SHDSL cable testing using WeConfig

Introduction

The DDW devices from Westermo are designed to reuse existing cable infrastructure for the
transmission of full duplex Ethernet at a data rate of up to 15.3Mbit/s on a single pair. Higher data
rates of up to 30Mbit/s by combining 2 pairs, please refer the DDW-142 and DDW-225/6 guides for
further details. To achieve the best performance, the cable pairs should be twisted. Cable used for
RS-485, telecoms or cabling used for bespoke data buses such as PROFIBUS, DH-484, MODBUS PLUS
or DH-485 are all ideal. It is possible to reuse non-twisted cores found in multicore cables or DH+ bi-
axial cable. However, it is strongly recommended when considering using non-twisted pairs that each
cable section is tested as detailed in this document, to check the performance before use.

This document outlines how WeConfig (version 1.4 or later) software tools can be used to verify the
suitability of a cable when using SHSDL. The SHDSL technology was originally designed to deliver high
speed data over existing POTS (Plain Old Telephone System) cabling. The POTS cabling is typically a
twisted pair, however non-twisted multicore cables can be reused for the transmission of SHDSL. This
document will give specific guidance when using non-twisted pairs as this cable type was not
originally designed for the purpose of data or voice transmission.

Cable definition twisted pair

In twisted pair cabling pairs of conductors are twisted together for the
purposes of cancelling out electromagnetic interference (EMI) from
external sources; for instance crosstalk between neighbouring pairs.
This type of cable is considered to be ideal for using with SHDSL.

Using SHDSL over multicore cables

SHDSL technology was originally designed to operate over twisted pair telecommunication cables to
deliver higher data rates than conventional modem technology over POTS cables. In some legacy
upgrade situations, the installed cables were designed to carry simple switched DC voltages inside a
multicore cable, so there was no need for twisted pairs. The lack of twisted pairs does not mean that
SHDSL will not work over a multicore cable, however the rate and reach may be reduced. The lines
will be prone to pick up more electrical noise than a twisted pair, due to the pair used being
unbalanced.
Where there is no other choice but to use multicore cables, it is advisable to test each core before
installation, as the characteristic will vary depending on the position of the individual core within the
bundle and any other signals present on other cores. The time spent testing should be seen as a long
term investment as the reliability of the complete system will be increased leading to a net decrease
in the whole life maintenance and operation costs.

Multicore with a quad inner

When using multicore cable, it is worth taking a close look at the inner four cores. These cores are
sometimes closely twisted together and may have a screen around them. This is often termed a quad
and is intended for use with voice communication. The quad cores are ideal for use with SHDSL and
will give at least the same performance as a regular twisted pair. Unlike any other type of cable
construction the pairs can be connected in parallel to for a single pair presentation.

In the above, cable Red - Blue are joined and Black – Yellow are joined. The tight winding around the
central filler gives the cable its exception noise immunity characteristic. When, making the cable off,
it is important to keep the conductors the same length and keep the unwound length to a minimum.

Testing multicore cables

When testing SHDSL over a multicore cable, ensure that the same pair of cores are used in the final
installation, as the position of the cores within a multicore will give different results depending on
the cross talk and level of noise that may be present in the cable.
 xKm

Never use two adjacent cores for SHDSL as the crosstalk will be extremely high. In extreme case, the
crosstalk will prevent a successful connection from being established. Selecting cores on either side
of the cable will yield better results when using non-twisted pairs. The presence of other signals in
the cable such as AC power or switched DC can also create interference. Therefore, consideration
should be given to testing the cable over a prolonged period so that all events that might affect the
performance of the connection are present for at least some of the test period.

A test that yields an unsatisfactory result does not mean that a reliable connection can never be
established, often selecting a different pair of cores further apart within the cable will yield a stable
result.

Improving results
Results can sometimes be improved by earthing one end of the unused cores within a multicore. It is
important that only one end of the cable is earthed or a circulating earth current could be introduced
through the cable. The circulating earth will have the net effect of introducing noise and make
communications almost impossible. This is also true for systems using twisted pairs.

xKm

What not to do to improve results!

