Application and requirement analysis of analog-digital converter in temperature measurement system

1 Introduction

    There are several types of sensors that measure temperature. Choosing the appropriate temperature sensor for a specific application depends on the temperature range to be measured and the required temperature. The system depends on the performance of the temperature sensor and the analog-to-digital converter that digitizes the sensor output. In most cases, because the sensor signal is very weak, a high-resolution analog-to-digital converter is required. Σ-Δ analog-to-digital converters have high resolution and are therefore very suitable for such systems, and such converters often contain built-in circuits required for temperature measurement systems, such as excitation current sources. This application note mainly introduces the available temperature sensors (thermocouples, resistance temperature detectors (RTD), thermistors and thermistors) and the circuits required to connect the sensors and analog-to-digital converters, and introduces the conversion The performance requirements of the device.

    Thermocouple

   001.jpg

    Figure 1. The analog circuit part of the thermocouple temperature system

    Thermocouples are composed of two different types of metals. When the temperature is higher than zero degrees Celsius, a temperature difference voltage will be generated at the junction of the two metals. The magnitude of the voltage depends on the deviation of the temperature relative to zero degrees Celsius. Thermocouple has the advantages of small size, ruggedness, relatively cheap price, and wide operating temperature range. It is very suitable for extremely high temperature (up to 2300°C) measurement in harsh environments. However, the output of the thermocouple is in the millivolt level, so it needs to be precisely amplified for further processing. The sensitivity of different types of thermocouples is also different, generally only a few millivolts per degree Celsius, so in order to accurately read the temperature, a high-resolution, low-noise analog-to-digital converter is required. When the thermocouple is connected to the copper printed wire of the printed circuit board, another thermocouple junction will appear at the place where the thermocouple is connected to the copper printed wire. The result is a voltage that cancels the thermocouple voltage. In order to compensate for this reverse voltage, we place a temperature sensor at the thermocouple-copper wire junction to measure the temperature at the junction. This is the so-called cold junction.

    Figure 1 shows a thermocouple system using a 3-channel, 16/24-bit AD7792/AD7793 sigma-delta analog-to-digital converter (6-channel AD7794/AD7795 can also be used). The on-chip instrumentation amplifier first amplifies the thermocouple voltage, and then performs analog-to-digital conversion on the amplified voltage signal through an analog-to-digital converter. The voltage generated by the thermocouple is biased near ground. The on-chip excitation voltage source biases it to within the linear range of the amplifier, so the system can work with a single power supply. This low-noise, low-drift, on-chip, band-gap reference voltage source can ensure the accuracy of analog-to-digital conversion, thereby ensuring the integrity of the entire temperature measurement system.

    The temperature of the cold junction is measured using a resistance temperature detector (RTD) or thermistor (RT in Figure 1). The resistance of these two devices varies with temperature. The on-chip constant current source provides the required excitation current. In this measurement, a ratio configuration method is used, that is, the reference voltage source of the analog-to-digital converter and the precision resistor use the same excitation current. The ratio configuration method can make the temperature measurement of the cold junction not affected by the excitation current, because the change of the excitation current can make the voltage change generated by the sensor and the voltage change generated by the precision resistor exactly the same, so there is no need for analog to digital conversion Any impact.

    Resistance temperature detector

    The resistance of the resistance temperature detector changes with the temperature. Commonly used materials for resistance temperature detectors are nickel, copper, and platinum. Among them, platinum resistance temperature detectors with a resistance between 100Ω and 1000Ω are common. The resistance temperature detector is suitable for temperature measurement with a nearly linear response in the entire temperature range of -200°C to +800°C. A resistance temperature detector consists of 3 or 4 wires. Figure 2 shows a schematic diagram of the connection between a 3-wire resistance temperature detector and an analog-to-digital converter, where RL1, RL2, and RL3 respectively represent the resistance of the lead wires of the resistance temperature detector.

002.jpg

    Figure 2. The analog circuit part of the temperature system of the resistance temperature detector

    In order to complete the configuration of the 3-wire resistance temperature detector, two fully matched current sources are required. In this 3-wire configuration, if only one current source (IOUT1) is used, the lead resistance will cause errors, because the excitation current flowing through RL1 will cause a voltage error between AIN1 (+) and AIN1 (–). We use the second resistance temperature detector current source (IOUT2) to compensate the error caused by the excitation current passing through RL1. The value of each current source is not important, but the complete matching of the two current sources is very critical. The second resistance temperature detector current flows through RL2. Assuming that RL1 and RL2 are equal (leads are usually made of the same material and have the same length), and IOUT1 and IOUT2 match, the error voltage at both ends of RL2 will be offset with the error voltage at both ends of RL1, so the difference between AIN1 (+) and AIN1 (–) No error voltage will be generated between. Although RL3 will produce twice the voltage, this is a common-mode voltage, so it will not bring errors.

