José Ricardo Cardenas-Valdez

Flexible test bed for the behavioural modelling of power amplifiers

Purpose – The paper aims to focus on the memory-polynomial model (MPM) as special case of Volterra series, implemented in hardware. The behavior of the MPM is fully proved through a comparison with AM-AM and AM-PM measured data. The results show that this simulation technique is able to prove the effectiveness of the MPM implementation as behavioural model for high power radiofrequency amplifiers. The system is able to compensate perturbations caused by modern communication systems. Design/methodology/approach – The implementation uses Matlab-Simulink, and its digital signal processing (DSP) builder. The first stage allows developing the model in Matlab using the DSP builder blockset through the signal compiler block. Then, the design is downloaded to the DSP board. Findings – The paper demonstrates a proper behavior of the MPM as a truncation of the Volterra series, with respect to different inputs. This is a key point, because the series truncations allow first to implement this model in real time and s...

Modeling memory effects in RF power amplifiers applied to a digital pre-distortion algorithm and emulated on a DSP-FPGA board

An emulation tool with the capability of modeling the nonlinearity order and memory effects for real power amplifiers (PAs) conversion curves, is introduced. The proposed tool comprises special cases of Volterra series as the Memory Polynomial Model and a learning technique like artificial neural networks. The proposed system is a novel integrated one able to model the real behavior of radio frequency (RF)-PAs under different memory models. The developed system starts from a test bed able to acquire AM-AM and AM-PM distortion curves measurements. This procedure can predict the behavior and improve the analysis using it as a design tool. Emulation tool for modeling the nonlinearity order and memory effects for real power amplifiers.It comprises Memory Polynomial Model and a learning technique like artificial neural networks.The new proposed tool reduces design time as well as realizing the implementation of a DPD.The overall performances have been properly verified by using AM as input signals.A novel technique by combining a GUI as control stage, Simulink and the DSP Builder tool.

Development and implementation of a fish counter by using an embedded system

Abstract The development and implementation of an instrument for the automatic counting of ornamental fish by using an embedded system, is introduced herein. The proposed instrument is tested with two marine species, the Guppies (Poecilia Reticulata) and Mollies (Poecilia Sphenops), under conditions of controlled lighting and specimens whose sizes vary from 0.5 to 2.3 cm. The counting is done by digital image processing obtaining an average accuracy up to 96.64% using different species of fishes and different sizes. The main contributions are the theoretical and experimental study to determine the aquarium background color and the algorithm of the proposed method implemented in a low cost and high performance embedded system, specifically in a Raspberry Pi 2 executing the free GNU Octave Scientific Programming Language, thus, allowing the counting instrument to be reliable, portable and easily migratory to different operating systems. The obtained results demonstrate that the proposed method is competitive with state-of-the-art ones.

FPGA-based test bed for measurement of AM/AM and AM/PM distortion and modeling memory effects in RF PAs

Using a field-programmable gate array (FPGA) development board, a digital signal processor (DSP) builder, and the phase-to-amplitude conversion principle, a low-cost system for measuring the amplitude-to-amplitude (AM/AM) and amplitude-to-phase (AM/PM) distortion curves of radio frequency (RF) power amplifiers (PAs) is presented. The state of the art based on the measurements and preliminary studies of AM/AM and AM/PM distortion curves is discussed. A full digital control of the test bed simulated/emulated in Matlab/Simulink is introduced to recalculate the known AM/AM and AM/PM measurements stored as look-up table (LUT). Finally, the low-cost system comprises the memory polynomial model (MPM) that involves the nonlinearity order and memory effects of real PAs. Low cost system for simulating/emulating/measuring AM/AM and AM/PM distortion curves in RF PAs.Comprises the Memory Polynomial Model that models the nonlinearity order and memory effects of real PAs.Measurements of AM/AM and AM/PM distortion curves by using an FPGA development board.Phase to amplitude conversion principle.Full digital control of the test bed simulated/emulated in Matlab-Simulink with DSP Builder.

