The circuit shown in Figure 1 is a complete thermocouple signal conditioning circuit with cold junction compensation followed by a 16-bit Σ-Δ analog-to-digital converter (ADC). The AD8495 thermocouple amplifier provides a simple, low-cost solution for measuring K-type thermocouple temperature and includes cold junction compensation.

The fixed-gain instrumentation amplifier in the AD8495 amplifies the small voltage of the thermocouple to provide a 5 mV/°C output. The amplifier features high common-mode rejection, rejecting common-mode noise that may be picked up by the long leads of the thermocouple. For additional protection, the amplifier's high-impedance input allows additional filtering to be easily added.

The AD8476 differential amplifier provides the correct signal levels and common-mode voltages to drive the AD7790 16-bit Σ-Δ ADC.

This circuit provides a compact, low-cost solution for thermocouple signal conditioning and high-resolution analog-to-digital conversion.

A thermocouple is a simple component widely used for temperature measurement. It consists of a junction of two dissimilar metals. These metals are connected at one end to form a measuring junction, also known as a thermal junction. The other end of the thermocouple is connected to a metal wire connected to the measuring electronics. This connection forms a second junction - the reference junction, also known as the cold junction. To derive the temperature of the measurement junction (TMJ), the user must know the differential voltage produced by the thermocouple. The user must also know the error voltage due to the reference junction temperature (TRJ). Compensating the reference junction temperature error voltage is called cold junction compensation. For the output voltage to accurately represent the hot junction measurement, the electronics must compensate for any changes in the reference (cold) junction temperature.

This circuit uses the AD8495 thermocouple amplifier and operates from a single 5 V supply. The output voltage of the AD8495 is calibrated for 5 mV/°C. When operating from a single 5 V supply, the output is linear between approximately 75 mV and 4.75 V, which corresponds to a temperature range of 15°C to 950°C. The output of the AD8495 drives the non-inverting input of the AD8476 unity-gain differential amplifier, which converts the single-ended input to a differential output for driving the AD7790 16-bit Σ-Δ ADC.

The low-pass differential and common-mode filters before the AD8495 input eliminate RF signals, which if allowed to reach the AD8495, may become rectified and manifest as temperature fluctuations. Two 100Ω resistors and a 1μF capacitor form a differential filter with a cutoff frequency of 800Hz. Two 0.01nF capacitors form a common-mode filter with a cutoff frequency of 160 kHz. A similar filter is used at the output of the AD8476 differential amplifier before the signal is applied to the AD7790 ADC.

The AD8495 input protects the device from input voltage excursions up to 25 V beyond the opposite supply rail. For example, in this circuit, with the positive rail at 5 V and the negative rail at ground, the device can safely tolerate input voltages from -20 V to +25 V. The voltage at the reference and sense pins must not exceed the supply rail by more than 0.3 V. This feature is particularly important in applications with power supply sequencing issues that can cause the signal source to become active before amplifier power is applied.

The theoretical resolution of the system can be calculated based on the bandwidth, voltage noise density, and gain of the AD8495. The peak-to-peak (noise-free code) resolution (in bits) is:

The AD8476 is an extremely low power, fully differential precision amplifier that integrates a thin film laser trimmed 10kΩ gain resistor for unity gain. It is ideal for such applications because it prevents relatively high impedance loads from being placed on the AD8495.

The AD7790 is a low-power, complete analog front end suitable for low-frequency measurement applications. It contains a low-noise 16-bit sigma-delta ADC with a differential input that can be configured in buffered or unbuffered mode.

Test Results

An important measure of the performance of this circuit is the amount of linearity error. The AD8495 output accuracy is within 2°C over the -25°C to +400°C range. To achieve greater accuracy when working within or outside this range, a linear correction algorithm must be implemented in the software. The CN-0271 evaluation software uses NIST thermoelectric voltage lookup tables to ensure output error within 1°C from 15°C to 950°C.

Figure 2 compares the performance of the AD8495 to the CN-0271 system and shows the results after linearizing the ADC output. For details on how to implement this algorithm in software, see the AN-1087 application note Linearizing Thermocouples with the AD8494/AD8495/AD8496/AD8497.

The noise performance of the system is also important to the accuracy of the circuit. Figure 3 shows a histogram of 1,000 measured samples. This data was obtained using the CN-0271 evaluation board connected to the EVAL-SDP-CB1Z System Demonstration Platform (SDP-B) evaluation board. See the "Circuit Evaluation and Testing" section for setup details.

The measured peak-to-peak noise is approximately 6 LSB (1 LSB = 4.9 V ÷ 65536 = 74.8μV), which corresponds to 0.449 mV pp and a noise-free resolution of 13.4 bits.

This shows that the converter is not degrading the noise-free resolution because the number of noise-free bits derived from the measured resolution at a fixed thermocouple input voltage is approximately the same as the number of noise-free bits predicted based on the theoretical output noise of the AD8495.

For a complete design support package for this circuit note, please visit www.analog.com/CN0271-DesignSupport