wire question on Cat5 or Cat5e cables

Hi Adam, It was a long message, so I post on top.

It looks like only two pairs in your cables were terminated, clearly for Ethernet application only. The rest of the pairs is going to be used if you change the cable's application. For example an analog phone would have to use pair #1 (conductors 4 and 5, missing in your test), and in you case would not work. Also, the additional pairs are going to be required if you are trying to implement Power over Ethernet, especially high-power version of it or if you are trying to go Gigabit Ethernet. So, in your case those two applications are out of reach. With a standard 100BASE-TX equipment it would not really matter if the "missing" conductors are shorted or crossed somehow: they are usually ALL connected to the ground plane of the Ethernet switch/hub.

[sbip]

I checked a 3Com NIC, and I found that the unused pairs were connected to ground as you said. The problem I see with this is that the newer Crisco switches may have POE, power over ethernet, for a VoIP phone, etc. If that is turned on, the shorted pairs could draw excessive current.

[snip]

The PoE specs allow for that possibility.

It is a part of the IEEE 802.3af protocol to poll the equipment on the other end to see if there is a short. However, it is an issue with DIY Power-over-Ethernet. I personally fried quite a few power supplies few years ago doing some DIY Power-over-Ethernet for early IP cameras we installed.

Adam, It sounds to me like these guys are going to deep into what you really need to do. If my understanding is correct, you only have a few cables that are not passing. I would reterminate the cables on both ends so that you know they are correct. After reading the other replies, I don't think you need to get into the "deep end of the electronics". The cable should be fine, just reterminate.

The unused pairs on Etherner card could be, depending on the card wired on one of the following ways:

1. left floating freely (not connected to anything)
2. free wired wired toghether and connected to ground (maybe though coupling capacitor)
3. wires are terminated together through Bob Smith Termination (several resistors and then through capacitor to card ground).

Options 1 and 3 seem to be the most common ways this has been done. This is based on looking at the Ethernet designd on Internet and looking at some Ethernet equipment I have opened (hubs, switches, Ethernet cards etc..)

The current POE standard (IEEE 802.3af) says that the equioment that feed power to the Ethernet wiring must first check that it is safe ty feed the power to the wiring. There is a specific identifiaction protocol that identified that there is equipment on the other end of the cable that is designed to take standard POE power and the protocol even identifies some product features (what is the power rating of those equipment). The protocol is such that that it really needs to be a POE supporting equipment on the cable, before the power supply will be turned on. And there is also over current protection if short circuit happens afterwards. Also earlier pre standard commercial POE impementations (for example from Cisco and PowerDsine) to my knowledge has some form of identification system to make sure that power is not fed to places it should not be fed.

The fact that the normally unused pair could be connected differently on different Ethernet equipment it is an issue with DIY Power-over-Ethernet that just simply feed the power to those free wires. Depending on the connection you could get a fried power supply or some fried Bob Smith Termination resistors on Ethernet card. In any case the power supply for this kind of tricks should be always so energy limited (either built in current limit or some fuses or similar) that short circuit on the wiring does not cause more current to flow on the wiring that it can safely handle. For example at equipment buit according POE standard (IEEE 802.3af) power supplying hub cutoff current limit is typically around 350mA.

Some more information on POE can be found at

connected

That's reassuring to know.

Right now, we are using WAPs that have an injector that plugs into a wall wart, then into the wall outlet. I would guess they have some form of protection, but I don't know. They're mostly Crisco, I believe.

connected

We use a digital PBX system that powers the phones over the single pair. Each port has a PTC (positive tempco) device made by Murata that is in series with the 48VDC line. If there's a short on the 66 block (easy to do) the blue spark alarms the tech, but the PTC gets hot real quick and reduces the current from over a hundred mA to just a dozen or less. It'll stay that way until the short is removed.

You ought to see the tech's eyes light up when he sees that fat blue spark! "WHOA!" What was that?!?!"

connected

Thank you very much for the good info.

However I don't agree with ths following statement: "When power-over-Ethernet technology becomes widespread, corporations and businesses will see at least one set of wiring hassles disappear or reduced considerably."

The problem is that the use of one pair for power makes it impossible to use that cable for 1000BT. Since most PCs are now coming with GB NICs, and switches will be, too, this means that any cable used for POE will not allow "full speed", and hencs another cat5 cable will have to be installed. PFFFT! There goes your cable minimization! You now have two cables. :-P

Not necessarily. It's entirely possible to send both DC and signal over the same wires. It's been done for decades, in the communications industry. Twisted pair ethernet uses transformers to connect the cables to the NIC. The power can be inserted on the line side of the transformer, without any effect on the operation of the NIC.

But as far as I remember, the one port Cisco power injectors has no protection and will just feed power.

Go for real PoE midspan devices that follov the PoE standard, then you are safe.

snip

I fully agree and think that PoE will get bigger impact than many think.

No, it uses 2 pairs. One pair for positive and another for negative.

No, the PoE standard has 2 options; to use Spare pairs or phantom power. For 1000BASE-T phantom power will be used. You will soon see Mid span devices for 1000BASE-T and wireless AP´s with

1G NIC. All clients shall accept both power options to comply with the PoE standard.

Read and learn :-)

Yeah. The only one I've seen installed is the spare pair affair, in which case what I said is true. If the phantom method is used, then the remaining two pairs should be useable. The phantom method requires more transformers with center taps, so it's possible that they are trying to save money and just use the spare pair.