With frequencies soaring to new heights, wireless and RF testing gets pushed to the limit in terms of complexity and cost. In fact, higher frequencies in the gigahertz range are now commonplace. Simple AM and FM/PM have disappeared, replaced by the more complex digital modulation methods.
Binary phase-shift keying (BPSK), quadrature phaseshift keying (QPSK), and quadrature amplitude modulation (QAM) are the norm these days. And, cell phones extensively exploit spread-spectrum (CDMA) technology. Meanwhile, other newer wireless methods have switched to orthogonal frequency-division multiplexing (OFDM).
Specialized wireless techniques like software-defined radio (SDR) and cognitive radio (CR), time-multiplexed protocols, bursty transmissions like radar, frequency hopping, wide-bandwidth technologies like Ultra-Wideband (UWB), and adaptive modulation further complicate the testing process. It isn't a pretty picture for the designer or the test engineer.
But that's not all. Testing speed is more important than ever before. Time-to-market still rules the roost in engineering today, and testing adds no value. It's just a cost borne to ensure that the product works and conforms to the guidelines in play. The more time spent in manufacturing testing, the greater the cost and the lower the margin.
That's a grim scenario in a high-volume commodity market like cell phones. With over 1 billion new phones produced this year alone, just think of the hours that go into testing. One manufacturer indicates that cutting the test time for one measurement by 10 ms can save $1 million in a production run.
On that front, though, there's some really good news. Test equipment manufacturers, always on the leading edge of technology anyway, recognized the problem and produced some excellent solutions that simplify and significantly accelerate testing procedures - at a price. Yet that price is a good tradeoff because time is, after all, still money.
When planning your wireless testing, make sure the standards of the technology you're using spell out the key parameters of what you want to measure. Whether the standard comes from an international standards organization or an industry alliance that certifies products, you must acquire that standard documentation and become aware of all its gruesome details. Here, you'll find the specific tests that need to be made as well as the required equipment.
Keep two facts in mind. First, RF measurements are measurements of power, not voltage. Meters and displays readout in power directly or, in some instances, in dBm (dB referenced to 1 mW). Table 1 shows the relationship between actual power and dBm. Since the goal in all cases is maximum power transfer, proper impedance matching within your circuits and between the test instruments and the device under test (DUT) is critical. Most RF measurements are made with a 50-Ω characteristic impedance.
Second, everything is a transmission line. If it isn't a coax cable, it's a strip line or microstrip whose impedance is crucial. Again, 50 Ω is the standard, and all impedances should match up for maximum power transfer. Also, impedances should match to minimize reflections and high voltage standing-wave ratio (VSWR), which can lead to inefficiency and circuit damage.
Generally, RF tests divide into two categories: those for transmitters (TX) and those for receivers (RX). Many other special tests exist, and beyond the ones listed below, companies consistently develop new tests to add to the mix (see "Six New Measurements You're Going To Need" at www.electronicdesign.com, Drill Deeper 17102).
The most important test is power output from the final power amplifier (PA). You can get a good measurement by using a spectrum analyzer or vector signal analyzer, though in most cases, greater accuracy of measurement is essential. This requires an RF power meter. It will give you the accuracy needed to ensure compliance with whatever standard or regulation you must meet.
The two common power measurements are average and peak. Your needs will be determined by the type of modulation you're using. A further complication is the requirement of a gated or timed power measurement in some applications. For example, the GSM cell-phone standard that uses TDMA requires you to measure the power in a burst of RF during the 524.6-µs time slot allotted. Another example of a pulsed RF application is radar, which has very narrow pulses and random and sometimes coded formats.
With CDMA, you will measure average power because the signal is similar to random or white noise. In a CDMA PA that must handle multiple signals concurrently, the signals (though random) can add together and produce higher power peaks 10 to 30 times that of one signal. A key measurement in such amplifiers is the crest factor, or the peak-to-average ratio, which may be a power or voltage ratio. Some RF power meters will measure and calculate the crest factor.
Another key measurement is the PA's 1-dB compression point. As the input power to a PA increases, the output increases linearly, up to a point. At some power level, the output will saturate, meaning the output power will max out and remain essentially constant despite increasing input power (Fig. 1). The 1-dB compression point is the point where the output power becomes 1 dB lower than what it should be on a linear output scale.
Of course, driving an amplifier into saturation stresses it. Even worse, the nonlinear response will produce harmonics and spurious signals resulting from intermodulation-distortion (IMD) effects. You can measure the harmonics and spurious signals with a spectrum analyzer.
Continued on Page 2
Third-Order Intercept (TOI):
IMD, another common test, measures the amount of nonlinearity in an amplifier. Two test signals are applied to the amplifier, and the output is measured. The two fundamental frequencies (f1 and f2) mix, producing the sum and difference signals as well as higher-level products. The sum and difference signals are usually easy to filter out, as are the second-order products. But the so-called third-order products, which are 2f1 - f2 and 2f2 - f1, are difficult to filter out because they fall so close to the two original signals (Fig. 2).
