Subject: Re: DECstation 3100 self-test failure
To: None <port-pmax@netbsd.org>
From: George Harvey <fr30@dial.pipex.com>
List: port-pmax
Date: 09/28/2004 22:21:52
On Fri, 24 Sep 2004 22:20:54 +0100
George Harvey <fr30@dial.pipex.com> wrote:

> > On Thu, 23 Sep 2004, George Harvey wrote:
> > 
> > > I've aquired a 3100 that I'm hoping to install NetBSD on.
> > > Unfortunately it fails the self-test:
> > >
> > > KN01 V7.01
> > >
> > > 7..6..5..4..3..2..
> > > ??
[ snip ]
> I found a web site which listed a lot of the ROM test routines
> (www.q.dyndns.org/~blc/DS3100/3100test.html) but I can't get any of
> them to run on my 3100, the console just hangs.

I've made some progress with the tests, seems that if I keep retrying
them they will sometimes work, usually if run immediately after a ctrl-C
but not always. Anyway, the significant results are:

>> t n
-022-00 wrt nram w/ 0000
-022-00 chk nram w/ 0000
?08c-00 NRAM - addr exp ret  b9000000 00000000 00001000
>>
>> t d
-02a-00 wrt dram w/ 0000
-02a-00 chk dram w/ 0000 
?08b-00 DRAM - addr exp ret  bb020000 00000000 00000010
>>
>> t s 
-047-01 SII bfr adr tst
?0bd-34 siiatst: bfr dta err: add exp rtn bb020000 00008000 00008010
>>

So the network RAM, disk RAM and SCSI test all fail with data errors,
which explains the POST failure. I've also run the main memory and cache
tests, which both pass, so my guess would be a bad buffer chip somewhere
in the bus interface. However, without a schematic that's as far as I
can go :(

Hmm, while writing this message there was a burning smell from the 3100
and it looks like a surface-mount capacitor near the power connector has
just self-destructed. It's some distance from the SCSI and Ethernet
chips, so it's probably unrelated to the test failures, but I could try
replacing it just in case. Unburnt ones are marked 'AS5', anyone know
what that translates into in uF?

George