Figure 2 (Source - Reference 4)
Here, we consider a scenario around an early detection of rotary position sensor failure.
In the aircraft, position sensors are connected to determine the position of a moving part. For instance, control surface such as rudder and ailerons are used to control movement of aircraft along roll, yaw, and pitch axis. Other control surfaces such as slats, flaps, elevator, balance tab, and stabilizer help in flight controls and maneuvers.
Control surfaces are driven through a motor where rotary motion of motor is converted to longitudinal motion such that surfaces moves up or down based on the pilot’s input. The position sensors will sense the motion of control surface and provide accurate and precise position details to the control software controlling these surfaces.
A physical rotary position sensor is shown in below with rotor and stator parts (Figure 3), and electrical representation of the same is shown in Figure 4.
Input to the rotary position sensor is a 10KHz AC voltage (~12 Vrms) called as excitation voltage. Based on the movement of surfaces, two windings S1/S3 (Cosine) and S2/S4 (Sine) generates modulated Cosine and Sine waveform as shown in figure 5. These modulated sine and cosine waveforms are demodulated and converted to digital position using algorithm that are fed to control software for controlling the surface position.
Figure 6 below shows the hardware and software building blocks used here. The setup contains aircraft surface, mechanical assembly used to rotate the surface, rotary sensor to detect the position of surface, Electronic control unit used for controlling surface and intelligent device used for sensing the sensor deterioration. Here sine, cosine, and excitation voltages are tapped, passed through hardware processing, and converted to a digital form.
The Controller will process this digital data and store it into memory, besides communicating with external world through wireless a medium.
Figure 3: Rotary Position Sensor (Source: Reference 3)
Figure 4: Electrical Representation of Rotary Position sensor (Source: Reference 3)
Figure 5: Rotary Sensor – Input and Output Waveforms (Source: Reference 3)
Figure 6: Case Study – Hardware and Software Building Blocks
Table 1 below shows normal mode peak-peak values of Sine and Cosine waveform for various surface position. For simplicity, only few sample angle values are listed below table. For every measurement there are accuracy requirements defined ( for example: ± 0.2 degrees) and these can be extrapolated to Sine and Cosine values.
Table 1: Normal Mode – Surface Position to Sine / Cosine Voltage Measurement
As described above, position sensor depends on excitation, sine, and cosine values. Any change in one will affect the position sensor measurement and that will lead to incorrect movement of control surfaces and can affect flight performance and safety. The intent of this case study is to identify as early as
possible the deterioration of PSU by detecting the change in excitation, sine, and cosine values. Following are the conditions that could lead to changes (or a degradation) in their values:
The focus is more on Case 2 above which in turn could lead to Case 1 or 3 and they occur over a period of time and can be easily detected. The Winding resistance can change due to environmental conditions (Hot / Cold / Corrosion) or due to electrical circuit parameters (voltage / Current). In order to see the impact of change in winding resistance, a 500 Ohm resistance is placed in parallel to secondary windings and readings are taken as listed in Table 2 & Table 3 :
Table 2: Degraded Mode – Reduced Resistance on Sine Winding
Table 3: Degraded Mode – Reduced Resistance on Cosine Winding
By reducing Sine winding resistance by placing 500 Ohm resistance in parallel to winding, the measured resolver angle for the same surface position reduced by 3 degrees for sine windings and 6 degrees for Cosine winding. This shows clear indication that change in resistance will impact the measurement and surface movement. In order to get better correlation between resistance change and error in measurement, a potentiometer was connected in parallel to winding and reading were taken at various angles as shown in figure 7 below.
Figure 7: Change in Cosine and Sine Winding amplitude with change in resistance
As suggested in the proposed solution, for this particular scenario, the raw analog output voltage (Sine and Cosine) from position sensor and raw analog input voltage (Excitation) were routed to separate sensor that keeps track of the output voltage levels & frequency of windings S1/S3 and S2/S4 periodically. These values are stored into an intelligent device. For periodicity, surface position movement was considered as we get real position data during surface movement and stop.
The intelligent device transmits the data over wireless medium to the ground station. For simplicity in this case, computer with a wireless receiver is used as ground station and also simple database is created within computer to store the data received from ground station.
The data stored in database is fetched in real time, and after pre-processing and noise removal, the sine, cosine, and excitation values are extracted and plotted into charts. These charts include values from previous flight legs. As defined in the case study above, changes in sine and cosine values at particular angle with respect to normal values, provides the indication of change in characteristics of position sensor and change in resistance values can be derived based on the deviation of sine / cosine from normal values. After considering the measurement in-accuracy and tolerances, two thresholds were defined. The first threshold would indicate that a deterioration is observed to start. At this point an alert indication is sent out (as defined in system). Based on this indication, an early check can be performed. Second threshold is set to indicate position sensor is continuously degrading and needs repair or replacement before it fails.
As more and more data is collected over a period time, the data values plotted provides a trend and distribution (as seen in image 8 below) and more insight into data. Prediction algorithms applied on these data to determine the remaining usable life.
Figure 8: Case study – Data values, trend charts