The micromachined monolithic silicon piezoresistive-bridge pressure transducer (SPPT) is a dominant technology in automotive, industrial, medical, and environmental pressuresensing applications. Available from many vendors in many variations, all share a similar architecture in which a thin (5 µm to 200 µm) micromachined silicon diaphragm incorporates an implanted piezoresistive Wheatstonebridge strain gauge. The applied pressure bends the diaphragm and imbalances the strain gauge, producing a differential, ratiometric output signal that’s proportional to the product of the pressure and bridge-excitation voltage.
But SPPTs are of little use unless supported by appropriate signal conditioning and calibration circuits. Monocrystalline silicon isn’t rubber, which limits the only finitely elastic SPPT diaphragm to relatively small deflections: it produces only ±1% modulation of the bridge resistance elements, and millivolt-level output signals. This creates the need for high-gain, low-noise, temperature-stable dc amplification to scale the output up to usable levels.
The signal-conditioning circuit also must include stable, high-resolution, preferably non-interactive, zero and span trims. Because automation of the sensor-calibration process can be of enormous benefit, good signal-conditioning circuit designs should be compatible with the automated production and testing environment.
Another complication of applying SPPTs is the large temperature dependence of both the total bridge resistance and the piezosensitivity (the ratio of the bridge output to excitation voltage × pressure). Bridge resistance increases with temperature while piezosensitivity decreases. Some SPPT designs (e.g., the Lucas NPC-410 series) equalize these opposite-sign temperature coefficients.
The payoff comes when such SPPTs are excited with constant current because the increase with temperature of the bridge resistance (and therefore of the bridge excitation voltage) cancels out the simultaneous decrease of piezosensitivity. These design ideas have been incorporated in the circuit in Figure 1.
The bridge bias is provided by op amp A1, which combines with voltage reference D1 and currentsense resistor R1 to generate a constantcurrent bridge drive of: IBIAS = 1.225 V/2k = 612 µA. The −10 mV/psi pressure-dependent signal is output differentially on sensor pins 2 and 4, superimposed on the commonmode bridge excitation voltage, which is 50 times larger.
The standard method for separating such a small differential-pressure signal from the much larger common-mode bias voltage would be to incorporate a relatively expensive, high-CMRR, difference (instrumentation) amplifier in the design, as illustrated in Figure 2.
Figure 1 employs a different scheme. A virtual ground for sensor pin 4 is generated by op amp A2, which closes a feedback loop that forces pins 1 and 8 of the bridge to whatever voltage is needed to drive pin 4 (VZ) into balance with A2’s non-inverting input. This action effectively nulls out the bridge excitation, leaving the pristine pressure signal signal (VB) on bridge pin 2.
Unlike the difference amplifier method, no precision resistor network is required, other than the precision network inside the SPPT bridge itself. A2 also allows precision adjustment, via DCP1, of any initial transducer null-offset error. To accomplish this, the bridge excitation voltage is programmably attenuated by the R2, R3, R4, R5 network and DCP1. Meanwhile, boosting the small VB signal to a convenient 1-V/psi output level is the job of noninverting gain-of-100 amplifier A3. Its nonvolatile precision calibration network comprises R7 through R9 and DCP2.
Combining the SPPT and the Figure 1 circuitry creates a signal-conditioned precision pressure sensor that’s compatible (thanks to DCP1 and DCP2) with a fully automated calibration process, very low in total power draw (less than 1 mA, most of which goes to transducer excitation) and low cost.
Reprints
