Protecting your cables from interference

"The big issue we've got right now is EMC - Shielding doesn't completely prevent EMI getting out or getting in. In some cases it may reduce its effects better than UTP, but not a lot better to make any significant difference. If you really believe you need it, don't muck about - go direct to fibre."

UTP vs ScTP and FTP
I believe the optical cabling medium to the desktop is UTP. If you can limit the route length to 100m, control the EMI environment and if security isn't crucial then UTP is going to give you the bandwidth to run as long as you want. Screened twisted pair of shielded twisted pair? The big issue we've got right now is EMC. Electromagnetic fields are known to be detrimental to humans. A two-year-old US study recently released states that no new subdivision homes should be built underneath high voltage power lines.

It also specifies that any school or day-care centre currently under high power lines should be closed. This is being taken to the next phase as to what happens with low power high frequency fields, which is what we're dealing with in datacoms. The first thing is there's a whole lot less power. We're using one or two volts as opposed to 500kV on these high tension lines. A big difference.

Also, the higher frequency signal has less power in the signal itself. Hi-fi buffs will tell you that your bass tones take a lot more amplifier power than higher frequencies. Regardless of the low transmitted power from higher frequency datacoms, we still don't know the precise effects. This is what's driving screened twisted pair and shielded twisted pair. In Germany, it's not illegal to install UTP, but nobody will do it because their insurance companies won't let them. The government hasn't legislated it - effectively the insurance companies have!

Basically what they say is, if you've got UTP installed and if in the future your workers need medical care because of EMI fields, then you're not covered. Consider the actual effectiveness of cable shielding. Most shields are about 70% effective at 100MHz. Go to your cable manufacturers and ask them for shield effectiveness rates. The shield effectiveness is variable according to frequency. They're very effective with low frequency, but not too effective with high frequency. If it's 70% effective, with voltage in decibels that's about 3dB.

If I give you Category 4 cable operating at 10MHz and compare its crosstalk with Category 5 cable at the same frequency, it's 7dB of difference. What I'm saying is a shielded Category 4 cable puts out more EMI than an unshielded Category 5 cable! In fact, the EMI that UTP cables emit is controlled, not so much by the cable, but by the quality of the signals that are on the cable. If I have a perfectly balanced, perfectly phased signal, the EMI from an unshielded twisted pair cable is zero. Shielding doesn't completely prevent EMI getting out or getting in. In some cases it may reduce its effects better than UTP, but not a lot better to make any significant difference. If you really believe you need it, don't muck about, go direct to fibre.

Fibre issues
The next question is what kind of fibre do you use? Singlemode or multimode? Multimode fibre is very expensive to build. Graded index multimode fibre has layers of varyingly doped glass, created so that we actually play with the propagation speed of light in glass.

Normally in a fibre, light travels about 180,000,000m/sec. In multimode fibre we've got a number of pathways. The shortest pathway goes down the centre of the fibre. Other pathways bounce off the sides and actually spiral around the fibre. These have longer path lengths. Fibre manufacturers make multimode graded index fibre so that the light travels the slowest down the centre of the fibre and the fastest on the edges. This ensures that whether you go straight down the middle or take the long route down the sides, all light winds up arriving at the other end at the same time. It ain't cheap.

Why do they do that? With multimode I can shove light in one end and get a lot of light out of the other. On the other hand with signal mode I've only got a little pathway. To get enough light in to get it out the other end I've got to use a laser. Multimode fibre was developed by phone companies for short haul communications within their central office, so that they wouldn't have to use expensive lasers. Once they got out of their building and wanted to go a long distance they used singlemode, because multimode cable's only good for a couple of kilometres and a couple of gigabits. Singlemode, on the other hand, is good for a bunch of gigabits and a lot of kilometres.

The theoretical limit of a multimode fibre is about 10gigabits per second. Compare this with the theoretical limit of a singlemode fibre at 25 terabits per second! Every singlemode optical fibre outputs the same bandwidth as the entire broadcast spectrum - from radio waves to microwaves. Singlemode is very difficult to connect because you have to line the fibres up very carefully. The core of singlemode fibre is only 4 to 10 microns in diameter. The core of multimode fibre, on the other hand, is generally 62.5 microns.

A 10% misalignment with singlemode cable is an entirely different issue to a 10% misalignment with multimode cable. This can particularly be a problem if you're trying to put a singlemode fibre and connector together. But we don't put connectors on that much any more. We either splice fibres together, or if we want to put a connector on we splice on a premanufactured pigtail. The singlemode connector is prefitted to the pigtail in a laboratory. Singlemode's not that expensive to install any more. It's not any more expensive than multimode if you've got the right tools. It's cheaper to buy the cable and the cable itself has higher bandwidth. There's only one reason I wouldn't use it - the source and receive equipment's a lot more expensive because of sales volume.We historically needed different printed circuit boards for single and multimode applications, because singlemode components have traditionally been different to multimode.

That's about to change.
Hewlett-Packard has released a complete line of pin compatible sources. Now you can buy a single chip pack that could be either a multimode connector or singlemode connector, complete with all necessary electronics. Electrical in, light out, light in, electrical out. And they all fit on the same footprint. So we'll be offered one printed circuit board - the buyer specifies what he wants to use, then the supplier simply plugs that connector on. Done! Singlemode and multimode products are now going to be the same price. My projection is multimode optical fibre will be obsolete in the next three to five years. Obsolete in much the same way as Category 3 is today.

Wireless systems
I believe wireless systems have some specific applications, but I don't believe they'll handle everything. Basically there are two applications for wireless. Either when one end of the channel's moving (which isn't our typical for the office environment) or where I cannot practically get to one end of the channels with a pathway. Examples of these are buildings which have asbestos insulation or buildings which incorporate impenetrable obstructions, like firewalls. It also has applications where a temporary service is required.

There is one more important factor. I could pretty easily give you wireless voice, wireless data, wireless video. The problem is, you can be sitting there at your desk, cranking along with your "wireless" laptop and your battery can go dead. You're out of luck. Sooner or later you've got to get power. Once I bring power in there, I've established a pathway. Once I have a cable pathway I can drop in a fibre optic line. I can put in a fibre optic line and the accompanying end equipment cheaper than I could put in wireless link.

So although wireless is valid, I feel it's only going to be appropriate for one or two percent of the applications in any environment.

Click here to go back to my Home Page

Enter your email address to receive email when this page is updated.

Your Internet email address:

This page has been accessed times.

Last revised: Sunday, 28 March 1999