Method for providing an inductive position sensor, IPS, and corresponding IPS sensor
The method improves inductive position sensors by using basis functions and neural networks to optimize coil shapes, reducing errors and enhancing design efficiency, making it suitable for diverse sensor geometries.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- EMC GEMS SRL
- Filing Date
- 2026-01-12
- Publication Date
- 2026-07-16
AI Technical Summary
Existing inductive position sensors face challenges such as high error rates, difficulty in design constraints, slow convergence speed, high memory usage, and difficulty in balancing convergence speed with design feasibility, particularly due to non-idealities in sensor implementation.
A method utilizing basis functions, such as sine and cosine, to regularize coil shapes and reduce non-idealities, combined with an artificial neural network to optimize sensor design parameters, and a cloud-based platform for end-to-end design and manufacturing, which includes an analytical calculation of the Jacobian matrix to iteratively refine coil shapes.
This approach reduces sensor errors from 5% to 1%, enhances design efficiency, and minimizes memory usage while being flexible across various sensor geometries, ensuring high performance and reliability.
Smart Images
Figure IB2026050221_16072026_PF_FP_ABST
Abstract
Description
[0001] "Method for providing an inductive position sensor, IPS, and corresponding IPS sensor"
[0002] DESCRIPTION TEXT
[0003] Technical field
[0004] The description relates to methods and systems for manufacturing position sensors, such as ratiometric sensors in which the position is detected by means of the voltage on circuit tracks sensitive to electromagnetic induction provided on printed circuit boards (PCBs).
[0005] One or more embodiments maybe applied in fields such as industrial automation, automotive, and consumer electronics.
[0006] Technological background
[0007] Inductive position sensors (IPS) comprise electromechanical components that employ electromagnetic induction.
[0008] IPS devices are known, for example, from one or more of the following documents:
[0009] - US 4737698 A discusses a sensor with an actuating coil to generate a front field, a conductive screen movable relative to a detection coil, and wherein, in the presence of the actuating field, eddy currents in the screen generate a counterfield that opposes the main field, changing the induction in the detection coil according to the relative displacement;
[0010] - B. Aschenbrenner and B. G. Zagar, "Analysis and Validation of a Planar High-Frequency Contactless Absolute Inductive Position Sensor," in IEEE Transactions on Instrumentation and Measurement, vol. 64, no.3, pp.
[0011] 768-775, March 2015, doi: 10.1109 / TIM.2014.2348631 discusses a precise, reliable, and low-cost inductive absolute position measurement system, in which the sensor operates on principles similar to those of contactless resolvers and consists of a printed circuit board with a rectangular antenna and a passive LC resonant circuit;
[0012] - Shao, Lingmin, "Automotive Inductive Position Sensor" (2017), Electronic Thesis and Dissertation Repository, 4569, https: / / ir.lib.uwo.ca / etd / 4569 discusses inductive angular position sensors(IAPS) widely used for high-precision, low-cost angular position detection in harsh automotive environments, for which a design optimization method based on response surface methodology is developed and used to create three types of sensors: a miniaturized sensor with a resonant rotor, a torque sensor for detecting the torsional angle, and a passive sensor designed to reduce energy consumption and electromagnetic emissions; and
[0013] - M. Passarotto, G. Qama, and R. Specogna, "A Fast and Efficient Simulation Method for Inductive Position Sensors Design," 2019 IEEE SENSORS, Montreal, QC, Canada, 2019, pp. 1-4, doi: 10.1109 / SENS0RS43011.2019.8956502 discusses an IPS coil design method that aims to efficiently evaluate the performance of each sensor prior to industrial production and application, in which the simulation is based on the sequential solution of various eddy current problems in the frequency domain, using a fast and efficient surface integral formulation instead of the more common commercial finite element codes (FE) codes; and
[0014] - Italian patent application no. 10202100023288 filed by the Applicant, which discusses a design method and related apparatus and computer program product .
[0015] Several techniques have been proposed in the literature to overcome some existing limitations with regard to the reduction of errors due to nonidealities in the implementation of IPS sensors, such as in the following documents:
[0016] - Lin Ye et al.: "Optimization of inductive angle sensor using response surface methodology and finite element method," Measurement, Volume 48, 2014, Pages 252-262, ISSN 0263-2241, doi: io.ioi6 / j.measurement.2oi3.ii.oi7 discusses a design method based on response surface methodology and the Maxwell finite element method, in which, by selecting key sensor parameters and setting the initial search domain, simulation experiments are completed using Maxwell software according to the central composite design;
[0017] - Ye, Lin et al.: “Nonlinearity Analysis and Parameters Optimization for an Inductive Angle Sensor”, Sensors 2014, 14, 4111-4125, doi: 10.3390 / S140304111 discusses the use of the finite element method (FEM) and particle swarm optimization (PSO) to perform a nonlinearity analysisbased on parameter optimization to design an inductive angle sensor;
[0018] - Dauth, R.A. et al.: “An Effective Method to Model and Simulate the Behavior of Inductive Angle Encoders”, Sensors 2022, 22, 7804, doi: 10.3390 / S22207804 discusses a sensor based on coupled coils, with position angle information modulated on the voltage induced in the receiving coils; since eddy currents complicate the calculation of physical quantities, a 3D finite element simulation is used, which requires long computation times; to reduce the computational effort, a method is proposed that uses regression models to replace simulations;
[0019] - US 7319319 B2 discusses a sensor that uses a digitally modulated excitation signal to detect parameters such as relative position and environmental factors (temperature and humidity), wherein the sensor comprises an excitation coil, a sensor coil, and a signal processor for processing the induced signals;
[0020] - WO 2020 / 171840 Al discusses a linear inductive position sensor that uses two detection coils to measure the position of a conductive target, wherein the coils are designed to induce opposing magnetic fields that change in the presence of a conductive target, allowing its position to be determined.
[0021] Notwithstanding the extensive activity in this area, as evidenced, for example, by various documents listed above, further improved solutions are desirable.
[0022] Existing solutions show one or more of the following disadvantages: use of a large number of simulations,
[0023] difficulty in providing viable solutions in the presence of design constraints to which the sensor is subject (for example, the footprint or air gap width between the target object and the PCB often cannot be chosen according to the optimization suggestion);
[0024] difficulty in effectively reducing error;
[0025] low convergence speed of optimization methods;
[0026] high impact in terms of memory occupied by simulations; and difficulty in determining trade-offs between convergence speed and design feasibility.
