RF Measurement Basics for
Non-RF Test Engineers

by Robert Green, Keithley Instruments

Wireless communications products that depend on RF principles are everywhere these days, and the growth rate is astonishing. From cell phones and wireless PDAs to Wi-Fi-enabled laptops, Bluetooth headsets, RFID tags, wireless medical devices, and ZigBee sensors, the RF device market is booming. This year alone, more than 850 million cellular phones will be manufactured and sold around the globe.

For thorough product testing and high test throughput, test engineers need to understand the basics of RF technology, what to test for, and which instruments are best suited for the job. Most engineers with experience in low-frequency applications below 1 MHz will not likely be familiar with high-frequency applications.

RF Terminology
Think in Terms of Power
RF signal strength can vary by vast amounts. As the signal propagates through space, the power per unit area decreases in proportion to the distance squared. Changes in power are measured in decibels (dB).

Using decibels for power measurements greatly simplifies calculations. Gains or losses in dB add or subtract. For instance, the multiplication operation reduces to simple addition. The formal definition of dB is:

dB = 10 log (P_out /P_in)

A dB value is a relative quantity. A related unit is dBm, which is the absolute power measured relative to 1 mW. Figure 1 shows dBm values and their corresponding values in watts. The power transmission range of a mobile phone is shown for reference as well as how low a signal a sensitive receiver can detect.

Figure 2 shows an equation defining the theoretical noise floor for RF signals at room temperature. Due to an RF signal’s lossy propagation through air as well as atmospheric interference and interference from other signals, the signal level that reaches the receiver can be quite low. It is not unusual for a receiver to detect signal levels below 0.1 pW.

Mismatch on Transmission Lines
At low frequency, the goal is to transfer voltages through circuits with minimal loss in magnitude. The most effective circuits have high input impedance and low output impedance.

With RF applications where a cable length can be a quarter wavelength, signals must be treated as waves. Any time a wave hits a discontinuity, some of the wave is reflected.

The goal of RF is to transfer all the power to the load without loss. Any reflection of power means not all of the power is getting to the load so mismatch is a critical parameter. Any difference in impedance between circuit elements and the transmission line causes reflections and loss of power.

In RF applications, transmission lines generally are coaxial cables external to circuit boards and microstrips within circuit boards. These components have a characteristic impedance. The expression for the characteristic impedance of a transmission line depends on the geometry of the conductors, the properties of the conductors, and the insulator holding or separating the conductors.

For RF applications, the characteristic impedance of the transmission lines and the input and output impedances of components are designed to be 50 or 75 Ω. A 50-Ω impedance is used to optimize power transfer in a system; 75-Ω systems are designed for minimum attenuation in applications such as cable systems. Most RF wireless transmission systems optimized for power transfer are 50-Ω characteristic impedance systems.

To minimize reflections, RF cables and components for wireless test and measurement applications are designed for 50 Ω. Conversely, the optimal power transfer takes place when impedances are matched.

A wave passing from one characteristic impedance to another causes reflection. If the impedances are the same, there is no reflection. In cases where there is a reflected wave due to an impedance discontinuity, there will be waves traveling in both directions on the transmission line.

At some point where the waves are in phase, a maximum voltage (Vmax) will occur and where the waves are 180 degrees out of phase (Vmin). The ratio of Vmax to Vmin is the voltage standing wave ratio (VSWR). This is one indication of how close a connector or a cable is to 50 Ω.

Figure 3 gives the formulas for determining the other measures of mismatch from 50 Ω. The reflection coefficient (ρ) is a direct indication of the percentage of the signal that is reflected at a discontinuity or a change in impedance such as a cable-to-instrument connector or antenna to low noise amplifier. The return loss is a measure of the attenuation to a reflected signal. A high return loss is desirable.

Figure 4 shows the relationship between the three parameters for the ideal case, a perfect match (no reflection), the ideal open circuit (100% reflection), and three values between the extremes. Test instrumentation typically has input or output VSWRs in the 1.2:1 to 1.6:1 range.