These third-order products are measured by determining the TOI. Also known as IP3 or IM3, the test indirectly measures the magnitude of the TOI. In the plot of output power versus input power as given in Figure 1, the slope of the primary curve is one. The slope of the curve for the TOI products is three, due to the mathematical nature of the trigonometric expression defining the nonlinearity.
Note that the curve intersects the main linear plot beyond the point where the amplifier goes into compression. That's why you can't measure TOI directly. The greater the difference between the linear plot and the TOI, the less the distortion and the lower the intermodulation products. TOI tests are also used with receivers.
Error Vector Magnitude (EVM):
EVM is a measure of modulation quality. It indicates how closely the transmitted signal is to an ideal version of the signal. Since most modulation methods use digital techniques (BPSK, QPSK, QAM, 8PSK, etc.) that put the signal into an inphase (I) and quadrature (Q) format, the output can be shown as a constellation diagram (Fig. 3).
Each point on the diagram represents one symbol of the output representing two or more bits. The actual phasor position may not be in the ideal position because of variances in the transmitter I and Q modulators. Common problems are I and Q signal-amplitude differences, phase error in the 90° shift between the local oscillator signals, and local oscillator frequency shift.
EVM is usually expressed as the ratio of the length of the error vector to the length of an ideal reference vector that's usually outermost symbol magnitude expressed as a percentage:
EVM = (error vector length/reference-outermost vector length) Ω 100
Adjacent Channel Power Ratio
ACPR is the ratio of the average modulated transmitter power to the power in an adjacent frequency channel. It's measured by passing the transmitter signal through a receiver filter set to the adjacent RF channel frequency. Sometimes called the adjacent channel leakage ratio (ACLR), it measures how much signal power leaks over into an adjacent channel.
Most often, ACPR is used in CDMA equipment. The signal usually is downconverted to an IF, digitized, and subjected to a fast Fourier transform (FFT) that's then displayed in the frequency domain. The resulting plot will show how far down in dBm the adjacent channel power is from the main signal power.
This key receiver test is usually measured by first feeding the desired frequency's signal into the receiver front end. Then it's attenuated with the signal generator attenuator or an external attenuator until the signal "drops out." Some definition of "drop out" is usually provided, such as the point where the receiver loses lock. Noise also is introduced into the signal to determine the signal-to-noise ratio (S/N or SNR), where the signal is no longer readable.
One accepted way to determine sensitivity is to perform a bit-error-rate (BER) test on the receiver. A pseudo- random bit pattern is modulated onto the signal from a generator and fed to the receiver. The recovered bits are compared to the demodulated received bits to compute the bit-error ratio. The signal input amplitude is continually reduced or the noise level is increased until the desired BER is exceeded.
Continued on Page 3
Adjacent Channel Rejection:
This test uses one or more signal generators to produce the desired signal and one or more interfering signals. It measures the ability of the receiver to reject the interfering signal in an adjacent channel.
Test Instruments of Choice
Many specialized RF test instruments are available. But the key devices, and the ones most commonly used, include the arbitrary waveform generator (AWG), the signal generator, the vector signal generator, the spectrum analyzer, the vector signal analyzer (VSA), and the power meter (Figure 4 and Figure 5). These instruments are vital for making fast and accurate measurements.
The vector generator and analyzer are based on an SDR architecture that makes them an ideal fit for today's wireless standards, as well as for speeding up measurements. That's because the SDR architecture makes these instruments flexible - they can be quickly changed, updated, and improved with a software or firmware addition.
A programmable DSP and/or an FPGA or ASIC generate the modulation in the generator and perform the demodulation, downconversion, and decoding in the analyzer. A high-performance PC often is used for the DSP and is built into the instrument. Specialized software or firmware may be added to the generator or analyzer to set it up for measurements on a specific radio technology or wireless protocol (Table 2).
Though they aren't commonly used in RF testing, oscilloscopes do play a role in some applications. For example, Tektronix's DPO/DSA70000 oscilloscopes are an ideal platform for very wide-bandwidth RF signals like UWB. Tektronix's UWB software makes it possible to fully test popular WiMedia UWB radios and other broadband wireless devices (Fig. 6).
Most test setups will need the proper probes and cables. Always use the manufacturer's matching probes, and have the necessary coax cables with the right connectors. Other common accessories for most tests include signal combiners or splitters, fixed and/or adjustable attenuators, and isolators.
I want to personally thank Jeff Owen, Ken Voelker, and Ben Zarlingo of Agilent Technologies, Mark Elo of Keithley, David Hall and Hon Yee at National Instruments, and Darren McCarthy of Tektronix for providing valuable information and insight for this article.
For More Information
RF and wireless measurement is a vast and considerably deep subject. For further detail, see the Web sites of the equipment manufacturers. They all have extensive lists of data sheets, application notes, articles, white papers, and the like. Even the equipment brochures are highly informative.
For a good reference book, see Production Testing of RF and System-ona- Chip Devices for Wireless Communications by Keith Schaub and Joe Kelly (2004), published by Artech House (www.artechhouse.com).