Laboratory Circuits: Multi-channel Thermocouple Measurement Solutions

　　ADT7320 : ±0.25°C Accuracy, 16-Bit Digital SPI Temperature Sensor

　　AD7793 : 3-channel, low noise, low power, 24-bit, Σ-Δ ADC with on-chip instrumentation amplifier and voltage reference

　　Evaluation and Design Support

　　CN-0172 Circuit Evaluation Board (EVAL-CN0172-SDPZ)

　　System Demonstration Platform ( EVAL-SDP-CB1Z )

　　CN0172 breakout board (included with the EVAL-CN0172-SDPZ evaluation board)

　　Circuit Function and Advantages

　　The circuit in Figure 1 functionally provides a high-precision, multi-channel thermocouple measurement solution. Accurate thermocouple measurements require a signal chain that uses precision components to amplify the small thermocouple voltage, reduce noise, correct for nonlinearity, and provide accurate reference junction compensation (commonly called cold junction compensation). This circuit solves all of these challenges for thermocouple temperature measurement with better than ±0.25°C accuracy.

　　The circuit in Figure 1 shows three K-type thermocouples connected to the AD7793 precision 24-bit Σ-Δ analog-to-digital converter (ADC) to measure the thermocouple voltage. Since the thermocouple is a differential device rather than an absolute temperature measurement device, the reference junction temperature must be known to obtain an accurate absolute temperature reading. This process is called reference junction compensation, often referred to as cold junction compensation. The ADT7320 precision 16-bit digital temperature sensor in this circuit is used for the cold junction reference measurement and provides the required accuracy.

　　These applications are very popular where cost-effective, accurate temperature measurements are required over the wide temperature range offered by thermocouples.

Figure 1. Multichannel Thermocouple Measurement System (Simplified Schematic: All Connections and Decoupling Not Shown)

　　Circuit Description

　　The circuit in Figure 1 is designed for simultaneous measurement of three K-type thermocouples using the ADT7320, a ±0.25°C accurate, 16-bit digital SPI temperature sensor.

　　Thermocouple voltage measurement

　　Thermocouple connectors and filters are used as the interface between the thermocouple and the AD7793 ADC. Each connector (J1, J2, and J3) is directly connected to a set of differential ADC inputs. The filters at the AD7793 inputs reduce the noise superimposed on any thermocouple leads before the signal reaches the AIN (+) and AIN (-) inputs of the ADC. The AD7793 integrates an on-chip multiplexer, buffer, and instrumentation amplifier to amplify the small voltage signal from the thermocouple measurement junction.

　　Cold junction measurement

　　The ADT7320 precision 16-bit digital temperature sensor is used to measure the reference junction (cold junction) temperature, with an accuracy of ±0.25°C in the temperature range of -20°C to +105°C. The ADT7320 is fully factory calibrated, and the user does not need to calibrate it. It has a built-in bandgap temperature reference source, a temperature sensor, and a 16-bit Σ-Δ ADC to measure temperature and perform digital conversion with a resolution of 0.0078°C.

　　Both the AD7793 and ADT7320 are controlled by the SPI interface using the system demonstration platform (EVAL-SDP-CB1Z). Alternatively, both devices can be controlled by a microcontroller.

Figure 2. EVAL-CN0172-SDPZ circuit evaluation board

　　Figure 2 shows the EVAL-CN0172-SDPZ circuit evaluation board with three K-type thermocouple connectors, the AD7793 ADC, and the ADT7320 temperature sensor mounted between two copper contacts on a separate flexible printed circuit board (PCB) for reference temperature measurement.
 

　　Figure 3 is a side view of the ADT7320 mounted on a separate flexible PCB, inserted between the two copper contacts of the thermocouple connector. The flexible PCB in Figure 3 is thinner and more flexible, which is an advantage over a small FR4-type PCB. It allows the ADT7320 to be neatly mounted between the copper contacts of the thermocouple connector to minimize the temperature gradient between the reference junction and the ADT7320.

