Re: [PARPORT] Scanner - Primax PagePartner

Grant R. Guenther (
Thu, 5 Feb 1998 13:45:36 -0500 (EST)

> I couldn't find anything like this on the PARPORT home page, nor does the
> SANE project mention anything. After monitoring this list for awhile it
> seems most traffic concerns storage systems, so forgive me if these
> questions are hidden in a FAQ somewhere (And if there is a FAQ, a pointer
> would be helpful).

The topic of parallel port scanners does come up here from time to time.
The bottom line is that unless technical documentation is available,
it is probably impossible to support them.

It is fairly straightforward to reverse engineer a parallel port protocol
if you have two things: a model of the application transactions and
the ability to capture traces of the parallel port i/o generated by
individual transactions.

In the case of storage devices, we generally have good models of what
is going on: most often we know that there's a SCSI or IDE device in
there somewhere. That gives us a framework for interpreting traces.
Some parallel port scanners are basically SCSI scanners with
parallel-SCSI adapters. If the adapter happens to be one that
we know, then it should be possible (with some work) to support
the scanner. Alas, I don't believe any known devices fall into this
category. It seems that many of the parallel port scanners use
ad-hoc protocols and also compress the data - in which case there
isn't a lot to go on.

Capturing traces is an easy task if the parallel port device has
a DOS driver. We can run it under DOSemu and use some built-in
features of the emulator to capture traces of specific operations.
Unfortunately, most of these devices require Win95 - and won't
work in DOSemu. Of course, if you have a good logic analyser
(capable of capturing a trace of 24 channels at a sample rate of
10MHz for 20 or 30 seconds) you could get a trace that way - but
if you can afford something like that, you'd buy the company
rather than try to crack their protocol ;-)

Grant R. Guenther

-- To unsubscribe, send mail to: --
-- with the single word "unsubscribe" in the body of the message. --

This archive was generated by hypermail 2.0b3 on Wed 30 Dec 1998 - 10:17:24 EST