"I also know that someday they may be able to raise the levels and allow the 56K instead of 53K as the max. If they do get to do this and the "level" they are referring to is dB into the line, they may be less susceptible to the noise problems as the S/N would be better. Any idea what "level" they are talking about? Think they're worried about harmonic interference with the regular traffic on the trunks?"
The level they are talking about is, as you suggest, the power level going onto the phone line from the modem. In order to get FCC certification, which is required to connect to the phone system (legally, at least), the output of the modem must be less than -11 dB into a nominal 600 ohm line. You guessed right;if this level is too high, they could have trouble with harmonic interference of sorts, because everything coming out of the modem will be at higher power levels than they are designed to handle. They have some pretty good "brickwall" filters to limit bandwidth, but I'm not sure how they handle excessive power. I assume they'd have limiters there, too, but I've never bothered to find that one out. I guess I will now!
The analog signal from the modem is converted to digital, as you know, before being placed on the backbone. If the signal level is too high, they get all sorts of harmonic problems, distortion, etc. It may not seem like much of a problem to allow modems to place a little more power on the line, but when you combine all the lines as they do, it can be substantial, and they will have to study the situation a while before deciding if they will make the exception. Probably not.
CHOPPED UP DESCRIPTION OF 56K: The PAM carrier used for 56K changes amplitude in response to bits being transmitted, which is how the carrier is modulated with information. The number of discrete levels possible in the carrier determines how many bits may be sent in each amplitude "pulse" or cycle, and the overall speed of the connection. The number of possible levels is in turn determined by the maximum allowable power level of the signal divided by the power level of one discrete step; the power used for each step is then determined by the signal to noise ratio on the line. Sooooo ..... if you have a quiet line, the power needed for each discrete step is lower (since the noise which interferes with detection of this level is lower) and you can place more steps, and more bits, in each carrier cycle; the total power in the carrier is limited to -10 dbm at the modem, and is in the mid-twenties, usually, by the time it gets to the other modem. Since the total power is limited "by law", the only way to get more bits per carrier cycle, and thus more speed, is to pack more discrete levels into each cycle, which requires good S/N ratio. If the noise anywhere in the system goes up (temperature?), the step size of each carrier cycle must be larger, to be detectable in the presence of the noise. This reduces the number of steps per carrier cycle, and thus, the speed of the modem. Anything which changes the carrier amplitude, and to some extent, it's phase, is "distortion". AWGN, thermal noise, RF energy picked up by the wires, 60 Hz from household appliances, anything which produces electrical impulses on the wire, reduces the S/N ratio and the speed of the modem.
By the way, you may wish to try to disconnect extension phones, fax machines, etc., from your phone line when you use the modem, and see if that increases connections speed. Occasionally, I've seen some phones and other devices interfere with modems and impair speed. If you try it once and don't notice anything, chances are you have a "clean" installation and won't need to bother with it. This may sound a bit silly but honestly, I've seen these types of problems often enough to make it worth trying, especially with 56K modems and their amplitude sensitivities.
I'll peek at the "other" thread once in a while just to see the fun.
Larry |