[0027] Object and scopeAn object of one or more embodiments is to help provide such an improved solution.
[0028] According to one or more embodiments, such object can be achieved by means of a method having the characteristics set forth in the claims that follow.
[0029] One or more embodiments may relate to a corresponding sensor. An inductive position sensor, IPS, may be an example of such a sensor.
[0030] One or more embodiments may include a computer program product loadable into the memory of at least one processing circuit (e.g., a computer) and including portions of software code for executing the steps of the method when the product is executed on at least one processing circuit. As used herein, a reference to such a computer program is intended to be equivalent to a reference to a computer-readable medium containing instructions for controlling the processing system to coordinate an implementation of the method according to one or more embodiments. A reference to "at least one computer" is intended to highlight the possibility that one or more embodiments may be implemented in a modular and / or distributed form.
[0031] The claims are an integral part of the technical teaching provided herein with reference to the embodiments.
[0032] Embodiments have the advantage of providing a fast and reliable technique for improving the shape of the tracks / coils / windings of the receiving circuit assembly of an inductive sensor.
[0033] One or more embodiments provide an integrated cloud platform for the production of inductive sensors with improved performance.
[0034] One or more embodiments facilitate the reduction of the impact of non-idealities in the inductive sensor as manufactured compared to the design.
[0035] One or more embodiments are flexible and applicable to a wide range of inductive sensor geometries, both linear and rotary, as well as their respective target objects.
[0036] One or more embodiments have at least one of the following advantages:
[0037] the use of basis functions (preferably orthogonal, such as sine and cosine) allows for natural regularization of the shape, counteracting the riskof obtaining "zig-zag" shapes and obtaining continuous-shaped coils; reduction in design time, since the solution can work from a single electromagnetic simulation;
[0038] memory savings, as the resulting system is smaller than those known in the art;
[0039] flexibility of the method in different contexts, taking into account any offsets without the need for adaptations to specific cases; this is achieved in particular by calculating the Jacobian matrix Jh analytically, i.e., by performing a directional derivative.
[0040] Brief description of various views of the drawings
[0041] One or more embodiments will now be described, purely by way of example, with reference to the accompanying figures, in which:
[0042] Figure 1 is an example of a linear inductive sensor;
[0043] Figure 2, comprisingcomprises portions a) and b), is an example of an ideal design of portions of the sensor exemplified in Figure 1;
[0044] Figure 3 is a flowchart exemplary of the method according to the present description;
[0045] Figure 4 is a further flowchart illustrating operations of the method exemplified in Figure 3;
[0046] Figure 5 is an exemplary diagram of principles underlying one or more embodiments;
[0047] Figure 6 is an exemplary diagram of principles underlying one or more embodiments;
[0048] Figure 7 is a diagram exemplary of a sensor made using the method exemplified in Figure 3, and
[0049] Figure 8 is a diagram exemplary of iterations of the method exemplified in Figure 4.
[0050] Detailed description of examples of embodiments
[0051] In the following description, one or more specific details are illustrated in order to provide a thorough understanding of examples of embodiments of this description. The embodiments may be obtained without one or more of the specific details or with other methods, components, materials, etc. In other cases, known operations, materials, orstructures are not illustrated or described in detail so that certain aspects of the embodiments will not be made unclear.
[0052] A reference to "an embodiment" in the context of this description is intended to indicate that a particular configuration, structure, or feature described with reference to the embodiment is included in at least one embodiment. Therefore, phrases such as "in an embodiment" that may appear in one or more places in this description do not necessarily refer to the same embodiment.
[0053] Furthermore, particular configurations, structures, or features may be combined in any suitable manner in one or more embodiments.
[0054] The references used herein are provided for convenience only and therefore do not define the scope of protection or the scope of the embodiments.
[0055] For simplicity, one or more embodiments are discussed below primarily with reference to linear IPS devices in which the conductive surface of the target is mostly quadrangular in shape, being otherwise understood that this type of IPS is purely illustrative and in no way limiting. One or more embodiments may also apply to rotating IPSs equipped with a target object with a semi-disc conductive surface, arc IPSs for detecting rotational displacement, arc IPSs, redundant IPS configurations, dual IPS configurations, and / or theoretically using any shape for the target and its reflective area (e.g., multilobed, multilobed with an inner ring, flat, square, and so on).
[0056] In various embodiments, portions of the sensor designed according to the method of the present invention may be implemented in a field-programmable gate array (FPGA) or in an application-specific integrated circuit (ASIC). As appreciated by those skilled in the art, various functions of the circuit elements may also be implemented as processing steps in a software program. Such software maybe employed, for example, in a digital signal processor, a microcontroller, or a general-purpose computer.
[0057] Figure 1 is an exemplary diagram of an IPS sensor (e.g., linear) io comprising:
[0058] a target object 14, for example, having a quadrangular shape, configured to be coupled to an electromechanical part (e.g., the shaft of an electric motor) and comprising a lower reflective area (not visible in Figurei) configured to reflect electromagnetic waves that reach it, for example, a metallic / conductive surface;
[0059] a transceiver printed circuit 20 comprising:
[0060] transmitting circuit track comprising a set of transmitting terminals TX configured to receive an oscillating signal at a frequency fo from an oscillator and coupled to a transmitting circuit path 200 (also referred to as a track, coil, loop, or winding) configured to transmit an electromagnetic wave to the target object 14 based on the oscillating signal at frequency fo received, and
[0061] receiving circuit track comprising a set of receiving terminals RxSin, RxCos coupled to at least one set of receiving circuit paths (also referred to as tracks, coils, loops, or windings) Sin202, Sin2O4, Sin2o6, and C0S202, C0S204 configured to receive respective reflected echo signals (in a frequency band around the transmission frequency fo) from the reflecting surface of the target object 14 when affected by the electromagnetic wave transmitted at frequency fo. For example, the receiving terminals RxSin, RxCos are coupled to a processing device (such as a microcontroller) configured to apply signal processing to the detected echo signal, in a manner known per se, providing a set of indicator signals to user circuits, for example to apply feedback control (in a manner known per se) to an electric motor to whose shaft the target object 14 is coupled.
