PDP-11/34a hung bus problem.

[Don North]

On 5/14/2016 2:41 PM, Noel Chiappa wrote:
>      > From Ed Groenenberg
>
>      > - insert both CPU cards, KY11-LB card & bootstrap card, 5 full grant
>      > cards, DL11-W  and bus terminator card.
>      > ...
>      > - cntrl + boot shows register dump at printer.
>      > ...
>      > All looks ok
>
> I'm surprised the bootstrap ran OK with no memory at all in the machine. I
> vaguely STR that I had a machine that would not work like that, but maybe I'm
> wrong. (DEC bootstaps tend to do things like set the NXM vector, in low
> memory, so they can size memory; and when it gets the NXM (since there is no
> memory) from trying to touch the NXM vector, and tries to push the old PS and
> PC to service _that_, and gets _another_ NXM, that 'double bus fault' often
> causes many -11 processors to do a cheap suit)

If the bootstrap card is an M9312 with the standard console PROM, it does NOT 
require any
memory to be present/accessible to get to the ODT prompt that prints out the 
registers and
waits for a command. Only until you execute a device bootstrap command with 
'diagnostics
enabled' (the default) does the memory sizing/test/diagnostic code get exectuted.

Listing:  http://ak6dn.dyndns.org/PDP-11/M9312/23-248F1/23-248F1.lst

So the system working to this level with no memory present is normal.

>
>      > power down machine, add memory (M7981, 128KW)
>
> What's an M7981? Did you mean an M7891 MS11-L?
>
>      > - power up machine -> run light is on, does not get cleared by
>      > cntrl + halt.
>
> This is where a UA11 would really help. I had similar issues with an -11/04,
> and the UA11 was a huge help in figuring out what's going on. One glance and
> you can see if a bus line is wedged, or something.

Agreed, a bus probe would really be helpful in further debug. Given that the 
system does not
even get back to the ODT prompt (which would not yet access memory) indicates 
that logic
on the memory card is hanging the bus. Could be a bad bus driver interface chip 
(8641, etc).

>      > - power down & replace memory with grand card -> run light is off.
>
> Well, that's good sign - the memory card didn't fry anything, at least...
>
>      > - tried a 2nd memory card (M8722, 128KW)
>
> Ooops. The MS11-M needs +/-12V, which is _not_ standard in most
> machines/backplanes). The EUB in the 11/24 and 11/44 (which this card is
> intended for) does have it. The really bad part is that those same pins
> usually carry +/-15V in most MUD backplanes. So hopefully you didn't fry it.
> It does have standard UNIBUS as well as EUB, but there's a jumper, IIRC.

Bad idea to put this card in an 11/34. As indicated the +12V VDD rail on the 
memory chips
is wired directly to the UNIBUS +15V line in the 11/34 (which can be set to +12V 
in the 11/44).
So you placed +15V (or more likely up to +15.5V or so) onto the memory chip VDD 
lines.
Some 16K chips had a 15V max spec on this pin, others only 13.2V. So it is 
possible you
toasted so memory chips, or not, depending on manufacturer and their sensitivity 
to OV.

In any event, DO NOT use this card again in the 11/34. It is not compatible with 
the backplane.

>      > So what could be the problem here? The bus works without the memory
>
> Two possibilities off the top of my head. i) The first memory card is bad (or
> configured incorrectly), or.. ii) The M7891 uses +/-15V as well as +5V? So
> maybe one of the other voltages is not so good? But you said the console
> worked, and I think that uses other voltages (at least, in EIA mode - not
> sure about 20mA, I never touch the stuff).
>
> 	Noel
>