25 Jul
2001
25 Jul
'01
11:37 a.m.
One item you mention certainly supports the notion that this board was used in a
multi-drop scheme of some sort, and that's that the termination resistors were
not populated. Only one station at each end of a transmission line should carry
the termination resistors, hence, it's likely they'd be external to the board.
I'm puzzled at the use of the 6850, however, since it's only async-capable. So
far, however, and maybe until you find out quite a bit more about this setup,
it's anybody's guess.
Since this setup is async, why would you believe that it transmits clock? I
guess it's conceivable that a master might provide a global clock, but, since
the data format is async, I don't see any benefit.
Dick
----- Original Message -----
From: "Tony Duell" <ard(a)p850ug1.demon.co.uk>
To: <classiccmp(a)classiccmp.org>
Sent: Wednesday, July 25, 2001 3:12 PM
Subject: Re: Apple ][ boards -- what have I found?
[MC3487 driver]
> > > I'm not convinced that there's a "disconnected" state
that would work to
> allow
> > > more than one transmitter on a pair. The line is intended to be
passively
> >
> > When the OE line is low, then both outputs of each transmitter in the
> > pair are high impedance (disconnected, etc). So you can have multiple
> > transmitters driving the same pair of lines, if you ensure only one is on
> > at a time.
> >
> I made that assumption once, and was immediately informed that one should
use
> the RS485-compatible drivers, back in the
"old" days. The guy who jumped me
was
> the app's engineer from MOT, and he was
pretty young, but he was only
spouting
> the party line. I'm not sure that it makes
any difference, but one does
wonder
My guess (without trying the chip) is that the 'disconnected'
(high-impedance) mode does work as described. But that you should only
use multiple drivers if you are sure there is no chance of contention (2
drivers enabled at once). The RS485 devices should allow for contention
without damage.
I would believe as well that you'd probably get away with short periods
of contention using most RS422 drivers. It's not a good design to do
that, but then a lot of things aren't well designed.
> > The data sheet gives a value for the short-circuit current. It also says
> > that only 1 transmitter should be shorted at a time, and not for more
> > than 1s. In a mulitdrop network type application, a collision would be
> > detected in a lot less than 1s (we hope!), so my guess is that it would
> > be OK to use the 3487 in this way. Not a good design, but a design that
> > would work (typical of a lot of Apple and IBM PC plug-in cards,
actually...)
> >
> > When I've got some time (too many other projects...) I'll see how the
> > 3486/7 are used on this card.
> >
> Until you learn what the target application was, you'll like be left in the
> dark. While one can figure out what a board might do, until one gets a
picture
of what the
mfg intended, it remains a mystery why things were done as they
were.
I had a quick look last night, and things are a little clearer. It
appears that 2 drivers and 2 receivers are used. The outputs of one
driver are connected to the inputs of a receiver (and ditto for the other
pair). There is space on the PCB for termination resistors (which are not
fitted), including a multi-turn preset between the lines of each
differential pair. The termination components are missing (never
soldered) on my board, so I have no idea what the values are.
There are 6 numbered solder pads at the front edge of the card. 2 are
obviously grounded, the other 4 are the 2 differential pairs. I would
assume these carried a cable to some external connector.
My guess is that one differentinal pair is the TxD and RxD lines of the
6850. The other might be clock. I didn't have any data sheets to hand, so
I can't be sure about this yet. And I've not looked at the enable inputs
to the 3487 driver (the 3486 receiver seems to be enabled all the time,
which makes sense). I can't even remember if the 2 drivers used would be
enabled together or not (i.e. do they share a common enable pin).
I need to do some more checking, obviously, but it's looking increasingly
likely that this card does interface to some form of multi-drop serial bus.
-tony