Piezoelectric Direct Drive Servovalve

Piezoelectric Direct Drive Servovalve Jason E. Lindler, Eric H. Anderson CSA Engineering 2565 Leghorn Street, Mountain View, California Industrial and Commercial Applications of Smart Structures Technologies San Diego, CA March 2000 Copyright 2002 Society of Photo-Optical Instrumentation Engineers. This paper was published in the Proceedings of SPIE Volume 4698-53, Industrial and Commercial Applications of Smart Structures Technologies 2002, and is made available as an electronic reprint (preprint) with permission of SPIE. One print or electronic copy may be made for personal use only. Systematic or multiple reproduction, distribution to multiple locations via electronic or other means, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.

SPIE Paper 4698-53, Industrial and Commercial Applications of Smart Structures Technologies, San Diego, March 2002 Piezoelectric Direct Drive Servovalve Jason E. Lindler and Eric H. Anderson CSA Engineering Inc., 2565 Leghorn St., Mountain View, CA 94043 ABSTRACT A single-stage servovalve using direct piezoelectric actuator drive is described. The single-stage servovalve design offers higher bandwidth than conventional two-stage valves. It takes advantage of the high energy density in piezoelectric materials while addressing the need for internal amplification of stroke. When used alone, the valve can regulate pressure, and when used in combination with a hydraulic output device it forms part of an effective servohydraulic actuator. Development of a direct drive prototype valve is described. Discussion includes design issues related to low stroke smart material actuators such as piezoelectrics. Component and subsystem testing and results are reviewed. Electronic drive and control of the piezoelectric and overall device along with performance in the control of fluid flow is discussed. The value of the new servovalve is shown in the combination of the valve with a hydraulic output device. Data are supplied for this servohydraulic actuator. The new actuator shows promise for a motion simulator application and more generally for motion control at higher bandwidth than is possible with currently available servohydraulics. INTRODUCTION While all-electric, or electromagnetic actuation is appealing for some applications, it isn t appropriate for others. For many high load systems, hydraulic devices have remained a necessary and desirable means of actuation. The control of these hydraulic actuators is commonly effected through the application of various types of servovalves. In most servovalves a mechanical or electrical signal is utilized to direct the position of a valve spool within a valve housing. The position of the valve spool determines the flow path(s) between valve ports to direct flow to the ports of a hydraulic actuator, thus determining the direction of force application and motion of the actuator. The need for higher frequency, more precise control of systems and devices has led to ever improving servovalves and other components. Servovalves have evolved from relatively slow acting two-stage devices, where the first stage consists of a pilot valve, to faster versions of the same architecture. The pilot valve is controlled by some low power means such as human force input or a low power electrical signal, and shuttles a pressurized fluid or gas supply in a pilot system which thereby displaces a valve spool in a larger flow, power control valve. The power control valve in turn determines the flow direction to a high force, large displacement actuator. The compliance and inertia in the hardware used for pilot valve operation and the compliance and fluid inertia in the pilot fluid system combine to significantly reduce the frequency response of the power control valve to the original system command input. The need for faster acting systems has led to the development of various single-stage control valves where a single, directly controlled system develops the force necessary to shuttle the valve spool in the flow control valve governing the high force actuator. This approach simplifies the internal component arrangement while retaining the fundamental multi-way action of the spool valve. One area of interest that motivated the development of the single-stage device described here was flight motion simulation. 1 A higher bandwidth servovalve would enhance overall simulator performance bandwidth, particularly in the case of small simulators. 2 The same approach could be used in pneumatic servovalves, but the immediate interest is in hydraulic systems. CONVENTIONAL TWO-STAGE SERVOVALVES Design and operation of servovalves is fairly well understood. 3,4 As described above, a typical servovalve consists of two stages, in which the first is an electromagnetic torque actuator, and the second is a clever multi-way fluid valve. In hydraulic control valves, mechanical motion of a spool directs fluid power between ports located on one side of the valve. In four-way valve operation, a manifold routes flow from a pressure port and a return port to the spool inside of a 1

