11/53 - Qbus backplane question

From: Carlini, Antonio <Antonio.Carlini_at_riverstonenet.com>
Date: Fri Jan 4 03:47:27 2002

Gooijen H wrote:

>The 11/53 has an H9278-A backplane. According to
>Megan's fieldguide the first 3 slots are Q22/CD
>and the last 5 slots are Q22.
>What does that mean?

The slots each allow for four "fingers"
on each card - a quad card uses all four
and a dual card uses only two (there are some
single cards too, but not many).

The holes on the backplane slot for the
fingers are labelled A,B,C,D from left
to right from the side the cards plug into.

Q22/CD slots have A/B wired as Q22 bus, C&D
are just wired straight through. C/D are not used
for Q-bus comms and are just for card-card
communication.

Q22/Q22 slots have Qbus on A&B and Qbus on C&D.

It sounds like in your machine, the path
followed by the Qbus is:
        slot-1: A/B
        slot-2: A/B
        slot-3: A/B
        slot-4: A/B
        slot-4: C/D
      slot-5: C/D
      slot-5: A/B
      slot-6: A/B
etc.

i.e. Qbus goes down A/B in slots 1,2,3,4
and then hops over to C/D on slot 4, drops
vertically down to slot-5 C/D, moves back
to slot-5 A/B ... repeat until you run out
of slots. The manuals refer to this as a
serpentine pattern. There is a diagram in
some of the MicroVAX manuals on
 http://208.190.133.201/decimages/moremanuals.htm

(I cannot remember exactly which ones but if you pick up
 the 630Z Owner's and Technical manuals, I'm 90% certain
 one of them has the diagram ... note that BA23 has 3
 Q22/CD slots and BA123 has four Q22/CD slots ... the
 principle is the same though).

The Q22/CD distinction matters in two cases that
spring to mind.

The first is that some Qbus signals need to
be passed on to the next card correctly
(interrupt and grant signals). An empty slot
breaks the chain here. So if your first three
slots have processor + memory and you put a
TQK50 controller in slot 4 (A/B) and an RQDX3
in slot 5 (A/B), the chain is broken because
slot 4 C/D and slot 5 C/D are empty - you
need grant cards (or other Qbus cards) in there
to let the RQDX3 work. Just to make this more fun,
the RQDX3 will show up on the bus (it's CSRs
are visible) it just won't work :-)

The other reason it matters is that quad wide
cards need to do the right thing depending on
the kind of slot they are in. Most seem to arrive
defaulted for a Q22/Q22 slot (which means they can go
straight into a BA23/BA123 chassis with no
additional jumpering etc). Putting one of these in
a Q22/CD slot is usually no problem except that the
card is passing on the C/D signals to the next
card. This may or may not matter. For those cases
where it does matter, cards provide a means of
preventing this from happening. On the KDA50
there is a zero-ohm link (looks liek a resistor)
that you remove. Of course, if you move a modified
KDA50 from a Q22/CD slot to a Q22/Q22 slot you
are in for some fun :-)

Later series chassis (certainly the BA200 series,
probably the BA400 series) were wired Q22/CD
throughout.


>At this moment, this is the configuration:
>slot 1 rows 1-4: M7554-02 - 11/53-PLUS CPU
>slot 2 rows 1-2: M3106 - 4-line async EIA MUX
>slot 3 rows 1-2: M7546 - controller for TK50
>slot 4 rows 1-2: M7555 - RQDX3
> rows 3-4: ???
>All other slots and not mentioned row positions are empty.

>In slot 4 rows 3-4 is a card with just one 8-pos DIP switch
>three 16-pin DIL resistor chips and some decoupling caps.
>At the top are two BERG connectors, one 40 pin and one 50 pin.
>These two flatcables connected to an external unit, brandname
>Dynafive. Inside that box are several "Dynafive" boards and
>one board of "VGScientific". The rear of the box has several
>BNC connectors with markings that pops 'video' stuff to mind.
>Like H-sync, Green-in, Red-in, Green-out, etc.
>This box and the 11/53 were connected to eachother. On the
>disk in the 11/53 I found .RNO files that describe the VGS5000
>and how to use the application (something with spectral analysis).
>So far for the system description.

Many years agon, on a PDP-11/23, we had an external box
(about the same size as the 11/23 cab - including RL02 and
RX01) which was just a frame buffer (i.e. video card).

Your peripheral may well be similar (but this is just
a guess based on the fact that it seems to have
video connectors!)

>1) Can I remove that ??? card

I guess yes. Obviously you loose a really
cool peripheral!

>2) Put a DELQA or DECNA at the same position?

You can put a DELQA there. You can put a DEQNA
there (which is what I guess you meant). But a
DECNA is (IIRC) the really *rare* ethernet card
for the Pro 300 series of machines. It won't
fit and you'd destroy a cool card if you
put that there :-)

>I seem to remember that the RQDX3 must be the last device in
>the backplane for some reason.

It doesn't *have* to be. The reasons for putting
cards in a particular order is described in
one (or more?) of the Micronotes that you can find
on the web
(e.g.
http://www.ibiblio.org/pub/academic/computer-science/history/pdp-11/hardware
/micronotes/numerical/ )

Basically, the RQDX3 will hog the bus given half
a chance, but has enough buffering not to mind
yielding the bus to other peripherals (within
reason). The nearer (electrically) you are
to the CPU, the higher your priority. So the
RQDX3 (and KDA50) traditionally go furthest
from the CPU.

>So, should the Ethernet card go in slot 4 rows 1-2 and should I
>move the RQDX3 to slot 4 rows 3-4?

That would be normal. It will probably work even if
you don't move the RQDX3 though.

It's worth mentioning that the RQDX*2*/RQDX*1*
are brain-dead and do not pass on one of the
signals (GRANT I think). This means that they
*must* go at the end of the bus (nothing after
them will ever be able to interrupt!)

Antonio
Received on Fri Jan 04 2002 - 03:47:27 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:34:52 BST