Can cable take the strain?

Just recently, when the new year festivities had hardly worn off, old Roget's home phone went sliver-side up. Pick up the receiver and - nothing. Now Roget, believing himself to have some influence with the local BT repair people through his office work, yelled at a bloke using a number that by-passes the umpteen-options call centre machine and, though he may not have speeded things up, did at least get an explanation that amazed him.

"They said," he said to anyone in the saloon bar of the Stopwatch and Clipboard who would listen, which as it turned out was only me, "they said an old lead cable had broken. Not the engineers' fault, they said, the cable wore out.

"Lead! I ask you. Was it insulated with gutta-percha, I asked him."

We did not dare ask what reply Roget received and in any case his face was now buried in his jug of Cask Matured Timely Motions.

Fortunately, let us assume, old lead cabling exists only as an oversight in rural areas these days - Roget living in a very rural area, which is fortunate for those spared from being his immediate neighbours. Our state of the art IT now relies on copper cabling - and perhaps optical fibre - manufactured with a precision those old lead cable engineers back in telegraph days would never have dreamed possible - or necessary. The result is that what looks like ordinary twisted pair cabling can take 10OMbps comfortably and lGbps by pushing it to the limit.

The 100Mbps local area network needs, as readers will all know, Category 5 cabling, a copper cable made with sufficient precision to ensure that the conductor is set exactly in the centre of the insulation, which is itself of reasonably uniform thickness and consistency. It also requires that the strands are twisted in a regular manner. All this suffices to keep line noises and signal degradation down to acceptable levels for the

100Mbps rating.

It is ten years, in fact, since CAT 5 cable first appeared before our astonished eyes, an anniversary partly to blame for my choice of cable as a subject. The other main reason, apart from Roget's episode jogging my memory, was to follow on from January's look at broadband services.

For eight of those ten years we had the simplicity of only one feasible choice. Now CAT 5 is obsolete - I stand by the word. 100Mbps LANs are now commonplace and many readers will be looking at, or already using, Gigabit Ethernet. To get lGbps through a copper cable you need more than CATS - and here the shopping choices start to get confusing. Cables capable of gigabit rates now exist but official standards for all of them do not. Indeed Standards bodies have been deliberating the details of what these new cables should adhere to for about four years now and still have not finalised their work. But then when did Standards bodies ever lead a market?

There is one standard, for something called CAT 5e - e for enhanced - that has been around for a while but, frankly, it required so little effort to dream up that even a Standards body could not delay it. In essence it is a matter of testing data cables to slightly higher levels than for CAT 5 and it is so lacking in breakthrough that virtually all CAT 5 cable manufacturers found their existing products satisfied it - subject, of course, to proper installation and the use of good quality components such as terminals.

For a LAN needing lGpbs in places, for instance on its backbone, therefore, many a management services type was relieved to find that there was no need to change out the existing CAT 5 cabling. Many have found that in testing the LAN to the necessary 100 Megahertz bandwidth, better results in near-end crosstalk (mutual interference between cable cores), enhanced capabilities of patch-cords etc, for CAT 5e they get

full or partial success. Sometimes part of the LAN will need upgrading, perhaps where a past sin of bad installation has been committed or where a component such as a terminal is to blame rather than the cable itself. (Incidentally do not confuse Megabits per second with Megahertz - the latter is a measure of the cable's bandwidth capability not

its data speed - hence a 32MHz cable is theoretically capable of 100Mbps.)

That previous paragraph, however, says 'lGbps in places' because CAT 5e is ratified for lGbps only over distances of 25 metres or less. Within one building, this may be sufficient for most of the runs but 25 metres soon goes when cables have to take circuitous routes around the premises.

It goes without saying, therefore, that testing a CAT 5 installation should be done carefully to find points of weakness before the IAN traffic does. Always test the entire LAN in-situ because odd things like quality of connection to components, tight bends, running cables near to electro-magnetic emitting plant etc can all affect the result.

The good news is that DIY hand held testers are cheap and plentiful and will test the integrity of each cable run, point to the location of any fault and confirm the length of the cable run to ensure that one is not asking more of it than the CAT 5e specification allows.

But CAT 5e, in its couple of years or so of life, has become looked upon as

an expedient. Some manufacturers marketed it widely at first - although it may have been their old CAT 5 product dressed up for all I know - but now the spotlight is on CAT 6. But it's a flickering spotlight because there is, as I hinted earlier, no formally agreed Standard for CAT 6 cable - but plenty of installations are going in.

CAT 6

The CAT 6 Standard, requiring the cable to run at 200MHz (actually performance is specified to 250MHz but we can regard that as a safety margin), exists only in draft form although the items still needing to be agreed are ones of detail that do not look likely to affect the designs and processes that cable manufacturers are using. To what extent are buyers of CAT 6 cable putting future compatibility at risk? Alan Cobb, the cable guru at the Telecommunications Industry Association says not at all, provided sensible purchasing precautions are taken.

"Deployment of systems prior to the final publication of recommendations is not new - it occurred for both CAT 5 and 5e. To make long-term investments in the generic cabling of buildings, it is imperative that systems have the headroom to allow for future protocols or they risk being soon out of date. We therefore recommend adopting

CAT 5e as a minimum and consider using (cabling) systems from suppliers who are following the latest available draft of

the CAT 6 Standard as the preferred solution."

The sting comes in that last line.

The TIA and such cable manufacturers who know they can accept it, recommend our getting written guarantees from installers and/or cable suppliers that the cable not only meets the current draft standard - in its entirety - but that the installation will be modified to meet the official standard when eventually published - should that prove necessary. Further, that it will be compatible with any post-standard CAT 6 cable added later in extending the LAN. This last point is not much of a risk since a mix of CAT 5, 5e and 6 cabling on the same LAN does not cause problems other, of course, than to drop traffic to the rate of the lowest type of cable encountered along its route.

