5 tips to improve power amplifier (PA) design verification efficiency
The importance of power amplifiers (PAs) in mobile devices is self-evident, especially with the development of communication technology. Broadband standards such as 5G, WiFi 6/6E, and UWB have put forward higher requirements on power amplifiers, more complex modulation methods, higher modulation orders, more carrier aggregation, higher frequency bands and bandwidths, which makes the complexity of test verification also increase accordingly.
I believe these are the troubles that everyone often encounters in their daily work. Based on this, we have summarized 5 typical problems that are often encountered, as well as tips for solving the problems. Friends, let’s take a look at these easy-to-step “pitfalls”. Have you successfully avoided them?
# Question 1 #
Inaccurate model during design and simulation phase
The design of broadband high-frequency PA is a complex task that requires the help of professional simulation tools. For designers, PA simulation faces two major challenges:
The first is how to obtain simulation results that are consistent and accurate with actual measurements;
Second, after the PA design is completed, how the model is used for subsequent system verification or DPD algorithm verification.
Tips
To solve these two problems, the most important thing is to obtain accurate simulation models, including large signal, nonlinear, and broadband models; also including accurate models of passive devices, circuit routing, and connectors.
For nonlinear models, you can use a vector network analyzer to extract the X-parameters of the device. For passive components, routing, connectors and other model extraction, you can design fixtures, de-embed, and test the component model. Use the measured model for simulation, multiple iterations, and finally the simulation is consistent with the test.
The document "Broadband Power Amplifier Design and Validation" at the end of the article details how to use ADS to design and validate broadband power amplifiers, as shown in the following figure:
Figure 2. Broadband power amplifier design using ADS
For system verification or DPD algorithm verification, the memory effect of broadband devices also needs to be considered. ADS simulation software can be used to generate FCE models for subsequent system verification or DPD algorithm verification. You can also use instruments to build a semi-physical test system, as shown in the figure below, to connect the designed PA to the system software through instruments to directly complete system performance verification or DPD algorithm verification.
Figure 1. Schematic diagram of Keysight's hardware-in-the-loop test system
# Question 2 #
Severe EVM distortion during testing
Modern communications place stringent requirements on the bandwidth and operating frequency band of RF systems. Especially for millimeter-wave and ultra-wideband power amplifiers, the distortion and errors introduced by the test platform will seriously affect the final test results.
The following figure shows a prototype test we have conducted. It uses FBMC modulation based on the 5G candidate waveform. The broadband vector source generates an original signal with a carrier frequency of 20GHz and a modulation bandwidth of 4GHz. The data transmission rate of its physical layer modulation reaches 10-20Gbps.
It can be seen from the spectrum curve that the amplitudes of different frequency components in the entire frequency range fluctuate greatly, and the frequency separation attenuation away from the center frequency increases, showing obvious amplitude unevenness. Because the signal is composed of many subcarriers, these frequency components with amplitude attenuation will reduce the signal-to-noise ratio of the subcarrier in which they are located, resulting in a decrease in EVM.
Although the prototype platform can still achieve a high throughput by relying on measures such as receiver channel equalization and error correction, if it is used for RF testing of PA or base stations, it will seriously affect the accuracy of the test EVM.
Figure 2. Example of an uncorrected UWB modulated signal.
Tips
Broadband correction compensation for test benches
Method 1: The instrument presets calibration data. The broadband distortion of the instrument itself is measured before leaving the factory and the calibration data is stored in the instrument. During the test, the instrument automatically applies the calibration data according to the frequency and bandwidth, and the test can be performed without additional calibration operations. (Note: This method requires that the instrument supports the built-in calibration function)
Method 2: System external calibration. Use a calibrator to perform width calibration on the instrument on site, generate calibration data in real time to compensate the instrument, and optimize the EVM of the instrument. The signal source, analyzer, and external devices are calibrated independently. The calibration data can be applied to the instrument test port, or can be calibrated together with external accessories or RF device modules used in the test. The calibration data can be applied to the input or output port of the device under test, and the impact of various environmental and working conditions on site will also be included in the calibration operation. Therefore, this method can always achieve the best EVM characteristics of the instrument on site.
Keysight Technologies' test platform,
An optimal solution combining the above two solutions is provided.
•
Signal generation part
,
M9384B VXG microwave signal generator has built-in calibration function, output signal is calibrated to the port, through user-defined automatic channel response correction and S parameter de-embedding, the signal calibration plane is extended to the PA input end face;
• For signal analysis , the U9361 RCal receiver calibrator is used to remove the frequency response caused by external accessories such as cable adapters, and the signal calibration surface is extended to the PA output end surface, as shown in the figure below. This is the currently recommended method.
