Where to cross RX/TX line in fibre optical cabling

Question

I'm in the process of connecting together some sites via optical fibre. The trans-site multimode fibres are terminated with SC jacks on patch panels from where they are patched to the corresponding switch ports.

In order to receive a working connection, the cables need to be crossed at one point to let RX point to TX and vice versa.

Should this cross-connection be made in the transit cable (i.e. between the patch panels) or one of the patch cables between the panel and the switch? Is there a common practice?

Answer 1

Cross on one of your patch cables. You want your installed fiber plant to be labeled correctly at both ends (Strand 1 at Site A = Strand 1 at Site B).

Answer 2

This is directly addressed by the ANSI/TIA/EIA 568-B Commercial Building Telecommunications Cabling Standard:

Connecting Hardware and Polarity

Optical fiber shall be installed with odd numbered fibers having Position A at one end and Position B at the other. Even numbered fibers will have position A and B reversed from the odd numbered fibers. When using the 568SC connector or other duplex connectors, the above polarity must be maintained.

When using copper cabling, it must be straight through from termination to termination, but optical cabling must be cross from one termination to another.

The reason fiber cabling is crossed is simple: when connecting fiber from one device to another, the fiber strands need to be cross so that Tx goes to Rx on each end. For this reason, fiber patch cables are crossed, but when you have horizontal cabling, you have two fiber patch cords which cancel each other out. When the horizontal cabling is crossed, you end up with three crossed cables, making the whole a crossed cable, leading to success.

Answer 3

Common practice would be just to cross everywhere since you will always get an uneven number of crosses which will result in an overall cross. But this only works if your fibres were installed by a sane company.

Basically you need to know that:

a) Couplings cross
b) You need crossed cables/trunks everywhere (which *should* be default)

So let's test that:

Device - Cable - Device: 1 cross
Device - Cable - Coupling - Cable - Device: 3 crosses
Device - Cable - Coupling - Trunk - Coupling - Cable - Device: 5 crosses

Problem is if you have straight cables somewhere, even worse if it is your site-to-site trunks. Depending on how the patchpanel is built you can maybe swap the fibres on one side. If you can't you have the only choice to define a point of crossing - depending on how your other fibres are setup.

Just a hint: I would label them. We have a single straight through panel on site out of hundred of crossed ones. It annoys the hell out of you! Then someone starts flipping calbes somewhere to get the connections to work. A year later someone else removes the tampered cables and puts them back into storage. Guess what happens next?

Answer 4

In addition to the already present answers here, the reason for copper being generally straight and fiber being generally crossed is fairly simple:

The major point of the initial twisted-pair Ethernet variants StarLAN aka 1BASE5 and its extremely popular IEEE successor 10BASE-T was the reuse of existing telephone cabling. This cabling already existed when Ethernet came along and it was straight through from end to end - analog POTS doesn't care for polarity. StarLAN/10BASE-T accounted for this and defined the required signal crossover into the interface variants MDI (used on NICs, routers and such) and MDI-X (used on repeaters, hubs and switches).

Fiber became popular with FDDI and it was for digital communication from the start. With fiber you need to cross transmitter and receiver in any case, so fibers were crossed - no need for defining two different interfaces.

Answer 5

run cable between building terminations straight thru. From patch panel to device cross tx rx on one side . then you can see what your doing