New Connectors, Cables, and Components
Cables with BNC connectors typically begin to degrade above 500 MHz. In the RF world, cables often are equipped with N connectors and SMA connectors. N connectors commonly are used on test instrumentation because they are rugged, can handle high powers, and perform well up to about 18 GHz. The SMA connector is much smaller and rated for lower power than the N connector, but it can be used well beyond 18 GHz.

All RF cables are coaxial. Coaxial RF cables can be inflexible or rigid, flexible for a limited number of bends, or flexible. Care of the cable is much more important for RF than low-frequency cables. Excessive bending of the cable and 90˚ bends can damage the cable and severely degrade performance.

At low frequencies, a good connection means that the conductors are in contact with each other. At RF frequencies, the importance of mismatch means that a good connection not only has the conductors in contact, but also that the connectors are properly torqued together. Manufacturers recommend about 7 ft-lb of torque to ensure good contact and minimal insertion loss between the connectors.

Maintaining the 50-Ω Line
Parallel connections or multiple signal paths in RF circuits are not as simple as in low-frequency circuits. Maintaining a matched circuit path to minimize discontinuities and signal reflections is critical.

RF switches are precision machined and designed to maintain 50-Ω impedance through the switch. To effect a parallel path, devices known as splitters or dividers separate an input signal path into two or more output paths, each with 50-Ω impedance. Combiners perform the opposite function by converting multiple input paths into a single output path.

These are just a few of the specialized components needed for RF test systems. If you are new to RF test, be prepared for sticker shock. RF components cost much more than their equivalent DC components.

What Do You Need?
As with the breadth of low-frequency test instruments, the world of RF test instruments is wide and varied, ranging from signal sources and power meters to spectrum and network analyzers. These instruments are used to generate RF signals and measure a wide range of signal parameters.

RF Power Meters
Power is the most frequently measured RF quantity. A power meter essentially measures the power of RF signals. It uses a broadband detector and reports absolute power usually in watts, dBm, or possibly dBμV. For the majority of power meters, the broadband detector or sensor is an RF Schottky diode or diode network that performs an RF-to-DC conversion.

Power meters provide the best accuracy of any RF instrument for measuring power. High-end power meters often requiring an external power sensor can measure with 0.1-dB or better accuracy. Power meters can operate down to near -70 dBm. Sensors range from high-power models to high-frequency models to high bandwidth models for peak power measurement.

Power meters are either single-channel or dual-channel instruments. Each channel requires its own sensor. Two channels provide the capability to measure input and output power on a device, circuit, or system and compute a gain or loss.

Some power meters have high measurement speeds of 200 to 1,500 readings/s. Some power meters can measure peak power characteristics of many types of signals including modulated signals and pulsed RF used in communications and other applications. Two-channel meters also make accurate relative power measurements. Power meters can be packaged into small enclosures designed for portability, making them suitable for use in the field.

The main limitation of a power meter is its amplitude measurement range. The wide frequency range is a trade-off for measurement range. In addition, a power meter will provide the most accurate measurement of power but will give no information on the frequency composition of the signal.

RF Spectrum or RF Signal Analyzer
A spectrum or vector signal analyzer measures RF signals in the frequency domain using narrowband detection techniques. The primary output display is a spectrum of both absolute and relative power vs. frequency. The output also can be a demodulated signal.

Spectrum analyzers and vector signal analyzers do not have the accuracy of power meters; however, the narrowband detection techniques used in these RF analyzers enable them to measure down to levels as low as -150 dBm. RF analyzers have accuracies typically at and above ±0.5 dB.

Spectrum and vector signal analyzers can measure signal frequencies from kilohertz to 40 GHz and beyond. The wider the frequency range, the greater the cost. The most common analyzers extend to 3 GHz. New communications standards that operate in the 5.8-GHz region require analyzers with 6-GHz and higher bandwidths.

Vector signal analyzers are spectrum analyzers with added signal processing capability that not only measure a signal’s amplitude, but also decompose the signal into its in-phase and quadrature components. Vector signal analyzers can demodulate modulated signals such as those generated by mobile phones, wireless LAN devices, and devices operating on other new and emerging standards. Vector signal analyzers can display constellation diagrams, code domain plots, and compute measures of modulation quality such as error vector magnitude.

