Piezo Amp

The 3DOF Micron uses a bending-mode piezoelectric actuator which can be driven with voltages of -240..+480 V for maximum range of motion. This requires some sort of high-voltage amplifier for drive. Piezoelectric actuators have the nice property that they can be driven to an approximate position by applying a voltage to the actuator (open loop). Voltage drive is the most common approach, largely because there are off-the-shelf power opamps that can be used in this mode.

There is obviously a severe safety concern with applying 480 V to a handheld tool. Fortunately, the actual currents required are small (we use 4 mA peak), so safety concerns can be largely addressed through robust current limiting. Though power opamps (from eg. Apex) do have current limit features, this is a secondary feature with low performance. Also, power semiconductors tend to fail shorted, so any malfunction in the amplifier will defeat the current limiting.

Another way of driving piezoelectric actuators is using charge control, which meters the total charge stored in the actuator. The main reason for using charge control is that it linearizes the actuator for open-loop applications. The most common approach to charge control is to use a voltage driver with a charge metering capacitor, which is if anything less safe than direct voltage drive, since capacitor is at the output, where it could potentially discharge into the user. However, charge control could also be implemented using a controlled current source. The problem is that this is not a standard component, but I had been thinking about implementing controlled current sources for years, so this seemed like a good excuse.

It has always annoyed me that you can get a wonderfully high impedance from the collector or drain of an output device, but common current source designs such as the Howland laboriously implement a poor current source using a voltage amplifier by relying on resistor matching, common mode rejection, etc.

You can see the custom high-voltage power amplifier used in Micron at the right front of the electronics box. I used a novel architecture based on open-loop current mirrors, which avoids the difficult problem of measuring a current at a floating high impedance node.

The amplifier performance significantly exceeds what was actually necessary in terms of current and bandwidth. The output impedance is so high that it was difficult to measure, and is for all practical purposes purely capacitive. The linearization resulting from charge control improves Micron high frequency performance somewhat, and I was also modestly successful in damping actuator resonances by controlling the current and then measuring the induced voltage, causing the actuator to also serve as a piezoelectric sensor. This feedback principle had been previously demonstrated by others using voltage control and current measurement.

I built the amplifier using SMT components with generous spacing on account of the high voltage. The actual high voltage supply comes from DC-DC converters, at the back of the board. The board is double sided, and has no ground plane and few bypass capacitors because it runs in current mode. Hardly any of the circuitry is referenced to ground.