The circuit shown in Figure 1 is a fully isolated low-power pH sensor signal conditioner and digitizer with automatic temperature compensation for high accuracy.

This circuit provides 0.5% accuracy readings for pH values ​​in the range of 0 to 14 with noise-free code resolution greater than 14 bits, making it suitable for a variety of industrial applications such as chemicals, food processing, water treatment, wastewater analysis, etc.

The circuit supports many pH sensors with very high internal resistances (ranging from 1 MΩ to several GΩ), and its digital signal and power isolation design makes it immune to noise and voltage transients commonly found in harsh industrial environments.

Basic principles of pH measurement

pH is a measure of the relative amounts of hydrogen ions and hydroxide ions in an aqueous solution. In molar terms, water at 25°C contains 1 × 10 ⁻⁷ moles of hydrogen ions per liter, and the hydroxide ion concentration is the same. A neutral solution is a solution in which the concentration of hydrogen ions is exactly equal to the concentration of hydroxide ions. pH is another way of expressing hydrogen ion concentration and is defined as follows:

Therefore, if the hydrogen ion concentration is 1.0 × 10 ⁻² mol/L, the pH is 2.00.

pH electrodes are electrochemical sensors used in many industries, but are of particular importance to the water treatment and wastewater industries. A pH probe consists of a glass measuring electrode and a reference electrode, similar to a battery. When the probe is placed in a solution, the measuring electrode generates a voltage that depends on the activity of hydrogen in the solution, and this voltage is then compared to the potential of the reference electrode. As the acidity of the solution increases (the pH value becomes lower), the potential of the glass electrode increases positively (+mV) relative to the reference electrode; as the alkalinity of the solution increases (the pH value becomes higher), the potential of the glass electrode becomes positive relative to the reference electrode Negative enhancement (-mV). The difference between these two electrodes is the measured potential. Under ideal conditions, a typical pH probe would produce 59.154 mV/pH unit at 25°C, expressed using the Nernst equation as:

The equations show that the voltage produced depends on the acidity and alkalinity of the solution and varies with hydrogen ion activity in a known way. Changes in the temperature of a solution change its hydrogen ion activity. When a solution is heated, the hydrogen ions move faster, resulting in an increase in the potential difference between the two electrodes. Additionally, as the solution cools, the hydrogen activity decreases, causing the potential difference to decrease. By design, under ideal circumstances, the electrode will develop a potential of zero volts when placed in a buffer solution with a pH of 7.

A good reference book on pH theory is pH Theory and Practice (Publisher: Radiometer Analytical SAS; Place of Publication: Villeurbanne Cedex, France).

circuit details

This design provides a total solution for temperature compensated pH sensors. There are three important circuit stages: the pH probe buffer, the ADC, and the digital and power isolators, as shown in Figure 1.

The AD8603 , a precision micropower (50 μA maximum) and low noise (22 nV/√Hz) CMOS operational amplifier, is configured as a buffer connected to the input of one of the AD7793 channels. The AD8603 has a typical input bias current of 200 fA, providing an effective solution for high internal resistance pH probes.

The pH sensing and temperature compensation system is based on the AD7793, a 24-bit (Σ-Δ) ADC. It has three differential analog inputs and an on-chip low-noise, programmable gain amplifier (PGA) ranging from unity gain to 128. The AD7793's maximum power consumption is only 500 μA, making it suitable for any low-power application. There is a low-noise, low-drift internal bandgap reference, and an external differential reference can also be used. The output data rate is software programmable and can vary from 4.17 Hz to 470 Hz.

The ADuM5401 (quad digital isolator with integrated dc-to-dc converter) provides digital signal and power isolation between the microcontroller and the AD7793 digital lines. Utilizing iCoupler chip-scale transformer technology, it is possible to isolate logic signals and power feedback paths in DC/DC converters.

pH sensor interface buffer

A typical pH probe electrode is made of glass, which creates an extremely high resistance, ranging from 1 MΩ to 1 GΩ, acting as a resistor in series with the pH voltage source, as shown in Figure 2.

