anderson@sapir.psy.jhu.edu (01/27/89)
I'm trying without success to attach a CDC Wren IV drive to a 4/110. Sun
tells me that the famous pin-26 problem on 4/110's is no longer a problem
as of the rev level of my machine, and nobody's power supply appears to
have fried, so that seems not to be what's wrong. The Sun itself doesn't
seem to be the culprit, since it recognizes a Sun SCSI shoebox snarfed
from another machine with no problem.
What happens is that when the drive is cabled to the SCSI connector on the
machine, it won't even spin up, though it does spin up when not attached
to the 4/110. Letting it spin up and then attaching it to the machine, the
Sun reports that it can't find any sd device.
One theory: the drive unit itself is cabled to its DB-50 SCSI connector
with a straight fifty-line ribbon cable. This appears to yield the
following physical arrangement of its pinouts:
-------------------------------------------------------------
\ 49 46 43 40 ............... 10 7 4 1 /
\ /
\ 48 45 42 39 ............ 12 9 6 3 /
\ /
\50 47 44 41 ............... 11 8 5 2/
---------------------------------------------------
According to Sun, however (and also according the numbers on a 50-pin
male-to-male connector I'm using between the drive and the Sun), the
physical configuration of the Sun SCSI connector has the following
pinouts:
-------------------------------------------------------------
\ 17 16 15 14 ............... 4 3 2 1 /
\ /
\ 33 32 31 30 ........... 21 20 19 18 /
\ /
\50 49 48 47 ............... 37 36 35 34/
---------------------------------------------------
If the logical interpretation of "1, 2, 3," etc. on the drive is the same
as that on the Sun, it stands to reason I need a cable to map the one onto
the other, rather than the 50-pin straight through cable I have. Does this
make sense? I'm having a hard time convincing the supplier I need a new
cable to do this, but I don't know what else to try. The drive is jumpered
to identify as drive 0 (i.e., no jumpers on the SCSI id block), to tie
ground to both signal and physical ground, and to get terminator power
from the supply in its box. As far as I can tell, these are the only
settings available to tweak, and they seem right. If my cabling theory
isn't right, what is? One (not generally very helpful or knowledgeable
seeming) person at 800-USA-4SUN suggested "the 4/110 doesn't support
embedded SCSI". Is this plausible, explanatory, and (perhaps most
important) correct?
If you're technically sophisticated about such things, it's probably quite
clear to you by now that I'm not. Help and/or advice would be very welcome
(apart, perhaps, from the suggestion that I trash the Wren and buy
something from Sun, at their prices....). Please send mail to me at the
address below (which your mailer will probably be able to find, as opposed
to my return address above). If information that looks more broadly useful
arrives, I'll summarize to Sun-Spots.
Thanks,
Steve Anderson
Cognitive Science Center
The Johns Hopkins University
<anderson@cs.jhu.edu>anderson@sapir.psy.jhu.edu (02/02/89)
Please cancel the messge I submitted last Friday (20 January). The problem turned out to be a matter of cabling, but one completely unrelated to the theory I presented in that message. Steve Anderson Cognitive Science Center The Johns Hopkins University <anderson@cs.jhu.edu> (DON'T use the return address at the top of this message! use this.) [[ Sorry, didn't see this cancel request in time. --wnl ]]
michael@uunet.uu.net (Michael McClary) (02/08/89)
