A hall effect sensor like the TMAG5170D can be used in angle detection systems like an electronic shifter where redundancy is necessary for critical system operation. This document examines the advantages of the TMAG5170D approach of stacking the dual dies on top of each other compared to implementations where the dual dies are placed side by side.
All trademarks are the property of their respective owners.
When using hall effect sensors in an electronic-shifter (E-shifter) application, there can be numerous approaches to achieving the system objective of detecting relative shifter stick position. Figure 1-1 shows two possible approaches that can be explored for an e-shifter. For those two approaches a dual die sensing design can be desired for redundancy which is not uncommon for automotive applications that frequently have functional safety requirements. Preferably for a redundant system, two separate sensors measure identical results. Yet this type of redundancy often is impossible because two different sensors must occupy two different places in space. This document explores the different Hall-element placements available for dual die sensors and illustrates how a device similar to the TMAG5170D-Q1 has both die arranged for reducing the discrepancy in measurements desired in a redundant system. To substantiate this assertion, let us go through the design process for the approaches below, then show the absolute error, standard deviation in calculated angle difference, and max difference in calculated angle for various mechanical errors that can occur in assembly.
Typical development for a design iteration of a particular application can follow a flow similar to what is illustrated in Figure 1-2. In this case, the system objective is to track position in a lever similar to what can be found in E-shifter. The approaches considered are those found in Figure 1-1. The constraints for a design can be sensing device location, the board size at the sensing location, magnet size, cost, resolution, and so on. The following analysis is only constrained by the above mentioned approaches and magnet sizes similar to what can be readily purchased. Post processing magnetic field values from sweeping a magnet through the expected path of motion can help determine a good magnet size and location for better placement. From a derived point, further processing of subsequent magnet parameter sweeps can be performed to quantify the possible error that can be observed for various mechanical assembly tolerances.
Let's first consider the approach with the diametric magnet axis at the shifter stick fulcrum. For a point centered under the magnet when the magnet is centered on the axis of rotation like in Figure 2-2, the magnetic flux density (B-fields) look similar to Figure 2-1. Taking the arctan of these fields we can get a linear slope of values directly related to lever position, shown as the angle in Figure 2-1. Assuming there is a flexibility in placement of our device, we can sweep the sensor offset from the magnet in the z-axis, like in Figure 2-3, to determine the bounds of where the device is too close or too far from the magnet. Neglecting the space occupied by the device or magnet, the device is too close when the b-field saturates the sensor for more than two singular angle values over a 360° magnet rotation. The device is too far from the magnet, when more than two singular angle values are beneath the device noise floor over a 360° magnet rotation. Figure 2-3 shows that an N42, 12.7-mm diameter, 3.175-mm thick magnet does not fall outside the sensing bounds of the TMAG5170 for any z-offset from the magnet origin within -8 mm and -2.5 mm.
Other notable metrics to gauge prior to proceeding with a specific design include the impact of magnet diameter and thickness on error when the device is offset in the xy plane from the ideal location, such as captured in Figure 2-4 and Figure 2-5. Whatever tolerance can be expected with assembly in fabrication is good to use for this analysis. This paper assumes ±1.5-mm offset. For Figure 2-4 the sensor z-offset was fixed at 7.5 mm. For the Figure 2-5, the diameter was fixed at 12 mm and the sensor z-offset was adjusted such that the air gap remained constant. The air gap is the distance between the magnet surface and the sensor plane.
Figure 2-4 indicates that smaller diameters are less forgiving in error for the same offset. Based on a large group of simulation data not shown, offsets less than 10% of the magnet diameter length frequently appear to provide less than 1° error. As for magnet thickness, Figure 2-5 suggests that only a slight change in angle error is observed for different thicknesses.
Figure 2-4 and Figure 2-5 are based on measurements from a single sensor. For automotive applications, there is often a desire to have redundancy to satisfy safety requirements. As redundancy requires multiple devices, and multiple devices cannot physically occupy the same space, at least one if not both sensors will measure fields different from the lone sensor above. Also with mechanical manufacturing and assembly tolerances, there can be increased discrepancies between sensor measurements. The deviation from the ideal behavior and the discrepancy between sensors is dependent on relative sensor placement. Two common sensor placements are side by side and stacked as shown in Figure 2-6. Stacked die is the arrangement found in the TMAG5170D-Q1 and those die are typically vertically separated by 0.123 mm. For the side-by-side die, around 1 mm of separation horizontally is not uncommon.
The conditions for the subsequent data are defined in Table 3-2. Data was collected through the Texas Instruments Magnetic Sense Simulator (TIMSS) tool and post processed in excel. With mechanical assembly, there is the possibility that the sensing device can be offset from the intended location, the magnet tilted, or the magnet can not be correctly centered on the axis of rotation. Consequently, multiple separate tests were performed to see the impact of each of these possible error sources.
All parameters in Table 3-2 correspond to input parameters available in the TIMSS tool in Table 3-2.
Simulation Parameter in TIMSS | Die Configuration | Test | TIMSS Simulation Parameter Value |
---|---|---|---|
Magnet Shape | Both | All | Diametric Cylinder |
Poles | Both | All | 2 |
Magnet | Both | All | Sintered Neodymium, N42 |
Outer Diameter | Both | Sensor Offset, Magnet Tilt, Magnet Offset, Diameter Tests | 12.7 mm |
Diameter tests | 6 mm, 8 mm, 12.7 mm | ||
Height | Both | All | 3.175 mm |
Magnet Position | Both | Sensor Offset, Magnet Tilt, Diameter Tests | x = 0 mm, y= 0 mm, z = 0 mm |
Magnet Offset Test | x = 0 to 2-mm steps of 0.5 mm, y= 0 mm, z = 0 mm | ||
Magnet Angle | Both | Sensor Offset, Magnet Offset, Diameter Tests | x = 0 ° y= 0 °, z = 0 ° |
Magnet Tilt Test | x=0° to 20° steps of 5°, y=0° , z=0° | ||
Arc Length | Both | All | 360 ° |
Sensor Position x | Stacked-die | Sensor Offset X offset | -1.5 mm to 1.5 mm, steps of 0.1 mm |
Sensor Y Offset, Magnet Tilt, Magnet Offset, Diameter Tests | 0 mm | ||
Side-by-side die | Sensor Offset X offset | -1.96 mm to 1.04 mm for one Hall element and -1.04 to 1.96 for the other Hall element, steps of 0.1 mm | |
Sensor Y Offset, Magnet Tilt, Magnet Offset, Diameter Tests | -0.46 mm, 0.46 mm | ||
Sensor Position y | Both | Sensor X-offset, Magnet Tilt, Magnet Offset, Diameter Tests | 0 mm |
Sensor Y-Offset Test | -1.5 mm to 1.5 mm, steps of 0.1 mm | ||
Sensor Position z | Stacked-die | Sensor Offset Tests | -2.9385 to -4.9385 for one Hall element and -3.0615 to -5.0615 for the other Hall element, steps of 1 mm |
Magnet Tilt and Magnet Offset Tests | -4.0615 mm, -3.9385 mm | ||
Side-by-side die | Sensor Offset Tests | -3 mm to -5 mm, steps of 1 mm | |
Magnet Tilt and Magnet Offset Tests | -4 mm |