Filipe M. Barradas

Agile Single- and Dual-Band All-Digital Transmitter Based on a Precompensated Tunable Delta–Sigma Modulator

In this paper, a new architecture for designing tunable single- and dual-band radio-frequency fully digital transmitters is proposed and validated. The proposed architecture excels the state of the art in terms of simplicity and flexibility. While its short critical path leverages the use of equivalent polyphase decomposition techniques to increase the global systems sampling frequency, the capability of changing the systems frequency response in real time enables its use in both single- and dual-band transmission scenarios. To mitigate a crosstalk in the dual-band scenario, a precompensation technique is also proposed. This novel concept has been successfully validated in a field-programmable gate array (FPGA) based transmitter. To validate both the proposed transmitter as well as the precompensation mechanism, spectrum and error-vector magnitude (EVM) measurements were obtained for two scenarios with a carrier frequency of 2.5 GHz: 1) single-band, using quadrature phase-shifting keying (QPSK), 16-quadrature amplitude modulation (QAM) and 64-QAM, with no intermediate frequency (IF), for different symbol rate (SR) values (from 3.125 up to 15.625 Msps) and 2) single- and dual-band, using QPSK and 16-QAM, with an SR of 3.125 Msps, for different IF values (from 2 up to 120 MHz). All the experimental results present EVM values below 2.6%, resulting in a well-defined constellation.

Higher locality non-linear basis functions of Volterra series based models to improve extraction conditioning

Extraction of Volterra based nonlinear systems models can be hindered by conditioning problems. This is especially true when working with low precision arithmetic, typical of hardware implemented DPD systems. Traditionally, monomials are chosen as the basis functions because they are continuous and easy to manipulate, although other bases could be used. Spline bases, for example, have been chosen instead of polynomials for their increased locality. However, we now show that it is also possible to construct polynomials with increased locality and thus, keeping continuity but still improving the condition numbers of the regression matrix. This polynomial basis can be seen as a bridge between LUT and polynomial models.

Compensation of Long-Term Memory Effects on GaN HEMT-Based Power Amplifiers

The long-term memory effects of gallium nitride (GaN) transistors have prevented its use in situations where the modulated envelope signal has a wide amplitude variation over time, such as in time division duplex systems. These long-term memory effects are generally attributed to electron trapping in GaN high electron-mobility transistors (HEMT), which have shown to be very difficult to compensate, especially in cellular base station transmitters known to be subjected to highly restrictive linearity specifications. On top of the electron trapping effects, we show that thermal effects can also induce long-term memory behaviors, which should also be accounted for when linearizing these devices. Because the conventional behavioral modeling approach has been incapable to compensate these long-term memory effects on GaN HEMT-based power amplifiers (PAs), we started by investigating the physical mechanisms responsible for these semiconductor impairments in GaN devices. This physics-based knowledge was then used to design new predistorter models that could effectively compensate those PAs subjected to GaN trapping and thermal effects. In this paper, we describe the new predistortion models for PA linearization, as well as the characterization methods used to determine their parameters. To validate the linearization effectiveness of the proposed model, several high power GaN-based PAs are tested with multicarrier GSM signals, and their linearization results are compared against other state-of-the-art models, evidencing a clear and significant improvement. In fact, to the authors’ knowledge, the proposed approach is the first one to reduce the PA distortion effects due to GaN long-term memory effects to such low levels, allowing a comfortable compliance with the imposed linearity masks.

Using spline basis functions in Volterra series based models

The choice of nonlinear description used in Volterra based models strongly influences the robustness of the parameters extraction. Traditionally, monomials are used as the nonlinear description. However, it has been shown that other polynomials provide easier extraction. Volterra-based models can also be described using spline interpolated look-up tables (LUTs). Splines have increased locality which improves conditioning, converging to polynomials as the spline order-1 tends to the number of LUT points. In this paper we first show how to describe Volterra models in terms of splines and then study the conditioning and error for increasing spline order in a practical example.