Since the CSA (Cross Sectional Area) of the cores has a very large influence on the rate and reach it
would seem logical to use multiple cores in a cable joined together to increase the CSA. Never be
tempted to double up the cores to increase the CSA to gain extra performance, doing so will create
unpredictable reflections and crosstalk within a twisted or non-twisted multicore cable. The
Exception to this rule is the QUAD cores (see Multicore with a quad inner, above).

Cable testing.
There are dedicated xDSL cable testing equipment available that will give a high level of detail about
the signals and the performance of a copper pair. The results can be hard to interpret and not give a
clear indication as to the long term reliability of the connection. Alternatively, the Westermo
WeConfig (1.4 or later) software tools can be used to test and record the performance of the line
under test. Using WeConfig has the advantage of being able to capture and plot the performance of a
line over a prolonged period of time.
WeConfig SNR monitoring
The SNR monitoring built into WeConfig can be used to measure and track the changes of the SNR
value using the WeOS managed DDW devices. WeConfig offers more flexibility to monitor the SNR
than the older DDW-Tool as it uses SNMP to retrieve the values from the device over the IP
management interface.

WeConfig can be connected locally to the WeOS managed devices or remotely via a suitable L3
secure VPN connection such as WeConnect. The logging of the SNR data when using the Ethernet
connection to the network opens up the possibility of leaving WeConfig monitoring a line from
anywhere on the network. Unlike the previous DDW-Tool, more than one device at a time can be
monitored by WeConfig simultaneously either by a local Ethernet connection or from a central
location.
Setting up
The managed DDW devices and WeConfig can be used to monitor the performance of a twisted pair
or multicore connection. Using the DDW devices has the advantage of using the actual equipment
that will deployed in the final application. WeConfig can be left to log and trend the SNR present in
the cable for as long as desired or there is available hard drive space for the log file. For example, the
test system can be left in place for 7 days to ensure that day to day conditions the multicore is
subjected to over a normal working week and the weekend do not cause the communications to be
unreliable. The long term test may point to the need to drop the configured data rate due to periodic
noise when a pump starts or a train passes down a little used track.

xKm

WeConfig interfaces to the managed DDW device via the IP interface in each unit. This means that all
the DDWs in a system need a unique IP address. The devices do not have to be on the same subnet,
but can be on a L3 wide area network. The WeConfig software can be used to discover managed new
out of the box DDW devices and assign unique IP addresses and backup the configuration. It is
outside the scope of this document to show how to fully configure a managed DDW, for more
information refer to www.westermo.com for the WeOS configuration manual.
Monitoring multiple locations
The WeConfig tool is not limited to monitoring a single device or line but can be used to monitor
multiple devices simultaneously over an extended period of time. To monitor the DDW devices
remotely some sort of wide area connection is required. This will most likely require the gateway
address to point to the local router. The output from the monitoring is stored in a CSV file and can be
used with applications such as Excel to display or analyse the data.

The display of the SNR can be filtered to show one or more of the monitored connections live. The
ability to monitor multiple lines reduces the time required to test and report.

WeConfig can log the SNR from multiple locations over a LAN or WAN connection

xKm

xKm
Central office

xKm

WeConfig can log the SNR from multiple

xKm locations over a LAN or WAN connection

xKm

SNR and reliable communications.

All the DDW range of devices support auto-negotiation on the SHDSL line and can be configured for
different minimum line SNR. The objective should always be to establish a reliable connection with a
good SNR rather than the highest possible bit rate with a low SNR.

The SNR modes are defined as reliable, normal and fast;

- Reliable is a 10dB difference

- Normal is a 6 dB difference
- Fast is a 3 dB difference
The greater the difference between the signal and the standing noise on the cable, the greater the
reliability and tolerance to disturbances.

Using auto negotiation on a new line will give an indication of the maximum data rate a line will be
able to support. In most use cases reliability as opposed to the maximum data rate is the most
important factor. The automatically negotiated data rate should be seen as a starting point. The
operational data rate should be set manually to a level one or two steps below e.g.

The auto negotiation may settle on a data rate of 6,200Kbit/s. Setting the line rate to 5,376 Kbit/s will
still give a good line rate but also delivers a greater SNR, making the line more reliable and doing so
increase the availability. It is strongly recommended that the target should be for a lower more
reliable date rate.

SHDSL Spectrum usage

SHDSL is a baseband technology and uses a very wide frequency spectrum up to 2.55MHz @15.3Mb.
A typical legacy modem using V.32 or V.34 modulation would use 300-4000Hz (POTS system voice
band) by comparison.

