Very Computer

We Installed an adaptec 1540CF into a Vtech Pentium-90Mhz Pc. It seems to work
fine for a while. Then, it starts giving messages of those types:

        Notice: aha: no controller response reading SCSI adapter dev 1/104
(ha=0 id=0 lun=0) block=58

         Sdsk: unrecoverable error reading SCSI disk
         Sdsk: 1 dev 1/104 (ha=0 id=1 lun=0) block 56

3 HDU are installed with this controller (id 0,1 and 3). The "unreoverable
reading" error is random on any of these HDU. We checked the jumpers (id=) on
the HDU and a proper SCSI terminator is in place.

So,

a) We thought the Controller was defective; we did change it.
   The problem is still alive !

b) We thought the motherboard could be the problem, so we did change it.
   The problem came back !

Any ideas ?

Thanks very much,

The fact that the "unrecoverable reading" error randomly occurs on all the disks
suggests a cabling or termination problem.

Adaptec controller's, particularly the 154x line is notorious for being
finicky about cable impedance and dead on accuracy of terminating resister
packs.  If any or all of your disks are external, this will aggravate the
problem.  

I would suggest that you first try replacing each cable, one by one, in the
chain of devices (assuming they are external) with a cable known to be of high
quality.  If this fails to correct the problem, try replacing the terminating
resister pack.

If your devices are internal, try replacing the ribbon cable, then if you still
have the problem, the resister pack.

Whatever you do, use quality cables.  Never use something the techie in the back
room of the computer store whipped up with a soldering iron.

--
------------------------------------------------------------------------

|  The Laitram Corporation     | Yoda of Borg am I.  Futile resistance |
|  New Orleans, Louisiana      | is.  Assimilated you will be.         |
------------------------------------------------------------------------

$       Notice: aha: no controller response reading SCSI adapter dev 1/104
$(ha=0 id=0 lun=0) block=58
$
$         Sdsk: unrecoverable error reading SCSI disk
$         Sdsk: 1 dev 1/104 (ha=0 id=1 lun=0) block 56
$
$3 HDU are installed with this controller (id 0,1 and 3). The "unreoverable
$reading" error is random on any of these HDU. We checked the jumpers (id=) on
$the HDU and a proper SCSI terminator is in place.
$
$a) We thought the Controller was defective; we did change it.

   I'll bet you didn't change the controller about which the complaint
was lodged.  You probably changed the controller at ID 7 instead,
by replacing the host adapter.

   In SCSI parlance, "controller" doesn't mean "host adapter"; each ID
needs a controller, which is basically a CPU of some sort which
interfaces the device(s) at that SCSI ID to the SCSI bus.  Since the
message points to the controller at ID 0, it's complaining about your
hard drive, not your host adapter.

   If you want more details on the exact definition of "controller"
in the context of SCSI, and what LUNs and IDs mean, lurking in
comp.periphs.scsi would probably fill in the details.

   Anyway, my guess is that this means that the host adapter's driver
has detected that it sent a command to the CPU which runs the hard
drive at ID 0, and that it didn't get a response back.  It could
be a communication problem, or a bug in the firmware on your hard
drive, or possibly even a bug in the firmware on your 1540CF or
a bug in the ad driver.  Since you say that the error moves from
one drive to another, I'd suspect it's not a physical problem with
one particular drive, and since I've used a variety of 154x
host adapters (including the 154xCF) over the years with a number
of drivers for a number of OSes, including several SCO ad drivers,
I doubt it's either of those, either.

   I know you already checked termination.  Check it again, just
to be sure.  Try changing which device is at the end of the bus
(I'm serious; I've seen simply swapping two devices, and altering
termination accordingly, cure seemingly random errors).  Power
problems can cause all kinds of odd errors in any part of the
system, though my gut feeling here is that power isn't the most
likely cause of this problem.

   Do you know the rules about termination power?  No, I don't
either.  Try adjusting it anyway, to see if that makes a difference.
I know there's been a treatise on it posted here once or twice
before but it just won't sink into my brain.
--

----------------------------------------------------------------------------
Stephen M. Dunn, CNE, ACE, Sr. Systems Analyst, United System Solutions Inc.
104 Carnforth Road, Toronto, ON, Canada M4A 2K7          (416) 750-7946 x251

(...)
:    Do you know the rules about termination power?  No, I don't
: either.  Try adjusting it anyway, to see if that makes a difference.
: I know there's been a treatise on it posted here once or twice
: before but it just won't sink into my brain.

Paraphrased and interpreted from the Book of ANSI,
SCSI-2 specifications and personal horror stories...

1.  There shall be two and only two terminations per
scsi cable bus be they internal, external, passive, or active.
2.  Thou shall apply terminator *POWER* only at the ends
of the SCSI cable chain.  
3.  If thy terminations be external, thou shall apply
terminator *POWER* from the adjacent device.
4.  If thy external terminator be powered by the scsi bus
instead of the nearby device, thou shall use an active terminator.

Numbers 3 and 4 are a bit obscure.  The idea behind #3 is
to *NOT* power the external terminator through the scsi bus.
The typical #28awg ribbon cable just can't carry the current.
The result is an offset from the nominal 2.85VDC that the
terminator/divider is trying to maintain.  If you must power
the external terminator from the scsi bus, try to get an
active terminator.  Passive terminators will dissipate
approximately 1.7watts if all 18 outputs go low simultaneously.
Active terminators "only" dissipate 0.98 watts.   The passive
terminator also dissipates 1.6 watts when doing nothing while
the active terminator doesn't dissipate any power in this state.
(Drivel: a differential terminator dissipates about 0.6 watts.)

All this power must be supplied through the terminator power
jumper or through the scsi cable +5V line.  The active terminator
creates less of a voltage drop and has a regulated 2.85V output
which is fairly independent of the actual +5V line.

So it is written, so it must be...

--
# Jeff Liebermann  Liebermann Design  150 Felker St #D  Santa Cruz  CA  95060

# 408.699.0483 digital_pager    73557,2074  cis [don't]