The IIoT is all about communication of information concerning the status of devices and being able to control them remotely. This imposes a whole new set of requirements on sensors. They have to be designed to be able to interface well with electronics and computers, with all of the complications of that, such as EMC — sensitivity to, and generation of, noise and interference. This can only be achieved with intelligent sensors. We interviewed K. Mark Smith, Ph.D., Senior Product Manager, Sensors, Microchip Technology, (Chandler, AZ) about the role of inductive position sensors in the IIoT.

Inductive position sensors have multiple uses in modern automobiles.

Tech Briefs: Why don’t we start with an overview of the technology and applications of inductive position sensors.

Mark Smith: In the industrial IIoT area, anytime you’re measuring a motion of some sort, there’s opportunity to use non-contact magnetic-based position sensors. Included in this general category, are inductive, Hall effect, and magneto-resistive. First of all, they’re robust — an important quality for the IIoT, where the quality and consistency of the information they convey is critical. Because they’re non-contacting, magnetic sensors are better able to withstand harsh environments. Among non-contacting magnetic sensors, inductive sensors have the particular advantage of high rejection of stray magnetic fields.

Tech Briefs: Could you give me some typical applications for magnetic sensors.

Smith: Most of our applications are in automotive, but we put automotive in our industrial category.

Most cars have a non-contact sensor in the accelerator pedal. This is an example of a rotary or linear sensor that is non-contact and is safety-critical. You put your foot on the accelerator, it measures the angle of the pedal as you apply your foot and it sends the information electrically to the Engine Control Unit (ECU). In response, the ECU will then control a motor that moves the throttle valve to open up in response to the angle of the accelerator pedal, to allow more air in. Another sensor sends the valve position information back to the ECU.

25 years ago, this was all done mechanically with an accelerator cable. Now it’s done electronically with a drive-by-wire system.

Tech Briefs: Is the communication between these devices done wirelessly?

Smith: It’s not wireless, it’s a solid connection. There’s an electrical cable that runs down the pedal and all the way to the ECU.

Tech Briefs: How is this sensor powered?

Smith: It’s a three-wire connection — most of our applications are three-wire. There’s a power, a ground, and an output signal. We do have the ability to do two-wire communication, which would be hot wire and ground. That uses a bus protocol called Peripheral Sensor Interface 5 (PSI5), which can modulate the current that we draw on that power line and that can be a communication protocol back to the ECU. The ECU actually measures the current — it’s kind of a current modulation. But most of our applications are three-wire.

Tech Briefs: What protocols do you use for transferring information from the pedal to the actuator?

Smith: The traditional way would be an analog signal between 0 and 5 volts. So, 1 volt could mean you’re at 5-degree depression of the accelerator pedal. Another mechanism is Pulse-Width Modulation (PWM). There are also a couple of digital protocols in the automotive space. In addition to PSI5, there is the Single Edge Nibble Transmission (SENT) protocol.

Tech Briefs: Are these ever used at high speed or are they always just used at low speed, like the angle of the accelerator pedal?

Smith: They are used at high speed — as replacements for standard resolvers on brushless dc motors. These can run at up to 50k rpm.

Tech Briefs: What kinds of applications would that be used in?

Smith: For an electric vehicle, it would be the main e-traction propulsion motor. For electronic power steering, although the rate of change is low, manufacturers want good response times to minimize latency. Resolvers on robotic motors in industrial automation applications also have to be high speed, to provide information on rotor position.

For an electric traction motor, you need to know the rotor position on idle condition. In the past, there were only two options: magnetic or optical resolvers. Inductive sensors can replace the fairly large magnetic resolvers with just a printed circuit board (PCB). The advantages are lower cost and less weight, while maintaining the same level of robustness that you would have with the resolvers.

Tech Briefs: These work in harsh conditions. Do they have special ratings?

Smith: They’re automotive (AEC-Q100) grade 0, which would be up to 150°C. There’s also a grade 1, which is up to 125°C. They have to work in the very hot conditions underneath the hood of the car.

Tech Briefs: Could you explain how inductive position sensors work.

Smith: Just like you excite the primary of a transformer winding, traces on the outside of the PCB generate a magnetic field. Traces on the inside are the “receive” coils, which detect that magnetic field. A target, which is a plain piece of metal, is attached to the moving component. When you place the metal target inside a magnetic field, eddy currents are induced, generating an opposing magnetic field. The net result is that wherever the metal is, the magnetic field drops to zero and wherever the metal is not, the magnetic field is still maintained. What that will do, is it will cause the voltages induced in the two receive coils to be different. The position is determined by the ratio of those two voltages.

