2 May
2013
2 May
'13
10:20 p.m.
On Wed, May 1, 2013 at 10:32 PM, Johnny Billquist <bqt at softjar.se> wrote:
Are you sure about Able? I can't remember seeing any such thing from them.
Quite sure - the "other KO" designed it - Ken O'h* (I don't remember
how to spell is last name). Ken O'h formed Able after Olsen shut him down with the
CalData machine (an 11/45 clone).
The Able 16 port serial product was called the DHDM. I wish I still had one. I do
still have the doc set for them in the basement, along with his original
"Enable" card he built for us UNIX guys - which was a slick take on a cache and
bus repeater. +
The DHDM could definitely do speeds of 19.2K but I think they could do 38.4K and may be
higher, I forget - have to look at the prints ;-) I put the code into BSD support the
higher speeds, base AT&T used the EXT A/B stuff. I'm pretty sure the original
DEC DH maxed at 9600.
I want to say it was at a sumer USENIX conference the late 1980's (Austin maybe) that
a number of us tried to get Ken O'h to put the DHDM on an ISA bus, but Ken O'h
did not think he could make enough money at it (he was used to PDP-11 peripheral pricing,
not PC community pricing). Eventually others created the "Rocket Port"
board became the RS-232C board for much of the early PC/386 based UNIX systems (I think I
do still have one them)..
As Ken O'h once said to a number of us. DEC's KO taught him a lesson and he was
careful to never create a processor again. DEC shut him down for
>>sourcing<< the UNIBUS. His Caches/Repeaters were as far as he would go
from a product stand point. A few years later, he did manage to splice a 68010 on the
the Unibus, but I don't he ever sold it because he was scared of KO. I want to say
he gave them to a few people to play with. I saw the board when I visited him in ~1984
but never programmed it. But I think I remember that some of the BTL folks might have
had a few.
As for your other comment about "enough for common use" WRT the modem control
lines - actually that was not true for the UNIX community. DZ was short pinned++ one
had enough to sort of support >>dial-in<< (i.e. off-hook/CD detection but
because if used wanted to run uucp, modems needed to control an autodialers and thus
needed to support the whole magilla. Remember UNIX comes from the TPC (The Phone
Company) - so base UNIX had all the support for AT&T communications equipment.
When modems started to add the autodial "in-band" (the @ command stuff - which
was not how the AT&T 212 stuff worked), that gave you dial out - but required XON/XOFF
for flow control. Trying to use an in-band flow technique like XON/XOF was an
anathema to an 8 bit protocol. So the fancy modems like the Telebit et al used an
out-band scheme keep from overflowing - i.e. the modem control signals of RTS/CTS used
s a handshake.
In fact on the very cool things that we got Ken O'h to do on the DHDM, was connect the
input (which I think I remember is CTS - I've forgotten) and the USART interrupt
through an NAND gate so the processor did not get the final "data sent
interrupt" until the downstream device was ready for it. This made a huge
difference on PDP-11s and Vaxen.
Clem
Footnotes:
+ I did the SW support for the "Enable" when I was at Tek. Johnny since you
have asked me for source, one of my earier academic papers was: C.T. Cole, S. Huxley,
"A Large Address Space UNIX for a PDP 11/40" that describes it.
++ You'll have to accept this as it or send Dave Cane a note directly if you want
verified. Dave as the head of the 750 project (had been part if the 780 did some 11
stuff before and thus did not realize that all the signals mattered). When Dave did the
first version of the Masscomp machine he stole the modem control signals from the console
to create a clock interrupt. I did not join the project until after was too late to fix
that. What a mess ;-) But Dave learned, RTS/CTS was very important and did not do
that again (at least that I now of).