[0062] In one or more embodiments, the windings of the transmitting circuitry track 200 and the receiving circuitry track Sin202, Sin2O4, Sin2o6, and C0S202, C0S204 in the transceiver printed circuit 20 are formed by etching or printing conductive traces on an upper layer of a printed circuit board (PCB).
[0063] A circuit for detecting the voltage at the receiving terminals RxCos, RxSin terminals and / or a sinusoidal alternating current source to be coupled to the TX transmitting nodes can be implemented in one or more semiconductor dies soldered to a printed circuit on which the coils of the transceiver 20 are traced to electrically couple the assembly of coils 200, Sin202, Sin2O4, Sin2o6, and C0S202, C0S204 assembly to one or more integrated circuit dies. The interconnections between one or more semiconductor dies may extend over different outer faces and / or withinother layers of the PCB.
[0064] The ability to integrate the circuit elements of the IPS io device onto a PCB 20 circuit makes IPS devices very compact and economical.
[0065] IPS devices are currently also referred to as "ratiometric" devices because the position of the target object 14 (and therefore of the drive shaft connected to it) can be determined based on the voltage ratio at the RxCos and RxSin terminals of the two receiving coils of the receiving circuit track.
[0066] For example, an oscillator produces a sinusoidal electric current at an initial frequency that flows in the track 200 of the transmitting circuit track; as a result, a variable magnetic field is produced around it. When a conductive object (commonly referred to as a target) is brought close to the transmitter 200, induced currents are generated inside it due to Faraday-Neumann's law. These induced currents generate a magnetic field that disturbs the field produced by transmitter 200: by reading this field or the inductance of transmitter track 200, it is possible to deduce the distance, and therefore the position, of the target relative to transmitter 200. The field disturbance can be read by measuring a voltage difference (with a voltmeter) at the receiving terminals RxSin and RxCos.
[0067] As exemplified in portion a) of Figure 2, a first set C0S202, C0S204 of receiving circuit tracks coupled to a first set of receiving terminals RxCos follows a cosine curve and is configured to ideally provide a voltage Vcos at the ends of the RxCos terminals, which can be expressed as:
[0068] VCOS=VCOS2O2+VCOS2O4=-V+V=O
[0069] As exemplified in portion b) of Figure 2, conceptually a second set Sin202, Sin2O4, Sin2o6 of receiving circuit tracks coupled to a second set of receiving terminals RxSin follows a sinusoidal pattern and is configured to ideally provide a voltage Vsin to the terminals RxSin, which can be expressed as:
[0070] Vsin=VSin202+VSin204+Vsin206=V / 2-V+V / 2=0 Under ideal conditions, a sensor outputs a value corresponding to the "ground truth" value of the measured quantity. Furthermore, based on this ideal model, the output provided by sensor 10 follows a theoretical or ideal curve that relates input and output within its operating range. For example, for a ratiometric position sensor, the position of target 14 is a function (inverse trigonometric) of the ratio between the voltages VSin / VCosdetected at nodes RxSin, RxCos of the receiving coils Sin202, Sin2O4, Sin2o6 and C0S202, C0S204, respectively.
[0071] The ideal behavior discussed above is difficult to find in practice: for example, the presence of the receiving terminals RxCos and RxSin causes the symmetry between the coils to be broken at least at their position; this is a first element of non-ideality that is difficult to contemplate in the model. Furthermore, it can be difficult to physically realize the tracks with sufficient precision.
[0072] The design of a sensor (or design) means specifying (in terms of layout and spatial coordinates on the printed circuit board) the geometric development of the paths taken by the TX transmission and RxSin and RxCos receiving coils, together with the geometry, position, and trajectory of the target.
[0073] If the shape of the transceiver printed circuit 20 is determined considering the ideal sensor (for example, using sine and cosine curves as the shape of the receiving circuit tracks Sin202, Sin2O4, Sin2o6, and C0S202, C0S204), this does not allow for the presence of non-idealities that can, once implemented in practice, compromise the performance of the sensor.
[0074] As appreciated by those skilled in the art, once a sensor is manufactured, it has non-ideal behavior that produces a deviation of the output from the "ground truth" value; the difference between the detected value and the "ground truth" value represents the sensor error.
[0075] As appreciated by those skilled in the art, the term 'linearity' refers to the property of a measuring instrument to output (read) values that can be linearly related to a quantity to be measured.
[0076] As appreciated by those of skill in the art, in the field of measurement systems, the full scale represents the maximum value that can be measured by a given measuring instrument.
[0077] An IPS 10 sensor based on an ideal model may, in practice, have a linearity error of more than 5% at full scale. In rotary IPS sensors, this greatly worsens the error, while in linear IPS sensors it can even compromise their operation.
[0078] As exemplified in Figure 3, the method of designing and manufacturing an IPS sensor as discussed herein can be implemented in aweb-based mode as an end-to-end (engineering-to-manufacturing) cloud platform for the creation of an IPS sensor with user-customized parameters.
[0079] For example, the method exemplified in Figure 3 comprises the following operations:
[0080] block 300: receiving, for example from an online web service user or through other communication channels, a set of constraint specification values S for the IPS sensor to be designed; for example, these S specifications may include information on the dimensions of the IPS sensor (if it is to be mounted around a 20 mm shaft of a brushless motor, for example, the radius / short side of the sensor maybe greater than 20 mm);
[0081] block 302: based on the set of specifications S received, apply an artificial neural network (ANN) trained using a training data set TD comprising a library comprising a plurality of different sensor designs to provide, based on the parameters of the given specifications S, a set of values for the remaining parameters P of the IPS sensor design (for example, the total number of variables for a sensor ranges from 10 to 20 and may include beam / side dimensions, air gap, i.e., distance between target and PCB, and so on, in a manner known per se);
[0082] block 304: the design of the coils / wires of the transceiver printed circuit 20 (e.g., provided by the user as a Gerber file) is then coupled to the chip part (e.g., starting from one of the chips for PCBs sold by suppliers selectable by the user via the web platform);
[0083] block 306: refinement method is performed on the shape of the coils of the transceiver printed circuit 20, in particular the receiving circuit tracks C0S202, C0S204, Sin202, Sin2O4, Sin2o6 as discussed below, providing an improved DRX design of the transceiver printed circuit 20;
[0084] block 308: starting from the IPS sensor design obtained in block 304 and the improved DRX design of the receiving coils 20 provided in block 306, a Gerber GF file is generated representing the printed circuit board (PCB) of the IPS sensor;
[0085] block 310: the system sends the Gerber file GF to a supplier (e.g., automatically selected by the platform) via known means of communication such as an application programming interface (API), which physically produces one or more PP prototypes, which can then be sent to the user as a sample of the IPS sensor with customized parameters.For example, a training database for training the neural network processing of block 302 can be generated with the results of a plurality of template simulations of a plurality of different sensor designs, for example generated by providing random parameters.
