Slashdot Mirror
A Brief History of Modems
Ant points out this two-page TechRadar article about the history of modems; the photographs of some behemoth old modems might give you new respect for just how much is packed into modern wireless devices.
← Back to Stories (view on slashdot.org)
how much I miss my original mod [NO CARRIER]
512 MB RAM, 20 GB disk, 200 GB transfer, five datacenters. $19.95/month.
There are still a few of us left who grew up in the acoustic coupler era, where modems connected to the (back then standardized) handset, and really whistled and purred into the microphone.
Speeds? We started with 110 baud (which back then was equivalent to bits-per-second, if you subtracted stop bits). Then came 300 baud.
Then someone had an epiphany, and figured out that no-one could possibly type faster than 75 characters per second, and even if they could, the printer(!) that spit out whatever you typed wouldn't be able to. So by reserving the low frequencies for upstream data and the high frequencies for downstream, you could achieve the blazing speed of 1200 baud down and 75 baud up. The 1200/75 modem was a workhorse for a long time, with way faster downloads than 300/300 could give.
Then came 1200/1200, 2400/2400, 4800 (which was really 2400 with compression), 9600, and then the Trailblazer, which was running at a ridiculously low baud rate (100 baud IIRC), but at so many parallel channels that it achieved ~18000 bps aggregate. That was lightning fast! Imagine almost 2 kB/s (unless something moved the other way at the same time, in which case speeds of course would drop). The ASCII porn didn't stand a chance against that speed monster!
Then came the short-lived 38400, and finally the ubiquitous 56k modem. Yawn.
In the mid-90s, we got BRI (ISDN, 2*64 kbps in most of the world, 2*56 kbps in the US). Which pretty much ended the modem era, except for in the US and UK, where 56 kbps POTS modems reigned supreme until well after the millennium.
I have mod points to burn but I have to post in here.
The traffic system I worked on had 300 baud modems attached to cheap leased lines (soldered in, mostly). Two modems per card. 8 cards on a bytecraft backplane. Up to 128 modems on a 19 inch rack. Each modem had three LEDs (carrier, TX, RX) and at the speed the system operated you could see the poll/response from the regional controller to the sites and back. In the dark it was a thing of beauty. Computers of old.
If something was wrong in the logic (say a checksum mismatch) then you could see it in the LEDs because one channel (slot) would not follow the nice pulse sequence. Several times I mucked up the checksums of a rack and took out a lot of sites. Maybe I shouldn't post about that...
Going back in time my 6502 system had a modem for the cassette interface. I knew you could overclock the UART and FSK modem driver and I had dreams of using my uncles reel to reel hifi system for storage. Never happened. Though I did find that you could use the cassette player as a sound card of sorts by locking on REC and PLAY.
http://michaelsmith.id.au
In "The African Queen," Katherine Hepburn's character asks Humphrey Bogart's character to make a torpedo. Bogart's character says something to the effect that "Lady, there ain't nothing so complicated as the inside of a torpedo. It's got gyroscopes, compressed air chambers, compensating cylinders..."
I remember once reading details about just how the signals in a 1200 bps modem worked... and modems at higher rates. It was just jaw-dropping how sophisticated it was. The reason why there was a distinction between "bps" and "baud" is that "baud" refers to the number of times per second the signal changes. Well, a 1200 bps modem only changes its signal 600 times a second... but it uses four different combinations of frequency and phase, so each signal combination signals two bits. That's bad enough, but the combinations literally increase exponentially. The 9600 bps modem actually requires the receiver to sense and distinguish sixteen different analog combinations (so that it can encode four bits at a time).
At the time I figured they had to be close to the theoretical limit, which depends on the bandwidth and the noise level. A phone line is only good up to about 3000 Hz. so the 2400 baud rate of a 9600 bps modem is changing about as fast as it can. The rest depends on how noisy the line is.
Theoretically, of course, you can signal at an infinite rate on a perfectly noise-free channel. Just send 3.141592653 volts on the end and measure it with a ten-digit digital voltmeter and, voila! You're sending ten digits at once. Except there aren't any ten-digit voltmeters.
I was frankly flabbergasted when they managed to cram 56 kilobits per second into a phone line. Of course, the 56 kb modems never really ran at that speed--they were always falling back to lower speeds because the phone lines were too noisy. Then they added compression, which didn't do much good because the ZIP files and JPGs you were sending were already compressed. In reality they were trying to cram 56 kilobits of data into a 33 kilobit bag, but it was amazing that it even worked some of the time.
But, lady, there ain't nothin' so complicated as the inside of a modem.
"How to Do Nothing," kids activities, back in print!
