8 May
2003
8 May
'03
6:30 a.m.
--- Rick Bensene <rickb(a)bensene.com> wrote:
In response to problems that I'd mention regarding
getting the RL11
controller to happily co-exist with the RX11 Ethan Dicks said:
Cards that use the circuit-board-based jumper plugs frequently do.
If possible, try putting the RX211 at the front,
then test.
So, I followed this advice, and greatly so, everything worked fine.
However, I later found that the position wasn't really the issue.
The problem was that the BR selection header on the RX211 controller
really squeezez the tolerance in terms of inter-board spacing. What was happening is that with the RX211 behind the
RL11 controller,
the BR selection header on the RX211 was apparently shorting some
connections on the back of the RL11 controller. So, *that's* what that
fully looking piece of semi-transparent orangish-material that I found
in the base of the cabinet when cleaning things up...apparently
it was some kind of insulator placed on top of the header to prevent
just such shorts.
Indeed it is. I must confess that the possibility hadn't occurred to me
(novice vs. expert expectations and what not). Yes. There's supposed
to be an insulator on the top of the BR selection PCB. The adhesive
does frequently dry out and fall off.
Very late model Unibus boards have a BR selector that looks like a DIP
resistor (yellow body, no exposed conductors). In many cases, though,
newer designs just didn't have a selector. The BR level was fixed.
Fortunately, the shorts didn't cause any damage.
Frequently the case with DEC stuff... short that cause damage tend to
show it right away (with magic smoke heralding the event!)
When I swapped the boards' positions, the
interference didn't exist
anymore, and everything works like a champ.
Given the boards in question, an admirable solution. It can be a problem
when you are trying to load out a DD11-DK and want to use the SPC slot
next to the terminator - I have tended to put things like LP11s there.
The RL01 drive works flawlessly, and both the RX02
drives also work
great.
Congrats.
Here's another question for you DEC folks:
The system has the main backplane... then there's the RK11 backplane...
(then) a 9-slot expansion backplane.
Question is: Is this correct? Should the RK11 backplane... be situated
BETWEEN the main bus and the expansion bus, or should it be located at
the end?
It kinda depends on what else is in there. There are some peripherals
that want to be very close to the CPU because they can't buffer
operations and would starve if some other device were in front of them.
The RL11 should have a "silo" (the RL8A does, I'm pretty sure) that
allows it to be somewhat patient about initiating transfers. I would
expect the RK11 to be less sophisticated, and more sensitive to timing
issues. Ethernet cards and streaming tape controllers are other classes
of devices that tend to want to be near the front. Terminal cards and
line printer controllers tend to be more tolerant, owing to the relative
slowness of the I/O (compared to disk and network).
So... if it's there right now, I'd probably leave it, unless you run
across a comment in a DEC handbook that suggests that it needs to be
somewhere else. Unless you are getting data errors with peripherals
behind it, it's probably not hurting anything.
-ethan
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com