[0086] It has been observed that, given that a linear IPS sensor can have an error of 10%, determining the parameters in block 302 facilitates the reduction of the error, for example, reducing the error to 1%. In general, having the suggestions of the neural network 302 available to determine the (not already constrained) parameters P of the sensor design helps to reduce the design time of the sensor.
[0087] As appreciated by those skilled in the art, artificial neural network processing refers to a method implemented on a computer that comprises a set of processing layers that perform operations on signals based on parameters (currently referred to as weights, activation functions, kernels, etc.) that are determined (automatically or guided by user interaction) during a "training phase." Based on the training signals used during this phase, the parameters (e.g., weights) of the artificial neural network processing are "calibrated." For example, a network that is trained to classify a set of images whose assignment to a certain class of objects (i.e., classification) is known a priori completes the training phase when it outputs classification predictions that substantially coincide (within a certain margin of error and processing time) with the classes known a priori. This training ensures that the processing provides reasonably reliable predictive results even when artificial neural network processing is applied to signals whose classification is unknown a priori (a phase currently referred to as "inference").
[0088] As exemplified here in Figures 3 and 4, operation 306 of improving the design of the coils of the receiving circuit tracks C0S202, C0S204, Sin202, Sin2O4, Sin2o6 employs one or more electromagnetic simulation tools applied to prototyping.
[0089] As will be appreciated by those skilled in the art, for a linear IPS sensor comprising two receiving circuit tracks, the ideal shape is that already discussed in relation to Figures 1 and 2, i.e., a first period of the sine and cosine functions that are mirrored with respect to a longitudinal axis that cuts the PCB lengthwise.As appreciated by those skilled in the art, to avoid intersections between paths, part of the paths is moved to different layers of the PCB (e.g., a top and bottom surfaces of the board).
[0090] As exemplified in Figure 4, the design method for the C0S202, C0S204, Sin202, Sin2O4, and Sin2o6 receiving circuit tracks comprises: block 400: an initial design D of the coils of the IPS sensor receiving circuit tracks is calculated from a preset template that adapts the shape (e.g., calculated according to an ideal model using sine and cosine functions) of the coils to the parameters P, S of the sensor geometry (e.g., length, width, thickness of the tracks, etc.);
[0091] block 402: based on the initial design D received, a shape matrix of shapes F is constructed which constitutes a basis of spatial shapes whose combination (e.g., linear) will provide the shape with improved flux of the receiving coils, as discussed below;
[0092] block 404: a measurement of the signals detected at the RXCOS, RXSIN terminals of sensor 10 is performed iteratively as the position of target 14 varies (e.g., for a finite number of target positions) while a signal is emitted from transmitter 200; for example, the measurement can be performed using electromagnetic simulation software discussed in Italian patent application no. 102023000005931 or other software, thereby providing the simulation results downstream of the simulation of a set of current values DJ[P] passing through transmitter TX, 200 and the target 14 of the simulated IPS sensor;
[0093] block 410: initializing a counter of the iterations of the simulation in block 404, and for each iteration:
[0094] block 412: applying post-processing to the result of the DJ[P] current densities in the elements of the calculation grid at the h-th iteration, analytically calculating the values of a magnetic flux vector h concatenated with the receiving circuitry, obtained for each P position of the elements of the calculation grid (this flux is proportional to the voltage Vi measured by the sensor according to the expression Vh =-10 h ) and also provides a Jacobian matrix Jh; the calculation of the flux and the construction of the Jacobian matrix are discussed in more detail below;
[0095] block 414: calculating an h-th flux error eh for each position Palong the IPS sensor as the difference between the simulated flux vector h with the current geometry and an ideal sinusoidal flux vector T, i.e., a theoretical flux OT calculated for a sensor free from non-idealities; for classic two-receiver sensors. The two fluxes are the first period of the sine and cosine functions, for which the maximum value of the sine and cosine functions is determined by the peak value of the initial fluxes O0(for example, an average or maximum and minimum value can be selected between the two);
[0096] block 420: verifying whether a norm 11 en| | of the h-th flux error eh (for example, an L2 norm, known per se) is lower than a threshold value s; if verified, the objective has been achieved and the end is reached, providing the improved DRX design of at least one of the receiving circuit track RxSin, RxCos of sensor 10 as output (for example, as the output of stage 306 of the procedure exemplified in Figure 3);
[0097] block 422: in the event of a negative verification in block 420 (i.e., if the norm | |eh| I is greater than or equal to the threshold value s), solving a system obtained as the product of the h-th Jacobian matrix Jh, the shape matrix F, and an additional system of unknowns matrix DAh that comprises a number M of multiplicative factors of the linear combination of shapes with which the shape of each of the receiving coils is constructed, as discussed below; the product performed in block 422 can therefore be expressed as: [Jh■ F] ■ AAh= eh;
[0098] block 424: determining a node displacement vector Dm as the product of the shape matrix F and the system unknowns matrix DAh , i.e., Dnh =FDAh; by adding all the Dm to the previous shape (e.g., the initial shape D), an updated shape of the receiving coils is obtained;
[0099] block 430: incrementing the counter h and iterate the procedure 412, 414, 420, 422, 424 until the condition 11 eh| | <e is verified, which ends the procedure for determining the positions of the coil nodes. It should be noted that the order of operations in the procedure is purely illustrative and not limiting: in alternative scenarios, the operation of providing the shape matrix F to block 402 could follow block 404.