Molex Premise Networks had its PowerCat 6 cable tested to the draft CAT 6 standard early last year and has had no need to modify anything since. Product Manager Carol Harris explains. "Following the Sydney ISO meeting in 2000 the parameters used to define a Category 6 (or Class E) channel were re-examined and agreement was reached. It is unlikely that these will be changed now, as a two-thirds majority vote would be needed. In any case, PowerCat 6 actually exceeds the draft standard in a number of important performance criteria, providing significant headroom for future expansion."

That being the case, the LAN will be running on CAT 6 cable, and appropriate network components - all capable of providing Gigabit Ethernet comprehensively, with full 90 metres runs allowed.

CAT 6 cable is dearer than CAT 5 or 5e by about a quarter but is capable of using the same 90 metres clear runs that CAT 5 uses for 100Mbps. Cost comparisons between CAT 5e and CAT 6 structured cabling systems, therefore, will probably be closer than 25%. It is more relevant, in fact, to worry about the physical problems one might encounter - CAT 6 cables have been a bit fatter than CAT 5, by 15-20%. That will matter not a jot for those readers enjoying comparatively new office buildings with raised floors you could hold a party under but for those of us in premises with older mouse-crouching raised floors or, shallow cable trays in false ceilings or older in-screed trunking, cable size is a factor.

The reason for this over-ample girth is that cable manufacturers used strips of various materials inside the cable to keep the individual cables apart to reduce the problem of crosstalk (mutual interference between cables in the sheath) that hinders the cable's meeting the exacting CAT 6 signal quality. These separators not only increased the girth of the cable, they were just another little job to waste the installer's time in clipping them off.

Plastic Insulated Cables now has a separator-free Cat 6 though still retaining its UTP - unshielded twisted pair construction. The company claims to have upgraded its manufacturing process and made changes to the cabling design resulting in a 10% diameter reduction to under 6mm.

So, CAT 6 cable, with guarantees, is the stuff for Gigabit Ethernet. But the Stopwatch and Clipboard saloon bar

chat already buzzes with 2.5Gbps - MultiGigabit Ethernet - and the near future prospect of 10Gbps. America's IEEE institute is already working on a draft specification for 10Gbit Ethernet, has produced discussion documents and what not and soon expects to publish what a spokeslady described as `advanced interim' findings. For these LANs a completely new form of cable is required - at least you must judge whether it is completely new or just the same old stuff in new wrapping. These speeds also bring closer the cost/performance characteristics of copper and optical fibre.

CAT 7

Unshielded twisted pair cable is the same technology as has been used for telephone cabling since telephones first started appearing in offices. Individual copper twisted strands are coated in a plastic insulation and the resulting wires are themselves twisted into pairs within an outer insulator cum protector sheath.

As I said earlier, quality of material and precision of cable extrusion have been the main advances. CAT 6 taking 1Gbps, however, takes that design to its limits as far as technology knows them at present. Although the cable's design principle is the same, how 1Gbps makes use of it is quite different. In 100Mbps, one pair of wires transmits and another pair receives. For 1Gbps, all four pairs in the cable both transmit and receive simultaneously. CAT 6 cables are tested against many more parameters than CAT 5.

For greater performance cable makers went back to the drawing board and came up with - foil wrapping.

CAT 7 cable is being manufactured now - some successful installations have already gone live in the US - although, much more than with CAT 6, its designers and makers are secondguessing to some extent what the final Standard specification will be. For now they are content that the cable can take 2.5Gbps, and in laboratory conditions only I believe 10Gbps, reliably over 90- metres runs. In these CAT 7 cables, each of the cable pairs is shielded in foil, something which is going to make the installer's termination work take much longer.

Not only have the cores to be unwrapped, those foil shields need earthing.

The cable, too, will be more expensive because of the complexity of its manufacture.

This complication is making structured cable installers talk of using exclusively pre-formed cable harnesses on site. Already used to some extent to cut down on site work - and more so in optical fibre cabling where terminations, joins, etc, are more complicated still - pre-formed cable looms may become the norm.

The problem is that, as with optical fibre, cable runs do not conform to pre-determined lengths. Some degree

of 'losing' excess cable under the floor can be tolerated but with anything susceptible to electro-magnetic emissions coiling up yards of the stuff is asking for trouble. Optical fibre does not have this problem and makers like Ortronics supply cabling systems pre-engineered and pre-tested. Typically the backbone will be pre-cut and fitted with terminals and supplied complete with a wall-mount cabinet or rack-mount panels and fibre cassettes as the earlier site survey required.

Not surprisingly, the optical fibre makers are loving this prospect, but would optical fibre to the desk be cheaper for our 10Gbit Ethernet LAN?

Optical fibre is already popular for LAN backbones where the highest speeds are required, linking with copper for the runs out to workgroups. Experience has shown that the cost of the cable and testing would be roughly the same between CAT 7 copper and fibre. In spite of pretesting of optical fibre sets, on-site testing cannot be ignored or skimped, of course.

I will also drop in here mention of the EN 50174 Standard to which it is a good idea to tie one's cabling installer - its aim is to ensure that the finished job does precisely what the work order laid down it should do.

Fibre's LAN components, however, can be four or five times the cost of those for copper. Some of this is electronics complexity but some is a matter of scale. If fibre were to become popular right across the LAN, in the majority of

LANs, as well as in the backbones of just some LANs, unit prices would fall.

By how much is a matter for guesswork at present.