Appunti completi Automotive Electronic Systems

Chapter 3 – Sensors

Sensor

A sensor is a device that transforms the variations of a physical quantity into a related variation of an electrical quantity like voltage, current, or resistance. It translates information from a physical domain to another.

Sensitivity

Sensitivity refers to how much the sensor is responsive to variations of the input quantity. It’s defined as the derivative of the output quantity with respect to the input quantity. S = (Vout2 – Vout1)/(T2 – T1).

Resolution

The resolution is the minimum variation of the input that causes a detectable variation in the output quantity.

Accuracy

Accuracy measures how close the measured input is to the actual value of the input quantity.

Selectivity

Selectivity is the ability of the sensor to make the output dependent only on the right input.

Temperature Sensors

To measure the temperature, we can exploit different physical mechanisms that allow measuring temperature change due to conduction, convection, and radiation. We would like the sensor not to influence the temperature measured with its presence, which requires a low thermal capability. There are different temperature sensors on the market:

Thermocouples

It’s the simplest sensor on the market. It’s made of two pieces of different metals joined together at one junction (hot junction). It uses the Seebeck effect to convert temperature into electricity. Their loose ends form the cold junction. The metals at the hot junctions are at the same temperature. Also, the loose ends are at the same temperature, even if they are not touching. The loose ends exhibit a voltage difference between them, which is proportional to the temperature difference between the hot and the cold junctions. Different types of metals generate different thermocouples in terms of cost, melting point, stability, and output.

To measure the voltage produced, it’s necessary to use a voltmeter connected at both loose ends. Its metal terminals could easily be different from the metal of the thermocouple. Fortunately, this doesn’t change the properties of the thermocouple if both voltmeter terminals are kept at the same temperature and are made of the same material. This law allows creating many junctions with the same metal, for example, to connect a thermocouple with the amplifier chip inside a circuit.

In general, they measure the temperature difference between hot and cold junctions, but not the absolute temperature at the hot junction. We can use two strategies to know the cold junction temperature:

• Ice-bath: we can expose the cold junction to 0 degrees, but this is impractical. However, we can force the cold junction to a reference temperature, not necessarily to 0.
• Using integrated circuit: they use a different kind of temperature sensor to know it.

Thermocouples are cheap, robust, stable, and have a wide temperature range.

Resistance Temperature Detectors (RTD)

They measure the resistivity variation caused by a thermal variation. They are very stable over time, precise, accurate over a wide temperature range, and sufficiently linear. However, they are expensive and affected by self-heating. To measure the resistivity, we must measure the resistance of the device. This is done by putting some current across it, but some power will be dissipated, and the temperature increases. Moreover, RTDs are slow because the typical materials used have large thermal inertia, and to measure low temperature correctly and generally with precision, they need a 4-terminal resistor.

Thermistors

They are similar to RTDs but are based on semiconductors rather than metals. They are precise, very sensitive, and cheap because semiconductors are cheaper than metals. However, they suffer from strong non-linearity and require complex electronics to compensate for it. Unlike RTDs, they show a positive or negative temperature dependence: the resistance may increase or decrease with temperature increasing. For this reason, they are classified in two ways:

• PTC (Positive T Coefficient): Typically based on silicon and exhibit a quadratic dependence on temperature. This causes the PTC to be very sensitive at high temperature. Usually used to detect if temperature increases beyond a threshold value. They suffer from self-heating.
• NTC (Negative T Coefficient): Typically based on metal-oxide semiconductors and show a strong exponential response. They are much more sensitive at low temperature and don’t suffer self-heating.

Integrated Temperature Sensors

RTDs and thermistors show the temperature variation by means of the physical quantity of materials (metals or semiconductors). Other kinds of temperature sensors show this temperature variation with the variation of a specific electric quantity in an electrical circuit. These are the integrated sensors. They are frequently used due to their reliability and precision. They can also integrate electronics to post-process the data. They are very small, linear, and cheap.

Pyrometers

This kind of temperature sensor operates without contact with the measured object. It’s based on radiation, so it’s sensitive to the electromagnetic waves radiated from an object. This phenomenon is governed by the Stefan-Boltzmann equation that connects energy emitted with the fourth power of its temperature: E = sigma*T⁴. This means that pyrometers are very non-linear and expensive, so not often employed. Moreover, the measurement could be altered by the radiations of nearby objects.

Force Sensors

They are transducers used to measure forces applied to an object. The most used types are: strain gauge sensors and piezoelectric sensors.