[0100] As will be appreciated by those skilled in the art, the Jacobian matrixof a vector quantity (such as a matrix) defined as a function of a vector is obtained by calculating the partial derivatives of each component of the vector quantity with respect to each component of the vector. It is organized by writing each row / column as the derivatives of a scalar function of the elements of the matrix with respect to all the variables of the vector.
[0101] For example, the shape matrix F comprises blocks, each representing one of the three spatial coordinates, concatenated vertically or horizontally, and can be expressed as:
[0102] F = [
[0103] Fx,
[0104] Fy,
[0105] Fz
[0106] ]
[0107] For example, the Jacobian matrix Jh comprises blocks, each representing one of the three spatial coordinates, for example concatenated in the opposite direction to that of the shape matrix F, and can be expressed as:
[0108] Jh = [Jx, Jy, Jz]
[0109] For example, given the flux at block 412 as a known term and wanting to obtain a measured flux increase of 10%, the following linear system can be expressed:
[0110] dPhi = phi* 1.1 - phi
[0111] At this point, given the shape matrix and the Jacobian, in an example case where only 4 nodes of a receiving coil are considered, a numerical problem is solved that can be expressed as:
[0112] dF = (J*F)\dPhi
[0113] dn_flat = F*dF
[0114] dn = [dn_flat(i:4), dn_flat(5:8), dn_flat(9:i2)]
[0115] In the example case where only four nodes are present, a node displacement vector can therefore be obtained, which can be expressed as:
[0116] dn: [ 0.0390 o o
[0117] 0.0000 -0.1922 o
[0118] -0.0390 -0.0000 o
[0119] -0.0000 0.1922 o]
[0120] As shown in Figure 4, the values of the template variables initializedin block 400 (for example, determined using the cloud-based design platform shown in Figure 3) comprise:
[0121] geometry of target(s) 14;
[0122] geometry of the TX transmitting coils;
[0123] geometry of the receiving coils (e.g., total length, position of the receiving nodes RXsin, RXcos, width, type of basis functions such as sine / cosine or Chebyshev polynomials);
[0124] geometry of any conductors located near the sensor.
[0125] As shown in Figure 4, these geometric elements are discretized using a mesh (e.g., 1D, 2D, or 3D depending on whether the conductor is linear, planar, or solid). For example, for receiving coils that are not traversed by currents, a 1D linear model is used.
[0126] To use the simulation system, the shape of the receiving circuitry is discretized as positions, for each receiving loop, on a Cartesian plane (matching the plane of the PCB board) of a set of coordinates (xn, yn) of N points that couple adjacent segments, currently called nodes, of the receiving circuit tracks C0S202, C0S204, Sin202, Sin2O4, Sin2o6. The shape of the C0S202, C0S204, Sin202, Sin2O4, Sin2o6 coils is then determined by successive approximations of the position of these N nodes with respect to an initial design, as discussed below.
[0127] As mentioned, the receiving loops can be modeled as consisting of a combination of consecutive segments forming a broken line, in which the consecutive segments are connected by a set of N vertices or nodes that define the broken line in a given receiving loop (e.g., the cosine loop C0S202, C0S204).
[0128] As exemplified herein, the shape matrix F processed at block 402 comprises a first integer P of elements (e.g., rows) corresponding to points of the receiving coil and a second integer M of fields (e.g., columns) corresponding to a certain number of shapes.
[0129] For example, as exemplified in Figure 5, the shape matrix F may comprise the coordinates of the points of a set of shapes in the space of a set of harmonics of the sine and cosine functions, for example, the first ten harmonics.
[0130] As exemplified in Figure 5:
[0131] portion a) of Figure 5 is a plot exemplary of a zero-order basisfunction;
[0132] portion b) of Figure 5 is a plot exemplary of a first-order harmonic of the sine function;
[0133] portion c) of Figure 5 is a plot exemplary of a second-order harmonic of the sine function;
[0134] portion d) of Figure 5 is a plot exemplary of a first-order harmonic of the cosine function;
[0135] portion e) of Figure 5 is a plot exemplary of a second-order harmonic of the cosine function.
[0136] For simplicity, one or more embodiments are discussed below, mainly in relation to the use of a type of harmonic functions of trigonometric functions for the construction of the shape matrix of shapes, being otherwise understood that this case is purely illustrative and not limiting.
[0137] One or more embodiments may conceptually employ any type of function (optionally, orthogonal to each other) in place of sine and cosine, such as Chebyshev polynomials (known per se).
[0138] As exemplified in Figure 6, it is possible to consider a longitudinal axis x starting from one end (e.g., left) on a first face (e.g., upper) of a PCB containing an IPS 10 sensor and which, upon reaching the opposite end of the axis origin on the same first face, then continues on the second face (e.g., lower) of the PCB comprising the IPS sensor and extends to the end on the same side of the axis origin but on the second face itself. In this reference system, which is convenient for defining the positions (x, y) of the receiving circuit track nodes with a one-to-one relationship with the Cartesian axes x, y thus defined. For example, it is therefore possible to transpose the curves exemplified in portions a) to e) of Figure 5 with those exemplified in the respective portions a) to e) in Figure 6.
[0139] In addition to or as an alternative to the regular shapes shown in Figures 5 and 6, to adapt the shapes of the portions of the coils close to the receiving nodes RxSin, RxCos located on the chip to provide a reading of the voltage induced in the receiving coils during operation of the IPS 10 sensor.
[0140] As exemplified in Figures 5 and 6, the amplitude of the shapes can be defined arbitrarily, for example, it can be set to a unit value, i.e., an amplitude equal to one.
[0141] As exemplified here, the waveform matrix F is configured to encodethe basis of the waveforms that will give the shape of the receiving circuit tracks RXSin and RxCos with their linear combination added to the initial waveform set by the user.
[0142] In light of what has been discussed above regarding setting of the axes on the two sides of a PCB, for linear sensors, the position of the nodes can also be reduced to a vector of the y-coordinate alone in sequence according to a travel order (for example, from the origin to the end of the x-axis) along the PCB. Alternatively, or in addition, for rotary or arc sensors, the F matrix (and consequently, also the solution with respect to the displacement of the nodes) is going to be calculated for each of the coordinates of the plane (x, y) or (x, z) in which the PCB is located.