The biggest problem with using modems was that you had to let everyone in the house know you were on the "modem". This meant, sticking post-it notes to every phone in the house, so that someone would tell you they needed to use the phone rather than just picking up the phone and dialing. You also couldn't tie up the phone for hours on end. There was very very few people that had an answering service (not an answering machine), like most do today with VOIP or CableCompany Provided Voice.
You also had to remember, if you were one of those people that had it, disable call waiting, as many modems would drop the connection when a call waiting signal came through. I believe you had to add a *70 after the AT.. so you had something like:
Today people can spend all day actively or passively (by leaving the computer on) online. Wit
Otherwise how could you think that v.32bis (14.4k) was introduced in 1980? I had to look it up to see what the hell they were on about, apparently the 1980 figure comes from a break through channel coding paper written in 1980 at IBM that didn't even get passed around for a few years. The reality is that the public had to wait nearly a decade before those techniques were out of the lab, and a few more years before a standard was ratified. Trying to figure out what niche this article fills - the wiki article on modems does a far better job at going over the same info. Hell, the author of TFA even put an old-time(tm) bw filter on a photo from the late 80s trying to make it seem like a shot with a laptop came from the 60s.
Its a shame that the article missed so much....
Like the times when much of the industry didn't want to license the Hetherington Patent from Hayes on the "guard time" surrounding the "+++" in the escape sequence, so Hayes ended all of their press releases with +++ATH0 (which would cause a lot of modems to hang up on the BBS systems of their day).
They also missed the interesting fact that the "56K" modem was an old idea that was rattling around Bellcore for years before 1996 and fairly common knowledge in the Bell system. [The big issue with getting there was the need to have digital trunks connecting all of the dial-in server pools with the telephone network.]
Probable never would have become a mainstream consumer device without AOL. Until AOL, you really had to be a geek to use one.
And, of course, the modem wars of 1996-1998, as the major technology companies duked it out, the vast majority of modem companies went bust, including Hayes.
]P.S. The word MODEM (as the article indicates) represents MOdulatorDEModulator. Hence it should be capitalized. This is also try of enCOderDECoder (CODEC). Slightly less related yet as correct LASER and RADAR....
Generally when an acronym is pronounced as a single word and has entered general usage, it is not capitalized. These days scuba, laser, and radar are not capitalized. Nor is modem.
The word MODEM (as the article indicates) represents MOdulatorDEModulator. Hence it should be capitalized. This is also try of enCOderDECoder (CODEC). Slightly less related yet as correct LASER and RADAR....
Okay, okay, fair point, but ...
People were tying up VOICE channels
Come on, that one you just made up.
Information theory is life. The rest is just the KL divergence.
I once worked at a place that had a DEC/VAX mini with a bank of about 8 modems for VT100-compatible terminals. If there were modem complaints such as dial-in problems, I had to first figure out which modem was connected to which phone number. Others didn't always keep the map up-to-date. Plus, it used busy-roll-over.
The test phone was a ways away from the modem bank for the VAX minicomputer, so I had to keep the modem trying to connect long enough until I got there to see which modem answered the call (via LED). The only easy way I found to do this was to manually whistle an acceptable modem tone into the phone in order to trick the modem into thinking I was a modem trying to connect. This would keep it trying long enough to allow me to run from the test phone to the modem bank. It had to be the right pitch and wavering to work most of the time. I got pretty good at it after a while. I learned to "speak modem" a bit.
A computer-room technician once saw me whistling modem sounds into the phone and running back and forth. I later told him why, and he told me I was nuts and mumbled something about whistling sweat nothings to my robotic girlfriends.
Table-ized A.I.
The American Heritage Dictionary of the English Language, Fourth Edition says, "acronym n. A word formed from the initial letters of a name, such as WAC for Women's Army Corps, or by combining initial letters or parts of a series of words, such as radar for radio detecting and ranging."
MOdulation/DEModulation certainly seems like it qualifies to me. It is using the initial parts of a series of words. I don't see how it is any different than RAdio Detecting And Ranging.
This is how I remember it. Hayes modems, using the patent, required a certain amount of delay time surrounding "+++", their escape sequence, before the modem would recognize it. Thus, "+++" in the text stream wouldn't trigger it under normal circumstances because it would come and go too fast.
But the patent was a patent on the delay; and to avoid paying for the "delay" royalties, other modem companies would just use "+++" without the delay for their escape sequence, which risks modem confusion if accidentally sent as text, but otherwise wasn't that common. However, to embarrass non-patent modem companies, Hayes embedded "+++ATH0" in their digital documents. This would cause non-Hayes modems hang up if they ever transferred such documents. The trick sounds rather Microsoftian.
I remember other vendors complaining to the press, saying "you cannot patent pauses. Next they'll patent Ummm's" or something like that. (Obama would have a big bill if they did.)
Table-ized A.I.