Trouble
Shooting Structured Cabling Systems
Structured cabling
systems account for between 30 and 70% of all network problems, yet only attract
between 0.5 and 5% of the LAN budget. A
professionally designed and installed cabling system will certainly pay for
itself. But if it doesn’t work,
what are the likely problems?
First of all
how do you know it doesn’t work?
If
the cable system has already been installed and working then the first
indication of a problem is a failure of the LAN to communicate to all of its
stations. This will be caused by:
failure of
LAN hardware or software
mechanical
damage to the cabling system
unauthorised
change or disconnection
intrinsic
failure of the cabling components
poor
installation practice
external
EMC/EMI problem
Problems Most likely cause
| Complete failure
of the LAN
|
Major server/switch/software fail |
| Individual
stations disconnected
|
Local failure of LAN cards/switches |
|
Localised mechanical damage to the
cabling |
|
Unauthorised
tampering / disconnection at patchpanel |
|
Cables too long |
| Long
response times
|
External electrical noise, EMC |
| Failure to
autonegotiate speed
|
Cable grade too low |
To
differentiate between active and passive equipment failure then the common
practices of swapping cards and ports can be used to isolate the problem.
There are specialist test equipment devices that can send packets of data
around a LAN to stress and exercise the system.
There is also a device called a Bit Error Rate Tester, BERT, which will
send blocks of data and compare what has been received to what has been sent.
The discrepancy is known as the bit error rate.
ATM and gigabit Ethernet require a bit error rate of 1010 or
better, i.e. for every ten thousand million bits of data sent, not more than one
bit is received in error.
Having
decided that it is the cable plant at fault, what are the likely causes;
|
Mechanical
damage
|
Cut
cable due to inadvertent removal, insufficient cable containment, sharp edges,
cable stepped on etc.
|
|
Flood damage from burst pipes, leaking roofs
|
| Cable
ties too tight
|
Nylon
cable ties are often applied too tightly and dig into the cable.
If this is coupled with very narrow ties, e.g. 2 mm, then this will
degrade high speed performance.
|
|
Bend
radius too small
|
High speed LAN
cables cannot go through 90 degree bends. The
manufacturer’s specification must be adhered to e.g. 25 mm is typical.
|
| Incorrect
termination |
The 8
wires of the cable must be correctly connected at each connector, with a maximum
of 13 mm ‘untwist’ on each pair. |
| Cables
too long
|
Category 5 and 6 cables are limited to 90 metres to guarantee all
protocols will be transmitted, Optical
fibre also has length restrictions for different protocols.
|
| Unauthorised
disconnection
|
Patchpanels can
be very vulnerable places, and it is very easy to mistakenly change the wrong
patchcord and disconnect a user or group of users.
At the work station end, patchcords can also be easily pulled out if too
exposed. The solution is to lock
all patching cabinets and telecom closets.
|
| Long
response times
|
Outside electrical
interference will cause the LAN protocol to keep on requesting retransmission of
the data packets. The user will see
this as long response times from the server or even unexplained logging off.
Ensure all data cables are separated from power cables, e.g. 75 mm and
130 mm from fluorescent lights. Severe
problems will need screened cable or optical fibre.
|
| Failure
to autonegotiate
|
Many
users now have 10/100 NICs in their PCs but are only communicating at 10 Mb/s to
the hub/switch. When the switch is
upgraded to 100 Mb/s, e.g. 100BaseTX, all the attached terminals will try to
autonegotiate up to the higher speed. If
the cabling is not up to this requirement then this exercise will fail and the
PCs will fall back to 10 Mb/s.
|
Modern
structured cabling needs to be 100% tested to ensure compliance with the latest
standards. These are now Cat5e (TIA 568A –A5, ISO 11801 Class D Amd. 1 &
2, EN 50173 Class D Amd. 1) or Category 6, Class E, to the latest draft
standards e.g. ISO 11801 2nd Ed 2001 Class E.
There is also a standard for testing cabling currently installed to see
if it is gigabit Ethernet compliant, TIA/EIA 568-A TSB 95.
There
are five manufacturers of suitable copper cable test equipment.
The cable must be tested from both ends to all the requirements of the
above standards or to IEC 61935.
Optical
cables can be tested with an optical power meter and calibrated light source.
This will tell if the cable link is satisfactory.
If there is a problem then the power meter will identify that there is a
problem but not where it is. For
this an instrument called an OTDR, Optical Time Domain Reflectometer, is
required. This device will give a
complete ‘picture’ of the fibre link and identify any problem areas, but
only after expert interpretation.
IEC
61935
Generic
specification for the testing of balanced generic cabling in accordance with
ISO/IEC 11801
The
tests required are;
|
IEC
61935
|
|
Wire Map
|
X
|
|
Attenuation
|
X
|
|
NEXT pair to pair
|
X
|
|
NEXT Powersum
|
X
|
|
ELFEXT pair to pair
|
X
|
|
ELFEXT Powersum
|
X
|
|
Return Loss
|
X
|
|
Propagation Delay
|
X
|
|
Delay Skew
|
X
|
|
DC Loop Resistance
|
X
|
Cable
length and ACR are also useful additions to this set of tests.
The
best way to ensure success in a structured cabling installation is to use
properly trained people to design, implement and test the system.
The RCDD qualification from BICSI is the only qualification which covers
all aspects of structured cabling design and implementation.
The
above information is offered as a summary of ISO 11801 and related standards.
It is not a definitive design guide and does not replace study and
implementation of the Standards themselves.
The publisher accepts no responsibility for inaccuracies or omissions.
To purchase the full Standards go to your national standards body, e.g.
British Standards Institution, Nederlands Normalisatie Instituut etc. or ISO.
