Measure power in the next channel
Wideband CDMA (W-CDMA) and digital television (DTV) transmissions produce relatively flat power spectra across their channels. If the signals on adjacent channels interfere with each other, they will affect the quality of cell-phone and (eventually) TV signals—something cell-phone subscribers and TV viewers won’t tolerate. To test a cell phone, base station, or DTV transmitter, you need to measure the amount of power that leaks from one channel to adjacent channels.
Transmitters use filters to limit the power that leaks into adjacent channels. You can measure a transmitter’s adjacent-channel power with a spectrum analyzer. Some spectrum analyzers have an adjacent-channel noise measurement feature that performs all the calculations you need for W-CDMA or DTV signals. If your instrument lacks that feature, you can still make the measurements, but you’ll have to calculate the instrument’s noise and compensate for it.
|Figure 1. A digitally modulated signal such as W-CDMA requires an adjacent-channel noise measurement at 65 MHz from the channel’s center frequency.|
W-CDMA transmissions must fit within a 4.096-MHz bandwidth (or 3.84 MHz depending on the system) where the center frequencies of adjacent channels sit 5 MHz apart. DTV uses 6-MHz channels because of its analog-TV legacy. For W-CDMA transmitters, you must measure a transmitter’s adjacent-channel power ±5 MHz from the channel’s center frequency (Figure 1). At these points, the adjacent channel power for some base-station amplifiers could be as high as 70 dB below a transmitter’s in-channel average power (Ref. 1). DTV signals should be 71 dB below the average signal power within the channel (Ref. 2).
Figure 1 shows a spectrum analyzer’s noise floor (the noise produced by the instrument) at –75 dB. If the adjacent-channel power you want to measure is in the instrument’s noise, you get intolerable errors because your signal’s power adds to the noise power produced by your spectrum analyzer.
If the signal power and instrument’s noise power are equal, then the power you measure will be 3 dB higher (double the power level) than the transmitter’s adjacent-channel power. If you increase the transmitter’s power by 1 dB, you’ll increase its adjacent-channel power by 3 dB. You can increase the adjacent-channel power to the point where it dominates the noise and improves measurement error, though. If you increase adjacent-channel power by 10 dB, then the spectrum analyzer’s noise will contribute less than 0.5 dB of measurement error.
Unfortunately, if you boost the signal power too much, the spectrum analyzer’s amplifiers will compress the in-channel signal power and create second-order harmonic distortion (HD) and third-order intermodulation distortion (IMD). The distortion will raise the instrument’s noise floor, which can place you back in the same situation you had before you increased transmitter power: The spectrum analyzer’s noise will mask the transmitter’s adjacent-channel power you’re trying to measure. The range of signal power that lets you measure adjacent-channel power depends on the spectrum analyzer’s phase noise, thermal noise, and distortion products.
Phase noise is wideband noise caused by jitter in the spectrum analyzer’s local oscillators. Phase noise power is constant regardless of frequency or a spectrum analyzer’s input-signal power. Thermal noise, inherent in amplifiers and other analog circuits, decreases (relative to the input) as the power level at the spectrum analyzer’s mixer rises. That’s good, but the sum of HD and IMD power increases as the spectrum analyzer’s input power increases. If you add the noise components as a function of the instrument’s mixer input power, you get a curve.
|Figure 2. The point of lowest total noise determines the optimum power at the spectrum analyzer’s mixer. Courtesy of Tektronix .|
The low point in that curve represents the spectrum analyzer’s minimum noise level relative to signal input power. That point is your best noise level for measuring adjacent channel power. Figure 2 shows the components of the instrument’s noise floor. The line marked “spectral regrowth” represents the sum of HD and IMD products.
Crunch the numbers
You can calculate the level of each noise component based on an instrument’s specifications. If a data sheet lists phase noise, the value will cover a bandwidth such as 1 Hz, 10 Hz, or 1 kHz. You’ll have to scale that value to match the channel bandwidth of your signal, using the phase noise value and the offset used for the adjacent channel measurement (i.e., 5 MHz or 6 MHz).
Assume a data sheet specifies phase noise over a 10-Hz bandwidth. In the W-CDMA example, you need to multiply the specified phase noise by 5 MHz/10 Hz, or 500,000, take the logarithm of that result, and add the 10-Hz phase noise. If a data sheet doesn’t list phase noise, you can connect a high-quality, low-noise reference oscillator to the spectrum analyzer and measure the noise.
To calculate thermal noise, you need to know your instrument’s displayed average noise level (DANL). Some data sheets simply specify this value as average noise level. A data sheet may specify more than one DANL value based on frequency, so pick the one best suited to your signal’s carrier.
You may have to calculate DANL for your frequency by plugging your frequency into an equation found on the data sheet. Because instrument makers specify DANL over a resolution bandwidth, you’ll also have to scale the DANL value up to your signal’s channel bandwidth. The equation is similar to that for scaling phase noise. (See Measuring ACPR of W-CDMA Signals with a Spectrum Analyzer [Ref. 1] for the scaling equations.)
Finally, you must calculate the power in the spectrum analyzer’s spectral regrowth. To do this, you must find your instrument’s second-order intercept (SOI) and third-order intercept (TOI) power levels. The TOI is more important because third-order IMD increases 3 dB for each 1-dB increase in input signal power. Second-order HD increases at a 2:1 ratio relative to an increase in the spectrum analyzer’s input power. You can plot HD and IMD from information on a spectrum analyzer’s data sheet. A manufacturer might, for example, specify IMD at –80 dBc for an input power of –30 dBm. Because you know the HD and IMD slopes of output power vs. input power, you can calculate them and add them to get spectral regrowth. If you need a refresher on intercepts, or if you need to calculate TOI, download Intermodulation Distortion (IMD) Measurements (Ref. 3).
Once you calculate the three noise components, plot and add them. If you plot the total, you should see the point of lowest noise. Set your signal’s input power so the power at the spectrum analyzer’s mixer, after accounting for any attenuators in your test setup, is about 10 dB below the spectrum analyzer’s 1-dB compression point.
If your instrument’s dynamic range won’t let you set the signal power that high, then use the information in Ref. 1 or Ref. 4 to calculate any additional error that will occur as the adjacent channel power gets closer to the noise floor. T&MW
1. Measuring ACPR of W-CDMA Signals with a Spectrum Analyzer, Application Note, Tektronix, Beaverton, OR, 1998. www.tek.com/Measurement/App_Notes/ACPR/2DA_12521_0.pdf.
2. Rowe, Martin, “Measure DTV Signals from Start to Finish,”Test & Measurement World , April 2000. p. 61.
3. Intermodulation Distortion (IMD) Measurements, Application Note, Anritsu, Morgan Hill, CA, September 2000. www.us.anritsu.com/downloads/files/11410-00257a.pdf.
4. Optimizing RF and Microwave Spectrum Analyzer Dynamic Range, Application Note 1315,
Agilent Technologies, Santa Clara, CA. literature.agilent.com/litweb/pdf/5968-4545E.pdf.
Martin Rowehas a BSEE from Worcester Polytechnic Institute and an MBA from Bentley College. Before joining T&MW in 1992, he worked for 12 years as a design engineer for manufacturers of semiconductor process equipment and as an applications engineer for manufacturers of measurement and control equipment. E-mail: email@example.com.