EE3901/EE5901 Sensor TechnologiesWeek 6 NotesSignal conditioning circuits for resistive sensors
How do you measure resistance? In the simplest case, Ohm's law may be all that you need. However, if a sensor experiences only small changes in resistance, then careful circuit designs are needed to maximise sensitivity and signal-to-noise ratio.
Interfacing with resistive sensors
There are two requirements for all resistive sensor interface circuits:
- The circuit must drive the resistive sensor with a voltage or current. Resistance can only be measured when there is a current flowing.
- The voltage and current must be low enough to avoid self-heating of the sensor (unless self-heating is part of the operational mechanism of the sensor, in which the interface circuit must ensure appropriate levels of power dissipation).
The two wire measurement circuit
The simplest approach is to measure resistance with Ohm’s law. Apply a current and measure the voltage, or vice versa. A typical circuit is shown in Figure 1.
The weakness of this approach is that the resistance of the cabling will affect the measurement. The cable resistance will vary with temperature so it not always easy to correct for. Hence, the two wire approach is generally unsuitable for precision measurements.
The four wire measurement circuit
A better design separates the voltage measurement from the excitation current, as shown in Figure 2.
If the voltmeter has a large input impedance, then the sensing current is very small in comparison to the excitation current. Consequently the voltage drop in the wires is reduced.
Measuring small changes in resistance
Suppose that we want to measure small changes in a large resistance. For example, imagine we have a sensor whose resistance changes from 10,000 Ω to 10,001 Ω. How can we design circuits to interface with such a sensor?
To formalise the problem, consider a sensor with a transfer function of the form
where . We are specifically interested in rather than the large constant . If we would measure directly, then we’d use much of our measurement precision on and make it hard to detect the interesting change in . Hence, we often need to design interface circuits that are sensitive to small changes in resistance.
An example of such a circuit is the Wheatstone Bridge.
The Wheatstone Bridge Circuit
A Wheatstone Bridge is two voltage dividers in parallel. It is often drawn in a diamond shape, like Figure 3.
There are two main ways to use the Wheatstone Bridge. These are called “balance mode” and “deflection mode”.
Balance mode
In balance mode, a current meter is connected across the bridge, as shown in Figure 4.
Resistor is the unknown resistance (e.g. the sensor). Resistance is a variable resistor (shown here as a potentiometer with its wiper shorted to one of the terminals). The idea is to adjust until a “balance” is achieved and the current meter reads . The balance condition is
The user adjusts until the bridge is balanced, then calculates the unknown resistance based upon the values of the other resistors.
Deflection mode
The second way to use the Wheatstone Bridge is in “deflection mode” where a voltage meter is connected across the bridge:
Analysing this circuit, the output voltage is
The Wheatstone Bridge is typically used with strain gauges or RTDs where a small change in resistance must be detected. Let be the sensor with transfer function
where is the parameter of interest. To ensure that the bridge is balanced when , we must choose resistor values that satisfy
Designing a bridge circuit is then a matter of choosing . In the tutorial questions you will prove that maximum sensitivity is achieved when . However, larger values of give increased linearity of at the expense of reduced sensitivity.
Using the Wheatstone Bridge with strain gauges
A common configuration is to place multiple strain gauges in a single bridge circuit. The configurations are called “quarter bridge” (one sensor), “half bridge” (two sensors) and “full bridge” (four sensors).
The half bridge
If the gauges measure equal but opposite strains, connect each sensor to the same side of the half bridge, as per Figure 6.
Analysing this circuit, we find
Simplifying,
Notice how the output voltage does not depend upon .
The full bridge
The full bridge circuit is shown in Figure 7.
Here the output voltage is
Notice that this is twice as sensitive as the half bridge.
Differential amplifiers
The voltages obtained from a Wheatstone Bridge are measured between two nodes in the circuit. For many applications we need to amplify the signal and convert it to a “single ended” voltage measured with respect to ground (for example, to connect to a microcontroller’s ADC). In other words, we want something of the form
where and are the nodes inside the Wheatstone bridge.
Your first year circuits textbook probably shows you an op-amp circuit that appears to solve this problem, most likely with something like Figure 8. However, there are some practical issues with this circuit that need to be explained.
Let’s first analyse this circuit to see how it’s supposed to work. Assuming the op-amp is operating within its linear regime, some elementary circuit analysis will show you that the output voltage is given by
Since we want an output proportional to , we split the output into common mode and differential mode components:
Solving for the input voltages:
Substituting into :
where the common mode gain is
and the differential mode gain is
To use this circuit as a differential amplifier we require . This places constraints on the values of the resistors. If we had exact resistors and an ideal op-amp then would be achievable. However in real life our resistors have some tolerance so we cannot exactly satisfy . Therefore, resistor tolerance makes this circuit impractical.
A desirable parameter for a differential amplifier is to have a high common mode rejection ratio (CMRR), defined as
There will some CMRR due to the tolerance of the resistors, as well as some CMRR due to the op-amp itself.
A further problem with this circuit is the input impedance. The resistors need to be in the range of 10s to 100s of kΩ for good op-amp performance. However, the resulting input impedance looking into and is far too small for some applications. For example, in tutorial questions you will analyse a resistance measurement where gigaohms of input impedance are required. This design is not suitable for many precision measurements. The solution is to use an instrumentation amplifier.
Instrumentation Amplifiers
An instrumentation amplifier (sometimes called an “in-amp”) is an IC that solves the problems of a basic op-amp differential amplifier. It has excellent CMRR and extremely high input impedance.
The basic design of an instrumentation amplifier is shown in Figure 9.
You should be able to recognise two voltage followers and the differential amplifier stage from earlier. Notice how the input impedance looking into and is extremely high because these signals are terminated directly into an op-amp.
The output voltage is
where is the gain of the amplifier. In the case of AC signals and dual power supplies, can be tied to ground. For single-supply operation, can be set to some nonzero value to shift the output. For example you might have a microcontroller ADC and you could set to shift the output to the middle of the ADC’s range.
Often all components except for are manufactured in a single IC. is typically external and is used to set the gain.
In next week’s practical, you will design and build a circuit using an instrumentation amplifier. Specifically, you will use the Texas Instruments INA826. Please refer to that device’s datasheet to familiarise yourself with its properties.
References
Ramon Pallas-Areny and John G. Webster, Sensors and Signal Conditioning, 2nd edition, Wiley, 2001.
Winncy Y. Du, Resistive, Capacitive, Inductive, and Magnetic Sensor Technologies, CRC Press, 2015.