So here goes the first part A
Back on July 9 ES posted about surplus piezo devices appearing on eBay for under $50. Many became curious, myself included. The stage I received was a Physik Instrumente PI P601K. Later I found out these are custom devices and no documentation available, so going forward on our own. No bargain OEM controllers were available, so an effort began to produce a custom Driver/Amp Controller at low cost to drive these PI devices for the macro community.
Physik Instrumente P601K Stage
The P601 stage is meticulously machined from a solid chuck (130X40X20mm) of Stainless Steel (SS) including the 25mm diameter lens mount, they are designed for remote focusing a heavy microscope objective. The inside of a section is machined out to allow a long 10X10X36mm ceramic piezoelectric element to fit and is held in position by natural stage compression at an angle of ~9 degrees. Various laser or electron beam precision "L" and "U" cuts are throughout the area around the elements, these cuts produce a "flexure" type bearing which has no noise, hysteresis or slippage issues if operated within bounds. They simply allow bending of the material around the cuts under stress. The stress comes from the piezo element expanding under a voltage influence, where the elements expand under a + voltage and contracts under a - voltage. So only a single possible moveable mechanical surface interface at each element end (normally doesn't move under expansion), no other surfaces involved to the lens mount!!
The general idea is to expand the piezo elements under applied voltage, which causes stress in the stage, that is translated by the multiple "flexure" bearings to move the lens mount. Since the elements are held in place by natural stage compression and + voltage causes expansion, you have a limited range under piezo element compression where the elements are no longer in contact with the stage walls, so this is the reverse or - voltage limit of the stage. Because of this compressive arrangement and the ceramic element details large - voltages are not recommended as this can dislodge the element and possibly damage the ceramic element. Later on this range was found to be about -30 volts, whereas the + voltage can exceed 100 volts, so a highly asymmetrical drive voltage is required from the Driver/Amp.
The ceramic piezo elements are very capacitive (>5uF), which is highly non-linear with significant asymmetry and long term memory effects, and are non-linear with multiple long time constants. The element expansion is also non-linear and experiences hysteresis and memory effects. These are complicated elements and these effects must be accounted for in the Voltage Driver/Amp parameters.
Position sensing on the P601K is by means of dual resistive Strain Gauges, each gauge having 2 identical resistors. The resistors change value with applied stress, linearly within range. These gauges are directly bonded to the top of the SS block above the cavity where the piezo elements are held and at each end of the element just above the "L" shaped cuts. These positions experience bending stress but in opposite directions under element drive voltage. The gauge is arrange in a distributed Bridge configuration where 1/2 the bridge is on each end of the element. The Bridge is connected with the resistors arranged in a cross coupled connection, so 4 wires are used to connect the resistors together and to the Bridge cable. Under applied voltage to the top and bottom of the Bridge the two "center" connections will have a voltage than is 1/2 the applied voltage. Under stress the opposite Bridge resistors will experience a different stress and thus produce a different resistance which produces a different voltage due to the voltage divider configuration.
For example, the resistors are 1000.0 ohms each, and the applied voltage is 5.00000 volts. Under no stress the 2 Bridge centers will have 2.50000 volts each. Under full stage stress the resistor value under stress changes to 999 ohms which will cause the Bridge center voltage to now be 2.50125 and 2.49875 volts respectively, for a difference of 0.00250125 volts. This is about the range to expect, and at such a small level difficult to process accurately.
P601K Driver/Amp & Controller
The driver amplifier design was started without any knowledge of the OEM Driver/Amps, later after the development some information on the OEM types revealed details. Since the amplifier "load" is the piezo element which is highly capacitive (>5uF) a different amplifier topology was selected rather than the usual low impedance "audio" type amplifier which would oscillate without additional compensation and stabilization.
The amp is based upon a complementary transconductance type with a high output impedance for driving the capacitive load. By using this technique a small current can be "forced" thru the high transconductance impedance to create a large output voltage and large voltage gain, both of which are necessary to drive the piezo element. Bipolar transistors were selected rather than MOS devices because the bipolar has a higher output impedance and higher transconductance. The core the amplifier is an op amp which has low offset and high gain and complementary PNP and NPN devices are the active output drivers, each with emitter ballast resistors to ground and +130V supplies respectively. Compensation is accomplished by a lead/lag lag network on the output and lead feedback so the amplifier is completely stable with all expected piezo loads. Some series resistance is employed to help protect things in case of a load mis-wire and the outputs are designed to current limit and self destruct in extreme cases to protect the output piezo load. Various LEDs are utilized to indicate output voltages and currents as well as supply voltages.
The higher supply voltage of 130V (adjustable) is created with efficient switch-mode boost type regulator, lower supply voltage of +9V is created with linear regulator as is the precision voltage reference. Everything works from a single external +12V Power Supply at less than 1/2 Amp.
Later the information on the OEM Driver/Amp was discovered. This showed a low impedance MOS "source follower" driver type amplifier with added compensation and slew rate and current limiting as well as series resistance to help protect the load, so a completely different approach. I'm glad I didn't find this before the design since likely would have just followed the OEM design approach with a conventional low impedance "audio" type amp.
The Controller is designed to interface with the popular Raspberry Pi (RPi) computer, and allow operation with routines including the BASIC like language Python. The decision was made to use inexpensive components rather than high precision, high cost components normally associated with this level of performance devices. A 12 Bit Microchip DAC is the heart of the Controller and is interfaced with the RPi using the I2C serial port. A precision 4.096 Volt reference supplies the DAC and thus with 12 bits each bit represents 0.001V. The Driver Amp is configured with a gain of 40 initially and later 30 (adjustable), so the scale factor is either 40 or 30mv per bit at the amplifier output. This proved to be a good choice and the Microchip DAC is surprisingly good for an inexpensive DAC, where normally one would use an expensive device. An added feature to the Controller is a 16bit PGA ADC that also operates from the I2C serial port, this can be configured to read the strain gauge output differentially and also help with setup calibration. Like the DAC the ADC can be programmed from Python making custom software routines somewhat easier to create than with C.
All in all we were able to create a very versatile custom Open Loop Controller/Driver/Amp for the surplus PI P601K stage at a low cost, but still achieve very good results. This system is operated "Open Loop" meaning that the feedback from the strain gauges is not employed to correct the stage position. Some initial work was done using the ADC to read the strain gauge and adjust the DAC to correct the error, sort of a pseudo-closed loop, but this is a work in progress and likely for others to follow up with.
Anyway, now comes part B for full Closed Loop Operation.
Best,
