原创 Port issues: Data and power

Designers and end users now consider USB as a general-purpose, somewhat-universal charging source delivering up to about 100 Watts under best-case conditions (USB 3.1/Type C). We're seeing many application notes, articles, and ICs on managing the latest USB standard with respect to its power-delivery and handling specification.

Similarly, power over Ethernet (PoE) has gone through three standards, each increasing the power-delivery capacity and also adding to the complexity of wiring and cabling issues. The original PoE specification, IEEE 802.3af-2003, provided up to about 13 W at the load; the updated IEEE 802.3at-2009 standard (also called PoE+ or PoE plus) doubles that to about 25 W, and the recently-released IEEE 802.3bt standard doubles and even triples PoE+, depending on the cabling specifics and connector wiring (CAT5, CAT6, and other factors).

In both cases, implementing the power-delivery function is not a simple design. Though they are very different, both USB and PoE topologies require significant "negotiation" and confirmation between the source and load, between intervening ICs and, in order to assure that power is available and acceptable by a peripheral, that there are no conflicts in hardware or software. While these may become simple "plug and go" power sources to users (similar to our basic AC/DC wall warts), there is a lot that has to go on inside those cables, connectors, and ports to make USB- or PoE-based power transfer possible and safe.

With present and even foreseeable technology, there's no way we can use an optical fiber for power transfer other than perhaps a few milliwatts. The photonic energy is too low, there are optical/electric/optical conversion losses, high-intensity optical energy would cause all sorts of issues with the fiber, and launching a very-high-power LED source would be a lossy situation. Though history demonstrates we should "never say never" when it comes to technology progress, I'll wager we won't see power over fiber (should I trademark "PoF"?) for a very long time.

But that doesn't mean that the fiber-optic link can't also carry power, in an adjunct way. There's a trend in the fiber installations to use what is called Powered Fiber Cable System (PFCS) which uses a hybrid optical-fiber cable which has a DC-supply copper wire alongside, housed in the same outer jacket. This makes a lot of sense since in the real world of both new and retrofit projects, much of the system's total cost is in installation and cable-laying, not the bill of materials. (Note that submarine fiber-optic links have long used an advanced form of this hybrid structure, with their optical repeaters getting DC power via a solid-copper wrap.)

What's interesting about PFCS is that in addition to the power it delivers, the topology integrates with PoE, below.

The Powered Fiber Cable System uses copper literally in parallel with optical fiber, thus letting each do what each does best: transmit data and power, respectively; it then segues to a PoE-based distribution scheme. Source: CommScope, Inc.

 

There are special connectors, below, for the hybrid optical/copper cable termination, and then the power port is split off to a PoE node. An integrated media converter and PoE chipset work in conjunction to deliver a basic PoE or PoE+ output into a standard RJ45 jack.

 

The hybrid optical-fiber plus copper-wire cable is designed toi serve its dual roles and address termination-connector issues. Source: TE Connectivity

PFCS seems to me to be a clever scheme, since it leverages the advantages of each medium (optical + copper, PoE) without either one getting in the way of the other one, and the associated complexity. In fact, PFCS claims a range of up to 3 Km, which is 30 times greater than the 100 meter limit of PoE – a very significant increase (yes, I²R loses and IR drop will affect the maximum deliverable power and voltage, but it’s still impressive). From a system standpoint, it seems to me that by using the long run of copper just for power and not for data, the technical difficulties will be reduced while the performance will be more consistent and easier to manage.

There's a little irony here as we try to add power-delivery capability on top of data in our network media. Historically, the availability of the AC power line spurred many efforts to use it as a data-transmission medium, with some success but only in limited situations. It never really succeeded for several reasons: the AC line is a terrible channel, with noise of many types and numerous disturbances; the processing power needed at each end to make it function as useful, consistent medium was much too intense and costly compared it achieved in throughput and BER; user's needs for ever-faster links dramatically outpaced whatever progress was made in using the power line; and the wired and wireless alternates improved rapidly in both performance and price.

What's your view of the viability of fiber plus copper hybrids, such as PFCS? Do you see it as a superior alternative to PoE alone?