sleeve. In addition, the manifold routes flow from two control ports, which normally are connected to an actuator, to the spool inside of the sleeve. In the center position, flow edges on the spool block the pressure and return ports from two control ports. Any displacement of the spool allows fluid flow from the pressure port to one of the two control ports. At the same time, the spool displacement allows fluid flow from the other control port to the return port. When the control ports are connected to sides of a hydraulic actuator, fluid flowing into one control port and out of the other control port creates motion of the actuator output shaft. Motion of the spool in the opposite direction reverses the direction of the flow and subsequently the direction of motion of the actuator shaft. The displacement of the spool relative to the sleeve regulates the amount of flow going into and out of the control ports. This proportional control of the fluid flow can be used to command a hydraulic actuator. Figure 1: Hydraulic circuit of a two-stage servovalve with pump supply and mass load In these standard hydraulic control valves, or servovalves, a low power electrical signal is hydraulically amplified to control the position of the spool. For the hydraulic amplification, an electromagnetic torque motor connects to a deflection tube that balances the pressure drop between two nozzles This action is shown schematically in Figure 1 where an external pump supplies pressurized flow. Application of current through coils in the torque motor angles the deflection tube, which then blocks flow from one of the nozzles. Blocking the flow through a nozzle creates a pressure differential that shifts the location of the spool in the sleeve. The location of the spool within the sleeve then directs the flow of hydraulic fluid. As pressurized fluid is supplied to one side, and removed from the other side of a hydraulic output device or actuator, the device moves and drives any attached load. DIRECT DRIVE SERVOVALVES While hydraulic amplification allows for low power electrical command signals, the response time of this amplification limits the bandwidth of the resultant actuation systems. To overcome the bandwidth limitation of hydraulic amplification, one possible approach is electrical amplification of the command signal. With electrical amplification of the command signal, an electronic amplifier and electromagnetic motor directly control the position of the spool. This is a viable approach that has been realized in specialized products. However, the weak force and energy density of electromagnetic motors limit the force and bandwidth of these direct drive systems for a given device size. High energy density piezoelectric materials present a possible alternative to electromagnetic actuation to further improve the response time of the spool in direct drive systems. A direct drive valve (Figure 2) is a device in which there is no hydraulic amplification of the electrical command signal. Instead of an electrically driven torque motor that allows flow to move the spool, the electrically driven actuator itself drives the spool. While the direct drive of the spool presents a solution to the problem of bandwidth, a number of other design challenges arise with direct drive servovalves. The force to accelerate the spool mass and flow forces on the spool can be high and thus require physically large electromagnetic actuation. Another important issue in direct drive devices is that the actuator must deliver the full motion of the spool, i.e. the spool lift must be large enough to allow significant flow of high-pressure fluid. 2

