Design of high-precision temperature sensing circuit

Temperature measurement sensor circuit design is an important component in most industrial measurement, control and monitoring systems. It is widely used in many specific environmental control process calculations. Some of the most common sensors can be used to measure absolute temperature or temperature changes, such as resistive temperature detectors (RTDs), diode sensors, thermistor sensors, and thermocouple sensors. In this article, we will introduce the key points of using these sensors for precision temperature measurement circuit design. Temperature sensing circuit design includes: correctly selecting the appropriate temperature sensor and the necessary signal conditioners and digital device products to measure temperature values ​​more effectively and accurately. Before we introduce the temperature measurement system , let's take a look at the advantages and disadvantages of common traditional temperature sensor design circuits. Traditional Thermocouple Sensor Design Circuit The working principle of thermocouple sensors is that when the temperature is different, a voltage (or electromotive force) is generated between the junctions of two metals of different compositions. A thermocouple is made up of two different metal ends connected, one of which is called the hot end. The other end is called the cold end and is connected to the temperature test circuit together. The difference in temperature between the hot end and the cold end causes an electromotive force. This electromotive force can be measured by the measurement circuit. Figure 1 shows a basic thermocouple sensor circuit.

Figure 1: Basic thermocouple sensor design circuit

The actual voltage generated by the thermocouple sensor depends on the relative temperature difference and the different types of metals used to make up the thermocouple sensor. The sensitivity and temperature measurement range of the thermocouple are also closely related to the two metals used. There are many types of thermocouple sensors sold on the market, which can be distinguished by the different metals used for the hot and cold ends: for example, type B (platinum/rhodium), type J (iron/nickel copper alloy), and type K (nickel chromium alloy/aluminum nickel alloy). You can choose the appropriate thermocouple sensor device according to the actual application.

The main advantages of thermocouple sensors are their robustness (the ability to restore normal operation of the system under abnormal and dangerous conditions), wide temperature range (minus 270 degrees Celsius to plus 3000 degrees Celsius), fast response, a variety of packaging, and low cost. Their limitations are mainly low accuracy and high noise.

Resistive Temperature Detection Sensor Design Circuit

The working principle of the resistance temperature detection sensor (RTD) is: since each metal has a specific and unique resistivity characteristic at different temperatures, the change in metal resistance is detected when the temperature changes, thereby obtaining a temperature measurement value. The resistance of a metal is proportional to its length and inversely proportional to its cross-sectional area. This ratio depends on the resistivity of the metal material of the sensor.

In order to measure temperature more accurately, the choice of metal material in the RTD construction becomes a key consideration. The main metals used for resistive temperature detection sensors are platinum, nickel and copper. Of these three materials, the resistance temperature detection sensor made of metal platinum is the most accurate and reliable. It is also not easily affected by factors such as polluted environments, which can ensure long-term stability and repeatability. The main advantages of these resistance temperature detection sensors include wide temperature range (-250 degrees Celsius to 900 degrees Celsius), high accuracy, and linearity. Its limitations include high cost and slightly slow response.

Thermistor sensor design circuit

Similar to the resistance temperature detection sensor RTD, the working principle of thermistor sensors is that the resistance value changes with the change of temperature. However, general thermistors have a calculable negative temperature coefficient. The main advantage of thermistor sensors is that they are low-priced and have acceptable accuracy. Their disadvantage is that the temperature range is non-linear. However, given that many microcontroller chips today have on-chip flash memory, a queryable error-corrected data table can be created to reduce the accuracy impact caused by nonlinear problems. If the temperature range to be measured is within -100 degrees Celsius to +300 degrees Celsius, thermistor sensors can still be used as relatively reliable and relatively accurate temperature measurement devices.

Temperature Measurement System

In a temperature monitoring system, the sensor must convert the temperature into an electrical signal, go through a signal conditioning stage (the signal processing depends on the different sensors), and then send it to an analog-to-digital converter (ADC) for conversion to obtain a numerical value. The system also requires communication peripheral circuits to interface with other large devices to provide feedback, or send the numerical value to the on-chip flash memory to store the measurement value or display it as necessary. Figure 2 shows the basic block diagram of the temperature measurement system.

