On Fri, Feb 28, 2020 at 6:40 PM William Pechter <pechter(a)gmail.com> wrote:
Most of the RS232 spec seemed to be designed for Sync
modems and their
management.
Most machines of the mini generation seemed to use either Async or Sync
interfaces. Stuff like the VT180 had a comm port that was a 8251 USART
for serial comm that could be either sync or async. I don't believe Dec
had anything like that in that PDP11 or early Vax days.
Anyone care to enlighten me?
I know some of the story having known some of the players and lived a
little of it, but I do not claim to know all of it. So I might be able to
fill in some holes, but there is still plenty missing from the complete
story.
First, remember RS-232 is an ECMA spec for connecting communication gear to
data gear. It's not so much about sync/async as it dealing with the phone
systems in the US and Europe and how to interface computer gear to it.
The driver of the spec was building and connecting RJE-like systems for
banks and financial institutions, airline terminals, *etc*. to connect to a
central (mainframe) computer. This is why it uses terms like "Data
Communications Equipment" and "Data Terminating Equipment" - as opposed to
modems, computer terminals, hosts and the like.
Different systems vendors had different ways of thinking about the computer
they were selling and how people would interface with them. And you can
see the differences in the choices they make in gear like the peripherals
that they interface at the time.
The AT&T Teletype Model 28 (
https://en.wikipedia.org/wiki/Teletype_Model_28
circa 1953 intro, 5-bit BAUDOT code, current loop) was the standard
terminal on DEC systems for many years. Gordon Bell invented (patented)
the UART to talk to it for the PDP-1 (maybe it was the PDP-6) sometime in
the early 1960s. What I do not know/understand is what was the work he
did at DEC and what when he had left to be a CMU prof. I was >>under the
impression<< the patent was granted during his CMU time; but I had thought
DEC originally built them as FLIP-CHIPS for the PDP-6.
Then in 1963, 7-bit ASCII was introduced. IBM and AT&T were to two
biggest firms
behind it (remember that the IBM System 360 was supposed to be an ASCII
system and has a lot of support for ASCII in its ISA; but due to the OS SW
being late that stayed with their earlier BCD – creating EBCDIC - read Fred
Brook's "The Mythical Man-Month
<https://en.wikipedia.org/wiki/The_Mythical_Man-Month>" for the details).
However AT&T, GE and DEC did switch to 7-bit ASCII pretty much as soon as
they could. The AT&T/Teletype Model-33 (
https://en.wikipedia.org/wiki/Teletype_Model_33 circa 1963 intro, 7 bit
ASCII code/current loop, but lacked shift key on the keyboard) was
introduced soon thereafter and that became the standard terminal (7-bit
byte, plus parity and 3 bits worth of start/stop -- 11 serial bits per
transmission). My memory is that AT&T did not make sell an ECMA RS-232
version, but the aftermarket had a ton of converters between the current
loop interface at the ECMA standard.
So at the time, you have IBM using primarily synchronous interfaces, while
AT&T (Teletype) used asynchronous. IBM liked sync because of the fact that
it needed no wasted start/stop bits. They liked 1/2 duplex because their
devices were primarily going one way at a time. In the 60s, IBM's big
business has them connecting RJE stations and they would only much later do
1/2 duplex synchronous* terminals*. DEC was more interactive much
sooner, used Teletype's and thus was async and full-duplex.
I was also under the impression (*i.e.* once was told) that Western Digital
obtained a license to make the UART as chips but it was never completely
clear to me who held the patent (CMU or DEC) *i.e.* who/how WD got the
license from. But after the chips appeared, DEC would buy those chips from
WD for things like the DL/KL-11's and DH-11 interfaces and I think they
made something like the DL11 for the PDP-8. If you look at the schematics
for the early serial ports for the PDP-11, they are all using the same WD
chip.
Soon after the UART, WD also starts making a USRT, which (the best I can
tell) they seem to be selling to IBM and the com vendors for IBM gear. I
personally never programmed them, but they are in an old WD book I once had
(may still). I remember seeing them in some Gandalf gear in the mid-70s
besides the IBM gear, but I don't known/remember much more.
In the early 1970s, CMU used the same WD UART chips as DEC was using in the
DL-11, but had designed their own serial board, which we called the ASLI
(there were other differences but mostly the SW could not tell the
difference).
Nat Semi was a second source for WD at some point in the late 1960s, and by
the early 1970s, they started to design there own UART (as was pointed out
eventually they created 8250 and it's follow ons). I'm not sure when Intel
and Moto started to make them, but I think both the 8080 and 6800 families
had UARTs chips. MOS Tech did not originally, although later when
Rockwell became their second source, a UART for that family appeared too.
At some point in the early 1970s, the first USARTs start to appear. I
was under the impression, WD was the origin of them, but I do not know.
By the time of the 16-bit micros, however, many of the better serial
interface chips could be either synchronous or asynchronous under program
control. With the 16-bit chips, a Zilog USART chip was fairly popular at
one point from Macs to UNIX boxes. As other pointed out, because of the
PC/AT the Nat Semi 8250 stuck around as it had ended up are part of the 'PC
support chip family', even though it was a bit of dog and notorious for
dropping characters at high speeds.
For completeness, the Unix folks at BTL used the Teletype Model 37 (
https://en.wikipedia.org/wiki/Teletype_Model_37 ß 1968 intro, 7 bit ASCII
code, full U/L case ) as their native printing terminal. IIRC the ASR-37
had an RS-232C option as well as a current loop one from Teletype as the
industry had pretty much dropped off of the current loop standard by
then. Interesting side note, the AT&T/BTL programmers often did not have
'hardwired' lines in their offices, but used modem (there was the phone
company of course).