STMicroelectronics – Signal conditioning for pyroelectric passive infrared sensors

Pyroelectric passive Infrared (PIR) sensors are frequently used in everyday life. They enable motion detection, and may be used in security systems, automatic doors, automatic lighting and many other applications.

Active sensors emit energy in forms such as ultrasound, light or microwaves, and determine that a change has occurred when the reflection of the emitted signal is disturbed. Passive sensors do not emit signals, but rather detect changes in the amount of IR radiation. These passive sensors consume less energy than active ones.

How does the sensor work?
PIR sensors consist of two halves which are sensitive to IR radiation. The configuration of the sensor enables it to detect motion: its signal represents the difference between the infrared levels detected by each half, as shown in Figure 1. As long as both halves see the same amount of background infrared radiation, the sensor is detecting no motion. But if one of the halves sees a different IR level from the other, the sensor’s output will go either high or low.

Fig. 1: PIR sensor’s operation

Fig. 1: PIR sensor’s operation

The area of the two IR- sensitive rectangles is small at around 2mm2 for each. A Fresnel lens is commonly used to increase this area, and thus to extend the sensor’s detection range.

Sensor signal conditioning
In normal operation, the temperature of a person’s body is higher than the ambient temperature, and so when a person moves across the PIR sensor’s field of view, it senses the increased level of IR radiation, and emits a small AC signal of around 1mVpp. This small AC voltage swing is around a DC signal which may vary significantly from one production unit to another.

This creates a signal-conditioning problem to be solved: the DC part of the signal must be cancelled out, and only the AC part amplified. As the signal will be disturbed by the environment, noise filtering will also be helpful. Operational amplifiers can help the designer to implement all these functions.

For the detection of human motion, the frequencies of interest range between 0.5Hz and 5Hz. A TSU102 dual-channel op amp may be used to amplify and filter signals in this frequency range, as shown in Figure 2.


Fig. 2: Schematics of a circuit for amplifying and filtering low-frequency signals

The AC signal generated by the PIR sensor is amplified by 69dB: 35dB at the first stage and 34dB at the second.

The TSU102 is well suited to this application as the gain-bandwidth product must be greater than 2.7kHz and the TSU10x series is rated at 8kHz. The calculation for the gain-bandwidth product is the maximum frequency x gain x 10:

5 x 53 x 10 = 2.7kHz.

In this calculation, the factor 10 has been included to provide some margin, ensuring that the system will not be limited by the device’s gain-bandwidth product. In addition, since the DC is cancelled in motion-detection applications, the device’s input offset voltage is not relevant. In a portable application, power consumption is an important criterion.

In the circuit in Figure 2, most of the power is consumed by the sensor, which draws 19μA. The rest of the application draws 3.6μA:
• 1.2μA drawn by the TSU102
• 2.4μA drawn by the divider bridge comprised of R6 and R7

The power consumption of the divider bridge could be reduced by multiplying the resistor values, but at the cost of reliability. When impedances are high, the impact of dust or moisture is higher. These perturbations can create parasitic impedances.

What about a digital output?
It might be beneficial to have a digital rather than an analogue output, making the implementation easier on the microcontroller side. To accomplish this, a final stage may be added to perform a window comparator. When a heat source is detected, the output of U3 or U4 will be at its low state. Figure 3 shows the schematics of this stage.


Fig. 3: Window comparator stage

The divider bridge comprised of resistors R6, R7, R8 and R9 is used to set the voltage reference of devices U3 and U4. These resistors replace R6 and R7 in Figure 2.

Since the TSU101 is an input/output rail-to-rail op amp, there is no constraint on the input common-mode voltage. This means that, for as long as the voltage references of U3 and U4 are within the supply-voltage range, the window comparator will work.

In this case, U3 will have a reference set to 0.84*Vcc:

When the signal (Vout2) is greater than this reference, which is equal to 2.77V if Vcc = 3.3V, the output of U3 will be at its low state, close to ground.

As with U3, U4 is used to detect when the signal is less than its reference. In this example, its voltage reference has been set to 530mV by the divider bridge. The voltage reference calculation is made in this way:

Thus when the signal (Vout2) is less than 0.53V, the output voltage of U4 will be in the low state.

Thanks to these schematics, we can see that a single quad-channel op amp can be used for this application: one channel for the first stage, another one for the second stage and two others for the last one. This means that, when using a four-channel TSU104, the circuit only needs one active component for signal conditioning.

It should be noted that, while the TSU104 is not a comparator, the use of an op amp for this function causes no problems when operating at such a low speed.

The circuit may be configured differently if it only needs to generate a single digital output: in this case, a NAND gate may be connected to the outputs of U3 and U4.

Hardware measurements
Figure 4 shows the signal at the output of the PIR sensor. The start-up time can be seen here, but this graph does not show whether anything has been detected.


Fig. 4: Output voltage of the PIR sensor

Figure 5 shows the signals after amplification and filtering. The detection events which were invisible before signal conditioning can now be seen clearly.


Fig. 5: PIR signal after conditioning

When motion is detected, output U3 or U4 goes low, close to ground. This means that the output of the entire signal chain can be directly plugged into a microcontroller’s input. This can be used to trigger an alarm, switch on a light or take some other action in response to the detection event.

Note that initialisation time is required for the sensor to warm-up and for the capacitors to be charged. Blank time therefore has to be taken into account in the MCU’s application software.

While PIR sensors are familiar and widely used, the signal which they generate is noisy and has a very small amplitude. Op amps may be used to amplify and filter the signal, and to compare the amplified signal with threshold voltages before sending them to the I/O of an MCU, with no need for an ADC.

By using a TSU104 quad-channel op amp, the application may be made compatible with 3.3V MCUs, and the whole analogue chain draws only 24μA.