Figure 2: Temperature measurement system block diagram

Although Figure 2 shows that signal processing is performed before the ADC, whether processing is required after the signal conversion depends on whether the system is analog or digital. The overall accuracy depends on the noise control, offset, pre-processing circuits and gain errors introduced by the ADC. Many applications require real-time temperature data acquisition from a remote location, such as mining, industry, and various automation applications. Serial communication protocols such as UART and I2C can be used to transmit this temperature data to the main system controller. How to Improve the Accuracy of Thermocouple Temperature Sensors Thermocouple sensor-based temperature control systems are widely used in industrial control due to their wide temperature range advantages. Its basic principle is to sense temperature by measuring the junction electromotive force. But it requires an assumption: the cold end is assumed to be exactly at zero degrees Celsius. However, it is not practical to keep the cold end at this temperature all the time. In order to achieve accurate measurement, a technical means needs to be applied, which we can call cold junction compensation (CJC). For cold junction compensation, a temperature sensor (mounted at the top of the cold end) is added to the precision temperature measurement system based on thermocouples to measure the temperature of the cold end. The most commonly used temperature sensor for cold junction temperature measurement is thermistor because of its low cost and temperature range that can cover the cold junction temperature and meet most applications. To measure the CJC voltage, first find the cold junction temperature and then check the thermocouple electromotive force to find the temperature. The CJC voltage is generated after adding the cold junction voltage, and its corresponding temperature is the actual temperature. The electromotive force generated by the thermocouple is only a few uV, which makes it very susceptible to noise interference. In addition, before this signal is transmitted to the analog-to-digital converter, it needs to be amplified (which also increases noise and offset). In precision measurement, this type of noise and offset should be removed. Let's take an example to illustrate how to use the correlated double sampling method (CDS) to eliminate offset and reduce low-frequency noise. CDS can reduce low-frequency noise and offset in the signal processing stage. First, measure the zero reference offset (which can be measured by shorting both inputs), and then measure the thermocouple voltage. When measured directly with the thermocouple signal, it includes the actual thermocouple voltage, noise voltage, and offset (see equation 1). The zero reference reading includes noise and offset (see equation 2). (Equation 1) VTCouple_Signal = VTC + VN + Voffset (Equation 2) VZero_Ref = VN + VoffsetThe relationship between the previous zero reference sample value and the current zero reference measurement value is: (Equation 3) VZero_ref_Prev = (VN + Voffset)*Z-1Then , the difference between the current thermocouple measurement and the previous zero reference level is: (Equation 4) Vsignal = (VTC + VN + Voffset) - (VN + Voffset)*Z-1Voffset is static, so its current value is the same as the previous sample value. VN is not static, because it is noise and drift, so it needs to be removed. Subtracting the previous noise value from the current sample value will remove the low frequency noise. Therefore, the correlated double sampling method CDS works like a high pass filter. EECOL_2011Mar09_DSP_TA_50.pdfThe analog-to-digital converter ADC itself has a low pass filter to remove high frequency noise. However, an IIR filter at the output of the ADC will help further attenuate the noise band passing through it or being transmitted to the ADC. Mixed-signal controllers available on the market can be configured with digital filters, which can save CPU cycles by filtering through the device's own hardware processing without filtering in the firmware circuit. Figure 3 shows a thermocouple-based temperature monitoring system that uses Cypress's PSoC5 and PSoC3 devices. These devices have on-chip 20-bit resolution delta-sigma ADCs, built-in programmable gain buffers for signal amplification, and built-in digital filter blocks (DFBs) for filtering. It provides a highly integrated temperature measurement system. However, due to the presence of thermocouples in the design, an additional gain stage may be required. This gain can be achieved through an amplifier, and the on-chip programmable gain amplifier (PGA) can be used.

Figure 3: Temperature measurement system circuit based on thermocouple sensor

In the system of Figure 3, the analog MUX, AMuxCDS, and AMuxCDS_1 are used to convert the positive and negative output signals of the sensor to the positive input of the analog-to-digital converter to implement correlated double sampling. Now the question is how to make both sensor circuits have the same zero reference value when using the same analog-to-digital converter. The answer is this - thermistors and thermocouples have different output voltage ranges, so different amplification is required. The ADC in PSoC3 and PSoC5 devices has multiple configurations that can change the operation time. For different gain settings, the offset is also different, so correlated double sampling is required in both sensor circuits. This will help eliminate the offset of the entire analog signal chain. AMux is used to select the sensor between thermocouple and thermistor. Direct memory access (DMA) reads the ADC value and writes it to the digital filter block (DFB) to filter the noise.

