20 ways to compromise a cabling system

02-05-2012 - John Hatcher

  1. One of the most frequent problems is simply not obeying the standards. Granted they are not the most exciting reading! But if people understand and stick to them, then the resulting installation will work exactly as it’s supposed to.
  2. In a similar vein, some installers manage to ignore what’s written on the customer specification. Again, it is important to make sure that its stipulations are transmitted to everyone involved in the installation.
    Chris told the story of one site he visited where an end-user was struggling because he couldn’t get any optical patch cords to go fully into the patch-panel “holes”. An investigation discovered that the installer hadn’t oven-baked the terminations to cure them.  “Oh we always use self-curing epoxy” he said.  Sadly the customer’s spec demanded oven-cured terminations.  Unlike this chap, the stores had read the customer spec and sent him the correct temperature-curing epoxy. Thousands of fibres had to be re-terminated at the installer’s expense.
  3. Choose your time to run in cables. Doing this while other trades are working will inevitably lead to cables being stood on, or ladders propped on them - which will crush the pairs – leading to out of balance and signal reflectors (return loss). Also damage can be caused to the sheaths allowing moisture ingress which could cause problems in three/four years.
  4. Really take care pulling-in cables. The standards call for a maximum pulling force of 25lb/ft or 110Newtons. In practical terms that means if the cable won’t pull gently then you need to do the pull in shorter stages. Also, avoid pulling round any tight corners as this will damage the sensitive cables. Note that the safe bend radius for pulling-in is much greater than for cables when installed.
  5. Conduit, much loved by electricians, is particularly prone to cause cable damage. Right angle bends and small circular “BESA” boxes cause significant friction and opportunities for accidental bending and kinking.
  6. Both the structural cabling standards and electrical cabling standards advise that cable containment systems (trays, baskets, trunking etc.) should not be more than 40 per cent filled at initial installation.
    We all know it’s really tempting to cut down on these expensive and labour-intensive elements – but if you ignore this advice, cables at the bottom are very likely to get crushed, pushing them out of balance – and those in the centre may well overheat – particularly now that high-power PoE (power over Ethernet) is being deployed.
    Most people don’t realise that the data performance goes down by 4 per cent for every ten degrees over 20°C.
  7. Remember when installing/dressing cables that neat is bad – and so are tight nylon ties. Parallel cables induce Alien Crosstalk and so bunches should be random and loosely tied, preferably with Velcro ties.
  8. A significant problem is due to the fact that most comms rooms are close to electrical switch rooms and cables from each have to share ceiling voids and cable risers. Parallel cables act like transformers – and so massive 50Hz currents, a well as noise spikes, can be induced into data cables if a minimum of 50mm separation is not maintained.
    Also avoid routing cables near “noisy” switched power supplies, motors and fluorescent lights. If noise is an issue on the site, use a shielded rather than UTP system.
  9. Beware “not quite standard” cables. Recently a lot of cheaper cabling advertising itself as Cat5e, 6 or 6A has come onto the market. These cables actually have smaller diameter copper conductors than dictated in their respective standards.
    These simply will not perform properly due to the extra resistance causing higher attenuation.
  10. Another non-standard cable (cheaper of course) has also entered the market; this time with the correct diameter conductors – but made of copper-plated aluminium. Again, because aluminium has a greater resistance these will not perform properly and data throughputs will be adversely affected. They also have a tendency to snap at connection points.
  11. Unbelievably, even though the data network is the foundation of most companies’ essential real-time financial and commercial systems, both installers and end customers “save” money by putting in a top brand system for the permanent links and then ruin the entire system’s performance by fitting cheap patch cords.
    Many published tests have shown that 68-80 per cent of such patch cords don’t actually meet the standards (despite what it says on their packaging). These then drag the previously tested system down to the level of the weakest link. You’ll probably also invalidate the warranty on the system too.
  12. How often have you seen the situation on site where, because another trade is running late, the installer has pulled-in all the cables as far as possible, coiling up the slack into a big hank and leaving it dangling off the end of the cable trays? The stress and strain this puts onto the cables is immense, pushing many of the pairs out of balance and often leading to a proportion of the channels failing tests with expensive replacement cables having to be pulled in.
    Try to avoid pulling in any cables which cannot go their full run. But if you are forced into this situation – ensure that the slack is laid on top of the ladder racking and fully supported so that no cable damage can occur.