In article <8901201904.AA00222@sapir.> anderson@sapir.psy.jhu.edu writes: >I'm trying without success to attach a CDC Wren IV drive to a 4/110.... >What happens is that when the drive is cabled to the SCSI connector on the >machine, it won't even spin up, though it does spin up when not attached >to the 4/110. Letting it spin up and then attaching it to the machine, the >Sun reports that it can't find any sd device. If a Wren doesn't spin up, the first thing to check is whether you've got the terminator plugged in upside down. That'll do it on our Jasmine boxes. (I think it results from asserting RST.) Also: Be sure you've got power to your terminator. We're using Wren IVs (packaged both by Jasmine and Relax) on several suns, including my IV/110. We use a cable from Monster Cable to go to the Centronix-style blue-ribbon connector on the Jasmine. On the Relax, we start with the Monster Cable, then a home-brew Centronix sex-changer, then an Apple-style Centronix-to-DB25. Pinouts of the SCSI-standard 50-pin header (typically found on the drive proper), SCSI-standard 50-pin "Centronix Style" connector, and the DB-50 (?) used by Sun. (I didn't check whether Sun does something strange with anything but varies/TERMPWR.): Header Cntrnx DB-50 What's connected ------ ------ ----- ---------------- 01 02 01 26 01 34 GROUND / -DB(0) 03 04 02 27 18 02 GROUND / -DB(1) 05 06 03 28 35 19 GROUND / -DB(2) 07 08 04 29 03 36 GROUND / -DB(3) 09 10 05 30 20 04 GROUND / -DB(4) 11 12 06 31 37 21 GROUND / -DB(5) 13 14 07 32 05 38 GROUND / -DB(6) 15 16 08 33 22 06 GROUND / -DB(7) 17 18 09 34 39 23 GROUND / -DB(P) 19 20 10 35 07 40 GROUND / GROUND 21 22 11 36 24 08 GROUND / GROUND 23 24 12 37 41 25 GROUND / GROUND 25 26 13 38 09 42 >varies / TERMPWR 27 28 14 39 26 10 GROUND / GROUND 29 30 15 40 43 27 GROUND / GROUND 31 32 16 41 11 44 GROUND / -ATN 33 34 17 42 28 12 GROUND / GROUND 35 36 18 43 45 29 GROUND / -BSY 37 38 19 44 13 46 GROUND / -ACK 39 40 20 45 30 14 GROUND / -RST 41 42 21 46 47 31 GROUND / -MSG 43 44 22 47 15 48 GROUND / -SEL 45 46 23 48 32 16 GROUND / -C/D 47 48 24 49 49 33 GROUND / -REQ 49 50 25 50 17 50 GROUND / -I/O Note that (neglecting right-left) the DB-50 is numbered: 1 2 3... 18 19 20... 34 35 36... the centronix-style is numbered: 1 2 3 4... 26 27 28 29... and the drive header is numbered: 1 3 5 7... 2 4 6 8... My Sun-4/110 doesn't provide TERMPWR, and leaves DB-50 pin 42 (= header pin 26, = "Centronix" pin 38) open. If your drive or terminator is looking for TERMPWR there, it's out of luck. (Did older Suns ground that? If so, Yike!) According to my copy of the SCSI spec, the "varies" pin has TERMPWR on the Header. (Alan Brunner (sp?) from Jasmine was here a couple days ago, and I think he said that the Wren leaves that pin open. I haven't tested it myself.) It is grounded on my Sun IV/110's connector, and the Monster cable propagates it through to the Centronix connector. The SCSI standard mandates that the pin be an open circuit on a Centronix-style connector, but warns that some equipment produced before the standard was promulgated may ground that pin. Thus, the combination of a Sun IV/110 and a monster cable violates the SCSI spec, but in a way the spec warns you to expect. So hook it up this way, check for inverted connectors, no TERMPWR on the TERMPWR pin at the terminator, TERMPWR on the "varies" pin, and if it still doesn't work, can I have the drive? B-)
tsui@uunet.uu.net ( iccad) (02/09/89)
We have about 10 (non Sun) SCSI shoeboxes, they all use the following
pinouts:
-------------------------------------------------------------
\ 49 46 43 40 ............... 10 7 4 1 /
\ /
\ 48 45 42 39 ............ 12 9 6 3 /
\ /
\50 47 44 41 ............... 11 8 5 2/
---------------------------------------------------
where pin 1 is the fist wire, pin 2 is the second wire, ..., of the flat
ribbon cable.
Alan Tsui
Mitel Corporation