Verify Wideband High-Power Amplifiers with Noise Power Ratio (NPR) Test
2020-07-17 | 10 min read
Wideband communications systems use wide channel bandwidths and complex modulation schemes to achieve faster data throughput. However, wider bandwidths gather more noise, and complex modulation schemes are sensitive to system noise. Proper signal-to-noise ratio (SNR) is critical to maintaining communication links. You need to increase signal levels and reduce system noise to sustain the links.
High-power amplifiers constitute a significant building block of radio-frequency (RF) and microwave communication systems. They provide high-output power but also introduce nonlinear distortion to the system. Therefore, characterizing nonlinearities of the wideband high-power amplifier is the key to making solid design choices and creating an excellent product. In this post, we will discuss how to characterize distortion with the noise power ratio (NPR) test.
Characterization of Nonlinear Distortion
System designers characterize and evaluate nonlinear distortion for each subsystem and component to achieve the required performance. Stimulus-response measurements provide a straightforward method for evaluating distortion. They require a stimulus input test signal and acquisition of the output response for analysis.
Stimulus-response tests help you understand the performance of the RF components under various conditions to determine the best trade-offs in your design. Common stimulus-response tests include complementary cumulative distribution function (CCDF), harmonics, third-order intermodulation (TOI), adjacent channel power (ACP), and error vector magnitude (EVM). To fully characterize your RF components, you need to know the power characteristic of the simulated input signal and the measured output signal. Perform stimulus-response tests, such as CCDF, harmonics, TOI, ACP, and EVM to understand the performance of the RF components under various conditions to determine the best trade-offs in your design.
In-channel and out-of-channel distortion
You can take a further step to classify the impacts of nonlinear distortion in the frequency domain: in-channel and out-of-channel distortion. The harmonics, spurs, and ACP measurements belong to out-of-channel distortion and these unwanted signals may interfere with other devices or systems. Test specifications and radio regulations specify test cases for these measurements.
The in-channel distortion affects the signal quality and causes a poor communication link. The EVM measurements examine the signal modulation quality. With vector signal modulation analysis, we can easily see the error vector spectrum caused by nonlinear distortion. Figure 1 illustrates a demodulation analysis and the error vector spectrum is at the lower-right of the figure. The error vector spectrum represents the spectrum of the residual signal after the desired modulation energy has been removed. The energy of the spurious signal is now clearly revealed, including an in-channel spur inside the signal spectrum.
For wide bandwidth signals such as satellite communications systems and 5G new radio systems, you require a wideband signal analyzer to analyze the signals. For example, Keysight N9021B MXA supports up to 50 GHz frequency range and 510 MHz analysis bandwidth.
Characterize Distortion with NPR tests
In the past, to characterize in-channel non-linear distortion for wide bandwidth signals was a challenge. Engineers explored a way to figure out the unwanted energy inside the signal. Noise power ratio (NPR) tests have been introduced for more than 40 years for the measure of the intermodulation distortion of active devices. Using NPR tests, you do not need a wideband signal analyzer for the distortion measurements. Let’s have a look at what NPR is and how to achieve the best measurement results.
What is NPR?
NPR is a distortion measurement that helps determine a maximum spurious-free dynamic range. To make an NPR test, you need to generate conditioned noise and analyze the change in the noise after it passes through the DUT. Figure 2 illustrates the NPR’s input and output signals and measurements. The NPR input signal consists of an additive white Gaussian noise and removes a portion of the spectrum with a notch filter. A DUT with nonlinearity leads to distortion components within the spectral notch at the output of the DUT (the purple area). NPR is the sum of all intermodulation products across the passband ratioed with the sum of all intermodulation products in the notch.
For accurate NPR tests, the notch must be deep and sharp for the stimulus signal (input signal), and the signal analyzer must have a high dynamic range. These test requirements ensure that the measured distortion is from the DUT rather than the signal generator and the signal analyzer.
Multitone distortion method
A deep and sharp notch filter may not be easy to access at high frequencies.
An alternative method is to create a large number of tones to simulate the wideband noise and disable some tones to create the notch using a single signal generator. However, the tones also cause intermodulation products inside the signal generator, which requires an advanced correction routine to suppress the distortion products both in-band and out-of-band.
Figure 3 illustrates the multitone distortion software setups for a 100 MHz bandwidth multitone signal with 1,001 tones and 100 kHz tone spacing. There is a 10 MHz notch at a 10 MHz offset to the center frequency for NPR tests (red area).
Figure 4 shows the measured result on a spectrum analyzer without connecting to a DUT. The noise power at the notch comes in at -43.18 dBm at a 9.8 MHz span. The in-channel distortion is mostly from the signal generator. To achieve the best measurement results, you need to minimize the distortion caused by the signal generator.
Apply pre-distortion to enhance signal quality and minimize test uncertainty
Keysight offers a correction tool to minimize intermodulation products. Once the multitone-generation software tool enables the correction routine, the spectrum analyzer measures the intermodulation products in the notch area. Then the tool recalculates and downloads the corrected multitone waveform into the signal generator to minimize the intermodulation products. The correction process will repeat until the intermodulation products are lower than the target suppression level. Figure 5 shows a reduction of noise power to -60.29 dBm (17 dB improvement) after the corrections.
The advanced correction improves the intermodulation products caused by the signal generator. You can extend the correction plane to the DUT’s input port, including extra active components in the test fixtures, such as external power amplifiers.
With the ability to generate a precise and repeatable stimulus signal for NPR tests, this digital method of determining NPR is in many ways superior to the analog method for evaluating the effects of intermodulation distortion in amplifiers and other communications devices. The digital method benefits from a calibrated flat amplitude accurate signal, predictable and repeatable sharp spectral shape, correction for improving notch depth, and extending the measurement plane to the DUT.
You can refer to my earlier posts for common stimulus-response tests, such as CCDF, harmonics, TOI, ACP, and EVM to understand the performance of the RF components under various conditions to determine the best trade-offs in your design.