Modeling PA linearity and efficiency in MIMO transmitters

Modern transmitter architectures rely on multi input multi output (MIMO) techniques, using several radio-frequency (RF) power amplifiers (PAs) to excite an antenna array. The array elements are often coupled, creating apparent variable loads at the output of each PA. In this system, the behavior of each PA cannot be fully described solely as a function of its input, as it will change according to the coupled signal. Moreover, the impact of the RF mutual coupling is felt not only on the RF output, but also on the direct current (DC) consumption, and thus efficiency, of the device. In this paper, we propose a novel PA modeling approach capable of predicting both the RF output and the absorbed DC current, dependent on the excitation and coupled signals. The proposed model is, therefore, suitable for system level simulations including efficiency and linearity predictions.

Setup and calibration procedure for LPE PA characterization with synchronous input-output excitations

Modern transmitter architectures make use of multi-input multi-output (MIMO) techniques, employing a high number of RF power amplifiers (PAs) to excite an antenna array. The array elements are often coupled, creating an excitation at the output of the PAs, which is dependent on the outputs of the other elements. Therefore, in this system, the nonlinear behavior of each PA cannot be fully described solely as a function of its input. In order to characterize a device that will operate in this arrangement, it is important to develop setups that are capable of exciting the PA at the input and output, synchronously at the RF carrier (phase control) and envelope (delay control) levels, with correlated or independent amplitude and phase modulations. In this paper we propose a setup for this type of characterization and show how it can be calibrated to correct the frequency response of the observation paths and the excitation paths in a simple way using broadband modulated signals, covering the full measurement bandwidth.

Digital predistortion of RF PAs for MIMO transmitters based on the equivalent load

Modern transmitter architectures rely on MIMO techniques, using several RF PAs to excite an antenna array. The array elements are often coupled, creating apparent variable loads at the output of each PA. In this system, the behavior of each PA is dependent on the coupled signal from the other array elements. This is felt in terms of device output power and efficiency variation and also as an overall change of the gain characteristic. When linearization of the RF PAs feeding the antennas is performed without taking into account this coupling, some degradation of the system performance in terms of linearity is expectable. In this paper, we analyze the PA operation on a dual antenna array in terms of distortion and propose a DPD structure with adaptive coefficients based on the equivalent load.

Using statistical information for fast static DPD of RF PAs

Modern radio-frequency transmitters require digital predistortion to fulfil linearity requirements while operating the RF PA in highly efficient modes. A least-squares process is typically employed to calculate the predistorter coefficients. The least-squares method requires complex mathematical operations and relies on the observation, at the input and output of the PA, of a specific realization of a telecommunication signal. With the increase of signal bandwidths faster DPD methods are required. In this paper, we use statistical measures to create an inverse of the PA. The proposed method avoids the least-squares formulation potentiating a much faster DPD extraction solution., Modern radio-frequency transmitters require digital predistortion to fulfil linearity requirements while operating the RF PA in highly efficient modes. A least-squares process is typically employed to calculate the predistorter coefficients. The least-squares method requires complex mathematical operations and relies on the observation, at the input and output of the PA, of a specific realization of a telecommunication signal. With the increase of signal bandwidths faster DPD methods are required. In this paper, we use statistical measures to create an inverse of the PA. The proposed method avoids the least-squares formulation potentiating a much faster DPD extraction solution.

DPD tuning with frequency selective distortion minimization

The determination of the coefficients of digital predistorter (DPD) models, for power amplifier (PA) linearization, is commonly performed through least mean squares techniques, where the residuals of the PA output signal are minimized as a whole. However, telecommunication regulation restrictions usually address distinct characteristics of these residuals, such as EVM and IMR. In this work we present a DPD extraction method for compensating specific characteristics of the PA behavior, discriminated in the frequency domain. This technique leads to DPDs more dedicated, for instance, to maximizing the IMR instead of the conventional minimization of the NMSE. A demonstration of the methods performance is presented in a multichannel GSM scenario.