On 4/7/10, Bob Armstrong <bob at jfcl.com> wrote:
For the right
price, I could part with my one 8200, but it would
obviously need a KA825 rather than a KA820.
I have an 8350 in the garage, complete and working, and I guess I'd part
with it too for enough money. I'd hate to see it go, though - it's a unique
machine in several respects. First, AFAIK it's the smallest VAXBI
"mainframe" and therefore the only BI bus machine that I'm ever likely to
have as a pet. It's also the only multi-processor VAX system that I have.
I'm right there with you. That's why I keep mine around (that and I
have a VAXBI peripheral board I helped design and did the ROM code and
device driver for, so I'd like to have a place to run it).
Gotta love that VAXBI mechanical design, too, what
with no cables
connected directly to the cards. It's a really easy machine to work on and
swap parts in.
I'm not so keen on that aspect of the design - we had numerous
problems with our 8300 that were resolved by reseating the cards. I
don't mind the cables on the back (though we did have a design-induced
problem because the lead hardware engineer dismissed my feedback,
assuring me that no customer would ever plug the cables in wrong - in
fact we had *every single customer* do that since in the DEC world, no
harm usually comes from inverted cables, so when something doesn't
work, it's common to flip the cable and try again), but the ZIF
backplane sockets were always a PITA to me.
You have to remove a ton of screws to get the cover
though, and after all that it won't run with the cover removed anyway (the
airflow sensors trip and shutdown the machine w/o a cover).
All true! When we were debugging our product, we needed to hang an
analyzer pod and an EPROM emulator pod on our board, and the cables
were too thick for the original lid to close enough. We ended up
making a 3/4ths "lid" out of cardboard and routing our cables out
through a new hole. There was enough airflow to keep the PSU happy,
and we were in a machine-room environment, so the ambient air was cold
enough to keep the boards happy.
My former employers paid $13K for that machine c. 1990, equipped with
1 CPU, 4MB of RAM (2x2MB, IIRC) and a KDB50. Later, we added another
KA820 ($100? $150?) and matched the microcode revs and made an 8300
out of it for SMP driver testing. We never did get a serial card or
ethernet for it, though I picked up a DEBNT for $50 or less in the
early 1990s. There used to be a scrapper down the road from my house
- besides things like MicroVAX Is, he occasionally got in VAXBI boards
- as a result, I picked up a 4MB board or two for under $50. My
configuration right now is one or two CPUs, variable amounts of memory
between 4MB and, IIRC, 12MB (depending on how many slots I have
available), usually a COMBOARD-BI (my product), a KDB50 (two slots), a
DEBNT (not yet working - need to check if I have the right cab kit for
the cables), and a DWBUA (not yet working - need to fix a UET module
before the DWBUA self-test will complete). My box has the two BI
cages; the DWBUA has external cables to a BA-11 in the next cabinet
over (which holds a couple of SDI disks). For logins, I use the
built-in four serial ports.
I just went googling for DEBNI/DEBNA/DEBNT information. Found this
... where Antonio Carlini mentions that the DEBNT might only be
supported in the VAXstation 8000. It's been years since I tried to
work on it - I don't recall where it shows up or fails to show up, but
I _think_ it passes self-test (amber LED), I think its registers show
up on the bus (there's an easy way to poke around on the console
reading physical I/O registers), but I don't think I have an ESA0: or
similar device at the VMS level.
If I had to replace the DEBNT with a DEBNA or DEBNI, it wouldn't break
my heart, but I do happen to have the DEBNT handy, so it'd be nice to
see it work. In the meantime, I have a long serial cable running
between the basement and my "office" room.