Let's shed light on inductive position sensors: 12 questions to dispel false beliefs

In recent years, artificial intelligence (AI) has increasingly entered our daily lives, although sometimes we do not realize it.

Think, for example, of new cars that drive themselves, or robots that look more and more like us humans. In short, AI is increasingly imposing itself in our lives. This is made possible by the fact that AI possesses information about the real world through the use of sensors that enable it to make decisions based on the conditions of its surroundings. A widely used type of sensors are position and motion sensors that provide indispensable information, for example, for controlling motors and actuators. These types of sensors are therefore indispensable for AI to interact with the real world. In this scenario, a new emerging technology has proven successful: inductive position sensors. 

They are very accurate, precise, reliable position sensors. A position sensor is a sensor that accurately measures a mechanical position. It consists of a fixed part and a part that can move relative to the fixed part. The moving part is called the target. So the position sensor measures the distance between a reference point taken on the fixed part and the actual position of the target. The movement of the target can be linear or it can rotate if an angular position is to be measured.

The sensors use electromagnetic induction of a magnetic field in a metal target. They measure the disturbance (or perturbation) that a conductive target has on a source magnetic field using Faraday-Neumann law. One can think of the sensor as a transformer in air where the TX transmitting coil is the primary, while the target forms the secondary circuit.

In addition to the TX transmit coil and the target, two secondary coils (also called RXSIN and RXCOS receive coils) are used that are placed in different physical positions to detect the magnetic field present. The paths are made to have a differential sensor. To determine the position of the target, the ratio of the two coil voltages is calculated. The arcotangent of this ratio gives an electrical angle that is related to the angle to be measured for rotary sensors, while in the case of a linear sensor the angle obtained is to be related to the linear length to be measured. The fact of using the ratio makes the sensor ratiometric (as well as differential). This fact is very positive since any variations in the geometry of the sensor (e.g., due to thermal expansion) tend to compensate when considering the ratio between two quantities both affected by the same variation. The main advantage of inductive position sensors is that they can be used in hostile environments, that is, containing dust or liquids, and in environments where electromagnetic fields are present. In fact, they are particularly valued for their immunity to static or electromagnetic magnetic fields. In addition, one and the same sensor can be mounted in a variety of ways, such as at the end of the shaft (end-of-shaft) or with the shaft running through it (through-shaft).

But, as is often the case, the biggest obstacle to mass adoption of these sensors is the fact that companies notoriously have great inertia in adopting new technologies, preferring to take refuge in the safety of an already used and tested technology. So we want to provide some clarity. Try to answer the most common doubts about how they work and introduce you to the many possibilities that these new sensors offer. In particular, we will compare inductive position sensors with the more common Hall-effect sensors, magnetoresistive sensors, and optical encoders.

A proximity sensor is a sensor that can detect the presence of nearby objects without any physical contact by emitting an electromagnetic field or beam of electromagnetic radiation such as infrared and, simultaneously, measuring changes in the field or return signal. Depending on the physical principle of object detection, there are four types of proximity sensors widely used: inductive, capacitive, ultrasonic, and ultraviolet proximity sensors. Proximity sensors are available in most cases with an output formed by a switch, since they are usually used as non-contact limit switches. In this case, the switch will be activated when the object is at a certain distance from the sensor. This detection range is therefore a characteristic of the sensor. The NPN configuration output is an output with a switch that switches the common or negative pole of the power supply to the load, which is connected between the sensor output and the positive of the power supply. The PNP output configuration, on the other hand, is an output consisting of a switch that switches the positive of the power supply to the load, which is connected between the sensor output and the common or negative pole of the power supply.Another family of proximity sensors outputs an analog signal (voltage or current) proportional to the distance between the moving object and the sensor. It must be said that usually the linearity error of this sensor is very large, so a linearization is usually required, which can be either internal to the sensor or done by the electronics downstream of the sensor.In the case of inductive proximity sensors, the moving object must necessarily be conductive to be detected. In this case, for both NPN/PNP configurations and analog outputs, the sensor reading is highly dependent on the conductivity of the material from which the moving object is made. For example, if the object is made of copper, the measurement will be larger than for an object made of steel, even if the distance of these two objects to the sensor is the same. 