The buffer amplifier bias current flowing through this series resistor introduces offset errors into the system. To isolate the circuit from this high source resistance, a high input impedance, ultra-low input bias current buffer amplifier is required in this application. The AD8603 is used as a buffer amplifier for this application, as shown in Figure 2. The low input current of the AD8603 minimizes voltage errors caused by bias current flowing through the electrode resistors.

For a pH probe with a series resistance of 1 GΩ at 25°C, the offset error is 0.2 mV (0.0037 pH) for a typical input bias current of 200 fA. Even at a maximum input bias current of 1 pA, the error is only 1 mV.

The cutoff frequency of the 10 kΩ/1 μF low-pass noise filter for the buffer amplifier output is f= 1/2πRC, which is 16 Hz.

Guarding, shielding, high insulation resistance pillars, and other such standard picoamp methods must be utilized to minimize leakage at the high impedance input of the AD8603 buffer.

ADC channel 1 configuration, pH sensor

This stage involves measuring the small voltage produced by the pH electrode. Table 1 shows the technical specifications of a typical pH probe. Based on the Nernst equation, the full-scale voltage range from the probe is ±414 mV (±59.14 mV/pH) (25°C) to ±490 mV (±70mV/pH) (80°C).

When reading the pH probe output voltage, the ADC uses an external 1.05 V reference source configured with a gain of 1. The full-scale input range is ±V _REF /G = ±1.05 V, and the maximum signal from the pH probe is ±490 mV (80°C).

Since the output of the sensor is bipolar and the AD7793 operates on a single supply, the signal generated by the pH probe should be biased above ground so that it is within the acceptable common-mode range of the ADC. This bias voltage is generated by injecting 210 μA IOUT2 ​​current into the 5 kΩ 0.1% resistor, as shown in Figure 2. The result is a common-mode bias voltage of 1.05 V, which also serves as the ADC reference voltage.

ADC channel 2 configuration, RTD

The second channel of the ADC monitors the voltage developed on the RTD driven by the IOUT2 ​​current output pin of the AD7793. The 210 μA excitation current drives the series combination of RTD and precision resistor (5 kΩ, 0.1%). (See Figure 1).

The temperature coefficient of pure platinum is 0.003926 Ω/Ω/°C. According to DIN Std. 43760-1980 and IEC 751-1983, the normality factor for industrial RTDs is 0.00385 Ω/Ω/°C. RTD accuracy is typically referenced to 0°C. The DIN 43760 standard recognizes two categories, as shown in Table 2, and the ASTM E–1137 standard recognizes two levels, as shown in Table 3.

The calculation formula for RTD resistance value is:

The RTD resistance varies from 0°C (1000 Ω) to 100°C (1385 Ω), producing a voltage signal ranging from 210 mV to 290 mV with an excitation current of 210 μA.

A precision 5 kΩ resistor generates 1.05 V as an external reference. When a gain is 1, the analog input range is ±1.05 V (±V _REF /G). The architecture forms a ratiometric configuration. Changes in the excitation current value will not affect system accuracy.

Although 100 Ω Pt RTDs are very common, other resistors (200 Ω, 500 Ω, 1000 Ω, etc.) and materials (nickel, copper, nickel-iron) can be specified. This application uses a 1 kΩ DIN 43760 Class A RTD for temperature compensation of the pH sensor. A 1000 Ω RTD is not as sensitive to line resistance errors as a 100 Ω RTD.

Use a 2-wire connection as shown in Figure 3. Apply a constant current to the RTD pin while measuring the voltage across the RTD itself. The measurement device is an AD7793, which exhibits high input resistance and low input bias current. The sources of error in this scheme are the pin resistance, the stability of the constant current source generated by the AD7793, and the input impedance and/or bias current and corresponding drift in the input amplifier.

Another possibility to eliminate line resistance errors is a 3-wire RTD configuration, see Current Note CN-0287 for details .

Output encoding

The EVAL-SDP-CB1Z demonstration platform board and PC process the data output by the AD7793.

