> but works. I should fix it though... I'll start digging for a source for
> bulbs.
Do so. They're something like 14V 80mZ T1+3/4 bi-pin bulbs. They're still
made AFAIK.
>
> > There are +5V on pin A2 of most of the connectors (maybe not the memory
> > clots), ground is pin C2 and T1. -15V on B2. You need both those voltages
> > for the CPU to run.
> You mean I need both these voltages on every card that makes up the cpu I
It can't hurt!. There are serveral separate 5V lines in that machine, one
from each power brick, which are not commonoed together. You need to make
sure that at least one card for every brick is getting its power. It
can't hurt to do that lot.
> assume, so I need to check each of those slots pins. I did check the
> regulators directly, but I guess it's possible there's a broken wire not
> getting those voltages up to the backplane.
Indeed. Or excessive voltage drop along the power harness, or you've
forgotten to plug something it (been there, done that!), or...
>
> > Can you tell me what boards (M numbers) you have in the CPU section of
> > the backplane. Do you have the FPU and/or the MMU?
> No FPU (and regulator configuration matches this), no MMU
OK,,,
>
> Slots are used as follows:
> 1 (A,B) M930C
> Slots 2-5 are open
> 6 M8100
> 7 M8101
> 8 M8102
> 9 M8103
> 10 M8104
> 11 M8105
> 12 M8106
> 13 - this slot is open
> 14 M8116
> 15 (c,d,e,f) M8109
> Slots 16-25 are open
> 26 (a,b) M9200 leg 1
> 26 (c,d,e,f) M7800
> 27 (a,b) M9200 leg 2
> 27 (c,d,e,f) M7800
> 28 (a,b) ribbon jumper to MF11-L slot 1 (a,b)
> 28 M7228
OK... That sounds right. I see you have some I/O devices (in particular a
pair of DL11 series cards). What I'd do next, I think, is disconnect the
memory (pull the jumper from slot 28 a/b and replace it with the
terminator) and try to access the registers on one of those DL11s from
the panel. You can deduce the address by looking at the jumpers and seing
what the do in the prints (be warned that in a lot of cases, a jumper is
fitted to make the corresponding address bit equal to zero).
The idea is to see if the processor can access anything on the Unibus --
in other words do you have a processor or a memory problem.
> > Yes, it should, but only if you don't get a bus error. If you get an
> > error, the address doesn't change.
> I did get address errors when examining, but not when doing a load address.
A LOAD ADDRESS will never give a bus error. That operation doesn't do
anything on the Unibus. EXAM or DEPosit might, if the device doesn't respond.
> > I assume you've set the rotary switches sanely. I think you want the
> > address one in one of the 'physical' positiions, the data one in the
> > 'data path' position.
> Yup, I read up on that. CONS PHYS... but I think I had the data one in the
OK...
> "bus address" position. The processor handbook made it sound like that was
> the right one to view addresses, and made it sound like the data path one
> was just for output of the ALU.
>From what I remember (and it's been a long time since I've done this (too
many machines to play with here...) the one you want is the one that
sounds less useful!
> > Do you have a KM11 maintenance PCB? In fact just what do you have?
> Nope, sure don't. The only cards in the system are the ones mentioned above.
Pity. I don't think you need it yet, but the time will come when you want
to single-step the CPU microcode.....
-tony
Received on Sat Feb 19 2005 - 18:05:48 GMT
This archive was generated by hypermail 2.3.0
: Fri Oct 10 2014 - 23:37:39 BST