The inductive position sensor (IPS), on the other hand, is a sensor that accurately measures the distance between a reference point taken on the circuit board relative to the sensor and the position of the moving target. Thus, it should be easy to understand that inductive proximity sensors and inductive position sensors are two completely different devices. They are both made differently but also measure a different things. The only thing they have in common is that they both use the physical phenomenon of electromagnetic induction/eddy currents described by the Faraday-Neumann law.

Until a few years ago, to get acceptable results, you had to have extensive experience in electromagnetic fields and the ability to perform very accurate finite element simulations, just to name a few difficulties. This is no longer true. First of all, suppliers of IPS sensor ICs usually offer evaluation boards and kits. For example, our company EMC Gems contributed to Renesas Electronics' sensor kit design for the ZMID520X and IPS2550 chips.

In addition, estimates of sensor linearity errors can be obtained using simulation results even before the PCB is tested in the real world. Chip manufacturers are organizing to make these electronic design automation (EDA) tools available to their customers. Pioneering this possibility was Integrated Design Technology (IDT), now Renesas Electronics, which, through collaboration with our company, first made available an EDA tool for IPS called ICOT (Inductive Coil Optimization Tool). Our company in particular provided software for electromagnetic simulation and developed technology for optimizing IPS sensors to reduce their linearity error. See the details in [1].

We are developing a novel tool called IPSmagic that will set new standard in the automatic design of IPS thanks to its unique features. The tool can be run by using a proprietary cloud-coumputing environment, or personalized to be installed on customers' servers. We will keep you posted once the final version will be available. We expect the final release in June 2023.

Inductive position sensors are very accurate especially at high temperatures, where, other systems that rely on magnets are far less accurate. The algorithm is designed so that temperature variation has minimal impact. First, the sensor is ratiometric in nature. Also, the primary oscillator creates yes a synchronous demodulation in the secondary receive channels but this will have no impact on the amplitude of the received signals.

Even where there are variations in temperature and mechanical assembly tolerances, error margins are kept below +/- 0.3%FS. In the case of absolute 360° rotary sensors, errors of 0.1%FS are obtained using sensor optimization techniques developded by EMC Gems. By constructing RXSIN and RXCOS receiving coils with more periods, the sensor resolution can be further improved, with the only disadvantage that the sensor becomes incremental and is no longer absolute.

Finally, it is worth mentioning that our company has developed a methodology for optimizing IPS sensors to reduce their linearity error. This methodology is used, for example, by Renesas Electronics in their EDA tool called ICOT. We are also constantly developing new methodologies for the optimized design of inductive position sensors. 

Sensors use the electromagnetic induction of a magnetic field in a metal target. They measure the disturbance (or perturbation) that a conductive target has on a source magnetic field using Faraday-Neumann law principles. One can think of the sensor as a transformer in air where the TX transmitting coil is the primary, while the target forms the secondary circuit. 

On the contrary, Hall-effect and magnetoresistive sensors rely on a static magnetic field generated by a permanent magnet, while inductive sensors use the time-varying magnetic field that is generated by the primary winding of a transformer (i.e., the TX transmitting coil). Usually a frequency range of 2 MHz to 5 MHz is used. Here the conducting target is placed in the source magnetic field inducing eddy currents within it that tend, by Faraday-Neumann law, to cancel the effect of the source magnetic field thus bringing the magnetic field strength to zero on the target. The target thus tends to behave like an electromagnetic shield. 

As mentioned above, in addition to the TX transmit coil and the target, two secondary coils (also called RXSIN and RXCOS receiving coils) are used, which are placed in different physical positions to detect the magnetic field and thus each will detect a different voltage. The ratio of the two coil voltages is used to determine the position of the target. 

