mdw
|
|
response 2 of 10:
| 
Sep 6 01:46 UTC 1991 |
The sun-2 serial ports (CPU board and scsi card) are all wired as DTE's.
A straight-through cable should suffice for a modem, in many cases.
Chances are good that the sun will ignore DCD (pin 8) and obey only DSR
(pin 6), so with some types of modems (for instance, Vadic), carrier
detect may not work correctly. If the modem can't be configured such
that pin 6 is asserted only when carrier is established, you may want to
cross pin 8 from the modem to pin 6 on the computer. The signals that
the computer apparently uses are:
1 fg
2 txd from computer
3 rxd to computer
4 rts from computer
5 cts to computer
6 dsr to computer
7 gnd
20 dtr from computer
Interesting note: this is different from the systech ports on
grex--those are wired up as DCE's.
|
mcnally
|
|
response 3 of 10:
|
Sep 6 04:57 UTC 1991 |
I suspect he wants information on the SCSI board serial ports.
Judging from the cable that came with mine, and seems to work fine,
there's no hocus-pocus. My cable consists of a 50-pin ribbon cable
with two quick-connect 25-pin D connectors (standard serial connector)
at one end and a quick-connect 50-pin connector on the other end (two
parallel rows of 25 pins, thin and rectangular.. I could pick it out
of a lineup but don't know what to call it exactly. It looks like
things are just wired straight through. Since there are no wire
crossovers, the only thing that could be reversed is which way the pins
cound on the DB-25s.
|
mdw
|
|
response 4 of 10:
|
Sep 6 06:54 UTC 1991 |
CPU board & scsi board serial ports are almost identical. Even use the
same chip and software driver, not to mention the same cabling
arrangement and pinout. The chip used is the Zilog Z8530 duart. Pin
assignments on any of the 50 pin connectors (CPU J1, SCSI J1 & J2) are:
port A/C/E port B/D/F
3 txd 28 txd
4 db 29 db
5 rxd 30 rxd
7 rts 32 rts
8 dd 33 dd
9 cts 34 cts
11 dsr 36 dsr
13 gnd 38 gnd
14 dtr 39 dtr
15 dcd 40 dcd
22 da 47 da
24 bsy 49 bsy
It might be worth tracing the lines out--it's not clear to me what, in
fact, is being done with dsr/dcd, or what da,db, and dd are. Or "bsy"
for that matter.
The Zilog doc I have says the chip supports, per port:
inputs: cts,dcd,rxd,rtxc,cts
outputs: dtr,rts,w/req,txd
input or outputs: sync,trxc
The zs driver for SunOs seems prepared to read the following signals
from the chip: dcd,cts,rts,dtr, and write: rts,dtr. dcd is also
apparently what is used for carrier detect - if they don't pull any
weirdnesses going to the port, dsr is ignored on the connector, and dcd,
pin 8, is what one wants connected. This contradicts the paper
documentation Sun supplies on constructing a null modem cable, which
says to swap 6 & 20, and to ignore 8. Oh well. Then again, the Sun doc
also states that carrier detect isn't likely to make sense except with
modems, so perhaps they realize, in some dim manner, that their
directions for a null modem cable doesn't connect anything useful to
dcd. So it may be best to ignore my earlier thought about crossing 8 to
6. It doesn't look like the driver does anything with rts/cts. So
using them for flow control is probably doomed.
|
choke
|
|
response 10 of 10:
|
Sep 10 17:18 UTC 1991 |
There, you can always trust vi.
Well, the straight cable seems to work fine.
The only problem is that my login process doesn't seem to be killing things
when it loses carrier. This means that if I hang up on it someone else can
call in and get on as me.
It's kind of sticky dialing out from the tty that has a getty process on it.
My system V machine had a very handy thing called 'ungetty' that would
make the getty sleep for a while until it got killed.
Anyways, point is it works acceptably.
|