56 kbps modems

[Norman Hardy]

At 5:09 AM 9/13/96, Asgaard wrote:
>>U.S. Robotics and Rockwell International are planning new modems with
>>speeds up to 56 kbps a second, almost double the speed of the fastest
>>rate now available.  The new devices should be available by the end of
>>the year, although their top speed initially may be less than 56 kbps.
>>(Wall Street Journal 12 Sep 96 B11)
>
>People who seemed to know used to say that 'the Shannon limit'
>set an absolute upper limit around 40 kbps. Has Shannon been
>proven wrong, or what?
>
>Asgaard

I imagine that both modems in a connection become phase locked with the
underlying 8K digital carrier. Then each modem signal element is carried by
just one 8 bit digital sample. That carrier moves 8 bit bytes 8000 times
per second. Stopping at 56 Kb instead of 64Kb means that a 7 bit DA
converter for the sending modem and a 7 bit AD converter in the receiver,
plus some fancy analog filters to undo the subscriber loop effects. Going
for 64Kb would require twice the signal to noise ratio on the local loops.

I think that modern PBXs have a digital link to the phone company. This
would mean that an ISP would not have to buy fancy modems. A modified PBX
could transmit and receive bits to a computer directly. It will impact the
phone compnany's ISDN service but I don't think that they can stop this.