On Thu, 28 Apr 2016, Christian Kennedy wrote:
On 4/28/16 17:45, Sean Caron wrote:
[big snip]
I find the design of the CBX really interesting. IMO, their appearance
belies that ROLM was a computer vendor first a a phone equipment maker
second. Not in a perjorative sense, just stylistically. Comparing them
against boards from WECo/ATT/Lucent/Avaya, Nortel and Harris.
I was a staff engineer at ROLM MSC between '82 - '86. By that time by
any reasonable measure MSC and telecomm were two utterly different
companies that happened to have common parentage; technology cross-over
between the divisions was for all practical purposes nonexistent
(although we did have the occasional employee move between divisions,
particularly after the IBM debacle) -- but it certainly seems that
experience building stuff on the MSC side informed *some* of the early
design decisions on the telcom side.
Very cool, thanks for the perspective. My understanding was that the
profits on the mil-spec computers side bankrolled the entry into the PBX
market but I was never clear on how much overlap there was between the two
divisions. Any awareness if the common control of the CBX is in any way
architecturally related to the DG Nova?
It seems like it would be advantageous to leverage the experience building
mil-spec Nova processors on the other side of the business, and looking at
the hardware, it is clear the CBX is a 16-bit machine. But never been able
to confirm.
One thing that would be a huge score for postierity and might help to
answer some of these questions is a copy of the first half of the System
Service Manual for the CBX. I have Part II which contains a command
reference and some discussion of peripheral cards, but everything relating
to system architecture and the design of the common control seems to be in
Part I. Do you know of anyone who may have preserved this? I've spoken
with a few old hands who were CBX switchmen in the past but all I've been
able to get is Part II.
IIRC the most interesting thing about the CBX was that it could do so
much with so little hardware (relative to other switches of the time)
thanks to TDM of the 12-bit bus through the "connection table", which
was a 384 slot recirculating command buffer that drove the codecs, dial
tone generators, tone decoders, ring generators and the like. Basically
the CPU would schedule the sender and receiver for the bus by dropping
commands into two parallel queues (one for transmit, one for receive),
so there was no need for bus request or arbitration logic and yet the
CPU could be slow, as the sequencer would just advance through the
buffer every 83usec processing the commands that it found. It was a
pretty clever way of substituting DRAM for bus control logic while
reducing processor requirements.
I had watched a Youtube video which discussed a little bit about the
design:
https://www.youtube.com/watch?v=A8J6CGI6HA0
However what you write is much more detailed. I believe there is an
apocryphal story behind that particular part and sampling rate although
the specifics elude me at the moment ... I understand it caused trouble
trying to interface the CBX to DS1 circuits.
MSC was effectively the alpha site for new builds and new hardware, and
we saw failures that at times left us without reliable phones for a day
or two. One of the more interesting was when the switch refused to
honor extension status changes and instead entertained itself by ringing
each extension *once* in ascending order, then repeating.
Sometimes my Definity gets confused and does that. Surprises the heck out
of my partner and me. Good thing it hasn't happened in the middle of the
night :O
Thanks,
Sean