CN0206

A Type T thermocouple is used in the circuit. This thermocouple (made from copper and constantan) measures temperature from −200°C to +400°C. It generates a typical temperature dependent voltage of 40 μV/°C.

The thermocouple response is approximately linear over a small portion of its entire temperature range, from 0°C to 60°C (see Figure 2). To allow accurate measurements over the entire temperature range, the CN0206-SDP-0 evaluation software implements a linearization routine.

Cold Junction Compensation

Thermocouples measure the temperature difference between two points, not an absolute temperature. To measure a single temperature, maintain one of the junctions (normally the cold junction) at a known reference temperature, and the other junction at the temperature to be sensed.

Having a junction of known temperature is not convenient for most applications; therefore, a thermally sensitive device is placed on the printed circuit board (PCB). This thermistor is used to measure the temperature of the thermocouple input connection. The thermistor fits inside the metallic tab found on the thermocouple connection, minimizing any temperature gradients that may exist.

Because the voltage from a known cold junction is simulated, the appropriate correction can be applied. See the Thermocouple Linearization section for more detailed information on processing and manipulating the thermocouple and thermistor voltages to produce an accurate temperature reading.

ADC Channel 1 Configuration, Thermocouple

The temperature range for the thermocouple is −200°C to +400°C. The typical temperature dependent voltage generated by the thermocouple is 40 μV/°C. This voltage generates a thermocouple voltage range of −8 mV to +16 mV.

When reading the thermocouple voltage, the ADC uses the external 2 V reference and is configured for a gain of 64. This defines the analog input voltage range as ±31.25 mV (±V_REF/Gain). For a gain of 64, the absolute voltage on the analog inputs must be between GND + 300 mV and AV_DD − 1.1 V.

Because the AD7793 operates from a single power supply, the signal generated by the thermocouple must be biased above ground so that it is within the acceptable range of the ADC. The bias voltage generator on-board the AD7793 biases the thermocouple signal so that it has a common-mode voltage of AV_DD/2.

ADC Channel 2 Configuration, Thermistor

The second channel of the ADC monitors the voltage generated across a thermistor being driven by one of the current output pins of the AD7793. A 1mA excitation current drives the series pair of thermistor and precision resistor (2 kΩ, 0.1%), as shown in Figure 1.

The thermistor value varies from 0°C (815 Ω) to 30°C (1040 Ω), producing a voltage signal range of 815 mV to 1040 mV. The precision resistor produces 2.0 V for use as an external reference. With a gain of 1, the analog input range is ±2 V (±V_REF/Gain). This architecture gives a ratiometric configuration. Any change in the value of the excitation current does not alter the accuracy of the system.

Assuming a linear transfer function between 0°C and 30°C, the relationship between the cold junction temperature and the thermistor resistance, R, is

Cold Junction Temperature = 30 × (R − 815)/(1040 − 815)

One other consideration is the output compliance of the IOUT1 pin of the AD7793. When the 1 mA excitation current is used, the output compliance equals AV_DD − 1.1 V. This specification is met because the maximum voltage at IOUT1 equals the voltage across the precision resistor plus the voltage across the thermistor, which equals 2 V + 1.04 V = 3.04 V.

Output Coding

The output code for an input voltage on either channel is

Code = 2^{N − 1} × [(AIN × Gain/V_REF) +1]

where:
AIN is the analog input voltage.
GAIN is the in-amp setting
N = 24.

The EVAL-SDP-CB1Z analog microcontroller processes the conversions from the AD7793.

Thermocouple Linearization

As mentioned in the Circuit Description section, the thermocouple is only approximately linear over a small temperature range. In fact, the thermocouple is highly nonlinear over the rest of the temperature range. For this reason, a linearization procedure is implemented in the CN0206-SDP-0 LabVIEW software.

The National Institute of Standards and Technology provides ITS-90 look-up tables for thermocouples. Each table contains a list of thermoelectric voltages (mV) and their corresponding temperatures.

To implement the cold junction compensation previously mentioned, the CN0206-SDP-0 evaluation software takes the Type T thermocouple look-up table and searches for the thermoelectric voltage associated with the cold junction temperature. It then subtracts this thermoelectric voltage from every element in the look-up table to produce a cold junction compensated thermoelectric voltage look-up table.

The evaluation software then searches the modified look-up table for the thermocouple input voltage, as sampled by the AD7793. When the CN0206-SDP-0 evaluation software finds what two points this thermoelectric voltage lies between, a linear interpolation is performed to precisely calculate the thermocouple temperature.

System Calibration

In addition to linearizing the thermocouple temperature, the evaluation software performs a two-point calibration. The user must input the lowest and highest temperature and use the CN0206-SDP-0 evaluation software to produce the corresponding temperature readings.

The software takes these readings and calculates the gain and offset, and then applies these values to any future thermocouple temperature calculations.

System Noise Considerations

For an output data rate of 16.7 Hz and a gain of 64, the rms noise of the AD7793 equals 0.086 μV (noise is referred to input). The peak-to-peak noise is

6.6 × RMS Noise = 6.6 × 0.086 μV = 0.5676 μV

If the thermocouple has a sensitivity of precisely 40 μV/°C, the thermocouple should measure the temperature to a resolution of

0.5676 μV ÷ 40 μV = 0.014°C

Test Data and Results

All data capture was performed using the CN0206-SDP-0 LabVIEW evaluation software. A CL540ZA source (and appropriate Type T cable) was used to simulate the thermocouple input. The CL540ZA is capable of simulating several different types of thermocouples (J, T, E, K, R, S, B, N, and so on).

By sweeping the CL540ZA input source from −200°C to +400°C in +5°C increments, the CN0206-SDP-0 evaluation software was able to capture, linearize, and calibrate the system according to the two user-defined calibration points.

According to Figure 3, the error over the entire temperature range is less than 1°C. However, over most of the range, the error is less than 0.5°C

The peak-to-peak noise of the AD7793 was determined by shorting the input pins of the ADC together and acquiring 1000 samples. As seen in the histogram found in Figure 4, the code spread is approximately 220 codes, which translates to a temperature spread of 0.02°C peak-to-peak.

Test data was taken using the board shown in Figure 5. Complete documentation for the system can be found in the CN-0206 Design Support package.