Insufficient spool motion does not allow fluid movement through openings and also results in a number of tolerance issues. Generally, spool lift is easy for an electromagnetic actuator to deliver at low frequencies. Direct drive arrangements are not necessarily well-balanced hydraulically. In hydraulically-amplified servovalves, side forces on the spool are balanced to prevent loads from creating excess wear on the spool that reduces the lifetime of the system. However, in most direct drive systems, the attachment point to the spool results in unequal fluid pressure being applied to the spool. As a result, careful design of the attachment point is required to prevent transmission of side loads from the mechanical drive to the spool. In addition, in the case of a power failure, the spool should return to the neutral position and not direct the hydraulic fluid. Figure 2: Hydraulic circuit of a direct drive servovalve with pump supply and mass load Direct drive servovalves do offer higher bandwidth. They also provide other advantages over conventional servovalves. First, typical hydraulic servovalves require careful balancing of the nozzle bridge amplifier to prevent a DC offset in the hydraulic command signal. However, directly driving the spool with electrical amplification allows for a simpler electrical balancing of the spool. Furthermore, directly driving the spool with feedback allows for advance positioning control strategies to improve the system s response time and linear performance. These advantages along with higher bandwidth provided the motivation for work on a direct drive piezoelectric servovalve. PIEZOELECTRIC MATERIALS The piezoelectric effect is exploited in numerous transducers. In the basic effect, a piezoelectric element generates a charge when it is subjected to an input mechanical stress. In the converse effect, the element develops an output mechanical deformation when an electric field is applied. For this application, the piezoelectric behaves as an actuator, although enhanced control may be possible if the element were to be used simultaneously as an actuator and sensor. Piezoelectrics were selected to be the primary actuator material for several reasons. First, these materials have high power density high power output per unit volume or mass. Servovalves should be compact to allow proper integration into larger control system hardware. A high bandwidth valve of the same size as conventional two-stage valves is desirable. Second, the materials are stiff, increasing the likelihood that the internal dynamics will not severely limit overall device bandwidth. Third, these materials respond well at high frequencies, far beyond what is required for this application. Finally, piezoelectric actuators are available in geometries that are readily integrated into other devices. Piezoelectrics produce a more linear response than electrostrictives or magnetostrictives. They provide a much smaller overall size than magnetostrictives. However, like other stiff smart materials, piezoelectrics do not produce large strokes. They are typically capable of only 0.1% strain, perhaps 0.15% for certain compositions. This constraint, and the lack of off-the-shelf high voltage, high current amplifiers, together impose the greatest potential limits on the use of piezoelectrics in many applications. In this area, advances in single crystal materials offer hope for higher strain and enhanced device performance. But for the demonstration device here, mechanical amplification was necessary. 3

DIRECT DRIVE PIEZOELECTRIC VALVE The piezoelectric-based valve is designed as a direct replacement for a conventional servovalve such as the ones described by ISO 4401-03-03-0-94. Therefore, one constraint on the design was that standard fluid porting be used. This fluid porting gave the valve a geometry that was similar in several overall dimensions to conventional two-stage valves. The design was completed using knowledge of piezoelectric actuator capabilities and consideration of applicable performance trades in traditional servovalve design including the total required spool motion. Figure 3: Piezoelectric direct drive servovalve solid model with cross-section The external view of the servovalve is shown on the left in Figure 3. A piezoelectric actuator replaces the electromagnetic torque motor and the hydraulic amplifier. In addition, there is no mechanical feedback in the piezoelectric servovalve. The cross-section on the right in Figure 3 shows detail of the interior. The piezoelectric stack in the upper section changes length and drives the vertical element on the right of the assembly. This lever element effectively amplifies the piezoelectric motion, trading force for a five times or greater increase in stroke. With a nominal stroke of up to 60 µm, the piezoelectric stack will move the spool by up to 0.3 mm. Up to twice as much spool motion may be desirable for large flows, but this level proved adequate for prototype demonstration. The pivot point of the lever arm seals the hydraulic fluid from the piezoelectric material, and the perpendicular location of the lever arm also prevents fluid pressure from loading the spool in the axial direction and causing an offset from the neutral spool position. A sleeve positioning screw allows for the axial positioning of the spool within the sleeve by compressing the sleeve preload spring. Careful attention must be made not to side load the spool, which could create excess wear on the spool and sleeve (Figure 4). Figure 4: Piezoelectric stack actuator (left) and spool and sleeve of the direct drive valve (right) 4

The piezoelectric stack (Figure 4) uses a special high strain PLZT composition from Rockwell Scientific. With dimensions 40 20 20 mm, and a room temperature capacitance of approximately 3.5 µf, the transducer presents a large reactive load that is difficult to drive with standard electronic amplifiers. Fortunately, an arbitrary drive waveform is not necessarily required. Periodic drive signals are usually adequate for most servohydraulic applications. Figure 5: The prototype piezoelectric servovalve Electronic control via feedback can be achieved by locating feedback sensors on the left end cap (Figure 3) to monitor the state of the spool. Furthermore, another servovalve configuration could include integrated electronics for local feedback control of the device. In the present research these sensing and control additions were not realized in hardware. EXPERIMENTAL RESULTS Two groups of tests were performed on the new actuator combination. First, the piezoelectric valve was characterized by itself in order to determine its dynamic characteristics. Second, the valve was used to control flow to verify that it could perform as a controllable valve, including when it was connected to a hydraulic output device. Figure 6: Frequency response of the piezoelectric valve driven at low amplitude with a broadband input 5

