Ethernet cable length?

[Ptyrider]

Hi,

I ran a CAT5 cable (I think) 100 meters (which I thought was right at the
limit) from a router to my computer and it didn't work. That is, it checked
out OK w/ the network cable tester, but the NIC didn't recognize it (no
light, no connection). (The NIC recognizes other short cables.) Then I cut
it in the middle and added a switch there and then neither segment worked,
though again the network cable says both are OK. Any ideas as to what the
problem might be?

 

[Grinder]

Hi,

I ran a CAT5 cable (I think) 100 meters (which I thought was right at the
limit) from a router to my computer and it didn't work. That is, it checked
out OK w/ the network cable tester, but the NIC didn't recognize it (no
light, no connection). (The NIC recognizes other short cables.) Then I cut
it in the middle and added a switch there and then neither segment worked,
though again the network cable says both are OK. Any ideas as to what the
problem might be?

What order have you put the leads into the connector? If you're not
careful to match particular ones, inductance can disrupt your signal
over long stretches, even if the wires are continuous. Here's a diagram:

http://www.duxcw.com/digest/Howto/network/cable/cable5.htm

I'm not sure how much difference it makes, but I always get Cat-5e.

 

[News]

Hi,

I ran a CAT5 cable (I think) 100 meters (which I thought was right at the
limit) from a router to my computer and it didn't work. That is, it
checked
out OK w/ the network cable tester, but the NIC didn't recognize it (no
light, no connection). (The NIC recognizes other short cables.) Then I cut
it in the middle and added a switch there and then neither segment worked,
though again the network cable says both are OK. Any ideas as to what the
problem might be?

Sure you don't used a twisted cable?

Bjorn

 

[Noozer]

I ran a CAT5 cable (I think) 100 meters (which I thought was right at
the

Sure you don't used a twisted cable?

Cat5 is twisted cable (UTP)... For a run like that I'd try and use Cat5e or
Cat6 though.

 

[John McGaw]

Hi,

I ran a CAT5 cable (I think) 100 meters (which I thought was right at the
limit) from a router to my computer and it didn't work. That is, it checked
out OK w/ the network cable tester, but the NIC didn't recognize it (no
light, no connection). (The NIC recognizes other short cables.) Then I cut
it in the middle and added a switch there and then neither segment worked,
though again the network cable says both are OK. Any ideas as to what the
problem might be?

The 100m limit is premised upon perfect transmitter, perfect cable, perfect
receiver, perfect connections. Sadly in the real world at least one of these
is usually missing. When designing distribution systems I used to routinely
take the manufacturer's claimed performance numbers and cut them in half to
account for real-world usage.

In your case, after determining that the cable and terminations are as close
to perfect as possible I'd start swapping NICs and routers to see what
happens. I've had at least one NIC recently from a normally reputable vendor
(3COM) which just wasn't able to hack it on runs of as little as 30m.

 

[Colin]

Sorry old chap but you have not made the correct terminations.

 

[w_tom]

Cable tester can verify that a wire connection matches
connector pins both ends. But tester cannot verify that wires
are properly twisted with its mate. Quite possible that pins
4 & 5 are not same twisted pair, or that 3 & 6 are not
properly selected to be twisted together. Again, a cable
tester cannot detect this failure. But the ethernet card
will.

 

[Wayne Stallwood]

Hi,

I ran a CAT5 cable (I think) 100 meters (which I thought was right at the
limit) from a router to my computer and it didn't work. That is, it
checked out OK w/ the network cable tester, but the NIC didn't recognize
it (no light, no connection). (The NIC recognizes other short cables.)
Then I cut it in the middle and added a switch there and then neither
segment worked, though again the network cable says both are OK. Any ideas
as to what the problem might be?

What type of CAT 5 cable did you use

Do the conductors each have a single solid core or lots of little strands.
The multi strand cable is meant for making patch cables and shouldn't be
used in lengths greater than 10 meters. The single core stuff is less
flexable and more easily damaged, however is has better electrical
properties and it's that which is rated for up to

I ran a CAT5 cable (I think) 100 Meters.

You think it was CAT5 cable or you think it was 100 meters ?

Cat 5 cable will always say so down the outer sheath, it will have 4 pairs
of wire each pair will be twisted together inside the cable.

As other posters have mentioned it is important to use the correct wiring
standards. Just wiring the plugs the same each end will pass on a simple
cable tester. but if pins 4+5 and 3+6 are not on twisted pairs it will not
work.

Also at that sort of length I would advise going for the foil screened type
(more so if it is being run near any other cables)

 

[Robert]

What type of CAT 5 cable did you use

Do the conductors each have a single solid core or lots of little strands.
The multi strand cable is meant for making patch cables and shouldn't be
used in lengths greater than 10 meters. The single core stuff is less
flexable and more easily damaged, however is has better electrical
properties and it's that which is rated for up to


You think it was CAT5 cable or you think it was 100 meters ?

