June 18, 1998

Prove that ADSL modems have the right stuff to vanquish interference

Peter Ziemann, Wandel and Goltermann

To transmit data over ordinary copper wires at 9 Mbps, ADSL modems must meet exacting requirements. Verification of the modems' analog-interface performance requires careful measurements. Learn here how to make these measurements and what test equipment you need.

Asymmetric-digital-subscriber-line (ADSL) technology for telephone and data-communications subscriber access is likely to experience enormous growth over the coming years, as new applications develop. With one ADSL modem in the telephone central office and another one at the subscriber site, you can achieve data rates of 9 Mbps (downstream) and 640 kbps (upstream) over a single copper-wire pair. These data rates represent a major increase in speed over conventional services available via normal telephone lines. With ADSL, the telephone infrastructure becomes a gateway to the multimedia world.

ADSL uses a discrete-multitone (DMT) analog-transmission technique (Figure 1). DMT enables highly effective bandwidth usage with a uniform power-density spectrum. DMT resembles the older frequency-division-multiplex (FDM) technique in that it divides the frequency band into 255 subchannels, with each channel consisting of a quadrature-amplitude-modulated (QAM) carrier signal. Dynamic data distribution among subchannels is new, however. This technology aims to reduce interference-induced transmission errors.

Plain-old telephone service (POTS) is also part of this concept. Telephony service loops through POTS splitters within the modems and is transmitted at the normal baseband frequencies. In this way, subscribers can continue to use terminal equipment under ADSL. With POTS splitters, conventional telephone service remains available even if a power failure interrupts the modem- or digital-interface power supply.

Simultaneous transmission of telephony services and DMT-modulated data streams over the same wires requires excellent selectivity in the POTS splitters, good dielectric strength, and minimal interference between data and speech. ANSI stipulates the relevant requirements in its T1.413-1995 norm. The following material describes the main measurements on ADSL modems' analog interfaces, POTS, public switched-telephone network (PSTN), loop interface­central-office end (U-C), and loop interface-remote­terminal end (U-R). (Telecommunications terminology uses many baffling and inscrutable acronyms. Table 1 explains the ones used here). The test approaches comply with ANSI norm T1.413 (Chapter 10, "Electrical Characteristics").

Voiceband characteristics

Insertion loss and attenuation distortion: Loss is the most important parameter in characterizing signal transmission (Figure 2); it is a reduction in signal amplitude that limits the length of the transmission path. By measuring insertion loss and attenuation distortion, you can make sure that the POTS splitters do not impair transmission quality in the speech band. Because the transmitting and receiving ends use different terminating impedances (600 and 900ohm), you must take into account a compensation factor of 1.77 dB when you measure the absolute loss figures.

Noise and distortion: The signal-to-C-notched-noise ratio and the second- and third-order harmonic-distortion products measure the subscriber's perception of transmission quality (Figure 3). To simulate how humans perceive sound, the ANSI standard uses a C-message filter that appropriately weights the measured noise frequencies. (Type-C is the North American standard for the frequency response of a direct-distance-dialed, voice-grade telephone connection.) A narrowband notch filter suppresses the 1004-Hz test tone or holding tone. The SNR is based on the difference in measured values with the notch filter switched on and off. To determine the harmonic distortion, use a narrow resolution bandwidth (for example, a resolution bandwidth of 25  Hz) to measure the level differences at multiples of the holding-tone frequency.

Longitudinal output voltage: Achieving good  interference immunity in signals transmitted over copper-wire pairs requires good balance on the line and also on the modem interfaces. Good balance reduces both the susceptibility to interference and the amount of emitted interference; that is, the line's influence on neighboring copper pairs. If you know the output balance, you can characterize the behavior of a source by generating a signal and measuring the undesired common-mode voltage (longitudinal voltage to earth).

To make this measurement, you need a simple passive network with sufficient intrinsic balance. Balance allows you to avoid corrupting the measured result. During the measurement, switch the modem to transmit mode. The test setup (Figure 4) shows the network that Bell Atlantic recommends. Contrary to the ANSI norm, this network uses the correct terminating impedance of 100ohms and provides a measurement bandwidth of 3 kHz. The additional capacitance of 0.15 µF provides more pronounced weighting at low frequencies. Based on the measurement bandwidth, the receiver's lower tuning frequency is only about 2 kHz.

Longitudinal balance (POTS/PSTN): POTS and PSTN interfaces have demanding balance requirements. The test bridge should measure at system impedances of 600 and 900ohms to match the impedances of the POTS or PSTN interfaces (Figure 5). For increased accuracy, normalize your measurements at 1 kHz. To normalize, replace the device under test with a reference element whose longitudinal conversion loss is 0, thereby producing a total unbalance. Use the measured receiving level as a reference in the subsequent measurements.

ADSL-band characteristics

Return loss: When you measure the return loss at the U-C and U-R reference points, you detect deviations from the required system impedance of 100V. You must use a suitable return-loss or impedance bridge to provide the system impedance over the required frequency range. Normalizing the test setup compensates for the bridge's insertion loss. Disconnect the device under test to produce total reflection, and use the measured receiving level as a reference value for the subsequent measurements.

Longitudinal balance (U-C/U-R): You must check the balance figures on the transmission-line interfaces over the entire system bandwidth. You should make the measurement with a high-frequency LCL bridge that is designed for measurements on systems whose Z=100V (Figure 6). You must compensate for the insertion loss by normalizing for LCL=0 at, for example, 1 MHz.

ADSL noise interference into the POTS circuit

Steady-state noise: During the idle-channel noise measurement, simulate a transmission path and determine the absolute speech-band noise power on the POTS interface. To simulate human hearing, use the C-message filter to weight the result in the receiver. Do not exceed the noise-power limit in the ADSL band, even during simultaneous data transmission. In addition to limiting the idle-channel noise, ANSI T1.413 limits noise power at any frequency below 15 kHz. To measure at these low frequencies, you need a narrow resolution bandwidth (for example, 25 Hz).

Impulse noise: The impulse-noise measurement counts the number of noise impulses that exceed a specified threshold of 47 or 65 dBrnC0 (dBrnC, referred to a zero transmission level point) (Table 2). According to ITU-T 0.71, this measurement is typical of transmission-impairment for FDM systems. Because this measurement takes longer than 15 minutes, it is practical only for type-approval and developmental testing. Test setups for the noise measurements are identical to the setups for noise and distortion (Figure 3).

Although you can make all of the measurements with general-purpose test equipment, test sets  designed for ADSL-modem tests are more efficient. Table 3 summarizes the equipment requirements.

References

1. ANSI T1.413-1995, "Network and Customer Installation Interfaces--Asymmetric Digital Subscriber Line (ADSL) Metallic Interface," American National Standards Institute.

2. Application Note 53, "Testing ADSL Modems," Wandel and Goltermann Inc, 1997.

| EDN Access | Feedback | Table of Contents |