CN0357

Figure 2 shows the schematic diagram of the electrochemical sensor measurement circuit. Electrochemical sensors work by allowing gas to diffuse through a membrane into the sensor and interact with the working electrode (WE). The sensor reference electrode (RE) provides feedback to amplifier U2-A to maintain a constant potential at the WE pin by varying the voltage on the counter electrode (CE). The direction of current flow on the WE pin depends on whether the reaction occurring within the sensor is oxidation or reduction. For carbon monoxide sensors, what occurs is oxidation; therefore, current flows into the working electrode, which requires the counter electrode to be at a negative voltage relative to the working electrode (typically 300 mV to 400 mV). The op amp driving the CE pin should have an output voltage range of ±1 V relative to V _REF to provide sufficient headroom for different types of sensors (Alphasense Application Note AAN-105-03, Designing Potentiostatic Circuits, Alphasense Corporation).

 
The current flowing into the WE pin is less than 100 nA per ppm of gas concentration; therefore converting this current to an output voltage requires a transimpedance amplifier with very low input bias current. The ADA4528-2 op amp has a CMOS input with a maximum input bias current of 220 pA at room temperature, making it ideal for this application.

The ADR3412 establishes a pseudo-ground reference voltage for the circuit, thereby supporting single-supply operation while consuming very low quiescent current (100 μA maximum).

Amplifier U2-A draws enough current from the CE pin to maintain 0 V between the sensor's WE and RE pins. The RE pin is connected to the inverting input of amplifier U2-A; therefore no current flows in it. This means that the current from the WE pin changes linearly with the gas concentration. Transimpedance amplifier U2-B converts the sensor current into a voltage proportional to the gas concentration.

The sensor chosen for this circuit is the Alphasense CO-AX carbon monoxide sensor. Table 1 shows typical specifications associated with this common type of carbon monoxide sensor.

WARNING: Carbon monoxide is a poisonous gas that is dangerous at concentrations above 250 ppm; extreme caution should be used when testing this circuit.

Table 1. Typical Carbon Monoxide Sensor Specifications

parameter	numerical value
Sensitivity	55 nA/ppm to 100 nA/ppm (65 nA/ppm typical)
Response time (t90, 0 ppm to 400 ppm CO	<30 seconds
Range (ppm CO, guaranteed performance)	0 ppm to 2,000 ppm
Exceeding the range limit (specs not guaranteed)	4,000ppm

The output voltage of the transimpedance amplifier is:



where I _WE is the current flowing into the WE pin and R _F is the transimpedance feedback resistor (shown as an AD5270-20U3-B varistor in Figure 1).

The maximum response of the CO-AX sensor is 100 nA/ppm and its maximum input range is 2000 ppm carbon monoxide. Based on these values, the maximum output current is 200 μA and the maximum output voltage is determined by the transimpedance resistance, as shown in Equation 2.



Applying 1.2 V to V _{REF of} the AD7790 allows ±1.2 V to be available at the output of transimpedance amplifier U2-B. Selecting a 6.0 kΩ resistor for the transimpedance feedback resistor provides a maximum output voltage of 2.4 V. Equation 3 shows the circuit output voltage as a function of ppm of carbon monoxide using a typical sensor response of 65 nA/ppm. The AD5270-20 has a nominal resistance value of 20 kΩ. Since there are 1024 resistor positions, the resistance step is 19.5 Ω. The AD5270-20 has a resistor temperature coefficient of 5 ppm/°C, which is better than most discrete resistors, and its supply current is 1 μA, which has minimal impact on the total system power consumption. Resistor R4 keeps the noise gain to a reasonable level. Choosing the value of this resistor is a trade-off between two factors: the magnitude of the noise gain and the settling time error of the sensor when exposed to high concentrations of gas. For the example in Equation 4, R4 = 33 Ω, which calculates the noise gain to be equal to 183. The 0.1 Hz to 10 Hz input voltage noise of the ADA4528-2 is 97 μV pp; therefore, the noise at the output is 18 μV pp, as shown in Equation 5. The input noise of a transimpedance amplifier appears at the output as amplified by the noise gain. For this circuit, only low-frequency noise is of concern since the sensor operates at very low frequencies. Since this is extremely low frequency 1/f noise, it is difficult to filter out. However, the sensor response is also extremely low; therefore a very low frequency low-pass filter with a cutoff frequency of 0.16 Hz can be used (R5 and C6). Even such a low frequency filter has a negligible impact on the sensor response time compared to the 30 seconds sensor response time. The system noise-free code is determined by the peak-to-peak output noise. The maximum output voltage of the ADA4528-2 is 2.4 V, so the noise-free number is: The noise-free code resolution is equal to: To utilize the full ADC range (±1.2 V), the AD8500 micropower, rail-to-rail input/output amplifier was selected to drive input to the AD7790. If the entire range is not required, the AD8500 can be removed and replaced with the AD7790 internal buffer. An important characteristic of electrochemical sensors is their extremely long time constants. When first powered on, it may take several minutes for the output to settle to its final value. When exposed to a target gas with a concentration step of half the range, the time required for the sensor output to reach 90% of the final value can be between 25 and 40 seconds. If the voltage between the RE and WE pins changes drastically, it may take several minutes for the sensor output current to establish its final value. This longer time constant also applies when the sensor is powered periodically. To avoid long start-up times, P-channel JFETQ1 shorts the RE pin to the WE pin when the supply voltage drops below the JFET's gate-source threshold voltage (approximately 2.0 V).