WIDEBAND TESTING By Florin Hodis, Member of Technical Staff The evolution of digital subscriber line (DSL) technologies and the increase of the bandwidth used present new testing challenges to service providers who deliver enhanced broadband services. In order to ensure quality of service (QoS) when delivering broadband services, wideband testing becomes essential. A wideband test is essentially any test method that involves measurement of parameters that depend on bandwidths up to 30 MHz, such as single-end frequency response measurement, PSD noise measurement, swept longitudinal balance test, etc. The Need for Speed Over the last decade, major technology advancements related to broadband communications have left their mark, which has influenced customers to demand more choices in the services that are provided to them. To meet the growing needs of customers, broadband services are gradually changing. The growth in popularity of Internet services, on-line gaming, voice-over-internet-protocol (VoIP), high-definition television (HDTV) and video-on-demand (VOD) requires Telcos to provide an increase in data rates for the access networks. Figure 1. Bandwidth requirements increase In the case of video services (IPTV or VOD), the bandwidth requirements vary based on the utilized compression algorithm (MPEG-2 SDTV requires about 4 Mbit/s and HDTV about 18 Mbit/s ). In the case of using H.264 (MPEG-4 Part 10), the SDTV rate is approximately 2 Mbit/s and HDTV is approximately 10 Mbit/s. In most cases, video services require delivery of multiple channels a requirement that can significantly increase the data rate. Telecom Test and Measurement
History of DSL While the theoretical capacity of copper to transmit data was well known, the practical use of telephone wires for high-speed data was first demonstrated in the late 1980s, when the feasibility of sending broadband signals was demonstrated through mathematical analysis. Also, the realization that many users would benefit from the higher data rates possible in one direction led to the introduction of asymmetry (represented by the letter A in ADSL) and the power it would bring. Then came the development of the discrete multitone (DMT) standard, which is the standard for most DSL circuits. By separating the signal into 256 sub-channels, many problems relating to line noise and disturbance can be minimized. Figure 2. History of DSL standards In June 1999, ADSL was standardized by the ITU-T. According to many experts, G.992.1 (G.dmt) and G.992.2, (G.lite) were the standards the industry needed to set the stage for worldwide mass-market deployment of broadband Internet services via copper and by 2002, the ADSL2 standard was approved. With the introduction of ADSL2+ (also known as ADSL2 plus) in 2003, carriers and service providers began to see the bandwidth numbers needed to effectively deliver IPTV. G.bond (the ITU standard that is also known as Bonded ADSL2+ ) was introduced in early 2005 as a way to double the downstream data rate of copper pairs. G.bond was most effective for customers 6 Kft or more from the central office (CO); however, it lacked the short distance and high data-rates that video demands. Very-high-speed DSL (VDSL) was ratified in 2003, but service providers concluded that more work was needed on the standard and that this new standard showed the potential of much higher data rates. VDSL2 (G.993.2) was consented in May 2005 and officially approved in May 2006. With the introduction of VDSL2, the industry has seen a major transition from high-speed Internet (which is primarily a data-only service) to broadband triple play: voice, video and data. Table 1 below illustrates a comparison of the xdsl technologies: ADSL ADSL2 ADSL2+ VDSL2 30a profile Standard G.992.1 G.992.3 G.992.5 G.993.2 Bandwidth 1.104 MHz 1.104 MHz 2.208 MHz 30 MHz Line coding DMT DMT DMT DMT Bins 256 256 512 3479 Tone spacing 4.3125 KHz 4.3125 KHz 4.3125 KHz 8.625 KHz Max downstream 8 Mbit/s 15 Mbit/s 24 Mbit/s 100 Mbit/s Table 1. xdsl technologies comparison As demonstrated above: Bandwidth has increased from 1.1 MHz for ADSL to 30 MHz for VDSL2 Same DMT line coding is utilized for ADSL/2/2+ and VDSL2 Number of bins increased from 256 to 4096 sub-channels for VDSL2 17a profile Maximum downstream increased from 8 Mbit/s up to 100 Mbit/s
Figure 3. ADSL2+ based on DMT Figure 4. VDSL2 bandplan VDSL2 is a true worldwide standard that provides numerous configuration profiles and bandplans to meet regional service provider requirements. The frequency bandwidth has increased to 30 MHz, with configuration options at 8.5, 12, 17.7 and 30 MHz. Wideband Testing Single-End Frequency Response Measurement With the introduction of VDSL2, the frequency bandwidth has increased to 30 MHz. When performing the frequency-response test, the spectrum of test tones needs to take into consideration the VDSL2 bandplans/profiles and must cover the spectrum up to 30 MHz. Twisted-copper, telephone cable is characterized by electrical impedance measured in ohms. The longer the cable, the more impact on the resistance (symbol R a real number) and the reactance (symbol X imaginary number) will have. The telephone cable manufacturing process will influence the characteristics of the cable resistance/length, capacitance/length and inductance/length. When all of these factors (L, R and C) are taken into account, it can be observed that the twisted-pair cable has complex impedance that varies with length. Therefore, twisted-pair cable has a non-linear attenuation higher frequencies are attenuated more than lower frequencies, and this effect is enhanced on longer circuits. The end result is that short cables have less attenuation overall, and the difference in attenuation between low frequencies and high frequencies is not as pronounced. On the whole, long cables have greater attenuation as compared to short cables. In addition, the higher frequencies on long cables are more attenuated when compared to lower frequencies. Figure 5. xdsl rate and reach as result pf loop attenuation
