Testing amplifiers for 5G with up to 2 GHz analysis bandwidth

Technology News | January 9, 2018

By Jean-Pierre Joosting

5G: wide bandwidths over multiple carriers

LTE-Advanced Pro currently offers theoretical data rates of up to 1.7 Gbit/s, whilst 5G is aiming for peak data rates of 20 Gbit/s and average user data rate measured in hundreds of Mbit/s. This can only be accomplished by utilizing more bandwidth: bandwidths up to 1 GHz are now under discussion. However, bandwidths of that size are unavailable for the 450 MHz to 6 GHz frequency band currently used for cellular communications. They are only available in the centimeter wave and millimeter wave frequency range.

With possible 5G frequency bands in the range from 24.25 GHz to 86 GHz, the focus lies on the 28 GHz and 39 GHz bands. Bundling multiple carriers will make it possible to achieve bandwidths of several hundred megahertz.

Flexible test

A flexible test and measurement solution for 5G has to support enough bandwidth at µWave frequencies. For transmitter tests and EVM characterization 800 MHz are required to simultaneously measure 8 bundled carriers of about 100 MHz bandwidth each. For component characterization like amplifiers even more bandwidth is desired to measure the out-of-band effects of non-linearities. Additionally software solutions are required for amplifier test and optimization and for a flexibly configurable OFDM-signal analysis.

The FSW high-end signal and spectrum analyzer is now available with the FSW-B2001 hardware option, providing the ability to characterize 5G devices and components with 2 GHz internal analysis bandwidth. The software options FSW-K18/K18D comprehensively characterizes amplifiers and the VSE-K96 OFDM analysis software flexibly analyses OFDM-modulated signals, such as the anticipated in future 5G standards.

Testing signals with 2 GHz analysis bandwidth

Characterizing components with that frequency and bandwidth also requires a vector signal generator. The SMW200A is capable of delivering signal generation up to 40 GHz with up to 2 GHz bandwidth. A test setup based on the SMW200A for signal generation and the FSW for signal analysis is easy to operate, while providing industry-leading performance and bandwidth.

To make these wideband measurements, test instruments need high dynamic range and low input signal distortion. Meeting these criteria, the FSW provides a spurious free dynamic range (SFDR) of 60 dBc at 2 GHz bandwidth, allowing users to precisely determine the signal modulation quality, for example with measurements of error vector magnitude (EVM). To ensure reliable measurements on signals with very good EVM, the EVM generated by the instrument must itself be minimal. As a case in point, the FSW-B2001 option, along with its VSE-K96 OFDM analysis software, ensures reliable EVM measurements in the order of –40 dB with 800 MHz wide signals in the 28 GHz range.

The VSE-K96 OFDM analysis software enables users to measure modulation on non-standardized OFDM signals and offers a high degree of freedom in defining the measurement parameters. This flexibility provides a tremendous advantage, as the specification of OFDM signals has not yet been finalized in the 5G cellular standard.

Digital pre-distortion compensates nonlinear effects in amplifiers

Component testing requires instruments with the appropriate measurement applications, such as the FSW-K18 firmware option to comprehensively characterize amplifiers, and the new FSW-K18D direct DPD measurements extension that makes it easier to compensate for memory effects.

Power amplifiers in base stations or smartphones must demonstrate high linearity over a wide frequency range to deliver the exacting transmit and receive characteristics. Unwanted nonlinear effects, however, generally occur in the upper power range and diminish signal quality. They manifest themselves as higher EVM values and increased interference in adjacent channels. Possible consequences are lower orders of modulation and – hence lower data rates – along with signal interference in adjacent channels.

Effective characterization of these effects makes it possible to provide digital compensation. Here, the signal is digitally pre-distorted upstream of the amplifier to counteract the distortion of the amplifier. This eliminates nearly all the artefacts caused by the amplifier, so that the developer receives an almost linear signal at the output of the amplifier or mixer.

Amplifier measurement applications that integrate functions for digital pre-distortion can be used to comprehensively characterize distortions caused by nonlinear output amplitudes or phase changes relative to the input signal (AM/AM and AM/PM). It is also possible to check the effects of pre-distortion. For example, the FSW-K18 initially compares an ideal reference signal from the signal generator against the acquired and distorted readings. Using this data, it generates a polynomial function that describes the pre-distortions by way of approximation then transmits corresponding correction data to the signal generator, which in turn generates the pre-distorted signal. Analysis bandwidths of triple, quadruple and quintuple the signal bandwidth are typically used in this process.

Image 1: Diagrams at the top: A signal distorted by an amplifier. The amplifier goes into compression when the power is increased (at approximately 1 dBm). Amplification is no longer linear and the phase is distorted. Diagrams at the bottom: A signal predistorted with correction data from the FSW. Compression starts at a significantly higher power level. The 1 dB compression point is approx. 1 dB higher, and the phase distortion is corrected perfectly. Correcting memory effects with the FSW-K18D option also reduces the scattering of test points; the displayed traces are narrower.

Correcting memory effects

As well as nonlinear effects, memory effects in the amplifier produce a frequency response: essentially a distortion of the amplitude or phase versus frequency. Until recently, correction of these memory effects has been very complex, with elaborate mathematics such as Volterra series required to describe this response. Today, options like the FSW-K18D DPD measurements extension greatly simplify the process. For every arbitrary signal, the software calculates the frequency response along with the EVM value that describes the signal quality. Instead of approximating with polynomials, it applies iterative approximation via the individual samples. The signal played back by the generator and distorted by the device under test is measured and the pre-distortion is adjusted. After multiple iterations, the signal generator outputs a signal with optimal pre-distortion. In this way, the option compensates both nonlinear distortion and frequency response for a predefined signal sequence. The result delivers the best possible reference for pre-distortion algorithms employed by the user.