Instrumentation Sensors Book

(c)

Figure 6.5 (a) Electron flow in a PNP magnetotransistor without a magnetic field, (b) electron flow in a PNP magnetotransistor with a magnetic field, and (c) cross section of a PNP magnetotransistor.

within the crystal structure of the material. When the material is exposed to an electric field, the charges try to align themselves with the electric field, causing a change in shape of the crystal. The same polarization mechanism causes a voltage to develop across the crystal in response to mechanical stress, which makes piezoelectric devices suitable for use in the measurement of force. Some naturally occurring crystals exhibit the piezoelectric effect, such as quartz, Rochelle salt, lithium sulphate, and tourmaline. Another important group of piezoelectric materials are the piezoelectric ceramics, such as lead-zirconate-titanate (PZT), lead-titanate, lead-zir- conate, and barium-titanate.

The differences between the characteristics of crystalline quartz and PZT make them suitable for use in widely differing application areas. Quartz, because of its stability, minimal temperature effects, and high Q, is an ideal timing device. PZT, due to its low Q, but much higher dielectric constant, higher coupling factor, and piezoelectric charge constant, gives a much higher performance as a transducer (see Table 6.3). These factors also make PZT useful for micromotion actuators, or micropositioning devices.

6.3.1Time Measurements

The measurement of time is so common that it is not considered as sensing. However, many process control operations have critical timing requirements. Highly accurate timing can be obtained using atomic standards, but quartz crystal–con- trolled timing devices have an accuracy of better than 1 in 106, which is far more accurate than any other type of parameter measurement. Such system can be integrated onto a single silicon chip with an external crystal.

Because of its stability and high Q, quartz makes an excellent reference oscillator. Thin sections of the crystal can be used as the frequency selective element in an oscillator. The crystal slice is lapped and etched down until the desired resonant frequency is reached. Thin metal electrodes are then deposited on both sides of the crystal. A voltage applied to the electrodes induces mechanical movement and vibration at the natural resonant frequency of the crystal, producing a voltage, which in an active circuit can be used to produce a sustained frequency. Above 15 MHz, the slice becomes very thin and brittle. However, a harmonic of the fundamental frequency can be used to extend the frequency range, or a phase locked loop system can be used to lock a high frequency to the crystal frequency. This type of circuit would be used in a high frequency transmitter, such as that used to transmit sensor data from a remote location. The orientation of the crystal slice with respect to the crystal axis is of prime importance for temperature stability. The pyroelectric effect can be reduced to 1 p/m over 20°C with the correct orientation, such as the AT cut referenced to the z-axis. The quartz crystal used in watch circuits has a low frequency (32.768 kHz), and is etched in the shape of a tuning fork. The tolerance is ±20 p/m with a stability of −0.042 p/m.

Timing falls into two types of measurement: first, generating a window of known duration, such as to time a process operation or to measure an unknown frequency, and second, timing an event, as would be required in distance measurement.

Window generation is shown in Figure 6.6, which shows the block diagram of a system for measuring an unknown frequency. The output from a 1 MHz crystal oscillator is shaped and divided by 2 × 106, which generates a 1 second window to open the AND gate, letting the unknown frequency be counted for 1 second and displayed. The divider can be a variable divider, and set to give a variable gate width for timing a process operation for any required duration.

Figure 6.7 shows a block diagram of a circuit for measuring an unknown time duration, such as the time for a radar pulse to reach an object and return to the receiver. The unknown signal is used to gate the output from the 1 MHz oscillator to

Unknown frequency

Divider output

1 Sec

Input frequency

Counter input

Figure 6.6 Block schematic of 1 second gate for measuring unknown frequency.

Figure 6.7 Circuit for time measurement.

the counter, and the indicators will give the time in microseconds. Oscillator frequencies can be chosen and divided down, so that the display gives a distance measurement directly related to the time being measured.

6.3.2Piezoelectric Sensors

PZT devices are commonly used for sensors. Because of the unique relation between force and voltage, a force is readily converted into a voltage. Size for size, PZT devices are approximately 100 times more sensitive than quartz. They can be used for dynamic measurements from approximately 1 Hz to well over 10 kHz, but they

Figure 6.10 (a) Backside etch of silicon pressure sensor die, and (b) position of conditioning circuits on the top side of the die.

Diffusions are made for strain gauge and transistors, interconnections, and metal resistors are then deposited on the top surface. Figure 6.10(b) shows the topological layout of a single die on the topside of the wafer, showing the following: the location of the conditioning circuits; the strain gauge, which is four piezoresistors forming a bridge; bonding pads; and nickel-chrome resistors that are laser trimmed to correct for offset and sensitivity.

After processing is complete, each die is tested and trimmed for offset and span. The wafer is then bonded to a second constraint wafer, which gives strain relief when the wafer is assembled in a plastic package. If the pressure sensor is an absolute pressure sensor, then the cavity is sealed with a partial vacuum If the sensor is to be used to gauge differential pressures, then the constraint wafer will have holes etched through to the cavity, as shown in Figure 6.11. The diaphragm is typically 3,050 × 3,050m in a medium range pressure sensor, and the signal compensated die size will be approximately 3,700 × 3,300m. The diaphragm thickness and size both will vary with the pressure range being sensed. After testing, the wafer is cut into individual dies and the good dies are assembled in a plastic package, as shown in Figure 6.11. Plastic packages come in many different shapes and sizes, depending upon the application of the pressure sensor.