The mechanically amplified piezoelectric response was measured using eddy current displacement sensors to determine the bandwidth of the piezohydraulic servovalve. These measurements were done with the valve end cap removed and the sensor sighting directly on the end of the spool. The dynamics were characterized in the absence of hydraulic fluid. A typical response plot is shown in Figure 6. The most distinctive feature of the plot is a very lightly damped mode at about 340 Hz. This is the first device mode involving the spool bouncing on the piezoelectric stack and lever compliance. The extremely low level of damping (Q ~300) is expected to increase significantly once the spool is inside the sleeve and surrounded by fluid. A low frequency amplification of about 7:1 was achieved. With the stack displacement of 50 µm, the spool motion is about.350 µm. As a conventional servovalve may have about 500 µm of motion, the piezoelectric valve s smaller displacement reduces the rated flow of the valve. A resonant frequency of about 500 Hz was hoped for in this first design. However, the additional compliance in the attachments between the piezoelectric stack and the lever, and between the lever and spool acted to reduce this frequency. Future designs would improve these interfaces. These designs could also make use of a slightly longer stack and the same lever ratio for larger overall valve stroke, or a smaller lever ratio for the same stroke and possibly higher natural frequency. A resonant frequency between 500 and 1000 Hz is necessary to offer significant advantage over existing two-stage valves. The piezohydraulic servovalve was then tested with fluid flowing through the valve. In the test setup, a variable orifice can be opened or closed to present a load (Figure 7). Pressure can be measured on both sides of the active valve. This test setup offers many advantages for characterizing a valve. Most of the advantages result from the fact that the control ports are connected to an extremely small volume. In a hydraulic actuator with large volumes of fluid, the performance of the actuator depends on the valve and the hydraulic pump. As a result, for certain frequencies the performance of the actuator is limited by the lack of flow and not the servovalve. However, with the test setup, the small volumes connected to the control ports prevent flow from limiting the response of the pressure. As a result, only the bandwidth of the valve limits the response time of the pressure to either side of the control ports. Figure 7: Test setup for measurement of valve response in control of pressure across variable impedance load To test the performance of the valve, the piezoelectric was driven with a square wave voltage. Pressure was measured at ports on either side of the valve. Results for four different drive frequencies are shown in Figure 8. Note that the valve drive is strictly open loop. There is neither a differential pressure loop, nor a position feedback loop closed. The quick rise time of the pressure demonstrates the fundamentally fast response time of the piezoelectric servovalve. The oscillations present in each response, but most apparent in the lower right plot, in which the square wave input is at 25 Hz, are due to the 340 Hz mode. The damping in the mode has increased significantly from the response measured 6

without fluid present. In the system with fluid, the damping ratio is about 2%, roughly 30 times that measured in dry air. The system response could be damped further with feedback control or by increasing passive damping in the critical mode. Figure 8: Pressure response of the piezoelectric valve driven with a square wave input Figure 9: Piezoelectric servovalve integrated with custom hydraulic actuator 7

