[Info-vax] DEC 3000/900 bootup problem

[johnwallace4 at yahoo.co.uk]

On Jan 8, 11:00 am, Vance Haemmerle <va... at toyvax.Glendale.CA.US>
wrote:
> Vance Haemmerle wrote:
> > johnwalla... at yahoo.co.uk wrote:
>
> >> On Jan 5, 10:46 am, Vance Haemmerle <va... at toyvax.Glendale.CA.US>
> >> wrote:
>
> >>> It looks like the sun has set on my 3000/900 system.
> >>> After removing and backing up my disks on another
> >>> system (the system disk had one problem with a decw font
> >>> that had a lost header - which I recovered from a backup
> >>> and then it passed ANALYSE/DISK), now my system has a scsi
> >>> hardware error on power up (?? 002 0008).  The console
> >>> "test scsi" fails the data Trans test...
>
> >>> ? T-ERR-SCSI A - Data Trans Test - nondma/virt inq data miscompare
>
> >>> but "show device" correctly sees all the drives and sizes.  I
> >>> removed all the drives I had pulled before and I still get the
> >>> error. Perhaps I did something to the cables though they seem ok.
>
> >>> --
> >>> Vance Haemmerle
>
> >> Before you write this off forever...
>
> >> Have you done the obvious things like reseating memory (and, if
> >> appropriate, IO) cards?
>
> >> Is there a more in-depth memory diagnostic on this box than the power-
> >> on default check?
>
> >> Although it's been said earlier that "you've proved that the hardware
> >> isn't at fault", that's not strictly true - you've just shown that the
> >> hardware doesn't have any problems picked up by simple diagnostics.
>
> >> Others may well have other suggestions but those would be my first
> >> recommendations.
>
> >   I did remove and reseat the memory.  I also put an external
> > tape drive on the B (external bus) and it also showed a data
> > trans test failure.  If the internal bus and external bus
> > are completely cable separate (I believe they are) this may
> > point to a controller problem.  Perhaps I disturbed it when
> > I cleaned out dust from the system.
>
> >   There is an extensive memory test available from the
> > SRM console ("test mem") that does read/write and address tests
> > all of which pass.
>
> > --
> > Vance Haemmerle
>
>    I got a hold of a copy of the DEC 3000 600/600S/800/800S
> Service manual.  Seeing where the Scsi controllers are makes
> me think that cleaning the dust was not the reason as I did
> not clean near the I/O board where both the scsi controller
> and the battery backed TOY/NVR chip are.  In fact they are
> on a board where the mounted components face inward and are
> not exposed.
>
> --
> Vance Haemmerle

OK, so given your last two posts you've done all the sensible stuff I
can think of. The sensible stuff which the system designers could
think of is covered in things like the DEC 3000 800/900 Owners Guide
eg at http://vt100.net/mirror/mds-199909/cd1/alpha/flmulogb.pdf You've
apparently also got the 3000 600/800 Service Guide (which I can't
actually find right now?) which I ass*u*me covers the same ground and
more. Was a 3000 Model 900 largely a Model 800 with an EV45 instead of
EV4? I'm thinking so, but would welcome confirmation.

If you feel that you've gone as far as you can with the built in SCSI
and still want to keep this box, you might want to check out the
possibility of getting a TURBOchannel SCSI card to replace the
embedded SCSI. Check with a VMS SPD of the relevant era, it would be a
PMAZ or similar iirc.

Good luck.