APPLICATION NOTE 182 In the case of the single-end, frequency-response measurement, the test equipment is placed at one end of the local loop, and the other end of the local loop is either on an open circuit or a short circuit. Test instruments utilize specific test tones that are transmitted at one end of the line. After some isolation, the transmitted signal is canceled from interfering with the receiver circuit. What remains are only the signals that have traveled the length of the circuit; they have been reflected back from the open or shorted end of the cable and travel once again the length of the local loop. The one-way attenuation is one half of the overall attenuation measured. With the results from this test, a technician can determine if various points of loss across the specific bandwidth of interest are too great to be able to transport xdsl signals. In addition, the technician can also view the roll off of the loop and observe notches that are caused by bridged taps and the ringing effects caused by loading coils. PSD Noise Measurement Noise is any unwanted signal that could adversely affect the desired signal as it passes through the transmission path. The end result might be a corrupted signal that can be misinterpreted by the receiver, which translates to errors in the digital bit stream. If the noise is corruptive enough, the far-end device may not be able to communicate at all; whereas at marginal noise levels, the data transmission rate is slowed down. The noise found on local loops is caused by many sources, including light dimmers, radio signals, neon lights, electric trains and adjacent power lines, to name a few. The most common and problematic types of noise are caused by the electromagnetic coupling of signals from one local loop to another within the same cable bundle; this effect is known as crosstalk. In fact, the reason that the telephone cable is twisted is to reduce crosstalk. A quiet terminated noise measurement is the best method for measuring noise. Again, it is desirable for the technician to be able to view a spectral display (spectrum analysis) of the measured noise. Due to the fact that ADSL, ADSL2+ and VDSL2 use many carriers (up to 4096 sub-channels, VDSL2 17a profile), as long as the total power of the noise is concentrated in a narrow band of frequencies, xdsl transmission remains possible. For this reason, a numeric measurement of the total noise power in the transmission bandwidth is absolutely meaningless. Only a measurement that shows the noise distribution in each of the carrier s bandwidths has any meaning. The telecom industry has coined the term power spectral density (PSD) to indicate the graphical display of noise power at various frequencies. Figure 7. Power spectral density noise measurement up to 30 MHz Quiet noise, background noise, noise margin or idle channel noise, as it is often referred to, can be determined by a measurement taken with no test signal impressed on the line. Any background signals found on the cable are therefore measured and considered to be noise and/or crosstalk noise. It should be noted that the signals used to carry xdsl might be very low in terms of received level, especially those carriers in the upper frequencies where the attenuation presented by the local loop is the greatest. For this reason, the pass/fail limits established for testing voice circuits are not valid for xdsl. Likewise, it is important that testing equipment used for the assessment of a local loop be able to measure noise power levels that are low enough. Normally, the test set should measure to as low as -140 dbm/hz (roughly equivalent to -110 dbm).
The effects of crosstalk can only be determined in live bundles that already carry several xdsl services. Consider, for example, an F2 facility with a pair count of 25. If no xdsl or any other higher-rate service has been deployed on that bundle, there is no useful crosstalk information to be measured. Once the first DSL circuit is deployed, the amount of crosstalk in the band of interest (for example up to 30 MHz for VDSL2) will drastically increase. Each subsequently added xdsl service will continue to add to the crosstalk level. As far as an unused pair in that count is concerned, every additional service added to the bundle is a potential disturber to signals that will be added to the unused pair; therefore, the maximum number of disturbers in a 25-pair count is 24. Many carriers have learned through experience that on long lines, the first xdsl deployed generally works very well until the second service is deployed. Figure 8. ADSL2+ signal interference detection A bundle of local loops may carry all the signals for an entire neighborhood or business park. It is a given that cable bundles have plain old telephone service (POTS) signals on the majority of the utilized loops. They may also contain traditional signals, such as those used for security-system alarm monitoring, T1/E1 transmissions, ISDN BRI, IDSL, DLC and HDSL. When adding one or more xdsl signal such as ADSL, ADSL2, ADSL2+, SHDSL, VDSL and VDSL2, it is important to take into consideration the signals that already exist in the bundle. In most cases, a B8ZS-coded T1 or HDB3 code E1 signals are not spectrally compatible with xdsl in the same bundle. The level and frequencies of crosstalk produced by these signals could inhibit the DMT-coded signals to coexist in the same cable. The power spectral density (PSD) signature of a T1 or E1 signal varies based on the type of line coding, the type of framing and the actual data carried. Being able to characterize and detect interfering signals allows technicians to isolate harmful interfering signals that originate from services that are not spectrally compatible with xdsl services. In addition to the type of signals in the bundle, the transmitted power levels are also important. In the United States, the Federal Communications Commission (FCC) has set limits to the amount of power that can be transmitted over a loop. The European community and many other countries have also set similar limits. Since cable bundles are shared between incumbent telephone companies and the competitive Telcos (CLECs), according to line-sharing and loop-unbundling regulations, some level of spectrum policing is needed. Appropriate testers for this application offer the ability to bridge onto in-service lines in order to determine the transmitted levels on the local loop. Note that signals should be measured at or very near the modem or DSLAM in order to measure the correct levels. In addition, the VDSL2 standard specifies multiple transmit power levels based on utilized profiles in order to accommodate the fiber-tothe-node (FTTN) installations where passive optical network (PON) technologies are utilized between the CO and a remote DSLAM, and then VDSL2 is deployed from the remote DSLAM to the subscriber. Profile 8b 8a 8d, 12a, 12b, 17a and 30a 8c Tx Power dbm 20.5 17.5 14.5 11.5 Table 2. VDSL2+ Tx Power levels The remote DSL access multiplexer (DSLAM) providing VDSL2 service may share the same cable with ADSL/2/2+ services provisioned from the CO location. ADSL/2/2+ signals will be significantly attenuated, and as a result, the transmission level of the VDSL2 signal needs to be at a level that will not interfere with the existing ADSL/2/2+ service.