Figure 1. External calibration method (for linear distortion)
For detailed usage of Rcal, please refer to the "Rcal Usage Guide" at the end of the article.
# Question 3 #
Poor EVM consistency in measurements
The selection of test accessories, such as adapters and cables, is an easily overlooked link in PA and broadband transceiver testing. In actual testing, test accessories will have a great impact on the results, especially the cables and connectors used in the millimeter wave band, which generally have greater linear distortion and unevenness than low-frequency bands below 6 GHz.
Tips
Method 1: Use high-quality adapters and cables to ensure test consistency.
Method 2: While selecting high-quality test accessories, adopt on-site external calibration to include the errors of the test accessories in the calibration data and remove the influence of these parts. Please refer to the previous article for the specific method.
# Question 4 #
After adding driver amplifier, EVM deteriorates seriously
A common problem encountered when testing high-power PAs is driver amplification. Since high-power PAs often require higher pins, and the optimal linear output level of the millimeter-wave vector signal source is usually lower than the requirement, it is often necessary to add a driver amplifier to the input of the PA under test. The following figure is a diagram of an actual test connection:
Figure 5G high-power PA test EVM connection block diagram
In addition to the signal source and analyzer used for 5G broadband signal generation and analysis, the driver amplifier itself also has a great impact on the test. Although the driver amplifiers generally used are broadband linear amplifiers, as long as the input and output power ranges are set appropriately, the amplifier works in the linear region and the nonlinear distortion is very small, it still has linear distortion, and the amplitude-frequency response and phase-frequency response fluctuations of the driver amplifier itself still have a great impact on EVM.
In our actual tests, we found that in the frequency range of 26GHz-29GHz and the modulation bandwidth of 800MHz, the EVM of the output signal of the signal source itself has been corrected to 0.8%, but after passing through the driver amplifier, the EVM will deteriorate to a maximum of 3%-4%. This not only causes the EVM of the final measured PA output signal to be very high, but also even exceeds the manufacturer's requirements for system-level EVM.
Tips
By using the external calibration method mentioned in the previous article, as shown in the figure below, the overall EVM of the signal source plus the driver amplifier can reach about 1%. Then, when the PA under test is connected for EVM testing, a relatively ideal result can be obtained, because the linear distortion of the driver amplifier will not affect the test.
Figure 1: RCal-based external correction method solves the problem of driver amplifier affecting EVM test
# Question 5 #
Multiple insertions during wafer testing lead to inefficiency and loss
In the on-chip measurement of highly integrated PAFEM, many parameters need to be measured, such as S parameters, noise figure, intermodulation distortion, compression, pulsed RF measurement, etc., and different parameters usually need to be measured using different systems.
It takes multiple needle insertions to complete the test of multiple systems, which will also leave marks on the PAD, affecting the test efficiency and accuracy to varying degrees. The following figure shows the PAD after 1 and 4 needle insertions. It can be clearly seen that after multiple needle insertions, obvious marks are left on the PAD, which is a great loss to the test board and probes. In addition, each measurement requires recalibration, which is time-consuming and labor-intensive.
Figure: After multiple injections, obvious marks are left on the PAD
Tips
The single connection and multiple measurements method, that is, connecting the device under test once and using one system to complete the work that originally required multiple systems, can reduce the complexity of connection and workload. At present, Keysight's PNA-X series high-performance network analyzers can easily achieve multiple measurements of active or passive devices with only one set of connections: S parameters, noise figure, gain compression, THD, IMD and spectrum analysis.
Conclusion
PA design and verification involves many aspects, and the testing methods on the R&D side and the production side are also different. In addition to the above-mentioned points, there are many other points that need attention:
For more information, click to watch Keysight Technologies’ video tutorial: From Design to Verification, RF/Millimeter-Wave Power Amplifier and RF Front-end Test Solutions. You can also download the corresponding technical materials for more in-depth study.
Data Preview
☚ Swipe left and right to view ☛
If you like this article, please click "Reading" at the end of the article.
Share it with your friends~
About Keysight Technologies
We are committed to helping enterprises, service providers and government customers accelerate innovation and create a secure and connected world. Since the founding of HP in 1939, Keysight Technologies has been operating independently as a new electronic test and measurement company on November 1, 2014. We continue to uphold the same entrepreneurial spirit and passion to start a new journey, inspire global innovators and help them achieve goals beyond imagination. Our solutions are designed to help customers innovate in 5G, automotive, IoT, network security and other fields.
Click "Read original text" to register now
Featured Posts
京公网安备 11010802033920号