Traditional spectrum analyzers are known as swept-tuned devices because a local oscillator is swept across a frequency span so that a narrowband filter can acquire the power content at the individual frequencies within the frequency span. Vector signal analyzers also sweep over a portion of the spectrum, but they capture wide frequency segments of data. As a result, vector signal analyzers can generate a spectrum more quickly than spectrum analyzers.

A key measure of a vector signal analyzer’s performance is its measurement bandwidth. The new high-bandwidth communications standards such as WLAN and WiMax generate 20-MHz bandwidth signals. The analyzer must have a large enough bandwidth to acquire the whole signal. If testing high-bandwidth, digitally modulated signals, make sure the analyzer has the measurement bandwidth to adequately capture the signal.

A spectrum analyzer will verify that a transmitter is generating the appropriate power spectrum. If distortion components such as harmonics or spurious signals must be tested, then a spectrum or vector signal analyzer is needed. Examples of other tests that require a spectrum analyzer or vector signal analyzer include testing for intermodulation distortion, third-order intercept, the 1-dB gain compression on a power amplifier or power transistor, and a device’s frequency response.

Testing a transmitter or amplifier that must process digitally modulated signals requires a vector signal analyzer to demodulate the signal. The vector signal analyzer can measure how much modulation distortion a device is creating.

The demodulation process is a complex, computation-intensive process. Vector signal analyzers that perform the demodulation and measurement computations quickly can save valuable test time and substantially cut test costs.

RF Sourcing Options
All RF signal sources generate continuous wave (CW) RF sine wave signals. Some signal generators also can modulate an RF signal while vector signal generators use IQ modulators to generate digitally modulated signals.

Types of sources can be further distinguished as fixed CW sinusoidal wave outputs, instruments that sweep over a range of frequencies, and analog signal generators and vector signal generators that add analog and digital modulation capabilities, respectively.

If test requirements call for a stimulus signal, an RF source is needed. Key requirements for RF sources include frequency and amplitude ranges, amplitude accuracy, and modulation quality for sources that generate modulated signals. Frequency tuning speeds and amplitude settling times also are critical for minimizing test time.

Vector signal generators are high-performance sources that often incorporate arbitrary waveform generators for digital signal generation. The arbitrary waveform generator enables the vector signal generator to produce any kind of digitally modulated signal.

Many waveforms can be generated internally, and in some cases, a waveform can be created externally and downloaded into the instrument. If the test specifications require a component, device, or system to be tested with the modulation that the device-under-test will process in its end use, a vector signal generator often is needed.

RF sourcing is used if test specifications call for receiver sensitivity tests, bit error rate tests, adjacent channel rejection, two-tone intermodulation rejection, or two-tone intermodulation distortion. The two-tone intermodulation tests and the adjacent channel rejection test require two sources. The receiver sensitivity test and the bit error rate test must have a single RF source.

A device used in the mobile phone industry most likely will require testing with the type of modulated signal required by the mobile phone standard. A mobile phone power amplifier will be tested with a modulated source such as a vector signal generator. Before selecting a vector signal generator, evaluate the speed at which the instrument can switch between different modulated signals to ensure the generator provides the fastest possible test times.

Network Analyzer
A third type of analyzer is a network analyzer. Network analyzers combine an internal RF source and either a broadband or narrowband detector to test RF devices. The output displays the device characteristics in X-Y rectangular coordinates, a polar display, or a Smith chart.

Essentially, a vector network analyzer measures the S-parameters of a device. A vector network analyzer can provide both magnitude and phase information and determine transmission losses and gains of these devices over a wide frequency range with good accuracy. It also measures return loss and impedance match as well as phase measurements and group delay.

Network analyzers are used primarily for analysis of components such as filters and amplifiers. Be aware that network analyzers work with continuous-wave unmodulated signals and that calibration of the analyzer is extremely important. A manufacturer’s calibration kit will keep the network analyzer in calibration.

