Eventually, any motion control system must achieve and maintain-a programmed, specified position. The positional servomechanism aids in achieving that end.
An actuator under load in a typical hydraulic application is equipped with a position transducer whose output is fed back to a controller and compared to a command input signal. The basic form of comparison is to subtract the feedback signal from the command signal, which, by definition, is the positioning error at any instant.
The explanation normally given is that if the error is not zero, the valve opens, the actuator drives the load and position transducer until feedback and command are equal, the control valve centers, and the actuator stops. This explanation serves a certain introductory, but idealistic, purpose. In this oversimplified scenario, barring transducer error, the feedback signal from the position transducer must be at the commanded value. In the ideal world, only transducer error puts the actuator at a position other than that desired.
Error is unavoidable
In the real world, servo loop closure guarantees only that conditions will automatically be found to ensure that the actuator will stop (assuming that servo loop closure is stable and the command signal is constant). Whether or not the actuator stops in the right place depends entirely upon the design of the servo system. In fact, the actuator will never be in the commanded position, only close. How close is a function of the loop design.
Positioning error is unavoidable because any real-world hydraulic control valve has internal leakage. At null (with the spool nominally centered), the valve meters pressure, not flow. So instead of blocking flow to stop the actuator, the valve causes a force balance on the actuator to stop it. This leaves the proportionally controlled servo loop susceptible to external influences, called disturbances.
There are eight generally accepted sources of disturbance (Group 1 errors) that the servoloop designer must take into account at design time:
- valve null shift from supply pressure variations
- valve null shift from temperature variations
- valve null shift from tank-to-port pressure variations
- load variations at the actuator output
- actuator and load breakaway friction
- valve dead zone (overlap)
- valve hysteresis and backlash within the loop
- valve threshold.
The last three are not disturbances in a strict sense, but the designer can treat them as such. Therefore, design the system to maintain positional error within the required tolerance for all anticipated operating conditions. Each disturbance can be resolved into an equivalent disturbance current, allowing the designer to take all of them into account as a lump sum. Doing so reduces the complexity of subsequent computations.
Other sources of error
Other sources of error (Group 2 errors) also must be considered. The following list is not intended to be exhaustive but merely to represent the range and consequences. These other sources of, or contributors to error include:
- transducer errors, which include calibration error (which is systematic) and non-repeatability (which is random)
- feedback or controller digital resolution
- command signal error and resolution
- external magnetic influences that can affect the valve null
- other electronic noise, whatever the source or sources.
Dealing with errors
The purpose of listing these error sources or contributors is to identify them in advance so they may be anticipated and dealt with. The Group 1 disturbances should be taken into account at the earliest stages of the design process. In fact, they are being considered by more and more value-added fluid power distributors at proposal time. It can be disastrous to build a system and then find that the accuracy criteria cannot be met simultaneously along with loop stability and settling time requirements.
Group 2 disturbances must also be considered at design time, but often they can be deferred until those from Group 1 have been resolved. All these conditions can be dealt with, but let us close this month's discussion with the transducer, because the strategy for correcting transducer problems is very simple.
Unlike the Group 1 error contributors, errors introduced by a transducer propagate undiminished, directly to the output position. This means that if an error in the transducer of, say, 0.015 in. exists, then 0.015 in. is the absolute minimum error that can be expected in the output position — even if we assume that all other error contributors add up to zero! The only way to improve the system accuracy by reducing error is to replace the transducer with a "more accurate" one — one with less total error. If the problem with the transducer is calibration, then it must be recalibrated. No servo loop changes can be made to improve a transducer's accuracy.
Any of several different actions can be taken to improve servo loop accuracy affected by Group 1 disturbances. Some can be implemented at commissioning time by tuning the controller.
Electrohydraulic pressure control
Author Jack Johnson says his newest book is the first dedicated to the electrohydraulic control of pressure. It opens with basic circuit laws that apply to hydraulic circuits. Further discussion covers the effects of orifices and how their pressures and flows interact with one another in real circuits.
Fluid compressibility and the fluid properties that affect it also are detailed. Because compressibility is so closely related to accumulators, complete, but user-friendly models for them are included. Also covered is the manner in which control loop gain becomes non-linear, including formulas for calculating the point of maximum loop gain.
Integral control is introduced, and both analytical and experimental results are used to show how challenging it is to achieve good control in systems that require a period of velocity or position control and then must switch to pressure control and back again. The book ends with a chapter on force control using cylinders.