Strain Gauge

This kind of sensor measures the force applied by sensing the local deformation of the body surface. The main idea is that a piece of metal changes its resistance while deforming. We can relate force and deformation in this way: ∆ = = . Therefore, in the elastic regime, we can achieve the force by knowing the Young modulus and the area. We know the resistance expression, and we can differentiate it to achieve the relation between the resistance variation and the shape deformation. However, only piezoresistive materials change their piezoresistivity with deformation, so the simplest strain gauge sensor is obtained by using a non-piezoresistive material. In this way, drho/rho = 0, and we can say: if uniaxial deformation is applied, the cross-section doesn't change. But in real life, the deformation is not perfectly uniaxial, so we must consider the Poisson effect. For this reason, we assume that the previous formula is equal to a constant G (Gauge Factor) and obtain: Usually, G is around 2 for most metals, and the deformation (epsilon) is measured in microstrain. The strain gauge sensor has typically 2 big tabs that allow connecting the cables and minimizing their contribution to overall resistance (S is big). There are alignment marks.

The main issue with the strain gauge is the dependence on the temperature. There are two strategies:

• Use a metal for the grid which has as low dependence between resistivity and temperature.
• Using a dummy gauge: If the strain is axial, we can put a strain gauge in the orthogonal direction. The second will respond only to temperature variations, so by subtracting the two resistance changes, we completely compensate for the temperature effect.

Another issue to be considered is the deformation caused by thermal expansion: if temperature changes, the object and the strain gauge may have different thermal expansion coefficients, so the object may change its shape and deform the strain gauge. Typical applications: load cells, structure monitoring, force, torque, pressure, acceleration transducers.

Piezoelectric Sensors

They are based on piezoelectric materials: they convert mechanical stress into electrical energy (voltage). They work similarly to strain gauges, but the output is not a resistance value; it’s a voltage. Moreover, the piezoelectric effect is reversible: if some voltage is applied, it will deform. So it can work as a sensor but also as an actuator. Usually employed in pressure sensors, vibration sensors, microphones, speakers, and proximity sensors.

Torque Sensors

They can measure torque by following different principles: strain gauge, magnetic coupling, optical.

Strain Gauge Torque Sensor

To measure the torque on a rotating shaft, it is possible to use two strain gauge sensors placed at 45 degrees with respect to the shaft axis and orthogonal to each other. If some torque is applied, the shaft will experience a compressive strain along a specific direction and a tensile strain along another direction. However, the strain gauge system needs a circuit to operate, and so wires attached to the sensors. A solution could be to place the entire circuit on the shaft, but it always requires power. You have these solutions:

• Place a battery on the shaft. The disadvantage is that it requires periodic replacement, so periodic shutdown of the system.
• Recover energy from the shaft, but it must be light and must guarantee a continuous power supply.
• Wireless signaling: a local circuit takes the sensors’ output and transmits it to a remote control unit.

Optical Torque Sensors

It uses a light sending unit and a light receiving unit. The light sending is mechanically connected to the rotating shaft and has a series of LEDs arranged in a circle. The light receiving is stable and made with a series of photodiodes arranged in a circle. By measuring the response of the photodiodes, it’s possible to obtain the rotational speed and, indirectly, the torque. Both units need power supply. This system is delicate and expensive.

Position Sensors

They measure the position of an object, which could be linear or angular position.

Potentiometers

They are analog position transducers, meaning their output is analog. They can be linear or angular. A potentiometer is basically a three-terminal resistor with a sliding (or rotating) element that forms an adjustable voltage divider. Two terminals are connected to the input voltage or current. The resistance seen from these terminals is fixed. The third terminal is connected to the sliding element which moves like the object to be measured. This means that it acts like a variable resistance. With a voltmeter, we measure the voltage across the sliding terminal and one fixed terminal. Ideally, it outputs very linearly. But this is not true if we consider that some power will be dissipated by the voltmeter. We can consider that by adding a new resistance which is the equivalent representation of the voltmeter. This causes a deviation from the ideally linear output. However, potentiometers are rarely used to control significant power since the power dissipated by them is comparable to the power in the controlled load. Since they use sliding contacts, they suffer from reliability issues and low precision. On the other hand, they are very cheap.

Encoder

It’s a position sensor with a digital output. It can be linear or rotational. It’s made with a disk or a strip of opaque material with a series of holes. It’s linked to the moving object, and on the two sides of the disk/strip are placed a LED and a photodiode. Encoders are divided into two categories: incremental and absolute.

Incremental Encoder

It’s an opaque ring with a series of identical holes equally spaced. As the disk rotates, the photodiode will receive light only when the disk is in such a position that one hole is aligned with the LED. The photodiode will emit a pulse each time the light hits it, so by counting the number of pulses, we can understand the angle of rotation. Specifically, each pulse corresponds to a rotation of r (resolution) = 360/n where n is the number of holes. Its name is due to the fact that it can’t measure the absolute angular position, but only the incremental rotation. The incremental encoder also needs a counter and starter that defines the starting position. Moreover, it can’t distinguish the direction of rotation. However, this problem is solved by putting another LED and a photodiode. The two photodiodes are placed in such a way that one of them is previously hit by light depending on the direction of rotation.

