I wear two hats. One belongs to the owner of an Atari 800; the other to an electronics engineer with a large, independent telephone company. As an Atari owner, I am intrigued by the doors that open when a modem is attached. Suddenly, my computer can talk to another computer across the country. In fact, if I assert myself after all the digital handshaking is complete, I can actually talk to the humanoid at the other end!
This is all very nice, until I put on my telephone engineer's hat. Then I'm appalled at the runaway abuse of the telephone network by some of the modem designers and users. Half-baked equipment, some of it very expensive, is flooding the market. Equally half-baked documentation accompanying it promises features that the telephone network cannot possibly guarantee. As a result, the deceived modem user turns against the telephone company, complaining that his modem sometimes works, sometimes produces errors; or it doesn't work at home, but works at his friend's house; or he can place a call to a bulletin board 1000 miles away, but not to another board 100 miles away, etc.
First of all, the telephone network is designed to supply a voice grade service. And "voice" means just that -- not a private analog data circuit, which is what too many of us seem to expect. All telephone companies meet national performance standards on voice quality. These standards are monitored at the federal level and by the state Public Service Commission. Because the voice grade network performs so well, some modem designers and users make false assumptions about what they can demand from it. They think of it as a network that exists only to transmit their two sinewave tones in the voice band, without error.
In particular, they are very hazy about such things as: A. Data signal send levels; B. Switched network losses from call to call, that depend on the path chosen by a central computer in another part of the country perhaps; C. Random noise appearing on the telephone pair; D. Hum appearing on the line (courtesy of the local or distant electric utility); E. Impulse noise level, duration and frequency (these can look like data signals); F. Ringing frequencies and voltages, which are often as much as 100 volts AC; G. Echo amplitude and delay from satellite links; and H. The so-called "ringback" tone. (This is what you hear while you are waiting for your party to pick up his phone. Note: you do not hear his phone ringing. His phone may, in fact, be disconnected. You will still hear the ringback. So don't complain that you heard his phone ringing -- you didn't.)
Armed with this lack of understanding, a modem designer can easily turn out a marginal design. The modem's performance will then vary with time and circumstance, in a manner beyond the designer's ken. Nevertheless, he sells it to the innocent consumer, whose dreams are soon shattered.
A specific case, if you will. The customer had just purchased a high quality portable computer (an Osborne), with an integrated modem and audo-dialer. It worked well from his residence, and from his friend's house, but not from either of his two business lines. I spent much time and money lookmg for faults in the telephone plant. The cable performance was in limits, but we were loathe to blame a prestigious computer manufacturer. We should have, because eventually we discovered that the modem, instead of waiting for the distant computer to come on-line, was trying to shake hands with the telephone ring-back tone. So when this ring-back tone stopped after one second, the modem disconnected. The customer's place of business was closer to the central office than was his residence, and the stronger ring-back was confusing the poor old modem. In effect, the better the telephone loop, the worse the modem's performance. The customer was extremely embarrassed by it all, and told me he would get rid of the computer.
There is an industry standard governing the maximum tone amplitude that may be placed on a telephone line by a modem. This level is -9dBm. (i.e., 9dB below 1 milliwatt.) Sometimes, a modem user who has trouble getting through opens up his unit, finds the LEVEL control, and cranks it up to the maximum. He knows nothing about crosstalk. He may care even less about the interference he is causing to other telephone users. He is determined to get through, and the heck with everyone else. Modem users should note that it is extremely rare to have a problem with signal level on a modem circuit. A receiver can operate down to a level of -4OdBm, and often lower. So, starting from a transmitted signal level of -9dBm at the far end, all tones will arrive at a minimum of -25dBm. In other words, with at least a 15dB safety margin. So why crank up the send level and interfere with someone else?
Other problems have occurred because some modem designers fail to guard against impulse noise on the line ... may not have given it a moment's thought, even. Impulse noise is caused by household appliances, power surges, CB or ham radio transmitters turning on and off, or lightning. An impulse is broadband, and some part of it will pass through any mark-space tone filter system. A poorly designed guard circuit will allow this impulse to false-trigger the pulse generator at the modem interface. An error results, and the telephone company is blamed. The consumer has been brainwashed into thinking that he has the right to a private analog data circuit, each time he hooks up. But he hasn't, and we all suffer.
Yours truly,
Peter Vaughan
St. Petersburg, FL