Figure 3. Side view of the ADT7320 mounted on a flexible PCB.

　　The small, thin, flexible PCB also enables the ADT7320 to respond quickly to temperature changes at the reference junction.

Figure 4 shows the typical thermal response time of the ADT7320.

　　Figure 4. Typical thermal response time of ADT7320

　　This solution is flexible and allows the use of other types of thermocouples, such as J or T. In this circuit note, K type was chosen due to its popularity. The actual thermocouple selected has a bare tip. The measurement junction is outside the probe wall and exposed to the target medium.

　　The advantage of using an exposed tip is that it provides the best thermal conductivity, has the fastest response time, and is low cost and lightweight. The disadvantage is that it is susceptible to mechanical damage and corrosion. Therefore, it is not suitable for use in harsh environments. However, in situations where fast response time is required, the exposed tip is the best choice. If the exposed tip is used in an industrial environment, it may be necessary to electrically isolate the signal chain. Digital isolators can be used for this purpose (see www.analog.com/icoupler).

　　Unlike traditional thermistors or resistance temperature detectors (RTDs), the ADT7320 is a completely plug-and-play solution that does not require multi-point calibration after board assembly, nor does it consume processor or memory resources due to calibration coefficients or linearization procedures. It consumes only 700 μW of typical power when operating at 3.3 V, avoiding the self-heating problem that reduces the accuracy of traditional resistive sensor solutions.

　　A Guide to Precision Temperature Measurement

　　The following guidelines ensure that the ADT7320 accurately measures the reference junction temperature.

　　Power Supply: If the ADT7320 is powered from a switching power supply, noise above 50 kHz may be generated, which can affect temperature accuracy. To prevent this defect, an RC filter should be used between the power supply and VDD. The component values ​​used should be carefully considered to ensure that the power supply noise peak is less than 1 mV.

　　Decoupling: The ADT7320 must be installed with a decoupling capacitor as close to VDD as possible to ensure the accuracy of temperature measurement. A decoupling capacitor such as a 0.1 μF high-frequency ceramic type is recommended. In addition, a low-frequency decoupling capacitor should be used in parallel with the high-frequency ceramic capacitor , such as a 10 μF to 50 μF tantalum capacitor.

　　Maximum Thermal Conduction: The plastic package and the exposed pad (GND) on the back side are the main thermal conduction paths from the reference junction to the ADT7320. Since the copper contacts are connected to the ADC inputs, the back side pad cannot be connected in this application because doing so would affect the biasing of the ADC inputs.

　　Guide to Precision Voltage Measurements

　　The following guidelines ensure that the AD7793 accurately measures the thermocouple measurement junction voltage.

　　Decoupling: The AD7793 must have decoupling capacitors installed as close as possible to AVDD and DVDD to ensure accurate voltage measurements. AVDD should be decoupled to GND using a 0.1 μF ceramic capacitor in parallel with a 10 μF tantalum capacitor. Additionally, DVDD should be decoupled to GND using a 0.1 μF ceramic capacitor in parallel with a 10 μF tantalum capacitor. For more discussion on grounding, layout, and decoupling techniques, refer to Tutorial MT-031 and Tutorial  MT-101.

　　Filtering: The differential inputs of the AD7793 are used to remove most of the common-mode noise on the thermocouple lines. For example, placing R1, R2, and C3, which form a differential low-pass filter, at the front end of the AD7793 removes possible superimposed noise on the thermocouple leads. The C1 and C2 capacitors provide additional common-mode filtering. Since both AIN (+) and AIN (-) inputs to the ADC are analog differential inputs, most of the voltages in the analog modulator are common-mode voltages. The excellent common-mode rejection (100 dB minimum) of the AD7793 further removes common-mode noise from these input signals.

　　Other problems solved by this solution

　　The following is a summary of how this solution addresses the other thermocouple-related challenges mentioned above.