Digital and power isolation

The ADuM5401 isolates the ADC digital signal while providing an isolated, regulated 3.3 V power supply to the circuit. The input to the ADuM5401(V _DD1 ) should be between 3.0 V and 3.6 V. Pay attention to the layout of the ADuM5401 to minimize EMI/RFI issues. For more details, refer to the AN-1109 application note: Radiation Control Recommendations for iCoupler Devices .

System calibration

The AIN3(+) input of the AD7793 is used to measure the voltage drop across a precision 5 kΩ 0.1% resistor. To accurately measure the RTD resistance, the ±5% variation in IOUT2 ​​current must be considered. On this basis, divide this voltage by 5 kΩ to find the exact current value of IOUT2. The RTD resistance is calculated by dividing the voltage in the RTD by the exact current value of IOUT2.

Calibrate the pH meter in the EVAL-CN0326-PMDZ evaluation software using the two-point calibration procedure shown in Figure 4 .

Users need to use at least two buffer solutions, of which a neutral pH buffer solution with a value of pH-7 is used to eliminate imbalances caused by the pH probe and system. Neutral buffer solutions can be used to set the first calibration point. The pH of the first buffer solution depends on the pH of the solution that needs to be measured. When measuring alkaline base solutions, you can use pH-10 buffer solutions; when measuring acidic solutions, you can use pH-4 buffer solutions. To improve the accuracy of measurements, a three-point calibration can be implemented. The method is to use two different sets of buffer solutions in steps 2 and 3, as shown in Figure 4, where the pH-7 solution is used to eliminate the imbalance.

The software includes a list of NIST recommended buffer solutions. Each buffer solution described in the list has its own temperature coefficient, ranging from 0°C to 95°C, which can be found in the book "pH Theory and Practice" published by Radiometer Analytical. The software uses this table to correlate the mV input from the pH probe to the correct pH value that corresponds to the temperature reading from the RTD sensor, using linear interpolation to fill in the blanks in the table. The user can choose to enable/disable the continuous temperature compensation option by clicking the green button as shown in Figure 4.

Buffer solutions used for pH sensor calibration are common in the market. Other NIST certified pH reference solutions can also be used for calibration. Since there are a variety of buffer solutions on the market to choose from, the software also provides users with an option to calibrate using the required NIST certified pH reference solution, as shown in Figure 4.

The software also provides the user with an option to use other RTD resistor values, but the default value is set to 1000 Ω.

System Noise Considerations

If the output data rate is 16.7 Hz and the gain is 1, the rms noise of the AD7793 is equal to 1.96 μV (noise referred to the input, from the AD7793 data sheet). Peak-to-peak noise is equal to:

6.6 × RMS noise = 6.6 × 1.96 μV = 12.936 μV

If the pH meter has a sensitivity of 59 mV/pH, the noise-free resolution pH level that the pH meter can measure is

12.936 μV / (59 mV/pH) = 0.000219 pH

This includes only the noise contribution of the AD7793. Actual system results are shown in the next section.

Test data and results

All data capture operations are implemented through CN0326 LabVIEW evaluation software. A Yokogawa GS200 precision voltage source was used to simulate the pH sensor input.

By sweeping precision voltages from −420 mV to +420 mV in 1 mV increments, the EVAL-CN0326-PMDZ captures data based on user-defined calibration options.

The peak-to-peak noise of the AD8603 buffer and AD7793 in a real system is determined by shorting the pH probe BNC connector and taking 1000 samples. As shown in the histogram in Figure 5, the code distribution is approximately 500 codes, which equates to a peak-to-peak noise of 31.3 μV and an equivalent pH reading distribution of 0.00053 pH peak-to-peak.

When testing the system, three different resistors were connected in series with the ADC input to simulate the different impedances of the high-impedance glass electrodes. The system was also calibrated and the result was 60 mV/pH. According to Figure 6, the linear error increases as the impedance of the simulated glass electrode increases. Figure 6 also shows that for a pH probe impedance of 200 MΩ, the linearity error is less than 0.5% over the entire simulated pH output voltage range.

Test data was collected using the evaluation board shown in Figure 7. Complete documentation for this system can be found in the CN-0326 Design Support Package .