2 May
2012
2 May
'12
12:16 p.m.
The problem arrose because you could link
'hosts' (computers)
directly togther with RS232.
And that's what a null modem is for. Conceptually, it's two
back-to-back modems, implemented by non-straight-through wiring in the
cable (or connector or whatever).
Twisted-pair Ethernet appears to have learned nothing from the
serial-line confusion; it went with two different wirings (`host' and
`hub') and thus repeated the straight-vs-crossover cable issue in a new
field. It would have made so much more sense, even more than for
RS-232, for everything to have the same wiring, so there's only one
kind of cable involved. Instead, we had two kinds of cables for a long
Which presumably would ahve bene crossed. That is, have the Tx pair and
Rx paor on the same pins of the 8p8c socket (not RJ45, please!), and have
the calbe link Tx at one end to Rx at the other.
time, and even now we have special hardware
capabilities (called
"auto-X" or "auto-MDI") to compensate for the whole mess.
RS-232 can, to a degree, be excused by its vintage; when it was
developed, the zeitgeist was much more heavily into the DTE/DCE
dichotomy than now. Twisted-pair Ethernet had much less excuse.
RS232 ws designed ot link a device (DTE) to a modem (DCE). It was
essentiall the description of how you could conenct to the modem you
rented from Ma Bell. If you;re goign to use it for soemthing else, it's
not suprising there are problems.
-tony