The distance most quoted for standard multi-twisted pair Cat cable is
100 meters. I assume this is worst-case.
How much can this be streached? Does it degrade in effective speed,
or does it just fail after a certain length? Will it fall back to 10
Mbps?
I'd like to run about 650 ft, outdoors, rural, with some cable that
plugs into RJ45 at each end without adaptors. I don't need more than
3 Mbps. What are my changes?
Phil
Possible solutions:
1) Wireless solution
2) Free space optical (I had to mention this, because it is so cool).
It seems the countries using this, can afford tall buildings, but
cannot afford wires
http://ronja.twibright.com/installations.php
http://ronja.twibright.com/
3) Power line networking - if the outbuilding and the main building
are powered from the same (North American) pole transformer, then
perhaps this would work. You would still have the danger of
lightning damage reducing these devices to ashes. This is the
latest generation of technology, and is more adaptive than the
older stuff. At the extreme distance you are planning on using,
this might not work. Just had to mention this option for completeness,
because you don't have to run any cables if the power lines are
already there.
http://www.theinquirer.org/default.aspx?article=31522
http://www.netgear.com/Products/PowerlineNetworking/PowerlineEthernetAdapters.aspx
4) Fiber optic Ethernet. Cures the lightning danger problem. Years
ago, the fiber I used to work with, cost ahout $1 per meter, but
the price will depend on how the fiber is jacketed. I have no
idea what a current price would be. Or how best to string
or bury it.
5) Your CAT5 plan. (Lightning danger, even if buried.)
10/100BT Ethernet has no ground. All four pairs of wires float.
Lightning can induce voltage on a cable, even without hitting the
cable.
A web search presents conflicting information on the topic of 100BT
Ethernet.
1) The length limit seems to be related to the Ethernet slot time. For
100BT this is 5.12 microseconds (512 bit times, 10ns each for 100Mb/sec
communication). This is to allow collision detection in "half-duplex"
applications. (The 5.12 microseconds is the max time of flight, from
one end of the cable to the other end, and then back again.) I don't
see any references to what improvements can be expected with full duplex
operation. If there was no Ethernet slot time/collision detection issue,
then you would need info on what limits are imposed by attenuation and
other cable effects.
http://support.intel.com/support/express/switches/sb/CS-014409.htm
2) This is a datasheet for an Ethernet PHY. The chip has adaptive equalization,
which is a means of compensating for cable length. A register on the chip
monitors the equalizer, and reads out what it thinks the cable length is,
based on the degree of equalization needed. See page 41 and page 9. On
page 9 it mentions "150 meters".
http://tech.am/data/AC101-DS01-R.pdf
This doc is slightly unrelated, but it does give a figure for the
propagation speed of CAT5 cable (71% of the speed of light).
http://marvell.com/products/transceivers/singleport/VCT_White_Paper.pdf
In view of the safety issues, I wouldn't spend any more time on
wired solutions. Something that doesn't use wires, is less likely
to burn something down or blow something up. Wireless technology
is a simple solution.
If you have radio spectrum to burn where you are located,
draft 802.11n is the latest wireless method. Apparently in
urban settings, 802.11n has had some impact on a neighbour's
wireless operation, so something like this is good in a
rural setting, as there may be fewer people to complain about
you messing up their Internet. But this might be overkill for
what you want.
http://www.linksys.com/servlet/Sate...513404&pagename=Linksys/Common/VisitorWrapper
Paul
