Piezoceramic (PZT), or "piezo," actuators are known to be excellent position transducers in the nanometer or micrometer range. These actuators are widely used in many precision applications. PZTs, which come in different shapes (tubes, disks, and plates), are cost effective and easy to use. Just glue them to the parts you want to move (mirrors, fibers), connect the wires to the amplifier, and go.
The "big" problem with PZT ceramics is that the maximum elongation (typically 0.1% of their size) is obtained with electrical fields on the order of 20 kV/cm. For common-size transducers, this can lead to full-scale operating voltages close to 1 kV, requiring expensive high-voltage amplifiers. Manufacturers partially solve this problem by building piezo "stacks," a pile of small piezoceramic pieces mechanically connected "in series" and electrically connected in parallel. These stacks have operating voltages one order of magnitude lower. The stacks are expensive, however, and are not as stiff as a single piece of ceramic. In practical applications, additional mechanics like steel parallelograms must be added.
The low-cost amplifier presented in Figure 1 permits the use of "high-voltage" piezos with excellent results. It is based on three high-voltage MOSFETs and a low-noise op amp. To understand how the amplifier works, consider that the piezo behaves more or less like a capacitor, having typical values of tens of nanofarads. Therefore, when driving the piezo with large voltage steps, the piezo needs high currents to charge and discharge itself.
Assume for a moment that Q3 is not present and D1 is a short. Q1, driven directly by the op-amp output, acts with R1 as a voltage-controlled current-sink. Together, Q2 with R2 and the 9-V battery act as a fixed current-source that pulls up the drain of Q1. Note that Q2's gate drains no current from the battery which can, therefore, last for years. (Caution: this battery is at the same high voltage as the output!)
When the amplifier needs to discharge the piezo, the output of the op amp goes up and Q1 sinks a current limited only by the set value of R1. When the amplifier needs to charge the piezo, the charge rate is fixed at the value of the current source Q2-R2.
Since this value is also the quiescent current of the amplifier, this quiescent current must be low (1 mA = 1 W) to "cool" the device at 1 kV. This limits the positive slope slew rate. For a 10-nF piezo, the slew rate is 0.1 V/µs for a 1-mA quiescent current.
Q3 serves as a current buffer that isolates the capacitance of the load when charging it, enabling higher slew rates at low quiescent currents. Diode D1 provides a path for the discharge current through Q1.
R3 and R4 set a positive gain of 100. Note that R3 and C1 must support a 1-kV voltage drop. A series of three 0.5-W, 330-kΩ resistors is effective. R5 isolates the capacitive load, improving the stability. The voltage divider R6-R7 provides a monitor signal that is 1/100th of the output (always advisable when high voltage is present).
The measured slew rate with a 26-nF piezo-tube load is 10 V/µs for a 1.5-mA quiescent current and 20 V/µs for a 3.5-mA quiescent current. The up/down slew rates can be set independently by trimming R2 and R1, respectively.
The output-noise spectral density of the amplifier is shown in Figure 2. At high frequencies, it is limited by the Johnson noise of the feedback network (chosen to be as low as 10 kΩ). The 50-Hz ripple is within 5 mV p-p (5 ppm) and can be reduced by further filtering the power supply.
The HV power supply is also quite simple. While a transformer with a 710-V secondary is used, the diode bridge is made with eight 1N4007s. The filter capacitor is made with a series of four 470-µF 250-V capacitors, each one paralleled with a 330-kΩ resistor to provide a discharge path when the HV supply is switched off.