Cat 5 cable will always say so down the outer sheath, it will have 4 pairs
of wire each pair will be twisted together inside the cable.

As other posters have mentioned it is important to use the correct wiring
standards. Just wiring the plugs the same each end will pass on a simple
cable tester. but if pins 4+5 and 3+6 are not on twisted pairs it will not
work.

Also at that sort of length I would advise going for the foil screened type
(more so if it is being run near any other cables)

Thanks everyone for these helpful replies! Unfortunately I won't have a
chance to pull the cable and check it until next week owing to a family
vacation. I'll get back to you w/ what I've discovered based on your
replies. I've got some good leads now on tracing down the problem. Needless
to say, this is the first time I've attempted my own cabling.

Good newsgroup here.

 

[Ptyrider]

What order have you put the leads into the connector? If you're not
careful to match particular ones, inductance can disrupt your signal
over long stretches, even if the wires are continuous. Here's a diagram:

http://www.duxcw.com/digest/Howto/network/cable/cable5.htm

I'm not sure how much difference it makes, but I always get Cat-5e.

OK, I'm back and working on this. It's Cat5e, so just 50M (now that I've
split it) should be just fine, right?

However, a friend did the connectors and (I just found) didn't follow the
568A/B standard. He just wired straight thru at random, like blue to pin 1
on both ends, orange to pin 8 on both ends, etc. His reason was that it
works fine for short cables so he'd never bothered w/ the standard. He
proved to me that it does work for short cables. So do you think it's
highly likely that inductance is preventing the NICs from recognizing the
cable, even though the (cheap) cable tester does? Thanks. I suspect this'll
be the last time I need to trouble the group w/ this question.

 

[kony]

OK, I'm back and working on this. It's Cat5e, so just 50M (now that I've
split it) should be just fine, right?

However, a friend did the connectors and (I just found) didn't follow the
568A/B standard. He just wired straight thru at random, like blue to pin 1
on both ends, orange to pin 8 on both ends, etc. His reason was that it
works fine for short cables so he'd never bothered w/ the standard. He
proved to me that it does work for short cables. So do you think it's
highly likely that inductance is preventing the NICs from recognizing the
cable, even though the (cheap) cable tester does? Thanks. I suspect this'll
be the last time I need to trouble the group w/ this question.

The standard is to a certain extend, for organizing it such
that it can be understood by any/everyone else who has to
deal with it. It can work fine and not adhere to the
traditional way of matching colors to pins but the ground
for each singal needs twisted with it. Essentially what you
need is for the pins used for TX +, TX -, to be a twisted
pair, and likewise with RX +, RX -.

So if you have blue on pin 1, you need white w/blue stripe
on pin 2, and then (whatever the other color is) the other
color plus it's white striped twisted mate for the pin 3 &
6. That is very important for longer runs or best
performance potential at any length... might as well have
the friend learn to do it right and get all 8 standard on
both ends. Then again, considering it has to be redone you
might consider just going with CAT6, and see if a single
piece wired properly will do the whole run without a
repeater or switch, hub, etc, in the middle.

 

[w_tom]

Your friend has demonstrated why they are not really
accidents or mistakes. They are problems directly traceable to
human failure. Just because something previously worked
experimentally is not sufficient. Such 'I know better and
don't need no stinkin education' reasoning even created
disasters such as the Challenger. Human must understand WHY
we do this and that. One first learns the underlying theory -
or follows the standards religiously.

Cut off connectors and redo cables properly as Kony suggests
- so they work both experimentally AND theoretically. Both
conditions are required for acceptable operation.

Notice how much time and trouble you have encountered simply
because your friend thought he was smarter than the standard.
He did not understand some very simple electrical principles
and yet still thought he knew better. How much time and
trouble did it cost you?

How to profit from this fiasco? You possess a perfect
example of why we teach these principles in junior high school
science. Something must work both theoretically and
experimentally to avoid failure. Even intermittent failures
are considered a 100% failure. Do those short cables work -
or is error tolerance within Ethernet covering up his
intentionally created failure? Reasons why he should wire per
EIA/TIA 568 are same reasons why so much engineering exists in
the design of CAT 5 cables.

We learn by making these little mistakes. These little
mistakes are disasters when we don't learn from them.

 

[Ptyrider]

Hi,

I ran a CAT5 cable (I think) 100 meters (which I thought was right at the
limit) from a router to my computer and it didn't work. That is, it checked
out OK w/ the network cable tester, but the NIC didn't recognize it (no
light, no connection). (The NIC recognizes other short cables.) Then I cut
it in the middle and added a switch there and then neither segment worked,
though again the network cable says both are OK. Any ideas as to what the
problem might be?

I'm glad this has led to some interesting and informative discussions.

I rewired the connectors according to 568B and then my 50-meter cable
worked fine. Splitting pairs definitely does lead to failure on longer
cables!

