[MiNT] Research report

[Jim DeClercq <jimd@panix.com>]

Looking for things that could be going wrong, but only sometimes,
with my cantankerous MiNT system, I googled the number of the
drive it is on, Seagate ST32550w, and found out a few things, and
reached a conclusion.

What I found, was that this drive is just the right size for any
Unix system, and is, or was, sold by the people who sell Cisco
stuff. I also found reports of fsck finding an impressive number
of bad blocks on a good drive, which is one of the things I
noticed, and that upgrading firmware did not help. I saw a report
that 8 of 12 such drives, which of course passed end-of-line
inspection, did not work in somebody's version of the real world.

Seagate seems to be aware that they had a problem, and automated
the process of getting a RMA (return materials authorization) to
get a replacement under warranty.

With some experience in failure analysis, that is a capacitance
problem. Capacitance of digital things is never specified, so it
cannot be tested to spec, and is uncontrolled. This is probably
input capacitances, which are measurable on a bridge, and there
will be a ratio of three to one between the ones that do not
work, and the ones that do. I have sorted out problems of this
sort with big SCRs and little dual comparators.

I have several of these drives, so that is one approach. And,
there are some pins for turning on the terminators, which might
swamp out the extra capacitance, and that might help.

This may be the reason for a few reports of the need to terminate
things in the middle of a SCSI bus.

But cables shorter than the ones I have are not commercially
available.

Now I go look for my little box of jumper blocks, and shut this
system down for repairs.

-- 
 /"\      Jim DeClercq--jimd@panix.com--Sylvania, Ohio, USA
 \ /      ASCII ribbon campaign | I'm a .signature virus!       |
  X       against HTML mail     | Copy me into your ~/.signature|
 / \      and postings          | to help me spread!            |
.