Tek-Tips is the largest IT community on the Internet today!

Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!

  • Congratulations SkipVought on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!

Cat5 Split Pairs? Showing 100 mbs/full is it really? 1

Status
Not open for further replies.
May 14, 2002
2
0
0
US
My network....older than dirt has some connections that are on split pairs. On my switches they are showing connections of 100 mbs/full. I always thought that you needed 3 pair to go 100/full. Someone correct me, so I can either correct myself or a Co-worker.

Thanx.
 
100baseTX only uses 2 cat 5 pairs
100baseT4 uses all 4 cat 3 pairs and is VERY rare

full duplex just uses both pairs at once while half duplex only uses one pair at a time I tried to remain child-like, all I acheived was childish.
 
Thanx. I guess the information that I recieved was for T4.
 
Cat5 Split pairs mean that the Wiremap is incorrect when you made up the cables. A good Wiremap is: 1-2, 3-6, 5-4, 7-8 at both ends(wirecode T568B states: 1-Orange/White, 2-Orange, 3-Green/White, 4-Blue, 5-Blue/White, 6-Green, 7-Brown/White, 8-Brown). If you follow this scheme with all of your cables, you will no longer have split pairs.
Split Pairs generate a tremendous amount of noise on the cable, but does not necessarily mean it will not work. Short links usually work with split pairs because ACR (Signal to Noise Ratio) checks out, however, as the length increases, so does Attenuation (signal loss) and the noise may be too much for the signal to be interperated at the far end.
Full Duplex is simply the ability to transmit(TX) and recieve(RX) on the same pair. In 10\100Mbps Full Duplex, this means a NIC can TX & RX on both pairs (1-2 & 3-6).
Only Gigabit ethernet (1000Mbps and beyond) uses all four pairs in a Full Duplex scenario.
So I would say that having split pairs on 5-4 & 7-8 in a 100Mpbs Full Duplex scenario would probably work (although NOT recommended) because there is no signal actually being transmitted on those wire-pairs.
There may be some switches out there that may utilize a third pair (eg: for auto link negotiation perhaps) but not for standard 100BASE-TX IEEE 802.3 ethernet traffic.

"In IT, 'Proprietary' is just a buzz word meaning: it's not a standard...yet"

RON GROULX
 
Good Description Ron. I believe by definition that full duplex means the ability to transmit and receive simultaneously independant of the number of pairs. I understood ethernet to transmit on one pair, and recieve on the other pair regardless of speed (10/100) or full/half duplex. Of course I can't find that in my books quickly, can anyone else? Daron J. Wilson, RCDD
Telecom Manager
LH Morris Electric, Inc.
daron.wilson@lhmorris.com
 
I agree with RON GROULX's excellent response, and wish to add just a bit more:

Network speed is hardware-dependent; hence, it is hardware-driven. For instance, you cannot get 100Mbps out of a 10Mbps NIC no matter how you cable it. The hardware design of a communications device dictates the cabling required to connect each of its circuits to the adjacent communications device to achieve a given transmission throughput and quality (quality defined here as number of packets successfully transmitted in a given unit of time). Maximum transmission distance is a limitation of the power that the hardware is designed to pass through the cabling, coupled with the transmission medium's capabilities for data integrity, and the two work together as a system. This is why Ethernet over fiber has a far greater transmission length threshold than Ethernet over copper.

Again, back to Mr. Groulx's response: attenuation happens when resistance through a wire (it's all about electrons) adds "noise", adversely affecting the signal-to-noise ratio. When insufficient signal passes through one end of a wire to the other end to enable reception of 100% of all packet data in a reasonable unit of time as specified in the design of the hardware, then we say that one of two things has happened: 1.0 the wire length has extended past its distance limitations, or 2.0 there is random noise being introduced into the cable. It is important to know that both of these conditions may exist simultaneously, but in widely varying states, and to know how to isolate them for testing and correction.

Splitting pairs is technically feasible, but was usually done to save the labor costs (wire is comparatively cheap) of running additional wires. This interference that it can introduce into the transmission of both carrier pairs is referred to as "cross talk". I have personally ripped out such an installation in the past and replaced these with dedicated pairs, and it was worth every penny of my time.

Another good reason to avoid pair-splitting is for future line diagnostic procedures. With dedicated cables, it is much easier to pick up your average cable tester and find out which cable is causing the problem, and even where in the cable this is happening. With cabling in which there are split pairs, this job becomes just that much more difficult.

Hope this helps a 'tad,

ThreeDots
 
Splitting pairs....I think we have a terminology problem here. Taking Cat5 and pulling off two pairs for a second connection is not splitting pairs. It is not recommended, it does work, and I run several 100 mbs connections that way when needed.

Splitting pairs is taking a pair and splitting it. This happens when people make cables incorrectly. If you just start at one pin and go pair by pair, you will be putting pin 3 on one pair, and pin 6 on another pair. This is taking the balanced ethernet signal on pins 3 and 6 and
'splitting a pair'.
It is only my opinion, based on my experience and education...I am always willing to learn, educate me!
Daron J. Wilson, RCDD
daron.wilson@lhmorris.com
 
There are 4 pairs in every cat 5 cable. Only 2 is needed for normal usage. That is typically 1,2,3 and the 6th pin. White Orange, Orange, White Green and Green respectively under 568 B standards(Swapping the green and orange will give you 568 A). Are the pairs really split or is the cable split by pairs? Split by pairs is acceptable but not really prefered as when the next generation beyond 100 Mbps hits, it will force you to use all four pairs and render any split cabling outdated
 
daronwilson is correct (and being an RCDD he certainly should be!!).

Splitting a single cable and wiring it into two ports is one thing and not really a good idea. If it was my network and I only had a single CAT5 run but wanted two ports I might consider doing it as a temporary solution. But, if my cable contractor did it on my network then I'd be pretty peaved off!! If you're gonna do something, do it right!

I think that the original post was about "split pairs", like Daron said using un-balanced pairs. You would normally put green/white and green (or oranges) on 3 & 6 as these pins are used as a "pair" on an ethernet connection. The wires on 3 & 6 would then be twisted together all the way down the wire. If you "split" the pairs so that you might have (for example) green/white on 3 and blue on 6 then your balanced pair no longer exists. Each wire from the ethernet pair would be twisted with another wire. The pair would be "split". If you get this and you test it with an Omni-Scanner or a Fluke or something then you can look at the wiremap on the tester and see this. It will fail CAT5 and CAT5e tests. Been there, done that!

Chris.
************************
Chris Andrew, CCNA
chrisac@gmx.co.uk
************************
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor

Back
Top