Date   : Sun, 29 Jul 2001 08:57:47 +0100
From   : Mike Tomlinson <mike@...>
Subject: Re: Help with BBC B fault

In article <df617fa14a%mikef@...>, Michael Foot
<mikef@...> writes

>> Was the machine known to be working previously?
>
>Yes. In fact it's condition is (was) mint and it still has the original 
>packaging.

And it's just failed spontaneously, not as a result of trying to fit
anything like a SWR/DFS upgrade?  Could anything have been plugged into
the 1Mhz bus/Tube sockets while the machine was powered up (this will
most likely result in damage)?

Some more things to try:

Check the power supply voltages and make sure the 5v rail is 5v +/- 10%.

Check all three power connectors are secure.

Check, carefully, that S21 (ROM select) and adjacent jumpers are set to
defaults.

Leave the machine powered up for 10 minutes or so, then (carefully!)
feel the RAM chips with the back of a finger.  If any are noticeably
cooler or hotter than the others, it may be a RAM failure.

I think the IRQ line appearing to be stuck high may be a red herring.  I
don't have a cct diagram with me at the mo, but if this line is active
low (has a bar above its' signal name on the cct diagram) it will
normally be high, and the transition from high-to-low-to-high on the
working machine you saw is normal (it's the OS or auxiliary ROM
initialising, for instance, the DFS hardware.)

If either the IRQ or NMI pins is stuck low, try removing the 6502,
bending the pins up, and reinserting it to see if the machine then
boots.  (Switching off in between, of course...) If it does, you can
trace the lines back to the culprit chip using the cct diagram.

Check the (4Mhz?) master clock crystal is working - I have replaced a
few of these.  It seems more prone to failure when the machine has been
in storage.  It is divided down to produce all the clock signals.

Check the 6502's reset pin is not stuck low and toggles when you press
break.  Check that the inverted reset pin is working. (goes high when
break is pressed)

If that doesn't help, put the dead and working Beebs side-by-side and
compare signals with an oscilloscope.  Check the address and data buses
on the 6502 are all active and wiggling.  A bad signal will "sag"
noticeably when you apply the probe - this can be a clue.  Check the RAM
data, RAS and CAS lines and see if you can spot anything unusual.  Press
break frequently and observe the behaviour on both machines while
monitoring signals.  Check the OS ROM is enabled and that its' address
and data buses are wiggling.

If all that gets you nowhere, then it needs professional attention.  As
you say the machine is in mint condition, it'd probably be best to swap
in the board from a working machine and use the chips from the faulty
board for spares.

Mike

-- 
"Security-wise, NT is a server with a 'Kick me' sign taped to it."
                      - Peter Gutmann