Power Amplifier Test in Five Measurements
11 June 2023
Figure 1: Diagram of a typical RF signal path including a power amplifier
Testing power amplifiers (PAs) requires numerous measurements, due to their crucial impact on RF path signal quality and battery life.
With increasing integration of components, like multichannel transmitters which require a dedicated PA for every channel path, the measurement complexity also increases. Beyond typical scattering parameters (S-parameters), Gabi Duncan of Keysight Technologies looks at why and how engineers make some of the most challenging PA characterisation measurements.
Historically, the key measurements required a lot of time and test equipment. Today, engineers just need a PXI vector network analyser (VNA) to handle these key amplifier characterisation tests.
Beginning with the basics, all VNAs essentially do the same thing. They send defined signals into a component, then receive the signal output. Engineers characterise components by comparing the input and output signals.
Characteristics include how much the signal reflected away from and transmitted through the device, as well as the transmitted signal’s condition. Engineers evaluate how amplifiers perform over a range of frequencies and power levels. Engineers visualise these results as S-parameters.
S-parameters are foundational VNA measurements. Typical RF amplifiers have one input and output port. Two-port devices have four S-parameters, shown in Figure 2 - the S11, S12, S21 and S22.
The S11 reflection coefficient shows the return loss. It indicates how much power reflects from port 1. Reflected power at port 1 reduces the total power available for transmission, in the forward or reverse directions, through the device. Engineers use S11 to optimise the power transfer of input-matching networks.
Figure 2: Set of two-port S-parameter measurement traces on a VNA
The S12 transmission coefficient captures the amplifier’s reverse isolation. It reports how much reverse power attenuation occurs in the amplifier.
The S21 transmission coefficient indicates the insertion loss or gain of the amplifier. It depicts the transmission ratio in the forward direction.
The S22 reflection coefficient shows reflected signal power at the amplifier’s output. It delineates how much power transfers through the output port. Engineers use S22 to create an output matching network that facilitates the greatest power output from the amplifier.
For multiport devices, the number of S-parameters equals the square of the number of device ports. Whether testing multiple two-port PAs, or several PAs supporting a multiport module, a PXI VNA provides the scalability needed for configurable measurements.
Altogether, S-parameters capture the linear behaviour of an amplifier. If the amplifier behaves linearly - as desired - the S-parameters remain constant regardless of input power. Though useful, S-parameters only signify the beginning of the PA evaluation process - there is in fact a lot more to it than just that.
Robust PA evaluation must contend with nonlinear amplifier behaviour. Particularly, nonlinear distortion presents significant characterisation difficulty. What does distortion look like for amplifiers, and how do VNAs evaluate it?
Figure 3: Diagram representing how the PXI VNA captures 2D gain data while sweeping over frequency and power ranges
Gain compression
Gain compression measurements traditionally sweep power at a fixed frequency plotting gain as a function of the supplied input power. Excessive power supplied to the amplifier causes gain compression. Once the gain trace begins to curve and decrease by 1dB - known as the ‘1dB compression point’ - the amplifier officially operates nonlinearly. Nonlinear amplifier behaviour means that the output signal cannot linearly relate to the input signal due to distortion.
To avoid nonlinear amplifier performance, engineers measure gain compression to pinpoint what power level pushes the amplifier into compression. Modern gain compression measurements enable engineers to complete 2D sweeps of both input power and frequency simultaneously while plotting gain, as described in Figure 3.
Capturing an amplifier’s gain compression point at various power levels and operating frequencies improves an engineer’s understanding of the nonlinear aspects of the amplifier’s behaviour. However, what causes the nonlinear aspect?
Harmonic distortion analysis
Amplifiers exhibit nonlinear behaviour when internal imperfections cause signal distortion, leading to compression when engineers supply more powerful signals. VNAs send a single-frequency tone into a nonlinear amplifier for basic distortion tests. The results of single-tone tests display the amplifier’s harmonic output distortion.
In telecommunications, components typically operate within comparatively narrow bandwidths. Because of the narrow operation bands, harmonic distortion falls outside the frequency band where the RF system operates.
Figure 4: The input modulated signal (yellow trace) has no distortion while the output signal (blue trace) has both in-band and out-of-band distortion
Communication engineers call distortion signals outside the system’s operating frequency ‘out-of-band’ distortion. Sending rogue distortion signals into other frequency bands threatens the operation of nearby devices, so designers use single-tone test information to filter away out-of-band distortion. However, when more than one signal tone enters the amplifier, the distortion products produced fall within the target channel as well. Assessing ‘in-band’ distortion requires additional measurements.
IMD analysis
Intermodulation distortion (IMD) occurs when two signal tones combine to create new frequency components. While these two signal’s harmonic products may fall outside the operation band, some distortion products will also fall inside the operating frequency channel. Filters cannot distinguish IMD from desired signals as they exist within the same band, so attenuating in-channel IMD also degrades the message signal.
To capture IMD, engineers need to send two separate frequency tones into the device and then measure it with spectrum analysis. Past IMD test setups required multiple instruments. However, measurement technology advancements enable engineers to quantify IMD by simply using a PXI VNA.
VNAs capture amplifier distortion at both two fixed frequencies and across the entire operating frequency range, allowing designers to robustly test how well the amplifier handles two-tone stimulus.
Modulation distortion analysis
While two-tone measurements like IMD help engineers estimate their amplifier’s capabilities, designers need confidence in the amplifier’s performance under real-world conditions. 
Figure 5: The error vector in an IQ constellation
Modern communication systems operate with modulated signal schemes that include thousands of frequency tones. This complexity means that modulation distortion introduces even more analysis challenges than IMD.
While IMD happens when two tones produce undesirable frequency products, modulation distortion occurs when hundreds of signal tones mix to create numerous combinations of harmonic products. Like IMD, these products fall both in-band and out-of-band. To stimulate modulation distortion in the lab, engineers send numerous tones into the amplifier.
Modern PXI VNAs streamline modulation distortion analysis, reducing test time and error. The PXI VNA accurately calculates the output’s error vector magnitude (EVM) by capturing both input and output modulation distortion data, as illustrated in Figure 4.
The telecommunications industry considers EVM the benchmark metric for in-band distortion. Think of EVM as the amplifier’s final grade - grouping all the errors, shown in Figure 5, into a single percentage. Modulation standards, such as 802.11ac and 5G NR, set minimum acceptable EVM percentages. As standard stringency increases so does the need to accurately capture and optimise amplifier EVM.
Conclusion
S-parameters, gain compression, plus intermodulation/modulation distortion analysis allow engineers to gauge PA performance. Until recently, amplifier characterisation tests like these required many instruments, time-consuming setups and calibration work. Fortunately, progress in measurement science enables robust amplifier testing without all that inefficiency. Now, engineers accomplish advanced analysis using the same VNA setup as for S-parameter measurements.
Contact Details and Archive...