RTD and Thermistor Temperature Sensor Design Circuit

When measuring temperature using resistance temperature detectors (RTDs) and thermistors, resistance needs to be measured, so the measurement method determines the accuracy of the system. In order to measure accurate signals, differential inputs should be used instead of single-ended inputs. Differential inputs can eliminate common noise and do a good job of achieving μV sensitivity (mV sensitivity is much better than single-ended inputs). Let's look at two different ways of connecting the -ve input to the ADC, as shown in Figure 4.

Figure 4: Two different -ve connection designs

The circuit design on the right side of Figure 4 is better than the one on the left side. In the circuit on the right side, -ve is directly connected to the reference voltage close to the voltage divider resistor. The circuit on the right side can help reduce noise during measurement and errors caused by PCB layout or trace impedance, etc.

The temperature measurement system based on thermistors can be said to be a combination of Figures 3 and 4. Now, let's look at the measurement system using RTDs. The temperature sensor made of metal platinum RTD is the most accurate and stable in terms of time and temperature, so it should be used in applications where precise measurements are required. The voltage drop across the RTD can be measured, and the measurement method is the same as thermistor, usually using the 2-wire method. When connecting the RTD to the measurement system, it is necessary to go through a long circuit. If a voltage source is used as an excitation, the circuit trace resistance becomes the main source of measurement error. Figure 5 shows the difference between the 2-wire measurement circuit and the 4-wire measurement circuit design.

Figure 5: Measurement circuit design for 2-wire connection and 4-wire connection

In a 2-wire circuit, the resistance of the RTD (RRTD) can be measured as per Equation 5. However, if we look at this circuit, there is another resistance, Rwire, that can cause a measurement error.

(Equation 5) RRTD = (Rref+Rwire)*( V2-V1)/(V-V2)

On the other hand, the resistance of the RTD in a 4-wire circuit can be calculated as per Equation 6. Because the measurement system has a high input impedance, there is no current in the measurement system, so the resistance between the voltage divider node and the measurement system is in series and has no effect. The resistance of the RTD (RRTD) can be derived as per Equation 6.

(Equation 6) RRTD = Rref*( V2-V1)/(V4-V3)

Let's look at Equation 5 and Equation 6 again. The accuracy of the measurement depends mainly on the accuracy of Rref. To overcome this problem in voltage excitation, the RTD uses a constant current source instead of a voltage source. When a constant current source is used, the voltage drop across the RTD depends only on its resistance value and the constant current source value. However, the accuracy of the measurement when using a constant current source excitation depends on the accuracy of the current source. Since the temperature measurement is accurate, the DAC current should be calibrated by the TIA. Figure 6 shows an RTD-based temperature measurement system implemented using PSoC3 and PSoC5 devices. These devices have on-chip current sources and do not require additional analog amplifier circuits. At the same time, these devices have on-chip TIAs that can be used to calibrate the IDAC.

Figure 6: Temperature measurement circuit design based on resistance temperature detector (RTD)

Below we summarize the basic points for designing precision temperature measurement circuits:

1. Choose the right sensor for the application.

2. CDS helps to get accurate sensor readings, avoid offset errors, and eliminate low-frequency noise.

3. For thermocouple systems, filters can be used to remove noise.

4. Current excitation systems can improve accuracy by eliminating inaccurate reference resistors in the circuit.

5. If voltage excitation is used, a 4-wire measurement system should be used.

6. The overall accuracy of the system depends on the accuracy and precision of the signal chain. Therefore, it is recommended to use a high-precision and high-resolution Delta Sigma analog-to-digital converter ADC.

7. In order to adapt to environmental changes while ensuring accuracy, it is recommended to use a mixed-signal implementation.

The temperature sensing circuit part is an important part of many industrial systems or embedded designs. We have discussed the various challenges faced in accurately reading sensor values ​​and how to use precise analog technology to improve accuracy. These are general techniques that are also applicable to other sensor interface circuits.