The development of the self-diagnostic accelerometer (SDA) is important to both reducing the in-flight shutdowns (IFSD) rate — and hence reducing the rate at which this component failure type can put an aircraft in jeopardy — and also as a critical enabling technology for future automated malfunction diagnostic systems. Critical sensors, such as engine sensors, are inaccessible to the operator during typical operation due to safety concerns and enclosed environment. The SDA can diagnose the sensor in-flight and remotely with minimal interference with the typical operation of the sensor. The SDA system utilizes programmed health algorithms that can automatically determine the health, therefore increasing the precision in diagnosing sensor faults by removing the erroneous perspective and opinions of a human operator. The health of the sensor could also be determined immediately, which would remove its erroneous effect on a system that depends on the sensor.
Improvements to the SDA system have been accomplished with a smaller, more robust FPGA (field programmable gate Authorized distributor array) system that reduces the size, cost, and power while also increasing the diagnostic system’s customization. The self-diagnostic accelerometer field programmable gate array (SDA FPGA) is a sensor system that utilizes a small and efficient, yet customizable electronics system to capture accelerometer diagnostic data. The diagnostic components of the system actively determine the accelerometer structural health and attachment condition. The SDA FPGA system sends an electrical signal to the accelerometer’s piezoelectric crystal. The physical state of the crystal and its surroundings influence the electrical response that is correspondingly received by the SDA FPGA.
Changes in the response are correlated to changes in the accelerometer health and attachment condition. Newly developed health algorithms were programmed into the SDA FPGA and utilize cross-correlation pattern recognition to discriminate a healthy from a faulty SDA. The results of the diagnostics are reported in real time on the LCD (liquid crystal display) of the FPGA. Recent aircraft ground testing demonstrated for the first time the robustness of the SDA in an engine environment characterized by high vibration levels.
The purpose of the SDA FPGA is to automatically determine the accelerometer structural health and attachment condition using an electronics system that is smaller, more energy efficient, and more cost effective than previously used diagnostic tools. The programmed FPGA utilizes cross-correlation health algorithms to diagnostically interrogate and automatically determine the health and attachment of the sensor in real time.
The SDA consists of the sensor, the FPGA, signal conditioning electronics, connecting cables, and power supply. The sensor is a piezoelectric charge accelerometer, and the FPGA is a commercially produced developers kit with an input/output daughter card. The signal conditioning electronics is a NASA-designed circuit containing filters and amplifiers to improve the output and input signals to the FPGA. Connecting cables used accelerometer-grade cables. The 12-V power supply feeds the signal conditioning electronics, and a separate power supply feeds the FPGA system.
The FPGA diagnostics system generates a sinusoidal wave that sweeps from 30 to 80 kHz. The FPGA outputs the diagnostic signal, which goes through a filter that reduces noise in the signal. The cleaned-up diagnostic signal is then amplified by an instrument amplifier and is fed into the accelerometer. The response is then coupled into the capacitor. The coupled response is filtered and amplified, then read through the FPGA input. The signal response consists of a signal pattern with resonant frequencies within the 30-to-80-kHz range, depending on the health and attachment of the sensor. The signal response for nominal sensor operation as well as sensor faults is recorded as references. These references are then cross-correlated with the existing signal response in order to diagnose the condition of the sensor in real time.
This work was done by Roger Tokars and John Lekki of Glenn Research Center.