　　Thermocouple voltage amplification: Thermocouple output voltage changes with temperature by only a few μV per degree. The common K-type thermocouple used in this example changes by 41 μV/°C. This weak signal requires a high gain stage before ADC conversion. The maximum gain that the AD7793 internal programmable gain amplifier (PGA) can provide is 128. The gain in this solution is 16, allowing the AD7793 to run an internal full-scale calibration function through the internal reference voltage source.

　　Thermocouple Nonlinearity Correction: The AD7793 has excellent linearity over a wide temperature range (–40°C to +105°C) and requires no user correction or calibration. To determine the actual thermocouple temperature, the reference temperature measurement must be converted to an equivalent thermoelectric voltage using a formula provided by the National Institute of Standards and Technology (NIST). This voltage is added to the thermocouple voltage measured by the AD7793 and the sum is then converted back to the thermocouple temperature using the NIST formula again. Another approach involves the use of a lookup table. However, to achieve the same accuracy, the size of the lookup table can be significantly different, requiring the host controller to allocate additional storage resources for it. All processing is done in software using the EVAL-SDP-CB1Z. The EVAL-SDP-CB1Z is done in software.

　　Common changes

　　For applications with lower accuracy requirements,  the AD7792  16-bit Σ-Δ ADC can be used instead of the AD7793 24-bit Σ-Δ ADC. For reference temperature measurement, the ±0.5°C accurate ADT7310 digital temperature sensor can be used instead of the ±0.25°C accurate ADT7320. Both the AD7792 and ADT7310 have integrated SPI interfaces.

　　Circuit Evaluation and Testing

　　This system uses the EVAL-CN0172-SDPZ and the EVAL-SDP-CB1Z. The EVAL-CN0172-SDPZ board comes with the CN0172 breakout board.

　　Equipment Requirements

　　The following equipment is needed:

　　An oil tank

　　EVAL-CN0172-SDPZ circuit evaluation board

　　CN0172 breakout board (included with the EVAL-CN0172-SDPZ evaluation board)

　　EVAL-CN0172-SDPZ circuit evaluation board

　　CN0172 Evaluation Board Software

　　A Datron 4808 calibrator

　　A Hart Scientific 1590 Super Thermometer

　　A Hart Scientific precision probe

　　GPIB cables (3)

　　A PC with Windows XP or later running LabVIEW and with a GPIB card and a USB 2.0 port

　　Setup and Testing

　　The test setup in Figure 5 was used to evaluate the performance of the multi-channel thermocouple solution. A Datron calibrator was used to provide a precision voltage source for the three thermocouple inputs. A Super Thermometer was used to measure the temperature of the oil bath and was controlled via the GPIB bus.

　　The CN0172 LabVIEW software controls the EVAL-CN0172-SDPZ evaluation board through the USB port EVAL-SDP-CB1Z evaluation board, the breakout board, and the SPI bus. The power supply of the EVAL-SDP-CB1Z evaluation board comes from the USB bus, and the 3.3 V output of the EVAL-SDP-CB1Z supplies power to the EVAL-CN0172-SDPZ evaluation board.

　　If oil bath measurements are not required, the EVAL-CN0172-SDPZ evaluation board can be used to measure the temperature of the three thermocouples through the USB interface of a PC using the software on the CD.

　　Detailed and detailed information regarding test setup, calibration, and how to use the evaluation software to capture data can be found in the CN0172 User Guide: www.analog.com/CN0172-UserGuide.

Figure 5. Test setup functional block diagram

　　Test Results

　　Table 6 shows the error curves of this solution at various thermocouple temperatures using different fixed values ​​of the cold junction (CJ) temperature. The overall solution error does not exceed ±0.25°C over a wide temperature range. Note that the solution accuracy can be further improved if a system calibration is performed on the AD7793 ADC.

Figure 6. Error vs. Thermocouple Temperature for a Fixed Cold Junction (CJ) Temperature

　　Figure 7 shows the error curve of the solution at various CJ temperatures using different fixed values ​​of the thermocouple temperature. The overall solution error does not exceed ±0.25°C over a wide temperature range.

Figure 7. Error vs. cold junction temperature for a fixed thermocouple temperature.