Friend of mine and I were doing this lil' home network and our deal was
that I'd do the computer configs if he did the cables. He'd never done
network connectors, though, just telephone connectors (which is more than
I'd ever done), and he'd assumed that straight thru would be fine . What
was confusing was that the short cables he did for practice worked OK and
the cheap network cable tester confirmed the current on each pin. So then
owing to the help received here and at alt.dcomp.cabling, I found out about
the standards etc. and the rest is history.

Well, now I also know that there's lots I'd rather be doing than fixing
connectors. Fortunately I won't be doing them often--if ever again!

Wanna thank everyone again for the help. Nice group here.

 

[w_tom]

Learning is a two way street. By reporting back, we all
have gained a little beneficial experience. Thanks for
posting back.

Why length is important - the longer cable was approaching
one-quarter wavelength. As cables get longer and approach a
one wavelength, then new problems arise - become more
dominate. Wavelength is why a longer cable can have more
problems.

 

[VWWall]

Learning is a two way street. By reporting back, we all
have gained a little beneficial experience. Thanks for
posting back.

Why length is important - the longer cable was approaching
one-quarter wavelength. As cables get longer and approach a
one wavelength, then new problems arise - become more
dominate. Wavelength is why a longer cable can have more
problems.

I has nothing to do with wavelength. A longer cable has more signal
loss, and the signal/noise ratio decreases because of cross-talk.
(Which is why using the proper twisted pairs fixed the problem!)

Read up on RF transmission lines. A properly terminated cable is not
affected by "wave length", as it has no standing waves.

Virg Wall

 

[glassman]

What order have you put the leads into the connector? If you're not
careful to match particular ones, inductance can disrupt your signal
over long stretches, even if the wires are continuous. Here's
diagram:

http://www.duxcw.com/digest/Howto/network/cable/cable5.htm

I'm not sure how much difference it makes, but I always get Cat-5e.

I want to thank you for your link showing the proper wire order. Thi
site came up through a Google search and I was just going to lurk fo
information. The diagram fixed my problem and I noticed the wealth o
other information throughout the other forums, so I decided t
register.

Thanks again

 

[VWWall]

Grinder Wrote:


I want to thank you for your link showing the proper wire order. This
site came up through a Google search and I was just going to lurk for
information. The diagram fixed my problem and I noticed the wealth of
other information throughout the other forums, so I decided to
register.

As long as the cable is wired with one twisted pair (color-white/color)
as 1-1 and 2-2 and the second twisted pair as 3-3 and 6-6, it doesn't
matter which wires go to which pins. The twist is to reduce cross-talk
and the same wire must go to the same pin at each end. As pointed out,
two pairs are all that are used. Many RJ-45 jacks are color coded to
the 566A/566B standard. It's always best to go with a standard. It's
easy because there's so many of them!

Virg Wall

 

[mad NATer]

As long as the cable is wired with one twisted pair (color-white/color) as

1-1 and 2-2 and the second twisted pair as 3-3 and 6-6, it doesn't matter
which wires go to which pins.

You are sooo wrong!

The twist is to reduce cross-talk and the same wire must go to the same
pin at each end. As pointed out, two pairs are all that are used. Many
RJ-45 jacks are color coded to the 566A/566B standard. It's always best
to go with a standard. It's easy because there's so many of them!

That would be 568 not 566. Use the standard. It's there for a reason and
that reason is to preserve the electrical transmission characteristics of
the channel throughout the entire link (XTLK, Zo, RL, etc.). Just providing
basic electrical continuity isn't nearly enough.

 

[VWWall]

You are sooo wrong!

Really? Maybe you just don't understand. Do you think the electrons
care about the color of the twisted pairs they're traveling along?

That would be 568 not 566. Use the standard. It's there for a reason and
that reason is to preserve the electrical transmission characteristics of
the channel throughout the entire link (XTLK, Zo, RL, etc.). Just providing
basic electrical continuity isn't nearly enough.

Exactly. Connecting the conductors as I mentioned will result in
preserving crosstalk (XTLK), line impedance (Xo), (etc?). RL, the
termination load, is provided by the device to which the line is
connected. To be sure, you must use a twisted pair for each circuit,
but the wire color doesn't matter.

Sorry for the typo. You're correct: Two of the EIA/TIA standards are
568A/568B. Another common one is USOC, which has an entirely different
set of conductor pairs, using pins 1-8 and 2-7 for two of the circuits.
The fact that all of these work proves that the only thing needed is to
connect the correct pair of transmit or receive "pins" in the proper
polarity with a twisted pair of conductors!

Following a standard is a good thing to do, but one must understand what
the standard does and which one to use!

I've even seen posts that claim the line length is dependent on the
"wavelength" of the signals, whatever that is. ;-) A digital signal can
not produce standing waves, unless it is modulated on a RF carrier, and
then only if Zo and RL are mis-matched.

Virg Wall