Resource optimal FPGA implementation of a Least-Squares-Estimator for fast and robust online measurement of current slope and absolute value

Abstract

This paper presents an improved Least-Squares-Estimator for the fast, non-recursive online measurement of the first derivative and absolute value of a piecewise linear oversampled current signal. Core topic is the partially online calculation of the pseudoinverse matrix in combination with an optimized small lookup table, minimizing memory requirements significantly.

Design guideline for PCB integrated, high bandwidth, current slope sensing based on a planar Rogowski coil

“…In recent years, new methods have been proposed that use the PWM excited current slopes of the inverter instead of additional test signals [1][2][3]. To measure the current slope, one possible solution is the derivation of the measured current signal using a least squares algorithm [4]. This requires high bandwidth current sensors, fast and precise Analog-Digital-Converters (ADC) and intense signal processing for stable calculation of the derivative and filtering.…”

“…This requires high bandwidth current sensors, fast and precise Analog-Digital-Converters (ADC) and intense signal processing for stable calculation of the derivative and filtering. Specialized Field Programmable Gate Array (FPGA) implementation reduces hardware resources significantly and minimizes calculation time to one FPGA cycle [4]. A different approach is the usage of a dedicated current derivative sensor based on a Rogowski coil [5].…”