    The analog-to-digital converter has a differential analog input and accepts a differential reference voltage, which can realize a ratio configuration. In Figure 2, the reference voltage of the analog-to-digital converter is also generated by a matched current source. This reference voltage is generated by the voltage across the precision resistor (RREF) and used for the differential reference input of the analog-to-digital converter. This scheme will ensure that the analog input voltage is proportional to the reference voltage. Any error in the analog input voltage caused by the temperature drift of the current source of the resistance temperature detector can be compensated by the deviation of the reference voltage.

    Thermistor

    The resistance of a thermistor also changes with temperature, but it is not as good as a resistance temperature detector. The thermistor usually uses a single current power supply. As with the use of resistance temperature detectors, a precision resistor is used for the reference voltage source, and a current source drives the precision reference resistor and thermistor, which means that a ratio configuration can be realized. This also shows that the current source is not important, because the temperature drift of the current source affects both the thermistor and the reference resistor, thus canceling the drift effect. In thermocouple applications, a thermistor is usually used for cold junction compensation. The nominal resistance of the thermistor is usually 1000Ω or higher.

    Thermal diode

    Of course, thermal diodes can also be used for temperature measurement. In this type of system, the temperature is calculated by measuring the base-emitter voltage of a diode-connected transistor. Two different currents are used to pass the thermal diode separately. Measure the base-emitter voltage in each case. Since the ratio of the current is known, the temperature can be calculated by measuring the base-emitter voltage difference between two different currents.

003.jpg

    Figure 3. The analog circuit part of the thermal diode temperature system

    In Figure 3, we set the excitation current source of AD7792/AD7793 to 10μA and 210μA (other values can also be selected). First, let a 210μA excitation current flow through the diode, and use an analog-to-digital converter to measure the base-emitter voltage. Then, repeat the above measurement with a 10 μA excitation current. This means that the current is reduced to 1/21 of its original value. In the measurement, the current is not important, but the current ratio is required to be fixed.

    Because the AD7792/AD7793 integrate the current source in the chip, it can ensure the matching of the current source, so that the current ratio remains unchanged. In order to eliminate the parasitic error that affects the temperature measurement, a constant current ratio is required. The measured two base-emitter voltage readings are transmitted to the microcontroller, and then the temperature is calculated according to the above formula.

    among them:

    n=ideal factor=measured,

    K=Boltzmann's constant,

    N = the ratio of IC2 to IC1,

    Q=the amount of electron charge,

    ΔVBE is measured by an analog-to-digital converter.

    2 Requirements structure for analog-to-digital converter

    Temperature measurement systems are usually low speed (up to 100 samples per second), so narrowband analog-to-digital converters are more suitable; however, analog-to-digital converters must have high resolution. The narrow band and high resolution requirements make the Σ-Δ analog-to-digital converter an ideal choice for this application.

    In this structure, the analog input of the switched capacitor front end is continuously sampled, and the sampling frequency is significantly higher than the useful bandwidth (see Figure 4). For example, AD7793 has a built-in 64kHz clock. The analog signal to be measured is close to DC, but it is oversampled at K times the signal frequency (KfS), thereby reducing the quantization noise in the baseband. The quantization noise is distributed from DC to half the sampling frequency (KfS/2). Therefore, the use of an increased sampling frequency increases the range of the quantization noise distribution and reduces the noise in the useful frequency band.

004.jpg

    Figure 4. The effect of oversampling, digital filtering, noise shaping, and sampling decimation on the noise spectrum

    The Σ-Δ regulator converts the sampled input signal into a digital pulse train, and the density of "1" includes digital information. Σ-Δ adjustment can also carry out noise shaping. Through noise shaping, the noise in the useful bandwidth is moved outside the useful bandwidth and reaches the useless frequency range. The higher the order of the regulator, the more obvious the effect on noise shaping within the useful bandwidth. However, higher-order regulators tend to be unstable. Therefore, a trade-off must be made between the order of the regulator and the stability. In the narrowband Σ-Δ analog-to-digital converter, a second-order or third-order regulator is usually used, so the device has good stability.

    The digital filter behind the regulator samples the regulator output to give valid data conversion results. The filter can also filter out-of-band noise. The digital filter image frequency will appear at multiple multiples of the main clock frequency. Therefore, the use of the Σ-Δ structure means that the required external component is a simple R-C filter to eliminate the digital filter image frequency at the multiple of the main clock frequency. The Σ-Δ structure enables the 24-bit analog-to-digital converter to have a peak-to-peak resolution of 20.5 bytes (20.5 stable or flicker-free bytes).

