Information Technology Reference
In-Depth Information
Ethernet, you cannot stack repeater hubs back to back. You will recall that you were
allowed to place up to four repeater hops on an Ethernet segment. Fast Ethernet lim-
ited that to two, and Gigabit Ethernet limits that to one.
Gigabit Ethernet (as well as other gigabit-class technologies) has a very finite
distance limit, depending on the medium. Part of this limit is due to the actual losses
in transmission relative to noise (copper, particularly) and part is due to the physics
of the transmission (fiber dispersion, for example).
Loss Measurements.
It seems self-evident, but you must use test and measurement
equipment that is appropriate for the gigabit range. Sure, you can use simple wire map
and optical continuity testers during initial installation. But to ensure operation at giga-
bit speeds, you must use cable scanners that are capable of measuring all the parame-
ters for the category of links you are installing. In this case, you should be testing to Cat
5e or higher, commensurate with the category of components you have installed.
For twisted pair, the measurements fall into four areas: attenuation, crosstalk,
impedance, and propagation delay. Attenuation and impedance problems cause
direct signal losses. Crosstalk, such as NEXT and ELFEXT, causes a lower signal-
to-noise level, and indirectly affects signal range. Finally, differences in propagation
delay degrade 4-pair performance. All these areas are critical in gigabit applications.
In fiber links, the essential parameter areas are attenuation and bandwidth, as
they are the primary factors that limit range. Other “soft” parameters may radically
affect fiber link performance. For example, multimode fiber may have to be mode-
conditioned to minimize the effect of normal fiber manufacturing characteristics
when used with the laser diode sources of gigabit technologies.
If you have a multimode fiber link that appears to pass testing, but will not
work in the application, the first place to look is at differential mode delay prob-
lems. Fiber testing is normally done with an overfilled launch (OFL) source, which
fills the fiber core with multiple modes.
DMD causes a problem when the core is illuminated with a laser source, which
uses only a few modes. This additional parameter is difficult to test for, so a simple
troubleshooting technique may be to keep a mode-conditioning patch cable handy.
If the link works with this special patch cable, you will need to provide one on a
permanent basis.
Another place to look on a fiber with excessive losses is a possible mode mis-
match. The connectors on single and multimode fiber are the same, yet there is a
substantial loss coupling from one to the other. A mismatch that is less disastrous is
a core-diameter mismatch. As we covered in the chapter on fiber-optic cabling, 50
and 62.5 mm core fibers may be coupled, but with a dB loss that may be unaccept-
able in a particular link.
