port soon.

BTW, the slave CPU's in the VME crate use NFM to get to the master CPU to find
spl, which uses NFM to get back to the slave to find a parallel port.  So your
idea has been tested and works well.

Finally, I suppose I ought to just  give the sh docs I have to somone to clean
up.  I am not getting to it very quickly.

Carl

There is 1 Reply.