    Gain

    Generally, the signal from the temperature sensor is very weak. For a small range of temperature changes of a few degrees, the corresponding analog voltage changes generated by temperature sensors such as thermocouples and resistance temperature detectors are mostly only a few hundred millivolts. Therefore, the typical full-scale analog output voltage is only in the mV range. If the gain stage circuit is not used, the full-scale range of the analog-to-digital converter is usually ±VREF. In order to improve the performance of the analog-to-digital converter, most of its analog input range should be used. When using this type of sensor to measure temperature, the importance of gain is extremely prominent. Without any gain, only a small portion of the full-scale range of the analog-to-digital converter is used, which will lose resolution.

    Instrumentation amplifiers allow the development of low-noise, low-temperature drift gain stage circuits. Low noise and low temperature drift are critical to ensure that the voltage change caused by temperature changes is greater than the noise voltage of the instrumentation amplifier. The gain of AD7793 can be set to 1, 2, 4, 8, 16, 32, 64, or 128. With a gain setting of 128 times and the generated reference voltage source, the full-scale range of the AD7793 is ±1.17mV/128mV or approximately ±10mV. In this way, the high-resolution characteristics of the ADC ensure that the effect can be achieved without any external amplifier components.

    Suppression of 50Hz/60Hz frequency

    The built-in digital filter of the sigma-delta analog-to-digital converter is very effective for suppressing out-of-band quantization noise and other noise sources. One of the noise sources is the frequency generated by the power grid power supply system. When the power grid supplies power to the device, it will generate a power supply system frequency of 50 Hz and its multiples (in Europe), or a power supply system frequency of 60 Hz and its multiples (in the United States). Narrowband analog-to-digital converters mainly use sinc filters. AD7793 has 4 filter options, and the analog-to-digital converter can automatically select the type of filter to be used according to the update rate. The sinc3 filter is used at an update rate of 16.6 Hz. As shown in Figure 5, the sinc3 filter has notches in the spectrum. When the output word rate is 16.6Hz, these grooves can be used to simultaneously suppress 50Hz or 60Hz frequencies.

    Figure 5. The frequency response when the update rate is equal to 16.6 Hz (chopping frequency)? ,

    Chopper

    Unfavorable factors such as offset voltage and other low-frequency errors will always appear in the system, and temperature measurement systems are no exception. The chopper is an inherent feature of AD7793 and can be used to eliminate these error signals. The working principle of the chopper is to alternate phase (or clipping) at the input multiplexer of the analog-to-digital converter. Then, perform analog-to-digital conversion for each chopping phase (positive phase and negative phase). Then, use a digital filter to average the two conversion results. In this way, any offset error in the analog-to-digital converter is eliminated, and more importantly, the effect of temperature on the offset drift is reduced to.

    Low power consumption

    Many temperature detection systems do not use electricity. In some industrial applications, such as temperature monitoring in factories, the entire temperature system including sensors, analog-to-digital converters, and microcontrollers are all in a separate circuit board, using a 4" 20mA loop power supply. Therefore, independent The current budget of the circuit board is 4mA. Portable devices, such as portable gas detectors used in mines, need to measure temperature and gas at the same time. This type of portable system uses battery power to extend battery life. In this type of application, low Power consumption is very important, but high performance is also very important. The power consumption current of AD7933 is 500mA, so it can continue to meet the high performance index requirements of the temperature system while consuming relatively low current.

    3 concluding remarks

    The temperature measurement system has very strict requirements on the analog-to-digital converter and the system. Each type of temperature sensor requires different components, but the analog signals generated by these sensors are usually very small. Therefore, it is necessary to use a low-noise gain stage circuit to amplify these signals, so that the noise of the amplifier will not overwhelm the weak signal of the sensor. A high-resolution analog-to-digital converter is required behind the amplifier to convert the analog signal output by the sensor into a digital signal. The sigma-delta architecture is very suitable for this type of analog-to-digital converter application, and an analog-to-digital converter with high resolution and high precision has been developed using this architecture. In addition to analog-to-digital converters and gain stages, temperature measurement systems also require other components, such as excitation current sources and reference voltage sources. In addition, these components must have low-drift, low-noise performance, so as not to degrade the system. Initial errors such as offset voltage can be calibrated outside the system, but the temperature drift of the components used must be very low to avoid introducing errors. In all portable applications, power consumption needs to be considered. Many systems that used to be powered by power grids now use independent circuit boards for power supply. Therefore, the issue of power consumption becomes more and more important.