Analysis of working principle of thermopile gas sensor based on NDIR principle

Non-dispersive infrared (NDIR) spectrometers are often used to detect gases and measure the concentration of carbon oxides such as carbon monoxide and carbon dioxide. An infrared beam passes through the sampling chamber, and each gas component in the sample absorbs infrared light of a specific frequency. The concentration of the gas component can be determined by measuring the amount of infrared absorption at the corresponding frequency. The reason why this technique is non-dispersive is because the wavelength through the sampling cavity is not pre-filtered; instead, the optical filter is placed in front of the detector to filter out all rays outside the wavelength that the selected gas molecules can absorb. .

The circuit shown in Figure 1 is a complete circuit of a thermopile gas sensor based on the NDIR principle. The circuit is optimized for carbon dioxide detection, but the thermopile with different filters can also accurately measure the concentration of multiple gases.

The printed circuit board (PCB) is sized with the Arduino expansion board and interfaces with the Arduino compatible platform board EVAL-ADICUP360. Signal Tuning Reasons The low noise amplifiers AD8629 and ADA4528-1 and the precision analog microcontroller ADuCM360 integrate a programmable gain amplifier, dual 24-bit Σ-Δ analog-to-digital converter (ADC) and ARM Cortex-M3 processor.

Thermopile sensors consist of a large number of thermocouples that are typically connected in series (or occasionally in parallel). The output voltage of the series thermocouple depends on the temperature difference between the thermocouple junction and the reference junction. This principle is called the Seebeck effect and is named after its discoverer Thomas Johann Seebeck.

This circuit uses the AD8629 op amp to amplify the thermopile sensor output signal. Thermopile output voltages are relatively small (from a few hundred microvolts to a few millivolts), requiring high gain and very low offset and drift to avoid DC errors. The high internal resistance of the thermopile (typically 84 kΩ) requires an amplifier with low input bias current to minimize errors, while the AD8629 has a bias current of only 30 pA (typ). The device drifts very slowly with time and temperature and does not introduce additional errors after the calibration temperature is measured. A pulsed source that is synchronized with the ADC sampling rate minimizes errors caused by low frequency drift and flicker noise.

The voltage noise spectral density of the AD8629 at 1 kHz is only 22 nV/√Hz, which is lower than the voltage noise density of the thermopile 37 nV/√Hz.

The current noise spectral density of the AD8629 at 10 Hz is also very low, typically 5 fA/√Hz. This current noise flows through the 84 kΩ thermopile, and the noise contribution at 10 Hz is only 420 pV/√Hz.


Figure 1. NDIR gas detection circuit

The low-noise amplifier ADA4528-1 is used as a buffered sensor with a common-mode voltage of 200mV, so the NTC and thermopile signal outputs meet the ADuCM360 buffered mode input requirements: ADuCM360 ADC buffer mode input is AGND + 0.1 V to approximately AVDD - 0.1 V. CN-0338 The Arduino expansion board is compatible with other types of Arduino compatible platforms with single-ended input ADCs.

The circuit's chopping frequency range is from 0.1 Hz to 5 Hz and is software selectable. Low Dropout Regulator The ADP7105 l generates a stable 5 V output voltage to drive the IR source and is controlled by the ADuCM360. The ADP7105 features a soft-start feature that eliminates inrush currents generated by cold-start sources.

The ADuCM360 integrates a dual-channel, 24-bit, Σ-Δ ADC that simultaneously samples dual thermopile cells over a programmable rate range of 3.5 Hz to 3.906 kHz. The data sampling rate range of the NDIR system is limited to between 3.5 Hz and 483 Hz for optimum noise performance.

Thermopile detector working principle

In order to understand the thermopile, it is necessary to review the basic theory of thermocouples.

If two different metals are connected at any temperature above absolute zero, a potential difference (thermoelectric EMF or contact potential) is generated between the two metals. This potential difference is a function of junction temperature (see thermoelectric EMF in Figure 2). Circuit).