[0143] As exemplified here, in the case of nodes that define outputs, the value of their respective positions in the shape matrix F is set to zero in block 402 in order to counteract the risk that the iterative method modifies these nodes.
[0144] In conclusion, operation 402 of constructing the shape matrix F comprises the operation of ordering vectors of the y coordinates of the various nodes of each receiving circuit track RxSin, RxCos in sequential order, so as to obtain the values in the columns of the shape matrix F.
[0145] As exemplified here, for electromagnetic simulations in block 404, it is possible to use a magneto-quasistatic model or even a full-Maxwell model, in a manner known per se.
[0146] As exemplified in Figure 4, block 404 comprises collecting, for example by performing measurements and / or electromagnetic simulations, a number P of density maps of currents induced in the IPS sensor target, one for each k-th position of the target with k ranging from o to P. For example, block 404 comprises storing the results of the set of electromagnetic simulations as a matrix of current density DJ[P] on each segment that discretizes the conductors, which represents the output value of the electromagnetic simulation.
[0147] The operations of block 412 exemplified in Figure 4 and comprising post-processing of the result of the current densities DJ[P], including analytically calculating the values of a magnetic flux vector h and a Jacobian matrix Jh, are discussed below.
[0148] As can be appreciated by those skilled in the art, the linked flux of awire conductor (which serves as a model for each receiving coil C0S202, C0S204, Sin202, Sin2O4, Sin2o6) for each simulated position pe [0,P] for the target (e.g., P=ioo positions of target 14 along the x-axis of the upper face of the PCB with a pitch of L / 100 between two adjacent positions, where L is the maximum length of the upper face of the PCB), the circulation of the magnetic vector potential can be expressed as:
[0149] N
[0150] 4>p = J A ■ dl = At■ It
[0151]
[0152] Li=l
[0153] where
[0154] Ai is a sampled magnetic vector potential, for example at the center of gravity of an i-th segment that makes up the trace of the receiving loop on the PCB,
[0155] U is a vector of an i-th segment of a receiving loop (e.g., C0S202) that can be expressed as:
[0156] b ~niA ~wiB
[0157] where
[0158] UiA and UiB are the coordinates of the two nodes (e.g., left end and right end) between which the i-th segment U extends.
[0159] For example, each i-th centroid bi for each i-th segment U of a receiving loop (e.g., C0S202) can be expressed as:
[0160] _ niA+ niB
[0161] b‘=—r~
[0162] where m and n2are again coordinate vectors of two nodes niA, niB corresponding to the ends of a segment U.
[0163] As a result, during the post-processing phase 412, it is possible to calculate the magnetic vector potential Ai from the 1D wire currents DJ[P] using, for example, the following expression, valid under the assumption that the conductors are far apart ("far -field approximation"):
[0164]
[0165] where
[0166] E is the number of segments with known current (in the TX and target coils);
[0167] Ij represents the current induced in the j -th segment Zj ;
[0168] | nj | is the distance between the j-th centroid bj of the j-th segmentand the i-th centroid bi of the i-th segment.
[0169] By applying a derivative operation with respect to the nodes n, it is therefore possible to express the contribution on an i-th segment li of the derivative of the flux <i relative to a movement of one of the nodes (for example, UIA) as:
[0170] E
[0171] dC*; d
[0172] dnAdnA
[0173]
[0174] Applying the formula for calculating the derivative of the product of functions to the previous expression, we obtain:
[0175] E
[0176] d
[0177] dnAl2-i dnAl|ri;| dnAl\\rlJ\JJ
[0178]
[0179] j=i71 Jj=i71 x J' Applying the rules of derivation and replacing the first sum with the definition of the magnetic vector potential A , we obtain:
[0180] E E
[0181] d^i V 1 ij
[0182] U,LA “ L I'tylJ
[0183] E
[0184] ij
[0185] -^ 1
[0186]
[0187] j=l Similarly, applying the derivative relative to node nB (and noting that there is a change in sign in the expression of the magnetic vector potential Ai) we obtain:
[0188] E
[0189] ij
[0190] dnn2 Z— i
[0191]
[0192] E j=1
[0193] As exemplified here, the procedure involves substituting the expression of A in the summation of the flux calculation, where the latter depends on the distance of the node from the center of gravity.
[0194] An entry of the Jacobian matrix at the h-th iteration of the procedure, corresponding to node of matrix J, is obtained by summing the two contributions of the i-th and k-th sides that share the vertex UIA = UkA which can be expressed as:
[0195] d<i\ d<i\ d<t>k
[0196] Jh[p> nJ - = ^=f + 7^
[0197] dnlAdnAldnA
[0198]
[0199] To improve the accuracy of the calculation in the case of wire, surface, or volumetric conductors, formulas can be used that provide Ai analyticallytogether with its directional derivative, as discussed, for example, in the document M. Fabbri, "Magnetic Flux Density and Vector Potential of Uniform Polyhedral Sources," in IEEE Transactions on Magnetics, vol. 44, no. 1, pp. 32-36, Jan. 2008.
[0200] at block 412, the calculation is then performed pairwise for adjacent segments, using the formulas discussed above. In addition, or alternatively, to speed up the calculation of a far field part (i.e., at a distance from the coils C0S202, C0S204, Sin202, Sin2O4, Sin2o6), a well-known procedure currently known as the Fast Multipole Method (FMM) can be used.
[0201] The operations performed in block 412, exemplified in Figure 4, facilitate the reduction of the calculation speed of the overall procedure, since the Jacobian matrix forms a system of dimensions PxM (for example, if P=5i and M=20, then J has dimensions P*M=i020, i.e., i6kbyte) instead of an NxN matrix (which could occupy Gbyte of memory, for example).
[0202] In Figure 7 it is exemplified the shape of a "candy-like" coil of one of the two receivers (for example, the one associated with a sine function) where the various portions of the receiver are shown as having:
[0203] the initial "theoretical" shape Sin202, Sin2O4, Sin2o6, before processing with the method exemplified in Figure 4, and
[0204] the final "improved" shape Sin202', Sin2O4', Sin2o6' obtained after processing with the method exemplified in Figure 4.
