Re: [nsp] Maximum length for RS-232 for serial console connections?

From: Jim Warner (warner@cats.UCSC.EDU)
Date: Mon Feb 04 2002 - 14:59:37 EST


> In an effort to do a 'poor man's out of band management', I would like to
> make long runs of cable from the console ports of various routers to a
> centralized location to connect to a terminal server. Anyone have any idea
> what the maximum length would be for such a cable?
>
> Thanks,
> Tim

=====================
RS-232 is ground referenced. If your long cable runs go outside
the building, you can fry the chips if lightning or a ground fault
breaks the assumption that ground is the same in both buildings.

All things being equal, you can go farther if the cable has each
of the Tx and Rx pins twisted with its own ground wire. If you
make up a minimalist cable that happens to have Tx and Rx twisted
together, you'll maximize cross talk and hurt your distance.
Using a long flat cable is probably worse than twisted-with-ground
but not as bad as twisted-together.

The reason this question is hard to answer quantitatively is that
RS-232 drivers are permitted to put out a signal that is +/-
15 Volts Pk-to-Pk to as little as +/- 5 Volts Pk-to-Pk. If your
drivers are of the higher voltage variety, they can send a signal
farther.

Having said all that, if the resistance of the copper along your
extended cable is less than, say, 100 ohms, you can assume that
the the console connection will work at _some_ speed. You may
not get 9600 baud, but 2400 baud is probably work.

If you read about Cisco's config register, you can set the power
on speed of your console ports lower than 9600.



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:31 EDT