Measure DSL power spectral density

You can simulate thousands of feet of wire to measure a transmitted signal's PSD.

By Dan Joffe, Adtran, Huntsville, AL -- Test & Measurement World, 12/22/2010 12:00:00 AM

Have a test idea you'd like to share? Publish it here and receive $150. Send us your ideas. Also: Read other Test Ideas Read Design Ideas published by EDN

A  DSL (digital subscriber loop) transmitter's PSD (power-spectral density) is its signature in a network. Industry standards specify the allowed power in dBm/Hz in different parts of the transmit-frequency spectrum. Fortunately, you don't need thousands of feet of wire to measure a transmitted signal's PSD. You can measure PSD with a spectrum analyzer such as the Agilent Technologies 4395A (which covers the DSL spectral range of 25 kHz to 30 MHz), and you can use a resistor network that simulates a DSL line.

DSL systems simultaneously transmit data from both ends because they have a DSL modem port at the customer premises and a DSLAM (digital subscriber loop access multiplexer) port at the telco's CO (central office). When you need to measure the PSD of the near-end transmitter, the signal will also contain an interfering signal from the far-end transmitter that interferes with near-end measurements.

See a plot of a typical PSD measurement from a spectrum analyzer.

You can build the resistor networks in Figure 1 or Figure 2 on a small circuit board. The networks simulate a local loop, and they attenuate interference from a far-end reflection, which provides more accurate PSD measurements of the transmitter in the DSL modem under test. Using a small circuit board makes the networks suitable for a production test or bench test.

To measure a transmitter's PSD, connect a spectrum analyzer to the end of the resistor network near the transmitter. The resistor network shows a 100-Ω impedance to the DSL modem's transmitter under test, a load that's similar to telco cables in the DSL frequency bands.

FIGURE 1.  A 10-dB insertion loss between the ends decreases the side-to-side interference in PSD measurements by 10 dB.

Figure 1 shows a basic approach. Assume that the modem under test is one designed for customer premises and another modem is designed for a telco's CO. Think of the customer premises as the near end in this example. Each end of the circuit has one port for a modem and one port for measurements.

In this example, the measurement port at the CO has a 100-Ω resistor to provide the proper load. Thus, the circuit terminates all four ports with 100 Ω because most DSL transceivers have a nominal 100-Ω output impedance.

A spectrum analyzer has 50-Ω input impedance. To get 100 Ω, you need a 100-Ω-balanced-to-50-Ω-unbalanced single-ended transformer, such as the North Hills 0300BB. The resistor network has 10 dB of insertion loss between the two DSL modems. At the measurement port, the circuit attenuates the near-end signal by 20 dB and the far end signal by 30 dB.

When measuring PSD, the spectrum analyzer responds to the sum of the power from both ends of the circuit, but with differing attenuation. Assume that a modem produces 1 mW at a given frequency. If you measure at the customer premises, the total power will be the sum of:

1 mW × 0.01 = 0.01 mW from the customer end (less 20 dB)
and
1 mW × 0.001 = 0.001 mW from the CO end (less 30 dB)
Total measured power would therefore be:
0.01 + 0.001 = 0.011 mW

This result is 10 × log(0.011/0.01) = 0.414 dB more than if there were no DSL modem at the CO end of the line. So, the interference from the CO modem creates a 0.41-dB error in measurements at the customer premises.

Figure 2. Adding a few components adds at least 40 dB rejection of side-to-side interference in PSD measurements.

The network in Figure 2 uses more components, but it provides greater attenuation of interfering signals from the far end. This circuit terminates all ports in the same manner as the circuit in Figure 1. The additional resistors cancel any component of the far-end signal that appears at the near-end measuring port. If the DSL equipment's output impedance is very nearly 100 Ω, then the interfering transmitter signal can be attenuated by at least another 40 dB at the near-end measuring port. The cross-coupled resistors R13-R16 add signals to the measurement ports that cancel the far-end interference.

The additional cancellation helps the noise floor of the measurement by saving dynamic range for the spectrum analyzer. In addition, it preserves the measurement noise floor in the case of overlapping PSDs. This greatly increases measurement accuracy. Figure 2 includes the cancellation from the cross-coupled resistors, so this network results in far less interference than the network in Figure 1. See a plot of a typical PSD measurement from a spectrum analyzer.