CAT 5 Question

[SgtTrojan]

Do a full box Monkehchops

 

[timbim]

Someone really needs to re-asses their network structure...

Urgently.

 

[Monkeh]

Do a full box Monkehchops

Don't think 305m will work. And I don't have one.

 

[kjacko]

Cat5e is fully capable of gigabit links over 25m.

May be capable in nigh on perfect conditions......but certainly not guaranteed or the norm in my experience.

I've just made up approximately 40-50m of cat5e for a test, and I can't call the conditions anything like perfect. I think the results speak for themselves:
[code:1][ 3] local 172.16.1.3 port 51347 connected with 172.16.1.1 port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 5.0 sec 398 MBytes 79.5 MBytes/sec
[ 3] 5.0-10.0 sec 400 MBytes 79.9 MBytes/sec
[ 3] 10.0-15.0 sec 399 MBytes 79.7 MBytes/sec
[ 3] 15.0-20.0 sec 399 MBytes 79.8 MBytes/sec
[ 3] 20.0-25.0 sec 398 MBytes 79.7 MBytes/sec
[ 3] 25.0-30.0 sec 400 MBytes 80.0 MBytes/sec
[ 3] 0.0-30.0 sec 2393 MBytes 79.8 MBytes/sec[/code:1]

If you're unable to achieve cat5e specifications beyond 25m, your installation is faulty.

Hmmm...fair ernough, thanks for the info monkeh. I must admit i didn't lay the cabling in our work floors, so can't speak for how well its been laid or terminated. It has been pushed and pulled around all over the place though.
Did you use cat5e spec for everything including modules, termination etc?

 

[Monkeh]

Hmmm...fair ernough, thanks for the info monkeh. I must admit i didn't lay the cabling in our work floors, so can't speak for how well its been laid or terminated. It has been pushed and pulled around all over the place though.
Did you use cat5e spec for everything including modules, termination etc?

Yes, all cat5e spec. Some of it thoroughly abused, there's a patch cable on one end which has been through wars and reterminated three or four times now, and the bulk of the length is in a coil on my floor (remains of a reel). The speeds are a bit subpar due to the NICs involved.

Oh, and just to throw a spanner in the works, the patch cable is actually a crossover, so there's an impedence mismatch there. 802.3ab is quite robust.

 

[RossR]

Oh, and just to throw a spanner in the works, the patch cable is actually a crossover, so there's an impedence mismatch there. 802.3ab is quite robust.

How does a crossover cable cause an impedance mismatch?

 

[Monkeh]

Oh, and just to throw a spanner in the works, the patch cable is actually a crossover, so there's an impedence mismatch there. 802.3ab is quite robust.

How does a crossover cable cause an impedance mismatch?

Difference pairs in a cat5e cable have different twist rates, and as such a different impedence. A crossover cable will swap pairs, causing a mismatch.

 

[RossR]

I didn't know about the different twist rates. Thanks. Learn something knew every day

 

[radweld]

The twist rate does not affect the characteristic impedance.

 

[RossR]

It does have a small effect:

http://qucs.sourceforge.net/tech/node93.html

 

[radweld]

Fair enough - I expected there must be some effect but couldn't find any references.

That said, the effect appears negligibly small. From what I can see from datasheets etc. all pairs in Cat5 cables have the same nominal impedance specified. I don't think you can attribute an impedance mismatch to the use of a cross-over cable: the impedance is much more sensitive to small variations in the insulation thickness for example. A 'good' cross-over cable will introduce less mismatch than an 'average' straight cable.

However, having never done any Cat5 work in a production environment I may be immediately contradicted.

 

[Monkeh]

I doubt it has much significant effect in a short run (you'd probably notice it if you exceeded 100m, things get pretty marginal past about 125m iirc), but it's there nonetheless.