[0205] As exemplified in Figure 7, the position of nodes niA, niB, ..., niA, niB identifying respective segments II, ..., li of the set of receiving coils Sin202, Sin2O4, Sin2o6 varies according to (for example, from niA to n’iA for a first segment h and from niB to n’iB for an i-th segment) according to the values of the node displacement vector Dm obtained 424 at the h-th iteration as the product of the shape matrix F and the system unknowns matrix DAh (e.g., Dnh =FDAh).
[0206] As exemplified in Figure 7, the receiving circuit track with the theoretical shape Sin202, Sin2O4, Sin2o6 for a linear sensor with a given length (e.g., approximately 200 mm) has an initial error (e.g., approximately 5%).
[0207] After processing with the method exemplified in Figure 4, the improved receiver Sin202', Sin2O4', Sin2o6' has a reduced error (e.g., approximately 0.43%).Figure 8 shows a graph (sensor error on the y-axis in millimeters, position index on the x-axis in arbitrary units) of the error for a sensor of length L (e.g., approximately 200 mm).
[0208] As exemplified in Figure 8, between the first iteration H=i and the fourth iteration H=4, the error is reduced (e.g., from an initial value of approximately 0.85 mm, equal to 0.43% of the length L, to a final value of approximately 0.039mm> equal to 0.02% of the length L).
[0209] As exemplified here, a (e.g., computer-implemented) method comprises providing an initial layout of an inductive position sensor, IPS, wherein the IPS comprises a movable target object and a transceiver printed circuit board.
[0210] For example, the movable target object has a reflective area configured to backscatter electromagnetic waves that reach it.
[0211] For example, the transceiver printed circuit board comprises:
[0212] a transmitting circuit track comprising a set of transmitting terminals configured to be coupled to an oscillator circuit for emitting electromagnetic waves therethrough toward the reflective area of the target object, and at least one receiving circuit track comprising a set of receiving terminals configured to provide at least one detection signal of electromagnetic waves backscattered from the target object.
[0213] As exemplified herein, the initial layout of the IPS sensor comprises a set of spatial coordinates of geometric parameters of the at least one receiving circuit track comprising circuit segments extending between pairs of nodes in a set of nodes starting from the set of receiving terminals, wherein two adjacent circuit segments are coupled via a common node.
[0214] As exemplified herein, the method further comprises the operations of:
[0215] providing a shape matrix of shapes comprising a set of spatial coordinates of nodes in the set of nodes arranged according to a set of basis function shapes;
[0216] collecting a spatial distribution of densities of electric currents induced in the circuit segments of the at least one receiving circuit track as the position of the target object varies during the emission of electromagnetic waves by the transmitting circuit track.
[0217] Preferably, the operation of collecting said distribution of electriccurrent densities comprises performing a set of electromagnetic simulations.
[0218] As exemplified herein, the method comprises initializing a counter of iterations of the method, and during each iteration:
[0219] based on the distribution of electric current densities, calculating a measured magnetic flux vector concatenated with at least one receiving circuit track and calculating the respective Jacobian matrix;
[0220] setting a target flux and calculating a flux error as the difference between the measured flux vector and the target flux vector;
[0221] performing a comparison of a flux error norm with a threshold value. For example:
[0222] if the comparison shows that the flux error norm reaches or exceeds the threshold value: solving a system in which the flux error is equal to the product of the Jacobian matrix, the shape matrix F, and a matrix of system unknowns; determining a displacement vector as the product of the shape matrix and the system unknowns matrix and increment the counter;
[0223] if the comparison shows that the flux error norm fails to reach the threshold value, terminating the iterations of the method and combining the displacement vector with the initial spatial coordinates of the nodes in a set of nodes, so as to provide a final layout of the IPS sensor.
[0224] As exemplified herein, the initial layout of the IPS sensor further comprises a set of spatial coordinates of geometric parameters of the transmitting circuit track and / or the target and / or any conductors in the vicinity of the IPS sensor.
[0225] As exemplified herein, the matrix of system unknowns comprises a set of multiplicative coefficients for performing a linear combination of shapes in the shape matrix.
[0226] As exemplified herein, the set of basis functions (preferably orthogonal to each other) in the shape matrix comprises: trigonometric basis functions, preferably sine and cosine functions, and / or Chebyshev polynomials.
[0227] As exemplified herein, the method comprises:
[0228] receiving from a user, preferably via a web platform, a set of defined IPS sensor parameters;
[0229] applying artificial neural network processing, ANN, to the set ofdefined parameters received, determining a further set of IPS sensor parameters,
[0230] wherein the artificial neural network, ANN, is trained using a training data set comprising a library (e.g., generated via the results of simulations performed with sensor templates populated with random values of the respective parameters) comprising a plurality of IPS sensor parameters.
[0231] For example, the method comprises, based on the set of defined parameters and the additional set of IPS sensor parameters, providing the initial layout of the IPS sensor.
[0232] As exemplified here, the method also comprises, based on the final layout of the IPS sensor, the set of defined parameters, and the additional set of IPS sensor parameters, generating a Gerber file representing the printed circuit board of the IPS sensor.
[0233] As exemplified herein, the method further comprises transmitting (e.g., via a web platform or portal) the Gerber file to a manufacturer and manufacturing a prototype IPS sensor having the final layout.
[0234] As exemplified herein, an inductive position sensor, IPS, comprises: a movable target object having a reflective area configured to backscatter electromagnetic waves that reach it, and
[0235] a transceiver printed circuit board 20 comprising:
[0236] - a transmitter circuit track comprising a set of transmitting terminals configured to be coupled to an oscillator circuit for emitting electromagnetic waves therethrough toward the reflective area of the target object, and
[0237] - at least one receiving circuit track comprising a set of receiving terminals configured to provide at least one detection signal of electromagnetic waves backscattered from the target object,
[0238] wherein the IPS sensor further comprises processing circuitry configured to drive the transmission of electromagnetic waves toward the reflective area of the target object and to receive the set of time-varying detection signals induced,
[0239] wherein the at least one receiving circuit track of the sensor comprises a final layout obtained by the method as exemplified herein.
[0240] For example, the IPS sensor comprises at least one sensor of a type chosen from: rotary, arc, or linear.Without prejudice to the underlying principles, the details and embodiments may vary, even significantly, from what has been described, purely by way of example, without departing from the scope of protection. The scope of protection is defined by the attached claims.