  13. Product managers will frequently suggest you terminate cables in the comms room while other trades are finishing off. Avoid this at all costs. RJ-45 outlet contacts will be damaged by excessive dust – so insist that you are handed over a clean, finished room.
  14. Remember that all multimode fibre is not the same. There are 50 micron and 62.5 micron legacy types around and plugging one into the other is just like pushing a 15mm water pipe into a 22mm one – water leaks everywhere. Well so does the light!
    More modern multimode fibres are 50 micron but, but whereas with OM3 a 10Gb/s signal will go 300 metres and on OM4 550m - it would only travel a mere 72metres on the legacy 50 micron fibre!
    Beware also, even if the fibre type is consistent in the permanent links, a patch cord with the wrong fibre core can bring a channel down to its knees,
  15. This one is probably everyone’s pet hate; those fiddly annoying dust caps on optical connectors. We really can’t stress enough how important they are.
    Push a dirty patch cord into a patch cord adaptor and you’ve not only introduced extra dB’s of loss, you’ve now contaminated the internal connector – and the only way to clean that is to open the patch-panel and jeopardise all of the other working channels.
    Always, always use the dust caps – and even more important, always clean every fibre connector before you insert it into an adaptor.
  16. Sadly we still see regular occurrences of “cost cutting” where an installer will try to use the wrong product to “save” cost. (Though in fact it’s frequently discovered and the re-work cost ends up being ten times the cost of doing it right in the first place).
    Here’s a superb example Chris saw on site recently. Correctly, the installer had run in a loose tube fibre cable across campus (it is easier to water-block and copes with extreme temperature variations better).
    Unlike “tight-buffered” fibre cables used for internal runs (where each fibre has a protective plastic coating) in a loose tube cable the fibres are completely bare. So you can imagine how fragile they are.
    The standards, and good practice, tell us to take the tubes into a protective splice tray and there to splice them onto pre-terminated “pigtails” – plastic protected fibres. But not this installer. He took these fragile little fibres right the way through the patch panel and straight onto the front panel adaptors! It was a disaster waiting to happen and “Re-do the lot properly” was Chris’s immediate command.
  17. Another area where installers regularly take “minor” liberties with the standards in order to “save costs” is in channel lengths. The standards specifically say 90m plus up to 10m of patch cords. Note carefully; they do NOT say 100m in the permanent link.
    On many jobs it would be possible to save having extra comms closets if channels would stretch just that bit further. Exceptionally some manufacturers will warranty longer links – but only with the agreement of their designers and the use of very specific products.
    But beware – if years later the user uses a “normal” patch cord instead of the specific ones dictated – there’s a good chance the channel will fail.
    Better then to stick to standard compliant 90m and in it in that extra comms closet!
  18. Beware, also, the temptation to run extra-long channels “just for video”
    It is true that video will work over a couple of hundred metres of Cat6. But if you run this cable as part of the structured cabling system, users in years to come will treat it just like any other channel – because it comes from the patch panel and ends on an RJ-45 outlet. Then of course there will be a warranty claim when they try to use it for data and it doesn’t work!
  19. Now when it comes to testing, there is still a culture of trying to get away with star passes.
    Let’s be clear, a star pass is so marginal that it’s a maybe/maybe not situation – and it’s simply not good enough.
    In 99.9 percent of cases, a star pass tells you there’s something wrong with the way the channel has been installed. What’s more, if it’s not investigated and rectified, the channel will almost certainly drift “out of spec” as the components age.
    If you value your reputation – and you should – make sure you hand over a system with zero star passes.
  20. With optical testing always use a light source and power meter. The older OTDR approach, because it has a manual rather than pre-set set up is too easy to “fudge” and if, as we hope you do, you want to hand over a rock-solid system to your customer, you will need to exclude this “opportunity to cheat” from the outcome.

In conclusion
It’s sad, says Chris, that in what is now a very mature mainstream market these things are still happening – and regularly too.
And they are bad all round. For the installer they frequently mean expensive re-works or call-backs to site which can wipe out the profit on a job.

For end customers it can mean unseen costs of “slow” connections and frustrated users – to financial trading situations where milliseconds of latency can mean £millions lost.

Brand-Rex and all of its top-tier competitors are determined to stamp out these practices – but every installer needs to adopt the mantra “No shortcuts”

March / April 2013 Issue

Click here to view latest issue online.

Don't miss an issue of Data Centre Management.
Subscribe Now