ANDY BALL wrote to KEN HRYNCHUK on 10-13-04 02:34:

  KH> Well, Jean, I tried this (approx. 20 foot) cable...

 AB> 6m is probably a bit long for a parallel cable.  I am surprised that
 AB> this worked at all, let alone in byte mode. For that kind of distance
 AB> it's probably best to use serial cables and nice, fast UART chips.

Do you know what the recommended maximum length is? This cable works
just fine in 4-bit mode.

  KH> ...I also tried setting the LPT ports on both machines
    > to 'standard parallel' mode.

 AB> Standard parallel ports are unidirectional (except for a few status
 AB> input signals), which is why things like PLIP and LapLink have to
 AB> transfer data a nybble at a time.  EPP and ECP ports are both
 AB> bidirectional.  ECP ports apparently can be DMA driven, given suitable
 AB> drivers.

The 8-bit cable configuration that I tried did, indeed, use status ports
for the second nybble (it threw me off a bit, when I first looked at
it). I did manage to dig up an old copy of LapLink Pro 4.0, and it
claims to be able to do 8-bit transfers (of course, it has no cable
documentation, though...).

  KH> I've yet to try it with 'real' legacy ports, though;
    > maybe that would solve the mystery (?).

 AB> I suspect that older parallel ports used more powerful line driver
 AB> chips. Todays are probably built into the chipset (making the
 AB> implications of harming your parallel port more severe).  I recently
 AB> picked up some parallel port expansion cards for experimentation,
 AB> because I don't want to risk the one built into the mainboard.

Thanks for the ideas, Andy. I've got a couple of old LPT cards, here. If
I decide to try another cable configuration, I think I'll put them to
use.

   Ken

 
--- MultiMail/MS-DOS v0.46