Squeezing Quarts Into Pint POTS

xDSL is a generic acronym used to describe any of a growing number of high-capacity data transfer schemes. ISDN was the first xDSL service and has done relatively well in Europe, although not so well in the United States. First envisioned to carry simultaneous dial-up switched-circuit data and voice or double-rate data, ISDN at its best only offers 128 kb/s—two 64-kb/s B channels that can be bonded at extra cost.

In contrast to the maximum 9.6-kb/s analog modem rate available when ISDN was introduced, 128 kb/s was an impressive improvement. It’s not such a big advantage today compared to V.90 modems that achieve 56 kb/s over a single twisted pair, although the modems are not capable of simultaneous POTS. At least ISDN provides a separate D channel for signaling. This means that its 64-kb/s B channels aren’t reduced to the 56 kb/s of V.90 caused by bit stealing to support in-band signaling.

Unfortunately, the several years of conservative rollout and comparative lack of demand for ISDN occurred during the time that internet usage was growing rapidly. Rather than treat data I/O as equivalent to a single-destination phone call, the model on which ISDN is based, internet users demand total interconnect flexibility. As a result, the fixed-circuit switched telephone network suits internet data transfer less well than packet-switched virtual networks. Internet users also don’t conform to the accepted norm for telephone network usage because typical surfing sessions are much longer than traditional phone calls.

The accelerating growth of internet usage has prompted timely rollout of xDSL services worldwide. However, although xDSL adoption will grow rapidly, estimates vary. Dataquest reported there will be 5.8 million xDSL lines globally by the end of 1999. On the same global basis, Communication Industry Researchers expect 5,152,200 xDSL lines for business use by 2006. Finally, Ovum predicts there will be 19 million xDSL lines worldwide by 2003.1

Regardless of the exact figures,


the internet is proving to be the killer application that LECs had earlier hoped high-speed video applications such as VOD and video conferencing would be.

Variations on a Theme

VDSL, focused on VOD with up to 52-Mb/s downstream rates, has a reach of 1,000 ft. For longer distances, the rate drops. Consequently, VDSL is limited to being part of FTTC installations or serving the immediate neighborhood of the CO. Either option is expensive and offers much greater speed than typical internet use requires.

Like some other types of xDSL, VDSL is asymmetrical because larger files usually are downloaded to the user than are uploaded to the server. The upstream rate for a 1,000-ft reach is 1.5 to 2.3 Mb/s.

There are symmetrical versions of xDSL that have equal upstream and downstream rates as does ISDN. Also like ISDN, HDSL uses two pairs but provides 1.5-Mb/s DS1 rates in each direction. HDSL does not support simultaneous voice services. SDSL uses only a single pair and allows voice services as well as data but is limited to a 10,000-ft reach at rates up to E1 (2.048 Mb/s).

Recently, lower-cost forms of xDSL have appeared. Because whole-scale replacement of the entire PSTN CO-to-user connection cannot be justified economically, all these schemes use the existing copper subscriber loop wiring—the so-called last mile. Actually, the last mile can be two or three miles.

Standard loop loss tests specify 12,000-ft and 18,000-ft distances with or without specific impairments such as bridged taps and splices. Typically, the feeder and distribution cables from the CO are not cut when a connection is made to a local subscriber. Instead, the subscriber’s drop wire is bridged to an unused pair in the distribution cable within a neighborhood which, in turn, is bridged to an unused pair in the main feeder cable.

But when service to the subscriber is discontinued, the bridged taps are not necessarily removed. The circuit simply is disconnected at the subscriber end. As a result, stubs of various lengths are formed, presenting few problems for POTS but wreaking havoc with the response of the overall loop pair to the high frequencies used by xDSL.

To extend the reach of the lossy copper pairs, POTS may use loading coils to approximate a lumped constant transmission line. Inductors are added in series with the line to counteract the line’s natural shunt capacitance. Voice frequency response is enhanced for distances greater than 18,000 ft, but high- frequency response is destroyed.

Most forms of xDSL have to work with some bridged taps because removing all the taps is impractical. On the other hand, loading coils present such a severe constraint to high-frequency working that they have to be removed before high data rates can be achieved.

T1.413 ADSL with DMT modulation occupies up to 256 4-kHz wide frequency bands (or bins) from about 25 kHz to 1.1 MHz. Each bin can be modulated up to 15 or 16 bits/s/Hz or 60 to 64 kb/s per tone, depending on the implementation. Said another way, DMT modulation has an efficiency of up to 15 or 16 bits per symbol.

Because DMT is adaptive, lower-frequency bins with better SNRs will be modulated more heavily and higher-frequency bands less heavily if impairments are encountered. This means that the highest specified downstream rate of 8 MHz will be exceeded on a short, ideal loop, or reduced on a longer loop with impairments. Conventional POTS is confined to the DC-to-4-kHz frequency band. The remaining 21 kHz to the bottom of the first bidirectional ADSL bin is used to separate POTS and ADSL signals.

Adaption also compensates for localized interference such as from AM stations that coincides with only certain bands. Modulation may be reduced to zero in these bins. In an ADSL system, the overall data rate can be adjusted in 32-kb/s increments.