A demonstration servohydraulic actuator was designed (Figure 9). The actuator includes a servovalve and a custom hydraulic output piston. Note that one end of the piston is contained within the actuator. The actual stroke is larger than 0.5 inch in this engineering prototype. To save space and weight, the manifold for the valve is directly integrated into the hydraulic cylinder. Incorporating an industry standard bolt circle and port layout, the new cylinder allows for control from either a commercial servovalve or the new piezohydraulic valve. Figure 10: Testing of the new actuator driving a mass simulator For our purposes, the ultimate test of the new actuator assembly is in an arrangement that captures the essential features of a motion control simulator. Figure 10 shows such a test setup with a 10-pound mass acting as payload. From an external pump, pressure and return lines were connected to the manifold of the hydraulic actuator. A high voltage amplifier was then used to drive the piezoelectric servovalve. Testing up to 200 Hz sinusoidal excitation demonstrated the feasibility of the piezoelectric servovalve for high bandwidth applications. One of the difficulties in testing was the interaction of the servovalve with the pump. The 0.5 gallon per minute (GPM) pump was unable to supply the necessary flow to the valve to allow high displacement operation. A larger 5 GPM pump was also used. Because the servovalve operated open loop, it was difficult to carry out tests with durations of more than a few tens of seconds. Further, pressures were limited to 1000 psi because the actuator prototype was designed to support up to 1500 psi safely. Figure 11: Measurements from the combined actuator tests 8

Typical data from the tests is shown in Figure 11. This data was measured at low drive amplitude of 100-200 V. The actuator can be driven up to 800 V. Sinusoidal motion was achieved, although the displacement amplitudes were low. The drift due to the open loop nature of the operation is not visible over these short time scales, but it was significant. Implementation of local feedback control is planned to enhance device performance. CONCLUSION The current piezoelectric valve is rated for 2 gallons per minute at 1500 psi. Modification of various O-rings and seals would allow for 3000 psi operation. Increasing the displacement of the spool would raise the maximum flow rate of the servovalve. However, achieving the increase in displacement by raising the lever arm amplification factor would reduce the bandwidth of the system. A better solution for increasing the stroke of the spool, without affecting the lever arm, is to incorporate a longer piezoelectric material into the servovalve. An alternate approach would use a higher output piezoelectric. The natural frequency of the direct drive valve limits the bandwidth of the servovalve. In the current design, more detailed manufacturing of the spool could reduce the mass of the valve by 1/3. Unfortunately, a 1/3 reduction in mass corresponds to only a 1/9 increase in the natural frequency. As a result, to raise the natural frequency of the valve it is preferable to improve the stiffness of the lever arm. Improving the attachment points and thickening the lever arm would increase the stiffness of the valve while only adding a small amount of mass. Other modifications would address internal stress concentrations at attachment points. For enhanced performance, a non-contact sensor, such as an eddy current sensor or a linear variable differential transformer (LVDT), could monitor the spool position. The sensor would allow for control strategies useful in high speed positioning of the spool. In addition, for spool position sensing, a housing should be incorporated into the servovalve to provide a space for signal conditioning close to the sensor. Finally, replacing the spool position sensor with an alternate pressure transducer would create a piezoelectric high-speed pressure control valve. The basic feasibility of a new piezoelectric servovalve was demonstrated. The new piezoelectric direct drive servovalve offers the potential for faster response compared to a traditional two-stage servovalves or even direct drive electromagnetic valves. The valve can be used for direct pressure control or as a flow controller. The servovalve and hydraulic output device comprise a servohydraulic actuator. The next generation device will incorporate several improvements to increase bandwidth and overall performance. ACKNOWLEDGEMENTS This paper reports on work conducted for the Air Force Research Laboratory under contract F08630-00-C-0058, Lt. Ben Smallwood, Program Manager. The authors thank Patrick Atkins for helpful input. REFERENCES 1. J. M. Carter, New HWIL Motion System Developments, Technologies for Synthetic Environments: Hardware-inthe-Loop Testing VI, SPIE Paper 4366-22, April 2001. 2. E. Anderson et al., Image Stabilization Testbed (ISTAT Technologies for Synthetic Environments: Hardware-inthe-Loop Testing VI, SPIE Paper 4366-24, April 2001. 3. H. Merritt, Hydraulic Control Systems, John Wiley & Sons, New York NY, 1967 4. J. Johnson, Designer s Handbook for Electrohydraulic Servo and Proportional Systems, IDAS Engineering Inc., East Troy WI, 2000 9