Current Instrumentation Amplifier Enhances Piezoelectric Accelerometers

A typical piezoelectric sensor consists of a disk of PZT-5A ceramic material with metallized electrodes on its surface. External wires are connected to the sensor by applying a conductive epoxy to the electrodes. An insulating adhesive connects the assembly to the structure being measured and isolates the sensor from the ground reference potential. The disk faces the direction of the expected acceleration. When mounted on the target structure, the piezoelectric disk produces a voltage proportional to the force acting parallel to the disk's polarization, thereby acting as a simple force sensor and accelerometer. The capacitive impedance of the piezoelectric disk presents a large reactance at low frequencies, making the disk and its wiring susceptible to interference generated by surrounding electrical equipment and power lines. Placing the sensor in a remote location requires a shielded interconnect cable, but even shielding is not completely effective in eliminating common-mode signals because noise pickup still occurs on the conductive surface of the disk.

One method of extracting the sensor signal is to use an instrumentation amplifier, which amplifies only the potential generated by the sensor and rejects the common-mode coupled noise potential that appears at the sensor terminals.

A typical miniature piezoelectric disk sensor is 0.125" in diameter and 0.0075" thick, presenting a capacitance of about 500pF. If the measurement application requires a dynamic response to force excitation frequencies at or below 10Hz, the sensor's output reactance can range into the tens of MΩ. The circuit's printed circuit board insulating substrate and ambient moisture impose a practical limit of about 10MΩ on the amplifier's input resistance.

Insulation materials must be carefully selected and guard potentials applied, and an amplifier with PA-grade input bias current must be used. Otherwise, the capacitance of the sensor and the amplifier’s input bias current resistors will cause a phase shift in the signal applied to the instrumentation amplifier. To eliminate the need for guarding and elaborate insulation requirements, the circuit in Figure 1 uses an instrumentation amplifier with feedback so that the sensor’s short-circuit current is measured rather than its open-circuit voltage. The common-mode voltage (VCM) between the sensor and signal ground comes from nearby noise sources, which in turn come from stray capacitive coupling. The following formula relates the sensor’s output current, i, to its open-circuit output voltage, ES:

Where A represents the voltage gain of IC1, and R = R1 = R2 in Figure 1. Resistors R1 and R2 provide feedback and input bias current return paths for IC1 (the INA121 instrumentation amplifier), while resistor RG sets the amplifier's gain. The INA121's 0.5pA input bias offset current produces a 5mV voltage offset across its 10MΩ feedback resistor. With an amplifier gain of 500, IC1's output offset equals 2.5 mV. Amplifier IC2 is a TL081 type that provides unity-gain signal inversion.

If 2A+1>>2RjwCS, then i≈jwCSES, and the input voltage VI of amplifier IC1 vanishes because the input of the amplifier acts as a virtual short circuit across the sensor. Summing the voltages around the loop consisting of the outputs of the instrumentation amplifier and the inverting amplifier, the two feedback resistors, and the input terminals of the instrumentation amplifier (whose potential difference is zero) yields eO=jwRCES, where eO represents the output of IC1 and is the negative of the output of IC2.

The operational amplifier based integrator IC3 provides the value for ES at its output, which is E' in the following equation.

For the component values ​​in Figure 1, IC1 provides a gain of 500. Resistors R1 and R2 are equal to 10 MΩ, and the capacitance of the piezoelectric sensor is measured as 500 pF. For the highest frequency of 10 Hz, the total amount 2RwCS = 0.6 << 2A + 1 = 501, and the output of the sensor, ES, appears in the form of E' with no phase error. This circuit can measure quasi-static force changes. Its ability to hold a charge on C1 places the ultimate limit on the frequency response of the circuit.