Using the existing copper loop for both data and POTS maintains the integrity of the existing telephone service, but depending upon the type of xDSL, additional hardware may be needed. T1.413 ADSL requires a splitter at each end of the loop to separate/combine POTS and digital data signals. Also, echo cancellation may be required because some of the lower-numbered bins can be used for both upstream and downstream data.

ADSL uses an HP filter in series with the modem itself and a LP filter in series with the local phone wiring. The HP filter protects the modem from POTS signals, including large transients from the ringing signal. The LP filter reduces HF modem signals on the local phone system that could disturb voice messages. Also, the LP filter minimizes changes in line impedance caused by phones going on- or off-hook. This means that less time is spent retraining—determining new line-equalization factors—each time the impedance changes.

G.Lite is a splitterless version of the T1.413 DMT standard that uses only the lowest-frequency 128 bins of the 256 ADSL bins, reducing the transmitted frequency range. Its 1.5-Mb/s downstream speed is far below the highest ADSL rate, but G.Lite should be less susceptible to line impairments because of its lower speed.

On the other hand, although splitterless xDSL implementations add the HP filter to the modem circuit board, they eliminate the LP filter. This means that the local POTS hardware receives the full-strength modem signal. To reduce the effect of the modem HF signals on POTS, the modem power is reduced. The modem upstream data rate also will be reduced, but not necessarily the downstream rate.

Test

xDSL testing begins with in- and out-of-service performance tests. An in-service indication that a problem has developed is an increasing rate of word, parity, or cyclic redundancy checksum errors within a frame. BERT determines the quality of service at a high level but is a good clue to what may be wrong below the data-link layer.

Fadi Daou, senior staff technologist at GenRad and chair of the ADSL Forum Test and Interoperability working group from 1996 to 1998, said, “Performance testing is done on closed systems, requiring both the ATU-C and the ATU-R. The tester emulates the various test conditions on the loop and environment while conducting a BERT or cell error rate test.

“This includes emulating various loop lengths and injecting intrusive noise and disturbances as crosstalk elements in NEXT and FEXT,” he continued. “You also must include various PSTN conditions such as ring, ring-trip, and off-hook tests with various loads. These tests are particularly critical for splitterless implementations of ADSL modems.”

Physical-layer testing is largely a matter of determining the performance impairments of the copper loop on which xDSL has been deployed. A line could have deteriorated since its initial evaluation, or additional POTS devices could have been added with a deleterious effect.

POTS uses differential signals carried by copper twisted pairs. Unfortunately, most signal generators, spectrum analyzers, and oscilloscopes have single-ended 50-W inputs and outputs. A good way to interface these incompatible circuit types is to use a balun.

A balun also can convert impedance, making it easy to drive the POTS 600-W system impedance from a 50-W output-impedance generator. A separate balun is required to drive the system impedance of the copper loop in the higher- frequency band used by xDSL. For example, ADSL has a 100-W system impedance. If you are using a dedicated TIMS, the source and receive ports already are differential and present the correct impedance. The list of tests includes the following:

·


Attenuation.

·


NEXT and FEXT crosstalk.

·


Longitudinal conversion loss.

·


Characteristic impedance.

·


Spectrum.

Attenuation varies with distance as shown in Figure 1. Increasing the reach from 12,000 ft to 18,000 ft represents an additional 40-dB loss at the highest ADSL frequencies. The corresponding additional loss for G.Lite operating up to 400 kHz is only about 20 dB. The figure also shows the very large 120 dB dynamic range of the signals xDSL modems deal with.

Because the loop exhibits such large attenuation, a modem must have high gain at high frequencies. This means that noise and crosstalk can become limiting factors if they are similar in amplitude to the desired signal. NEXT testing determines the effect of other nearby copper pairs carrying large signals.

When using two baluns to simulate a NEXT condition they should be shielded from each other by a metal plate. Otherwise, the coupling between the baluns could be larger than that of the two copper pairs. Because FEXT tests usually are run on installed cabling, and because the U-C end of the loop is inaccessible to the remote operator, two test sets are required.


2

LCL is a measure of how well balanced the two wires are in the copper pair. For example, consider the description presented by Consultronics of the theory behind using copper pairs and the impairments that can arise:

“The two wires are purposely twisted around each other to ensure that the pickup is identical. Noise that is equally induced onto both conductors of a pair is referred to as ‘Common-Mode Noise.’ Input circuits of xDSL receivers (modems) are designed to reject most of this common-mode noise so it will not interfere with the differentially transmitted signal.

“In real cable pairs, unfortunately, the balance between the two conductors in a pair does not always remain equal. Resistive splices, water in the cable, uneven pair twisting, and the slow degradation of protectors and insulation over time lead to imbalance. This results in a translation of the common-mode noise into differential noise because of the imbalance caused by the anomaly. Under these conditions, xDSL receivers cannot differentiate between the signal and this type of noise…and data errors may occur.”


3

References

1. Harrison, J., et al, “ADSL: Prospects and Possibilities,” Center for Telecommunications Management, University of Southern California, 1998, p. 2.

