Date   : Sat, 29 Jun 1996 20:23:27 +0100
From   : Mike Tomlinson <mike@...>
Subject: Re: Hardware problems

In message <19960628174543.069a773c.in@...>, steve
<steve@...> writes
>Update:  (Sorry if anyone's getting bored with this)
>
>I went ahead and swapped the ULA for the VTi, no difference.
>
>I have discovered the following:  If I run Mike's soak program after the
>machine has been on for a while, it crashes after a few seconds, with
>several horizontal lines on the screen.  If I change the GCOL line to read
>GCOL 0,RND(7) (instead of RND(15)) it runs fine.  Does this suggest
>anything to anyone?

All I can say is: !! (speechless) :)

That shouldn't make any difference. Have you tried changing ic14?

Contrary to Pete's experience, I have found this ic to be a common
failure point, and I agree with his comments about the timing
relationships within the machine. However, I have almost always found
that substituting a F245 or ALS245 reduces reliability: it seems to
exacerbate timing problems, and I always used a pukka LS245 in my
repairs.

Please accept my apologies for tha ballsup I made: Pete is quite correct
when he states that there are six 81LS95s, not three. The words "pairs
of" somehow got omitted from the sentence "..there are three 81LS95s.."
etc.

Though quite how Pete can describe a memory circuit with sixteen DRAMs,
six 81LS95s, and some glue logic, plus two RASes and CASes, as "pretty
standard", escapes me :)

cheers,

-- 
Mike Tomlinson
constructive email to mike@...               ; flames to /dev/null.