Because network analyzers combine sourcing and measurement in one instrument and because the analyzer has a wide frequency range, they are expensive instruments.

Typical Application
An example of an application that requires four major RF test instruments is power amplifier (PA) testing. A source can provide the input signal, and either a power meter or a spectrum analyzer can measure output power. If accuracy is critical, such as in a maximum power measurement, then a power meter is needed for the output measurement.

The input impedance match of a PA is a key parameter for a designer developing an RF transmitter. It is important to amplify all the power supplied to the PA and not lose a substantial amount due to reflection. For this reason, PA manufacturers will specify and measure return loss.

Alternatively, if only the scalar magnitude is required, then a source and a spectrum analyzer or power meter can combine with a coupler to measure the magnitude of the reflected power. The setup is more complicated compared with the use of a network analyzer because additional passive RF components are required. The power meter will provide the more accurate power measurement for the return-loss scalar measurement.

The capability of a PA to deliver power to a load whose input impedance is not matched to a typically 50-Ω output impedance is a key measure of the amplifier’s capability to perform in real-world conditions where loads such as antennas may not have exactly a 50-Ω characteristic input impedance. In such cases, a non-50-Ω resistive load is switched to the output of the PA.

The load can force the PA to output into a VSWR of up to 20:1; a 50-Ω load would result in a VSWR near 1:1. The PA must be able to function properly and deliver some power to the load in the presence of a large amount of reflected power.

Some output measurements require spectrum analysis. RF PAs used in broadcast or mobile phone applications must not generate excess power in frequency channels adjacent to the channel where the PA is operating.

Adjacent channel power, intermodulation distortion, and harmonic distortion are measures of power that a PA generates outside the intended transmission channel. For these measurements, dynamic range, the capability to measure a small signal in the presence of a large signal such as a carrier signal, is an important spectrum analyzer specification.

For example, consider the case when a PA has a specification that its adjacent channel power is -60 dBc. The dynamic range of the spectrum analyzer must be at least 6 dB greater than the minimum allowed power level for the harmonic, the adjacent channel power level, or the intermodulation product.

The adjacent channel power measurement must be performed with a modulated signal, which means the source’s adjacent channel performance also has to be considered. The source’s adjacent channel power output must be at least 6 dB less than the maximum allowable adjacent channel power that the power amplifier can generate.

For harmonic measurements, the analyzer should have a frequency range three times greater than the maximum operating frequency of the PA to adequately capture the power in the 3rd harmonic of the maximum operating frequency. Again, the noise floor of the spectrum analyzer must be at least 6 dB lower than the 3rd harmonic component to have a reasonable signal-to-noise ratio for an accurate and repeatable measurement. The harmonic measurements indicate the amount of distortion the PA creates. Excessive distortion can negatively affect modulation performance.

Intermodulation distortion determines how much distortion the PA generates when signals at different frequencies or components of a signal at different frequencies are at the PA’s input. Two sources are required to generate the test signals. One dual-output source is inadequate due to insufficient isolation between the two outputs. The source would create its own intermodulation distortion, which would lead to higher and incorrect amplifier distortion measurements.

Modulation quality measurements often are made on PAs designed for the mobile phone market and other market segments such as WLAN applications where complex modulation schemes are used. This usually involves measuring the error vector magnitude.

Conclusion
This overview of RF test instrumentation provides overall guidance on what types of test instrumentation are needed to meet test requirements. In the vast majority of cases, any one or a combination of these four instruments will be needed: signal sources, power meters, spectrum analyzers, and network analyzers.

About the Author
Robert Green is senior market development manager at Keithley Instruments. In addition to having more than 10 years of experience in the wireless market, Mr. Green earned a B.S. in electrical engineering from Cornell University and an M.S. in electrical engineering from Washington University, St. Louis. Keithley Instruments, 28775 Aurora Rd., Cleveland, OH 44139, 440-248-0400, e-mail: rgreen@keithley.com

Published by EE-Evaluation Engineering
All contents © 2008 Nelson Publishing Inc.
No reprint, distribution, or reuse in any medium is permitted
without the express written consent of the publisher.