[lug] comparing cat 5e and cat 6 cables

[Nate Duehr]

On Tuesday 27 December 2005 05:18 pm, Lee Woodworth wrote:

> Has this changed? Netgear says 'UTP Category 5 or better cables'
> required for the GS108, a full duplex 1000BT switch.
>
> The wall jacks are a bigger concern for signal degradation than
> CAT5 cable, even for 100BT. So is Cat5E/6 just marketing hype?

The reality is "all of the above", to some extent.  Cat 5e is rated to 
carry Gig Ethernet signals for a shorter distance than Cat 6, etc.  

Because it's all about power levels in individual wires, cross-talk, and 
impulse noise between wires/pairs.  

At Ethernet speeds, any tendency for the cable to have high resistance 
will also "round off" the nice square signals (on/off, ones and zeros, 
etc...) passing through the cable.  At some length, the data signal 
would be unrecoverable.  You can see this on an oscilloscope that's 
fast enough.

Virtually ANY piece of wire will work over very short distances.  The 
greater the distance, the more effort is put into having a better 
signal at the far end.

Twists in wires tend to keep high levels of cross-talk from happening, 
because the cables cross at known distances and angles.  It's an effect 
caused by the angles the E & M fields cross at.

If you lay the wires side by side (Category 3 cabling standards), they 
can couple together via inductance/capacitance and pass each other's 
signals. 

Each "category" rating gets more and more strict about the internal 
layout of the cable, but they're all the same basic thing -- pairs of 
wires twisted together.  The higher categories can get into things like 
the actual distances specific wires must remain from each other, etc.

The standards are just standards for number of twists per foot, minimum, 
etc.  

In installations where the cabling is important, the real cable pros 
*test* every cable end-to-end in the installation to see how much 
crosstalk and noise is introduced on each cable installed.  Same thing 
with high-priced RF feedlines at broadcast, cellular and 2-way radio 
sites.

Time-domain reflectometers that also are programmed to know the "rules" 
for the various cable types are used -- they can tell if an 
installation is botched or a cable has an internal manufacturing 
defect, almost instantly.

It takes a *good* test set to do this stuff.  Some cost a few thousand 
dollars -- to really do it right.  They can give you printouts of 
graphs and also see other unforeseen problems like that run of Cat 6 
that just happens to pass a foot or two from a noise fluorescent light 
ballast, and it is picking up all sorts of electrical and RF noise... 
etc.  

Anywhoo... to *try* to answer the question about home stuff:

In short runs (residential), Ethernet... even Gig Ethernet... is fairly 
forgiving.  You'll probably "get away with it" most of the time if you 
have 5e.  

The longer the runs or the cheaper the connectors and wall plates at 
either end, the more chance induced noise and/or cross-talk will affect 
the signal and degrade it.  

An example of degrading signals at the termination points: 

Many building installations of Ethernet include punch-down blocks like 
"110 blocks" or "66 blocks".  Neither is rated to reduce cross-talk 
enough to truly be used in a 100Mb Ethernet or faster connection, but 
they're used all the time, and tend to work... 

Punchdown blocks like "3M blocks" *are* rated for at least standard Cat 
5, but are a giant pain in the backside to use and require expensive 
bits for your punch tools that can run $50-$80 for a single bit.  (It's 
because they mechanically have a pair of scissors integrated into 
them... etc.)  Most companies take the "performance hit" and choose to 
use the common cross-connect blocks anyway, and realize that they can't 
run their cable runs out to the maximum distance of the specifications.

Hope that helps, 
-- 
Nate Duehr, nate at natetech.com