The strain gauge can be four individual piezoresistive devices forming a conventional dc Wheatstone bridge for improved sensitivity, or can be an X-ducer (Motorola patent), which is effectively four piezoresistive devices in a bridge. The integrated amplifier used to amplify the sensor signal, and the trimming resistors, are shown in Figure 6.12. The amplifiers are connected to form an instrument amplifier circuit. The circuit shown is typical of piezoelectric strain gauge elements in pressure sensors. A number of resistors are trimmed to adjust for temperature, offset, and span. After trimming, the operating temperature range is from −50° to +100°C, giving an accuracy of better than 1% of reading.

Gas flow sensors have been developed using bulk micromachining techniques. The mass air flow sensors utilize temperature-resistive films laminated within a thin film of dielectric (2 to 3 m thick), suspended over a micromachined cavity, as shown in Figure 6.13(a). The heated resistor also can be suspended over the cavity. Heat is transferred from one resistor to another by the mass of the gas flowing. The

Conditioning circuits

Glass seal

Silicon wafer

Silicon constraint wafer

Cavity

Optional pressure port for differential sensors

Figure 6.11 Cross section of micromachined absolute pressure sensor.

Figure 6.12 Integrated pressure sensor amplifier.

imbalance in the resistance caused by heat transfer is directly proportional to heat flow. The advantages of this type of anemometer are its small size and low thermal mass. It does not impede gas flow, and its low thermal mass reduces the response time to approximately 3 ms. However, the sensor is somewhat fragile and can be damaged by particulates. Figure 6.13(b) gives an example of the control circuit used with the mass flow sensor.

6.4.2Surface Micromachining

In surface micromachining technology, layers of material are deposited (e.g., polysilicon) or grown (e.g., silicon dioxide) on the surface of the silicon, and then shaped using a photolithographic process. The sacrificial oxide layers are then etched [9]. Leaving freestanding structures.

Acceleration

Attached to substrate

Figure 6.15 Topological view of the comb structure of a surface micromachined accelerometer.

•Holes are then etched in the oxide, so that the next layers can make contact with the diffused areas.

•Lightly doped polysilicon (approximately 2m thick) is then deposited on the oxide. Resist is applied and patterned, and the polysilicon is then etched to the pattern shown in Figure 6.15.

•A metal layer (aluminum) is then deposited; resist is applied and patterned; and the metal is etched to form the leads to the control electronics.

•The structure is annealed to remove stresses in the polysilicon, so that when released, the fingers in the polysilicon will lie flat and not bow or buckle.

•The final step is to remove the sacrificial oxide under the polysilicon. This is done by masking the wafer with resist, so that only the polysilicon structure is exposed, and an oxide etchant is used to etch away the exposed oxide, but not the polysilicon. This also will remove the oxide under the polysilicon, leaving a freestanding structure, as shown in Figure 6.14(b).

This is a very simplified process. Since the control electronics around the sensing structure are also included in the processing steps, these steps are omitted. Figure 6.15 shows the direction of movement of the structure during acceleration. Movement of the seismic mass causes its fingers to move with respect to the fixed fingers, changing the capacitance between them. The change in differential capacitance between C1 and C2 then can be amplified and used to measure the acceleration of the device. The values of C1 and C2 are of the order of 0.2 pF for full-scale deflection. The movement of the seismic mass gives about a 10% change in the value of the capacitance if open loop techniques are used for sensing. The integrated control electronics have temperature compensation, and are trimmed for offset and sensitivity. The control electronics can use open loop or closed loop techniques for sensing acceleration. Using open loop switched capacitor techniques, capacitance changes of approximately 0.1 fF can be sensed. Using closed loop techniques, electrostatic forces can be used to balance the forces produced on the seismic mass by

acceleration, thus holding the seismic mass in its central position. At the distances involved, electrostatic forces are very large. Approximately 2V is required for balance at full acceleration. The dc balance voltage is directly proportional to acceleration. This dc balance voltage can be achieved using techniques such as width modulation or delta-sigma modulation, where the width of the driving waveforms applied to the plates is varied, giving an electrostatic force to balance the force of acceleration. The delta-sigma modulator output can be either a serial digital output or an analog output.

An alternative accelerometer layout is shown in Figure 6.16. The seismic mass is made of polysilicon with a polysilicon lower plate, with the spacing between the plates of approximately 2m. The processing steps are the same as in the previous structure. The center of mass of the top plate is displaced from the mounting pillar or anchor, the mass has torsion suspension, and the displacement under acceleration is sensed using differential capacitive sensing. A test plate is used to simulate acceleration using electrostatic forces. A difference of approximately 2V between the test plate and the seismic mass will give the same displacement as approximately 20g, for an accelerometer designed to measure up to 50g acceleration.

Filters also have been developed using surface micromachining techniques, and are similar in layout to the accelerometer. Figure 6.17 shows the topology of a comb microresonator, and the circuit used for a bandpass filter. The Q of the filter is controlled through negative feedback, and is the ratio of the feedback MOS devices. These devices typically have a rejection of 35 dB. The microfilters or resonators are small in size and operate in the range from 20 to 75 kHz. Other practical devices using surface micromachining techniques are vibration sensors and gyroscopes.

6.5Smart Sensors Introduction

The advances in computer technology, devices, and methods have produced vast changes in the methodology of process control systems. These systems are moving away from a central control system, and towards distributed control devices.

Lower plate 1

Figure 6.16 Surface micromachined accelerometer.