Design Ideas: January 20, 1994
You can use many tricks to make temperature-compensated resistive full-bridge transducers, such as piezoresistive pressure or strain-guage sensors. Most of these tricks use some additional thermosensitive components that produce a compensating voltage or current. The disadvantages of such solutions are obvious: low accuracy due to the inevitable temperature difference between the sensor's body and resistive bridge, increased circuit complexity due to the extra parts, and demands for accurate adjustment of analog values.
The alternative in Fig 1 uses an ADC and takes advantage of the thermal properties of resistive solid-state sensors such as pressure sensors or strain guages. Without any additional temperature-sensitive components, the output of the circuit, which is in a form of a digital code, includes the temperature compensation. This technique is applicable to all full-bridge resistive sensors.
The equivalent resistance of the full-bridge pressure sensor depends on the temperature and, for a first approximation, is independent of the pressure. If you apply a constant voltage to the bridge, the current will be inversely proportional to the bridge's resistance, and the current's sign will be the inverse of the temperature effect in comparison with the bridge's resistance. The op amp will produce an output voltage proportional to the full bridge current and the resistance of RFB. The circuit applies this voltage, which includes the temperature effect, to the reference pin of the ADC and applies the sensor's output to the ADC's differential input.
Most solid-state pressure sensors have negative temperature coefficient of span (TCS) and positive temperature coefficient of resistance (TCR). The absolute values of TCS and TCR are very close to each other. For example, the SLP004D from SenSym Inc (Sunnyvale, CA) has typical values of TCS equal to -2400 ppm/°C and TCR of +2300 ppm/°C. The MPX-50, -100, and -200 pressure sensors from Motorola (Phoenix, AZ) have typical values of TCS equal to -0.19%/°C and TCR of +0.24%/°C. So, the output voltage of IC1 will have negative temperature coefficient very close to the TCS of the sensor.
The ADC provides the perfect opportunity to output, in the form of a digital code, a precise ratio of two analog signals applied to the ADC's input and reference pins. So, the analog-to-digital conversion reduces the total TCS of the transducer to approximately the algebraic sum of TCS+TCR.
The inverting input of the op amp connects to ground, so the sensor will operate exactly as if it were really grounded except for the error introduced by the input offset voltage and current of the op amp. In most modern op amps, this error is negligible when you compare it to the commonly 5 or 10V VREF applied to sensor. Note that the TCR of RFB also affects the total TCS. For better compensation, it is possible to use a feedback resistor with small TCR approximately equal to TCS+TCR and locate the resistor as close to the sensor as possible.
Although this circuit uses an ADC from National Semiconductor's (Santa Clara, CA) ADC0801 to ADC0805 family, many others will also work. For single-ended ADCs, you can add any differential amplifier, such as Analog Devices' (Wilmington, MA) AD620. Many op amps will also work with this technique, and the choice of both ADC and op amp depends on the required accuracy.