As with all communications using RF signalling, the signal is attenuated over distance and the higher
the frequency the greater the attenuation. The net effect is the rate and reach will drop over
distance. The effect of the attenuation can be seen in the SNR dropping over distance as the level of
usable signal drops in comparison to the background noise on the cable under test. Configuring a
lower data rate between the DDWs will reduce the frequency range the SHDSL protocol will demand
from the cable. In doing so, the very high frequencies which are more attenuated over distance will
no longer be used, leaving the less heavily attenuated part of the spectrum to carry the SHDSL
protocol and in doing so will increase the SNR value. The Reach calculator described later can be used
to give a prediction of data rate a cable with a given CSA (Cross Sectional Area) will be able to carry.

The relationship between the data rate and the frequency range present on a cable can be calculated
using the following formula;

(Bitrate + 8192 (overhead constant)) / "number of bits per symbol"

"Number of bits per symbol" is according to the PAM (Pulse Amplitude Modulation) mode used in
selected speed.

- <= 2.3 Mbit/s: PAM16 = 3 bits per symbol

- <= 5.7 Mbit/s: PAM32 = 4 bits per symbol
- <= ~10 Mbit/s: PAM64 = 5 bits per symbol
- <= 15.3 Mbit/s: PAM128 = 6 bits per symbol
Example:

(192000+8192)/3= ~66000 Hz

The used spectrum will be from DC – 66 kHz.

At data rates of 15.3 Mbit/s, the spectrum will be: 15,300,000 + 8192 / 6 = 2.55 MHz

The used spectrum will be up to – 2.55 MHz

The calculated frequency range is an approximation. Under certain conditions.

Note: under certain scenarios the PAM encoding could be lower or high than the rates indicated
above this will have an impact on the frequency range e.g.

10,000,000+8192 at 5 bits per symbol = 2 MHz

10,000,000+8192 at 6 bits per symbol = 1.66 MHz (fixed data rate and a very quiet background noise
level)

Knowing the frequency spectrum can be useful when working out if existing lightning or isolation
transformers are compatible with the SHDSL signals. This is often overlooked and can be the source
of an unexpectedly low performance over a line. The majority of the line isolation transformer are
designed for a POTS frequency range of 300-4000Hz. The design of the transformer will
disproportionately attenuate the higher frequencies. See the Surge and lightning protection section
for more details.

Reach calculator
The rate reach calculator built
into WeConfig can be used to
decide on a starting value for
the manual line rate selection.
The calculator gives an
indication of the possible data
rate and distance a line will
reliably sustain when using a
cable of a given CSA (Cross
Sectional Area). The
environmental noise expected
can be adjusted to give a more
realistic result.
The reach calculator is designed to give an indication for the likely performance of SHDSL on twisted
pair cables. If the target cable is a multicore, there will be a great deal of variance depending on
which core are selected and where they are positioned in the cable in relation to other signals that
may be present.

EMF (Emergency Freeze)

As with any communications that use modulated analogue signals to transmit data over a copper
pair, there will be a possibility of external noise interfering with the transmission. The SHDSL protocol
used to modulate the data has multiple mechanisms to detect and correct any errors caused by
interference or corruption of the baseband signals. However, if the interference continues for too
long or the level of corruption overwhelms the error correction, then the synchronisation between
units will be lost. Normally, this would force a renegotiation of the link between the DDW units. To
help overcome the short duration glitches on the line, the managed DDW devices support emergency
freeze. The emergency freeze triggers when the communication between two units has broken down
due to a loss of synchronisation. The last known good synchronisation values are frozen in memory.
After the glitch or burst of noise has passed, the DDW units will attempt to re-establish the link using
the frozen values. The difference this makes is to potentially limit a break in communications to a few
milliseconds or tens of milliseconds, as opposed to a full renegotiation, which could take 60 seconds
or more. The emergency freeze is enabled as default and rarely, if ever, needs to be disabled.