We can therefore conclude that inductance is not used at all. In fact, the position measurement is made by two voltage measurements on the RXSIN and RXCOS receiving coils, which are not run through by current.

Inductive position sensors, on the contrary, work much better when conductors such as copper, aluminum, or steel are used. This is because the sensor works by detecting a change in the electromagnetic field disturbed by a metal target, which absolutely does not have to be a magnet. The only basic requirement is that the material of the target allows eddy currents to flow, and the best conductors are, indeed, copper and aluminum.

IPS inductive position sensors, in order to be immune to the increasingly intense parasitic magnetic fields created by electrical machines and electronic boards placed in their vicinity, use active demodulation. This allows the voltage at the ends of the two receiving coils RXSIN and RXCOS to be filtered in order to measure only the frequency corresponding to the frequency of the source magnetic field created by the TX transmitting coil. Thus IPS sensors, unlike Hall-effect and magnetoresistive sensors, are very robust with respect to the presence of external electromagnetic disturbances. They are therefore suitable for particularly harsh environments or where the required immunity with respect to spurious electromagnetic fields is very high, such as the automotive industry. In particular, by not using any magnetic material (when the target is realized in non-magnetic conductors like copper, aluminum or non-magnetic steel), the sensor is not affected by any static magnetic (DC) field.

It is true that inductive sensors show the best accuracy when the target length is one-half or one-fourth the measuring range. However, this does not mean that they cannot measure linear positions of lengths ranging from 5 mm to 600 mm and beyond. The only limiting factor for the measurable length is the ability of the oscillator to generate the correct LC resonant signal. In particular, we must take into account the effects due to the parasitic capacitance of the TX transmission coil and those due to the fact that the transmission coil, given its length, behaves like a transmission line. On the other hand, if we use Hall effect sensors, the magnet must be moved from one position to another and several sensors must be multiplexed. This makes crossover management difficult and complicated and makes it susceptible to temperature changes. These problems do not occur with inductive sensors.

They can be used very well for arc and rotary measurements such as car pedals, rotor position, water and air valves. A 360-degree rotary sensor is in fact almost like a linear sensor whose ends are curved to meet. Indeed, 360-degree rotary sensors provide by far the best performance in terms of linearity error.

IPS sensors are very cost-effective compared to other technologies, such as optical encoders. The cost of the IPS sensor is formed mostly by the chip. The price of the chip is just over a US dollar for purchases in thousands. To the chip, we must add the price of the PCB and target (which can equally be made in PCB technology). The PCB can be made in most cases using only two layers. So with a few tens of cents, both the target and the static part of the senor, which includes the transmit coil and the two receive coils, can be produced. To conclude, the total cost of the sensor is between 1 and 2 USD.

Let us remark that inductive position sensors present a new way of implementing sensing, but the underlying technology is well known and proven. To the best of our knowledge, the first description of the technology, along with much of its development, was introduced in [2]. The principle using inductive position sensing is related to LVTD, linear voltage differential transformer. LVTD position sensors use the same technique i.e., the ratio of two voltages induced by magnetic field perturbations by a conducting element to detect position. But mind LVDT uses primary and secondary coils, IPS sensors use a simple PCB board to make the transmit and receive coils.

Two redundant inductive sensors can share the same space (and thus magnetic field) while providing galvanic isolation. The target will also be shared by both sensors. The only difference is that the redundant sensor thus made needs a 4-layers PCB, which is a bit more expensive than the 2-layers PCB.

Inductive proximity sensors and LVDT sensors have a large linearity error. They therefore need linearization either inside the sensor or externally from the microcontroller or computer that manages the measurement. IPS sensors are already born with an error certainly inferior to the other mentioned types of sensors. 

Moreover, the error can be reduced by using a sensor optimization methodology we developed, see for example [1]. Thus, for most applications, no post-processing of the measurement is needed, saving time for product setup. In case higher accuracy is needed, linearizing will be more effective on an optimized sensor. Moreover, it further accuracy is needed, one can add more periods and go from an absolute position sensor to an incremental position sensor.