help needed: problem with UNIBUS access on my PDP-11/70

From: Gooijen H <GOOI_at_oce.nl>
Date: Thu Jul 11 02:39:01 2002

Hi all.

I am having trouble with my 11/70 because it works only partial.
Since things are quiet on the list, I have something to ask.
It is a long story, but as said on the list this week, the more
info you give, the less not relevant suggestions are typed.

First the description of the machine.
The machine has only all CPU boards (with FP), the DL11-W and
3 MASSBUSS interfaces. The fourth MASSBUS interface has in the
correct slot a G727. All other UNIBUS slots have a G727 in the
card position D. The MOS memory box is also connected to the CPU.

This is what I have done so far.
When I turn on the machine everything "looks" fine. With the switch
ENABLE/HALT on HALT and pressing START, the machine sort of resets.
With the panel I can dump data in the MOS memory at address 00000000.
I did this also at 00001000 and 00010000. It 'works' and when I read
the contents back from those addresses, it is the correct data.
So, my first conclusion is that (part of) the CPU is OK and that the
address and data path to the MOS memory and the MOS memory box itself
are all OK as well.

Here is the part that worries me.
In slot #40 (IIRC) is the DL11-W (M7856) console interface.
I am trying to write to the transmit buffer address (XBUF - 17777566).
I set the knobs on the 11/70 console to "CONS PHY" and "DATA REGISTER"
so that the address on the switches is the real physical address and
the data on the switches is what I want to store.
After the reset of the machine (HALT/START) I set the switches to
17777566, and press LOAD ADRS. On the ADDRESS leds appears 17777566.
Now, I set the switches to 00000071 (should give a "9" on the VT220).
When I toggle the DEP button, the PAUSE led goes on.
According to the handbook that means the the CPU tries to finish the
instruction as far as possible and then waits for an event to finish.
The event could be (I assume) an interrupt or, in this case, access
to the UNIBUS section.

Next test was checking the DL11-W interface.
First I switched on my 11/34C. With address 165020/START, I get the
dump of the registers on the VT102. So, I mad sure that the M7856
in the 11/34C is OK.
Now, I swapped the M7856 of the 11/70 with the M7856 of the 11/34C.
First, I started the 11/34C again. I get the register dump on the
screen. So, the M7856 from the 11/70 was OK.
Just to make sure I tried the 11/70 again (with M7856 of the 34C),
but I get the same result: PAUSE led goes on.

Yesterday, I had a long conversation on the phone with Edward. We
talk about all kind of (PDP-11) things, and also the 11/70 problem.
One suggestion was that the UNIBUS map is not yet initialised, and
that would cause the UNIBUS accesses to fail. My guess is that when
the console panel knob is set to "CONS PHY", I have full access to
all addresses without any mapping taking place. Is this correct?

Anyway, to test the "UNIBUS map initialisation" theory, I read the
M9312 bootstrap manual. The 11/70 (and 11/60) have a different PROM
that stores the diagnostics. The 11/04--11/55 PROM also contains a
console monitor, but the 60 and 70 PROM only has diagnostics.
The manuals says to load address 17765744, then set the switches
8-0 to the device code and then press START.
I checked the M9312 to look at the diag PROM. According the M9312
manual the PROM code must be 248F1 (IIRC) for the 11/04-11/55, and
233F1 for the 11/60-11/70. My M9312 has a PROM with code 616F1.
Is that a new type for the 11/70? The M9312 is the card that was
in the 11/70 when I got it, and another 11/70 (with remote console)
that I have has the same PROM. Edward's 11/70's (he has also 2)
one has a 233F1, the other has a 616F1.
So, I did that test, but the PAUSE led goes on again.

Trying to boot from an RX01 should need at least the RX11 card in
a UNIBUS slot ... I did that, so I put the M7846 in slot 41 (IIRC),
next to the DL11-W. Tried the test again: 17765744 - LOAD ADRS - set
the device code on switches 8 to 0 - START. Alas, PAUSE on again.

What puzzles me is the "device code" you must set on the 9 least
significant switches. I have set that to 170 as that are the LSB's
of the start address fro the bootstrap of the RX11. Is that correct?

Come to think of the good advice from Tony: I will check the
power supply voltages this evening. Especially the one that supplies
the cards that deal with the UNIBUS section ...

tnx for reading all this,
and TIA for all responses,

- Henk.
Received on Thu Jul 11 2002 - 02:39:01 BST