Failure to negotiate
In some cases, the DDW units will not be able to establish a connection. This may be due to the core
or cores selected being faulty or the line having a particularly high level of background noise. DDW
units can do nothing about a faulty cable, but the managed versions can be adjusted to compensate
for the noise level on the cable. A failure to negotiate is most likely due to a “quiet” period not being
detected after the initial G.hs probe tone has been sent by one of the DDW units to the unit on the
other end of the line. During the negotiation, a sequence of tones are exchanged between the units
to detect if there is a unit at the other end and the SNR of the line. A high level of background noise
during the test phase means that no quiet period is detected and the negotiation is halted while the
DDW retries the probe (G.hs) tone. Where the noise floor of the cable is at a higher level than the
pre-set, the G.hs quiet period detection can be adjusted to cope with this high level of noise using
the G.hs threshold setting. The level can be set using CLI, GUI or using WeConfig 1.4 or later, on the
managed DDW device. Adjusting this parameter should be seen as the last resort after all other
means of establishing a connection have been tried.

Surge and lightning protection

As with all cables between two geographic locations, there will be voltages induced into the cable
from external sources. Normally, this can be seen as the base level of interference or background
noise on the cable. Under some circumstances, such as a fault condition on the local electricity grid,
the energy induced in the cable can be much higher than the normal background. This is particularly
the case in the railway environment, where it is entirely possible to induce 10s to 100s of volts in the
cable.

Lightning strikes are also a source of interference and can potentially cause catastrophic damage to
the line interface and even connected equipment.

TBU (Transient Blocking Unit)

The DDW range of device have a state of the art TBU and Gas discharge on the line interface.
However, this should not be confused with lightning protections or protection against all type of
surge conditions that can occur on a cable. The TBU will very effectively protect against Differential
mode, also known as Transverse, Normal or Metallic modes. This is where the potential on one of the
conductors in a pair will rise in respect to the other.

The inbuilt protection will have little or no effect on Common mode, also known as Longitudinal
mode type surges. This is where the potential on both conductors rise by the same amount
simultaneously.

Differential Mode Common Mode

Transient
V1 Voltage V1
DDW Line DDW Line
interface interface
V2 Transient
Voltage
Earth Earth

It is strongly recommended that external line protection units are always installed where lines are
running between two geographic locations. The cost of the protective devices is far smaller than the
cost of a replacement DDW or the system down time.

External surge suppression devices

To mitigate the effects of a Common mode, or differentially induced high voltages from a lightning
strike or induced surges, an external protection device can be fitted to each end of the line. The
selection of the correct device is important as a device with the incorrect characteristics will clip or
degrade the signal. See the required characteristics below.

Required characteristic:
- Must have a line clamping voltage above 15 Vpp “Voltas peak to peak” (SHDSL signal can be
as high as 12 Vpp)
- Dynamic frequency range up to ~2.55 MHz
There are many devices of this type available from;

- DEHN
- FURSE
- ADC
- TYOCO
- 3M

The following units can be used from DEHN.

Up to 5.7Mbit/s:

- BXTU ML2 BD S 0-180 (2-wire)

- BXTU ML4 BD 0-180 (4-wire)

Up to 15.3Mbit/s

- BXT ML4 BD HF 24 (4-wire)

Singularly the most important point is to ensure that the suppression device is connected to a good
low resistance earth point, meeting the suppliers’ cross sectional area requirements.

NOTE: The protection devices will give protection against a near strike, but there is very little that can
be done to mitigate a direct lightning strike on the cables or installation. In areas where lighting is a
common occurrence, consideration should be given to airborne lightning suppression using lightning
conductors.
Hot site isolation
Where the DDW is located in a high voltage substation (hot site), it
is recommended that a 10-15 kV line isolation transformer is
installed. The isolation transformer not only protects the
equipment but also anybody who might be testing or maintaining
the line during a fault condition. There are very few hot site
isolation transformers that are capable of passing the frequency
range required for SHDSL. The majority of the line isolation
transforms are designed for a POTS frequency range of 300-4000 Hz. The design of POTS transformer
will disproportionately attenuate the higher frequencies and result in unexpectedly poor
performance.

A suitable hot site isolation transformer is available from;

Kenton Research
LIU102A and LIU202A
Line isolation up to 15 kV

The use of isolation transformers will add losses by 3 dB per transformer. Since the loss is close to
linear, the background noise and signal will be affected equally so the SNR should remain the same.

Author:
Ray Lock, Network Technology Director
Westermo Group

WeConfig software can be downloaded from the following location;

WeConfig Download