Claims
CLAIMS1. A method, comprising:providing (400) an initial layout of an inductive position sensor, IPS (10), wherein the IPS sensor (10) comprises:a movable target object (14) having a reflective area configured to backscatter electromagnetic waves that reach it, anda transceiver printed circuit (20) comprising:a transmitting circuit track (TX, 200) comprising a set of transmitting terminals (TX) configured to be coupled to an oscillator circuit through which electromagnetic waves are emitted toward the reflective area of the target object (14), andat least one receiving circuit track (RXSIN, Sin202, Sin2O4, Sin2o6, RXCOS, C0S202, C0S204) comprising a set of receiving terminals (RXSIN, RXCOS) configured to provide at least one detection signal of electromagnetic waves backscattered by the target object (14),wherein the initial layout of the IPS sensor (10) comprises a set of spatial coordinates of geometric parameters (D) of the at least one receiving circuit track (RXSIN, Sin202, Sin2O4, Sin2o6, RXCOS, C0S202, C0S204) comprising circuit segments (II, li) extending between pairs of nodes in a set of nodes (niA, niB, niA, niB) starting from the set of receiving terminals (RxSin, RxCos), wherein two adjacent circuit segments (II, li) are coupled via a common node,wherein the method further comprises the operations of:providing (402) a shape matrix of shapes (F) comprising a set of spatial coordinates of nodes in the set of nodes (mA, niB, niA, niB) arranged according to a set of basis function shapes;collecting (404) a spatial distribution of current density (DJ[P]) induced in the circuit segments (II, li) of the at least one receiving circuit track (RXSIN, Sin202, Sin2O4, Sin2o6, RXCOS, C0S202, C0S204) as the position of the target object (14) varies during the emission of electromagnetic waves by the transmitting circuit track (TX, 200), preferably wherein the step of collecting said current density distribution (DJ[P]) comprises performing a set ofelectromagnetic simulations;initializing (410) an iteration counter (H=i) of the method, and during each iteration:based on the current density distribution (DJ[P]), calculating (412) a measured magnetic flux vector ( h[P]) linked with at least one receiving circuit track (RXSIN, Sin202, Sin2O4, Sin2o6, RXCOS, C0S202, C0S204) and calculating the respective Jacobian matrix (Jh);setting a target flux and calculating (414) a flux error (eh) as the difference between the measured flux vector (4>h) and the target flux vector;performing a comparison (420) between a norm of the flux error (eh) and a threshold value (e), and:if the comparison shows that the flux error norm (eh) reaches or exceeds the threshold value (e):solving (422) a system equating the flux error (eh) to the product of the Jacobian matrix (Jh), the shape matrix (F), and a matrix of system unknowns (AAh);determining (424) a displacement vector (Anh) as the product of the shape matrix (F) and the matrix of system unknowns (AAh), and incrementing (430) said counter; if the comparison shows that the flux error norm (eh) does not reach the threshold value (e), terminating the iterations of the method and combining the displacement vector (Anh) with the initial spatial coordinates (D) of the nodes in the set of nodes (niA, niB, niA, niB), thus providing a final layout (DRX) of the inductive position sensor, IPS (10).
2. The method according to claim 1, wherein the initial layout of the IPS sensor (10) further comprises a set of spatial coordinates of geometric parameters (D) of the transmitting circuit track (TX, 200) and / or the target (14) and / or any conductors in the vicinity of the IPS sensor (10).
3. The method according to claim 1 or claim 2, wherein the matrix of system unknowns (AAh) comprises a set of multiplicative coefficients to perform a linear combination of shapes in the shape matrix (F).
4. The method according to any one of the preceding claims, wherein the set of basis function shapes in the shape matrix (F) comprises: trigonometric basis functions, preferably sine and cosine functions, and / or Chebyshev polynomials.
5. The method according to any of the preceding claims, comprising: receiving (300) from a user, preferably via a web platform, a set of defined parameters (S) of the inductive position sensor, IPS (10);applying an artificial neural network (ANN) processing (302) to the received set of defined parameters (S), determining an additional set of parameters (P) of the inductive position sensor, IPS (10),wherein the artificial neural network, ANN (302), is trained using a training dataset (TD) comprising a library including a plurality of IPS sensor parameters.
6. The method according to claim 5, further comprising:based on the set of defined parameters (S) and the additional set of parameters (P) of the IPS sensor (10), providing (304) the initial layout of the inductive position sensor, IPS (10).
7. The method according to claim 5, further comprising:based on the final layout (DRX) of the IPS sensor, the set of defined parameters (S), and the additional set of parameters (P) of the IPS sensor (10), generating a Gerber file (GF) representing the printed circuit board (PCB) of the IPS sensor.
8. The method according to claim 6 or 7, further comprising: transmitting (310), preferably via a web platform, the Gerber file (GF) to a manufacturer and fabricating (PP) a prototype of the IPS sensor having the final layout (DRX).
9. Inductive position sensor, IPS (10), comprising:a movable target object (14) having a reflective area configured to backscatter electromagnetic waves that reach it, anda transceiver printed circuit (20) comprising:a transmitting circuit track (TX, 200) comprising a set of transmitting terminals (TX) configured to be coupled to an oscillator circuit through which electromagnetic waves are emitted toward the reflective area of the target object (14), andat least one receiving circuit track (RXSIN, Sin202, Sin2O4, Sin2o6, RXCOS, C0S202, C0S204) comprising a set of receiving terminals (RXSIN, RXCOS) configured to provide at least one detection signal of electromagnetic waves backscattered by the target object (14),the IPS sensor (10) further comprising processing circuitry (30) configured to drive the transmission of electromagnetic waves toward the reflective area of the target object (14) and to receive a set of time-varying detection signals induced,wherein the at least one receiving circuit track (RXSIN, Sin202, Sin2O4, Sin2o6, RXCOS, C0S202, C0S204) of the IPS sensor (10) comprises a final layout (DRX) obtained with the method according to any of claims 1 to 7.
10. The IPS sensor (10) according to claim 9, comprising at least one sensor of the following type: rotary and / or arc and / or linear.