An oscillator on the sensor chip generates the ac excitation voltages and an analog front-end inputs the two received voltages. A microcontroller is embedded in the chip to calculate the position based on the ratio of the two received voltages. That information is then sent out to the ECU.

Figure 2. (Left) A simplified diagram of the coil and target setup — CL1 and CL2 are the two “receive” coils. (Center) The primary coil is shown in red and the two received waveforms are shown in orange and green. (Right) A sketch of the magnetic coupling between the oscillator coil and the conductive target.

The system is illustrated in Figure 2. On the left side there is a simplified diagram of the coil and target setup — CL1 and CL2 are the two “receive” coils. The diagram in the center shows the primary coil in red and the two received waveforms are shown in orange and green. The right side is a sketch of the magnetic coupling between the oscillator coil and the conductive target.

Magnetic position sensors have also been based on the Hall effect. A Hall effect sensor’s block diagram is similar. Although it also uses a microcontroller, there is no oscillator and instead of inductive sensing, it uses a Hall sensor in the front end, which responds to a field produced by a magnet. The difficulties with magnets include their nonlinear properties, which cause accuracy problems.

Another issue is noise from stray magnetic fields. Inductive position sensors have very good rejection of this noise. This is because we know precisely the frequency that we are exciting the primary windings with — we excite the primary winding with a given frequency, generally in the range of 1 – 5 MHz. Since everything is on the same chip, we can just tune in to the exact frequency with which we are exciting our primary winding and filter out all others. We also don’t have to worry about noise that’s coming from brushless dc motors or large amounts of currents in an application like an electric vehicle that might have 100s of amps flowing.

You could even put the sensor inside the brushless dc motors, in spite of the fairly strong magnetic fields and magnets in motion.

Tech Briefs: Could you put numbers on the accuracy.

Figure 3. Typical sensor accuracy as a function of angle of rotation.

Smith: Let me answer that with a typical application — say, for a throttle body valve. It opens, there’s a target at the end, and then that target, which is a piece of metal (It looks like a PCB board but we’re just using the copper on top of this one layer) gets placed over the sensor. We then measure the accuracy. Refer to the chart in Figure 3. The measured angle in degrees is at the bottom. The vertical axis on the left is the output. The blue line represents the ideal transfer function. The red trace is the deviation of the measured output from the ideal. We divide the absolute error by the full-scale range and that gives you the accuracy shown on the right. The measured error is below 0.1% full-scale, but it increases with temperature and air gap variations. We like to say that the error for this type of inductive position sensor over all operating conditions is below 0.3%. In comparison, Hall sensors have a tough time staying below 1.0%.

Tech Briefs: Could you describe some other industrial applications.

Smith: All sorts of IIoT applications, such as motor controls, actuator arms, etc. Anything related to an actuator, linear or rotary, is applicable in IIoT. One example would be high-end audio mixer boards. In the past, most of them have used linear potentiometers. One disadvantage of linear potentiometers is that they wear out and are susceptible to impairments such as things spilling on a mixer console, or dust. These kinds of things would require the potentiometers to be cleaned or replaced, however inductive linear sensors would not be impaired.

Tech Briefs: When I think IIoT, I think networking, where the sensor would be networked throughout the building, to send information back to the office, or to a control board for a number of systems. Information about things like motor speeds or position of actuators — all of this in an industrial setting — would be networked together. How might that kind of application work?

Smith: We send an output signal that measures position to a microcontroller that would then communicate it over an information bus. Inside cars, this is typically done with a CAN bus. In a building or industrial setting, this could be wired or wireless and ultimately would connect to the Internet.

Tech Briefs: So, the signal that would go to the microcontroller is analog, digital, PWM?

Smith: It could be any of them. It depends on the sensor you’re designing. Analog and PWM are two typical ones. The other protocols that are specific to automotive are the SENT protocol or PSI5. The microcontroller calculates the position and communicates it through a gateway and onto a standard network. It might then go to the Internet, a central hub in a building, or to the factory floor, to give you information about what’s going on there.

Tech Briefs: I could see applications for this in building automation, like HVAC controls, where the angles of dampers are automatically controlled.

Smith: For cooling, that’s exactly right. In the past, it might have been done using potentiometers, but they wear out. An HVAC vendor wants this thing to last for 20 years — a potentiometer’s going to have problems.

We continue to explore new applications for inductive position sensors. With the 5G expansion of the IoT, we expect there will be many more.

This article was written by Ed Brown, Editor of Sensor Technology. For more information, visit here.