Absolute Encoder

The absolute encoder follows the same principle as the incremental encoder, but number, shape, and position of each hole are completely different. Here the holes are not identical to each other and are not placed on a single ring. There are many concentric rings. The innermost has only one hole that covers half the ring (half opaque, half transparent). The second innermost has two holes, each covering a quarter of the ring. Therefore, even here, the ring is half opaque and half transparent. So, moving from the innermost to the outermost ring, the size of holes decreases but the number increases. If k rings are present, there are k LEDs and k photodiodes. In this way, each position has its unique combination of photodiodes output. In this way, we can achieve immediately the absolute angular position without using a counter and a starter. Its resolution is 360/2^K. There are many possible arrangements of holes’ positions. The most frequently used is the Gray code: two successive values differ in only one bit. In this way, there is even an error detector. On the other hand, the absolute encoder is more expensive and could even be magnetic: a permanent magnet is mounted on the rotating element, and a magnetic sensor detects the direction of the magnetic field.

Resolver

It’s an angular position transducer based on magnetic induction. It’s very robust and reliable, for this reason frequently used in the automotive sector. A coil driven by an AC current generates a magnetic field that changes over time. Due to the magnetic field property, it generates on a secondary coil an AC voltage (and current). This AC voltage on the secondary coil has the same frequency as the first AC voltage, and they are in phase. The magnitude of the second voltage depends on the winding, the angle between them, and the distance. The resolver is made by two parts:

• The rotor: that generates the AC magnetic field.
• The stator: composed of two orthogonal windings on which some AC current is induced, and from which two output voltage waveforms are read.

The rotor generates this voltage wave: V_EX(t) = E₀sin(wt). On the S2 coil, the voltage will be: V_S2(t) = E₀Tcos(theta)sin(wt) where T is the transformation ratio. While on S1: V_S1(t) = E₀Tcos(theta-pi/2)sin(wt) = E₀Tsin(theta)sin(wt). So the two output voltages include information about the sine and the cosine of the angle between the stator and the rotor. We can achieve the angle theta with no ambiguity. However, to know V_EX, we need to put some wires on the rotor. If we are not able to know V_EX, the resolution of the resolver is limited to 180 degrees. To have a resolution of 360 degrees, it needs a sophisticated algorithm.

NCAPS

Non-Contact Angular Position Sensor. It’s a general class of sensors that can retrieve the angular position in a contactless way (the resolver is in this class). They are typically used in the automotive sector; however, they require complex electronic components to post-process the output signal. Typically made with two parts: one attached to the moving object and the other fixed.

RCDT/LCDT

Rotary/Linear Capacitive Displacement Transducer. For example, the LCDT is made with two metal electrodes, with a dielectric in between. One electrode is a metal rod which is free to move in linear motion and it’s covered by the dielectric material. The rod and the dielectric can slide in and out of the cylinder. This is basically a capacitor: since the capacitance is proportional to the overlap length of the two cylinders, it’s also proportional to the position of the object. It has a linear output, but this characteristic falls when we reach the maximum displacement.

Acceleration Sensors

They are used to measure accelerations. They can be linear (mono-, bi-, tri-axial) or rotational. There are many typologies: piezoelectric, piezoresistive, mechanical, capacitive, and integrated. The main idea is to understand the acceleration by using a “seismic system” composed of a spring, a mass, and a damper. Its dynamics is ruled by a second-order differential equation based on the mass position. So, by doubly differentiating the position, we calculate the acceleration. Another strategy could be to apply a strain gauge to the spring and get its deformation. Here we will talk about piezoelectric, piezoresistive, and MEMS acceleration sensors.

Piezoelectric Acceleration Sensors

It generates a voltage while deforming and can’t monitor continuous accelerations. It suffers from bandwidth limitation: the minimum acceleration frequency that could be sensed is limited by the crystal time constant, while the maximum frequency is limited by mechanical resonance (not employed in airbags). On the other side, it has a large sensitivity, a large thermal operating range, and a good time response (small delay between input and output).

Piezoresistive Acceleration Sensors

It can monitor continuous accelerations since there is nothing limiting their bandwidth on the low-frequency side. It is quite immune to the electromagnetic field but is more sensitive to temperature variations.

MEMS

Micro Electromechanical Systems are the technology of microscopic devices, particularly those with moving parts. They usually consist of a central unit that processes the data and various components that interact with the surroundings. They are built with a large surface-to-volume ratio, for this reason, in the design process, surface forces are more important than in large devices. MEMS technology allows fabricating different types of sensors like inkjet printers, accelerometers, magnetic sensors, and microphones.

The MEMS acceleration sensor is made by complex moving and standing masses, which could be made of silicon, creating a seismic system. A moving mass, which has a finger-like structure, is linked to a fixed base. Between the moving fingers, there are fixed fingers. Movable and fixed fingers are separated by air (dielectric) and form the plates of a capacitor. The motion of the mass varies the distance between the fingers, so by implementing the electronic circuits needed to evaluate the capacitance variations, we achieve the position of the mass and then we can calculate the acceleration.