On Jun 5, 2019, at 11:42 AM, Grant Taylor via cctalk <cctalk@classiccmp.org> wrote:
> Why did it require a micro? Could the host not perform the function that the > micro would do? […] Why couldn't that state machine be implemented in > software on the host using the modem & auto-dialer? Character-based I/O on mainframes and even minicomputers was fairly rare at that point. On some systems it was impossible; on others it was very CPU-intensive. I remember trying to do character I/O on a Tandem NonStop; it wasn’t easy, or effective. Having a dedicated microcomputer solved a lot of problems. My dad actually built a successful company in the 1980s to do just that: he build front-end controllers, which were essentially large number of serial ports connected to a rack of Z80 boards, which would then be connected to a mini (a Tandem) or mainframe via some bulk-oriented format. For example, I wrote code for him that let a typical terminal (eg, Wyse, VT-100) emulate a Univac block-style terminal. All the input, screen management, etc. was done on the Z80 machines, then shipped over in a multiplex fashion first to the Tandem and then to a Univac as block forms. > Reading the links that Ethan provided, it sounds like some auto-dialers did > use a second port, but it was not a second (recommended) standard 232 port. > Instead it was an RS-232 and RS-366. […] Or was the modem actually hard wired > in with no plug / jack at all? You’re getting warmer. ;-) Modems started out as straight modulators-demodulators, connecting two remote devices (computers, terminals, printers) over a point-to-point leased line specially installed from the local telephone company — basically like a T1 line would be installed today. The POTS line was hard-wired (via screw-down terminals) on one side of the modem and the digital data connector (eg, DB-25) on the other. Usually the connections were always on; I suppose the billing was probably by data usage, or even fixed price, instead of by time. > I have this mental picture, which I think is based on something I've seen at > some point in the past, that was a device that attached / actuated / ??? a > traditional rotary dial phone. As in it had a finger that interfaced with > the dial and something that could rotate it to dial the digit in question, > rewind (term?), and dial the next digit in question. Maybe someone created that monstrosity (;-), but the typical usage was that you used an acoustic coupler modem that had cups where a typical handset would fit. The modem itself only had a data connection to the terminal (or printer or card reader/punch). Next to the modem was a regular telephone — you dialed the number on the phone, and once you heard the carrier squeal, you’d quickly set the handset into the coupler. Usually you’d see a spurt of random characters on the screen which was the modem getting confused by the carrier being gradually synced up. To hang up, you’d simply pull the handset out of the coupler and hang it up as normal. I do recall a little handheld device with a touchtone keyboard that you could fit over the microphone of a normal handset. It wasn’t automated, but at least you didn’t have to use the rotary dial. (This presumed, of course, that the telco switch was DTMF-compatible.) —John