Coefficients Estimation of MPM Through LSE, ORLS and SLS for RF-PA Modeling and DPD

This paper shows and compares three techniques based on the least squared error for the estimation of the constant coefficients of the memory polynomial model used for the modeling of power amplifiers for radio-frequency and for the construction of a pre-distorter. The first technique is the conventional linear regression using the least square error method. The second technique is the order recursive least squares which can be used for exploring the most adequate nonlinearity order and memory depth of the memory polynomial model by comparing subsequent errors. The sequential least squares method is useful when the measurements of a system are coming sample by sample and the parameters of the model should be adjusted on-line. The mathematical background of the three methods is shown; as an experimental validation of this methods they were simulated in Matlab for the measurements of a 10W NPX Power Amplifier based on the transistor CLF1G0060 GaN HEMTs. An NMSE of \(-19.83\) dB was reached for the best model. Also in order to linearize the power amplifier a pre-distorter was constructed through indirect learning architecture achieving a 50 dBm spurious free dynamic range and a 25 dBc reduction in the adjacent power ratio.

Coefficient extraction for MPM using LSE, ORLS and SLS applied to RF-PA modeling

Three methods for extracting the behavioral modeling coefficients of the memory polynomial model are compared herein. The first one is the ordinary least square regression, which is widely used for adjusting model parameters; the second is the order recursive least squares, which is suitable for exploring the optimal nonlinearity order and memory depth by comparing subsequent errors while increasing the complexity of the model; and the third is called sequential least square, which is very attractive to be implemented and it only requires identifying the behavior of a power amplifier, and calculating the most accurate model coefficients for each measurement. The equations of the three methods were simulated in Matlab for the NXP 10W power amplifier with complex baseband data, and their implementation was evaluated with normalized mean square error. Also a comparison of their computational complexity based on Halstead metrics is given herein.

FPGA-Based Modeling and Design Methodology of a Digital Pre-distortion System for Power Amplifier Linearization

This paper presents the design methodology of a complete digital pre-distortion system that enables the power amplifier linearization. This system employs the memory polynomial model for its realization. The performance of the linearization is validated by using an LTE carrier signal in the band of 10 MHz. This integrated solution is capable of linearizing any real power amplifier from measurements of AM/AM and AM/PM conversion curves. Furthermore, this development test bed is able to predict the behavior and facilitates the design analysis of a pre-distorter. The experimental results are implemented employing a DSP-FPGA by using DSP Builder tool to obtain the VHDL hardware description. The proposed model shows a spurious-free dynamic range of 50 dBm and an adjacent channel power ratio reduction of 25 dBc for the NXP 10W power amplifier.

Behavioral Modeling for Power Amplifiers Comparing MPM, Wiener and Hammerstein with FPGA-Based Implementation

This paper aims on three different behavioral models with memory for radio frequency power amplifiers. These models are based on the principle of Volterra series, which were simulated in the Matlab-Simulink environment and implemented on a DSP-FPGA Altera Stratix III board. The MPM, Hammerstein and Wiener models were compared based on the distortion curves AM-AM and AM-PM of a RF-PA 10W through different levels of nonlinearity and memory depth. The results show the metric NMSE in the range of -30 dB for the three models of a NXP 10W GaN HEMT @ 3.0 GHz PA demonstrating its high accuracy during the FPGA implementation.

Test bed for low-cost measurement of AM/AM and AM/PM effects in RF PAs based on FPGA

This work presents a new proposal FPGA-based for low cost measurement of AM/AM and AM-PM distortion curves in RF Power Amplifiers through DSP Builder tool into a Development Board. A practical comparison between state-of-the-art reported works based on measurements of AM/AM and AM-PM distortion curves is discussed, in terms of low cost measurement, by using a FPGA development board and a PC to obtain the PA distortion curves adding flexibility to read the PA behavior. The main contribution is the full control of the digital PA behavior based on the phase to amplitude conversion principle simulated in Matlab-Simulink. This test bed is able to recalculate the known AM/AM and AM/PM measurements stored as LUT in a previous development board but also a real PA as DUT can be added.