2. “ADSL Copper Loop Measurements,” Hewlett-Packard, Product Note 4395-1, 1999.

3. “Longitudinal Balance Measurements for xDSL,” Consultronics Auto-TMS III, Application Note #10, 1998.

Additional Resources

Lane, J., Personal Broadband Services: DSL and ATM, Virata, 1998.

“G.Lite: Making the Internet Fast Enough for Consumers,” Aware, Inc., 1999.

 

xDSL Test Products

Flexible ATE Platform

The GENEVA Model 4000 Functional and Parametric Tester uses an extensive library of voice- and xDSL-frequency tests to determine the integrity of POTS splitters and verify transmission performance in both the POTS and ADSL bands. For ADSL, return loss, longitudinal conversion loss, longitudinal output voltage, power spectral density, and total power measurements are made. For POTS, ringing and DC loop characteristics, group delay, and channel noise and distortion are evaluated. Multiple channels can be tested concurrently to improve throughput. Starts at $74,000. GenRad, (978) 589-7000.

Portable ADSL Tester

The CoLT-250 Hand-Held ADSL Test Set supplies upstream and downstream parameters as negotiated with the central office ATU, and can be used during installation or maintenance of ADSL circuits. The CoLT-250 features single-button ADSL analysis, the capability to perform tests at various points between the local exchange and the customer, and a lightweight design. The tester can quickly confirm the quality of service delivered to the customer. Under $1,500. Consultronics, (905) 738-3741.

Line Tester Upgrade

The Wide-Band Test Pack (WTP) can be deployed in the field as an upgrade to the manufacturer’s existing test systems. The WTP option extends the copper-pair testing capabilities of Models 107A/F and 105A remote test units (RTUs) into the higher frequency ranges used by broadband services such as ISDN, ADSL, SDSL, and HDSL. The enhanced test systems measure high- frequency noise, send and receive high-frequency tones, and locate bridged taps on a copper pair. The Bandwidth Analysis Tool is a software application. It analyzes RTU and WTP data to determine the fastest bit rate and type of xDSL service for a particular loop. Call company for price. Harris, (800) 442-7747.

Hand-Held Test Set

The ADSL Installer™ is designed for use by field technicians during the troubleshooting, service qualification, and turn-up of an ADSL facility. The tester provides physical layer ADSL testing and ATU-R emulation conforming to the ANSI T1.413 specification. Displayed information includes upstream- and downstream connect speed, noise margin, attenuation, output power, attainable bit rates, and DSLAM configuration data. One-button testing, a 20-character × 16-line LCD, and an intuitive softkey menu system are featured. Optional graphing and expert analysis software are available. Call company for price. Nitech, (732) 553-1650.

Modular DSL Tester

The hand-held E8560A DSL Service Installer provides one-button ADSL loop testing. Functions include load coil detection; DC voltage and current measurements; tip-to-ring, tip-to-ground, and ring-to-ground resistance; and generation of 404-Hz, 1,004-Hz, and 2,804-Hz POTS tones and 196-kHz, 392-kHz, and 1.5-MHz DSL tones. E8571A or E8551A modules adapt the instrument for DMT or CAP transmission protocols; other modules will deal with future DSL variants. A weather-resistant membrane front panel, a backlit LCD, test-result storage, and an RS-232 port also are featured. E8560A base unit: $1,995; E8551A CAP ADSL test module: $1,995; E8571A DMT ADSL test module: $1,995. Agilent Technologies, (800) 452-4844, ext. 6834.

Modem Test Set

The Veritas 2000 DSL Test System performs functional testing without requiring actual DSLAM equipment. Two standard-compliant CO ports are provided for testing and analyzing remote-terminal ADSL modems and chipsets. Full-rate ADSL ANSI T1.413, G.dmt ITU G.992.1, and G.lite ITU G.992.2 ADSL standards are supported. Based on the Analog Devices 20msp918 chipset, the product interfaces via V.35 or ATM25, can be configured to work with other ADSL test equipment such as the Consultronics DSL 400, and supports future technological change via its modular architecture. Call company for price. Aware, (781) 276-4000.

 

Glossary of Terms

ADSL asymmetric digital subscriber line

ATU-C ADSL modem, CO end

ATU-R ADSL modem, remote end

B ISDN bearer channel

BALUN balanced-to-unbalanced transformer

BERT bit error rate test

CO central office

D ISDN signaling and control channel

DMT discrete multitone

DSLAM digital subscriber line access multiplexer

FEXT far-end crosstalk

FTTC fiber to the curb

HDSL high-speed DSL

HP high pass

ISDN integrated services digital network

LCL longitudinal conversion loss

LEC local exchange carrier

LP low pass

NEXT near-end crosstalk

POTS plain old telephone service

PSTN public switched telephone network

SDSL single line OR symmetric DSL

SNR signal-to-noise ratio

TIMS transmission impairment test set

U-C loop interface—central-office end

U-R loop interface—remote-terminal end

VDSL very high bit rate DSL

VOD video on demand


Copyright 2000 Nelson Publishing Inc.

January 2000


Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!