Swept Longitudinal Balance Test Twisted-copper telephone cable consists of two conductors both of which are floating and balanced with respect to ground. In other words, neither is connected, not even partially, to ground. Measuring the longitudinal balance of a pair gives an indication of the different voltages caused by imperfect balancing of the tip and ring with respect to ground. The measurement is given as a level in db: the higher the db reading, the better the balance of the cable pair under test. A poorly balanced pair invites greater near-end crosstalk (NEXT) or far-end crosstalk (FEXT) that may disturb the carried signals. Longitudinal balance measurements are extremely useful for identifying loops that will suffer from crosstalk once a cable bundle is loaded with broadband signals. This is a good way to prevent future problems on bundles where DSL services have not yet been deployed. It should be noted that loops that have improper longitudinal balance act as an efficient crosstalk receiver or transmitter, and for this reason, a single, poorly balanced loop can bring down an entire bundle. A well balanced loop will minimize the effects of impulse noise. The impulse noise can be narrow or wideband, depending on the spectral signature of the disturber signal. Impulse noise is the number one enemy to all DSL-based services and must be minimized. Longitudinal balance is affected by short circuits (or partial short circuits) between either conductor and the cable sheaths and other ground sources water in the cable is one of the principal sources of partial conductivity to ground. For xdsl deployments and loop fault-finding, it is important to measure longitudinal balance at many frequencies across the utilized bandwidth. Although a DC fault (i.e., conductivity to ground) will cause problems with longitudinal balance across all frequencies, AC faults (i.e., capacitive conductivity to ground) may only present itself as high-frequency imbalances. Capacitive faults occur when cable bundles are squeezed or kinked during installation or with frost heaving, earthquakes, construction loading, etc. As a result, measuring longitudinal balance at all frequencies covering the specific xdsl technology up to 30 MHz in case of VDSL2 allow the detection of faults that may not be visible at certain low frequencies. Conclusion In order to deliver reliable broadband services, it is very important to invest in improving the access network infrastructure in order to eliminate, as much as possible, the malfunctions at the physical layer. The investment in cable plant wideband conditioning needs to be taken very seriously in order to avoid future customer complaints and unnecessary truck rolls. EXFO Corporate Headquarters > 400 Godin Avenue, Quebec City (Quebec) G1M 2K2 CANADA Tel.: 1 418 683-0211 Fax: 1 418 683-2170 info@exfo.com Toll-free: 1 800 663-3936 (USA and Canada) EXFO America 3701 Plano Parkway, Suite 160 Plano, TX 75075 USA Tel.: 1 800 663-3936 Fax: 1 972 836-0164 EXFO Europe Omega Enterprise Park, Electron Way Chandlers Ford, Hampshire S053 4SE ENGLAND Tel.: +44 2380 246810 Fax: +44 2380 246801 EXFO Asia 151 Chin Swee Road, #03-29 Manhattan House SINGAPORE 169876 Tel.: +65 6333 8241 Fax: +65 6333 8242 EXFO China No. 88 Fuhua, First Road, Central Tower, Room 801 Shenzhen 518048 P. R. CHINA Tel.: +86 (755) 8203 2300 Fax: +86 (755) 8203 2306 Futian District Beijing New Century Hotel Office Tower, Room 1754-1755 Beijing 100044 P. R. CHINA Tel.: +86 (10) 6849 2738 Fax: +86 (10) 6849 2662 No. 6 Southern Capital Gym Road APPNOTE182.1AN 2008 EXFO Electro-Optical Engineering Inc. All rights reserved. Printed in Canada 08/04