Testing is an integral part of every cable installation. George Georgevits offers a guide to diagnosing and solving cabling problems quickly and effectively.

For more than 20 years I have been running a test and measurement lab, and in that time I’ve seen it all – from bad cable and components to dodgy workmanship.

Clients often ask me to troubleshoot data communications problems by checking cable and component performance in the lab (because you can’t do it with a field tester), and also to confirm field test results.

Most office data installations today use unshielded twisted pair cabling of Category 5e or Category 6 grade, so let’s concentrate on the associated performance issues.

Here is what to do when the data cabling field tester indicates that a newly installed link or channel fails to meet the Standard.

Field testers

If data cabling is installed professionally, testing for Standard compliance of link/channel performance is usually included as part of the contract.

I have used several types of tester, and for basic link/channel testing up to Cat 6 there is not much between them. These days, Fluke is the make you are most likely to come across. Such devices have one thing in common – you have to tell it which Standard you are employing, and whether you are performing a link or channel test. If you get this wrong, then it’s simply garbage in, garbage out.

So, check that the tester has been set up correctly before you start work. While you are at it, check that test leads and any adapters are in good condition and that the tester is in calibration.

This takes only a few minutes, and it can save you a lot of angst, time and money.

When the field tester says ‘fail’

The link or channel under test is made up of various components connected in a chain, and any one of them could have a problem.

The tester goes through a sequence of transmission performance tests. When finished, it reports the results in summary form on the screen. If any parameter fails the test limits (which you have entered when setting up), the device reports accordingly.

Before analysing faults in detail, note that field testing assumes the various elements have been designed to be component level compliant according to the relevant Standard (Cat 5e or Cat 6).

A link or channel may show a ‘pass’ during the test even if it is not fitted with compliant components. However, it may not support the required application (eg: Gigabit Ethernet) when put to use.

Clients pay for guaranteed protocol support. They don’t care what your cable tester says – they just want the installation to work.

For this reason, it is very important to use components from reputable manufacturers and/or suppliers. Before you decide on a cable or jack brand for a project (particularly a big one), ask for an independent test report showing compliance with the relevant Standard.

I have often been called upon to test a cable or jack sample after a site has been cabled and terminated, only to find that many links/channels fail the transmission testing.

By that time it is too late. Someone has to pay for troubleshooting the installation, then pay even more to have noncompliant cable or components ripped out and new hardware installed. Things can get very ugly, particularly if a lot of money is involved.

Analysing field test failures

Here are some likely problems associated with various test parameters.

Wire map fail – will show up with crossed wiring and open/short circuits. The solution is obvious: visually inspect all punch-down connections. If they are OK, punch them down again. If that fails, re-terminate the link or channel from scratch.

If it still fails, look for damage to the cable. I have seen a screw driven through a data cable inside a wall when work was being done by another contractor.

I have seen cheap cable from China, and during manufacture the cable-forming machine had run out of a particular colour of wire. Someone at the factory simply joined the next roll by twisting the two sets of pairs together.

It showed up as a short on the wire map test. Clearly, the quality control department was out to lunch when that box of cable was made.

Insertion loss fail – can be caused by a poor punch-down connection or a faulty jack or patch cord plug. It is very rare for a link or channel with good terminations and connections to fail insertion loss.

Water in a cable will raise its insertion loss, particularly at high frequencies. Water will also show up in the NEXT and FEXT results, as these will be seriously degraded.

The sheath material used for data cables is not waterproof. If the cable is left in a wet environment (typically for several days – there is no hard and fast rule), water will eventually pass through the sheath and transmission performance will be permanently degraded. Return loss fail – this is most often a physical problem associated with the cable.

Return loss is a measure of how well the impedance of each pair in the cable matches the system characteristic impedance (100O).

Characteristic impedance is a function of the wire diameter, the pair spacings and the dielectric constant of the insulation material.

One common problem with Cat 5e cables in particular occurs when the coil-winding machine in the factory is not set up correctly. It places regular deformations on the cable as it winds the coil into the box.

Each dent in the cable causes a small reflection when tested. Because there are many such equally spaced dents (one for every turn of the coil), they add up and cause a spike in the return loss response. It occurs at the frequency where half the wavelength in the cable corresponds to the spacing of the dents.

This problem is evident as the cable is drawn from the box. If you see regular kinks – beware.

Cat 6 cable is usually supplied as a reel in the box, so the problem doesn’t occur.

A non-compliant jack can also have poor return loss characteristics. When tested in a link in combination with a low return loss headroom cable, it may show a ‘fail’ in the link or channel test.

NEXT or FEXT fail – two common causes are bad cable and poor termination technique when punching down the jacks.

Crosstalk occurs wherever two unshielded signal circuits are run in proximity to each other, and in a data cable four high-frequency signal circuits are very close.

Crosstalk in a cable is minimised by ensuring the exposure from each leg of each pair matches that of the opposite leg of each pair. This is done by forming each pair in the cable with a different, carefully calculated twist rate.

Pair-forming machines in the factory take care of this. If something goes wrong with the machine settings, the NEXT and/ or FEXT performance will degrade.

When terminating a jack, the twist of the cable pairs must be maintained right up to the point of punch-down. If this is not done, significant crosstalk can be induced into the open loops of each pair. Visual inspection is the easiest way to check this.

I have seen a link fail the NEXT and return loss tests because of shoddy installation.

A young apprentice had run cable on a difficult extended route and the cable was not long enough. So he terminated it with an RJ45 plug and used a telephone-grade RJ45/RJ45 joiner to connect it to another piece of cable similarly terminated, then hid the lot inside a wall.

Needless to say he was soon looking for a new job.

Cable or connector?

With a hard fault such as an open or short – or very bad crosstalk (indicating a short between legs of different pairs) – the fault usually lies somewhere in a termination.

Shorts and opens inside cable are very rare.

If a transmission performance issue involves some parameter failing by a decibel or two, it is most likely to be a cable problem, as cable performance dominates link or channel performance. A connector has to be quite bad to fail a link or channel that uses good-quality cable.

George Georgevits, BE (Hons), manages his own communications engineering consultancy Power and Digital Instruments, which was established in 1980. PDI specialises in lab and field transmission testing and troubleshooting of cabling systems and components, as well as general electronics and communications engineering. Contact PDI on +61 2 9411 4442.