Subject: Re: Hardware problem
To: Ian Clark <ic@smoke.com.au>
From: Chris Rupnik <chrisr@geeknet.montreal.qc.ca>
List: port-sun3
Date: 09/09/1998 09:32:04
Hi
In my case, the parity error is ALWAYS the same address. I believe the
problem to be with VME cards instead.
If i have no VME cards installed, then the machine boots fine. (only 4
megs though)
If i put in any memory boards into the machine, it fails with the parity
check.
If i put in the xyc VME board, the kernel fails on boot (bus error)
I spent a good 2 hours last night going over jumper settings on both
the CPU and the back of the 12 slot cardcage, and nothing seems to be
wrong.
Anyone out there with a working 3/160 want to email with me their exact
jumper configuration?
Thanks!
Chris
On Wed, 9 Sep 1998, Ian Clark wrote:
> On Mon, 7 Sep 1998, Alfred Arnold wrote:
>
> > >It looks like the parity error is happening just below the 4MB boundary,
> > >which would seem to indicate that it's a problem with memory card #1 (the
> > >one in slot #5). Try removing that card and replacing it with the one
> > >from
> > >slot #3. You'll have to change a jumper on the card to mark it as RAM
> > >CARD
> > >#1 (or #0, depending on how they numbered those...). That will take you
> > >down to 8MB of RAM, but it's probably your only solution if that mem card
> > >is bad.
> >
> > If it's really a 3/160, the first 4 Mbytes are on the CPU board and I
> > don't know if there is a jumper on the board to disable the on-board
> > RAM...so this might be a reason why the machine was thrown out :-(
> >
>
> It is possible to replace the chips, I have a 3/140 which had a similar
> problem (parity faults, lots of them at different addresses). Finding the
> dud chip was a bit of a problem , and its here I come clean with the
> nasty thisgs I did to my machine! To find the chip I cut the data-out pin
> of the 4116? 4164? memory chips in sequence to try to find a pattern as
> to address range /bit for each chip. I eventually found the right one and
> replaced it. Probably more by good luck than anything else.
>
> In this case I figured the machine was dead anyway and I coudn't make it
> worse.
>
> Of course if someone has details of how to map the parity fault details
> to a chip, its a piece of cake
>
> Ian (I love the smell of flux in the morning) Clark
>