If the two wires are connected at two locations, two junctions are formed (see the thermocouple connected to the load in Figure 2). If the temperature of the two junctions is different, a net EMF is generated in the circuit and a current flows, and the current is determined by the EMF and the total resistance of the circuit (see Figure 2). If one of the wires is broken, the voltage at the breakpoint is equal to the net thermoelectric EMF of the circuit; and if the voltage is measurable, it can be used to calculate the temperature difference between the two junctions (see thermocouple in Figure 2). Voltage measurement). Remember that a thermocouple measures the temperature difference between two junctions, not the absolute temperature at a junction. The temperature at the measurement junction can only be measured if another junction (often referred to as a reference junction or cold junction) is known.

However, it is difficult to measure the voltage generated by the thermocouple. Assume that the voltmeter is connected to the first thermocouple measurement circuit (see Figure 2 for the actual thermocouple voltage measurement of the cold junction). The wires connected to the voltmeter form more thermocouples at the junction. If these additional junction temperatures are the same (regardless of temperature), the intermediate metal rule indicates that they have no net contribution to the total EMF of the system. If their temperatures are different, an error occurs. Thermoelectric EMF is produced for each pair of different contact metals – including copper/solder joints, kovar/copper sheets (can be an alloy for IC lead frames) and aluminum/corinable (welding in ICs) ) - In practical circuits, the problem is more complicated, and it is necessary to be extremely careful to ensure that all junction pairs of the thermocouple peripheral circuit (except the measurement node and the reference node itself) have the same temperature.


Figure 2. Thermocouple principle

The thermopile is made up of a large number of thermocouples connected in series, as shown in Figure 3. The thermoelectric voltage generated by the thermopile is much higher than that of a single thermocouple.


Figure 3. Multiple thermocouples make up a thermopile

In NDIR applications, filtered pulsed infrared light is applied to the series active nodes; therefore, the junctions heat up, producing a smaller thermoelectric voltage. The temperature of the reference junction is measured by the thermistor.

The positive or negative charge center transients or steady states of many gases do not coincide. In the infrared spectrum, gases can absorb specific frequencies, which can be used for gas analysis. When infrared radiation is injected into the gas, and when the self-resonant frequency of the molecule matches the infrared wavelength, the gas molecules will resonate with the incident infrared light according to the energy level transition of the atom.

For most infrared gas detection applications, the composition of the target gas is known and therefore does not require gas chromatographic analysis. However, if the absorption lines of different gases overlap, the system must deal with the mutual interference between these gases.

Carbon dioxide has an absorption peak between 4200 nm and 4320 nm, as shown in Figure 4.


Figure 4. Absorption spectrum of carbon dioxide (CO2)

The output wavelength range of the infrared source and the absorption spectrum of the water also determine the choice of detection wavelength. Below 3000 nm, and between 4500 nm and 8000 nm, water has a strong absorption. If there is moisture in the target gas (high humidity), the detected gas will be affected by strong interference within these ranges. Figure 5 shows the carbon dioxide absorption spectrum overlapping with the absorption spectrum of water. (All absorption data comes from the HITRAN database).


Figure 5. The absorption spectrum of carbon dioxide and water overlaps

If infrared light is applied to the dual thermopile sensor and a pair of filters is installed such that one of the filters has a center wavelength of 4260 nm and the other center wavelength is 3910 nm, then the voltages of the two thermopiles are measured. The ratio of carbon dioxide can be measured. A filter whose center wavelength overlaps with the absorption wavelength of carbon dioxide is used as a measurement channel, and a filter whose center wavelength is outside the absorption wavelength of carbon dioxide is used as a reference channel. When the reference channel is used, the measurement error caused by dust or radiation intensity attenuation can be eliminated. It is important to note that carbon dioxide and water vapor do not absorb almost all of the 3910 nm infrared light; this makes the area an ideal location for the reference channel.

The thermopile used in the NDIR test has a relatively high internal resistance, and 50 Hz/60 Hz power line noise is coupled into the signal path. The internal resistance of the thermopile may be around 100 kΩ, causing thermal noise to become the dominant noise in the system. For example, the thermopile sensor voltage noise density selected in the system of Figure 1 is 37 nV/√Hz. In order for the system to have the best performance, the sensor should be output with as large a signal as possible and a lower gain in the circuit.

The best way to maximize the signal from the thermopile sensor is to use a chamber with high reflection characteristics, which ensures that as much radiation as possible enters the detector without being absorbed by the chamber. The use of a reflective chamber to reduce the amount of radiation absorbed by the chamber also reduces system power consumption because a low power radiation source can be used.