Choose and use your valves wisely to achieve precise control.
Good performance requires more than just a great motion controller. Even the best controllers cannot compensate for poorly designed systems and poorly-selected system components. As explained in the March 2006 issue (Choosing the right valve, pg. 30), servo-proportional valve characteristics can have a huge effect on the performance of a closed-loop motion system. Components such as counterbalance valves can interfere with the operation of servo and proportional valves. Tight schedules sometimes lead to poorly thought-out designs and the selection of incorrect components. The result is often many frustrating hours spent trying to get the desired performance from the system. A better understanding of some common valve issues can shorten system setup time and achieve more precise motion.
A problem with drift
Drift can be a tricky problem in a hydraulic control system. We will discuss two drift topics, the relatively straightforward constant actuator drift and the more elusive issue of null drift. Actuator drift occurs when a valve is out of null, resulting in a piston moving slowly or drifting when there is no control signal (e.g. when the electrical power is off). In some cases, this drift is desired — such as when null is adjusted so that the piston rod retracts to a safe position upon loss of the control signal.
Problems arise when the rate of drift is too high or in the wrong direction. For example, with a high drift rate, as much as a 10% control signal to the valve could be required just to compensate for the out-of-null valve. If a 10% control output is required just to hold position, only 90% is left to make the actuator move in the direction opposite the drift. Consequently, the actuator may only get to 90% of full speed in that direction. Therefore, in applications where quick moves are needed, a strongly biased null valve can keep the actuator from reaching the desired full speed.
Adjusting the valve null is as easy as turning a screw on the servovalve or turning a pot on a proportional valve amplifier. With the control signal to the valve set to zero, the screw or pot is adjusted until the actuator stops drifting. Alternatively, with the axis holding position in closed loop, you can adjust the null screw or pot until the control signal to the valve is zero volts. The null can also be compensated for in the motion controller by adjusting the bias or null parameter. Keep in mind, however, that this has the speed-limiting disadvantage mentioned above.
If the motion controller has an integrator term in its closed-loop algorithm, the integrator will automatically compensate for the null while it is in closed-loop control mode. However, misusing the integrator as a null compensator can lead to some unexpected behavior. For example, because the integrator is not applied in open-loop mode, the null will not be corrected during jog moves or any part of a cycle where an open loop move is used. So, it is best to adjust the null at the valve or adjust the bias parameter in the motion controller instead of relying on the closed-loop compensation of the PID algorithm.
Beware of the drift
A varying null condition, called null drift, is a more serious problem. This can be caused by backpressure, flow forces, or valves with non-existent or poor spool control. Null drift requires the controller to be constantly changing the output to the valve to hold a position or maintain a pressure.
Spool control is important to minimize null drift. A well engineered servo-proportional valve controller has an inner control loop that moves the spool position to correspond to a control signal, Figure 1. Ideally, the spool position would move to a +50% flow position when the controller sends the valve a 50% control signal.
Now assume the controller is sending a 0% control signal to move the spool to the null, or 0%, position. As the spool gets closer to the 0%, position the error becomes small, so the force to correct the error becomes small. This force may not be large enough to overcome real world friction or flow forces so a small null error will remain.
Valves with only a proportional control will not reach the desired location, because there isn't enough force from the spool controller to reduce the error to zero. A PI (proportional with integrator spool) controller has an integrator term that will eventually reduce the error between the desired and actual spool location to zero and minimize null drift.
Dealing with deadbands
It may be tempting to eliminate null and null drift by using a closed-center valve. These valves have spools that are cut such that there is a deadband region around zero where no oil flows at all, Figure 2. Some valves have deadbands as great as 20%, which simply means that the controller must output a 20% control signal just to get to the point where oil begins to flow. These valves can be a problem in applications where the spool must be shifted back and forth across the zero point to hold a position or pressure.
Many motion controllers have deadband parameters and can partially compensate for the deadband region. In the example above, the output could immediately step up to 20% in the direction that the spool needs to move. Unfortunately, this does not help much in many applications since the spool still takes time to move.
Often, the controller must command the valve to quickly route oil to and from each side of the piston to maintain a position or a differential force across the piston. This can mean that the spool must be rapidly shifting as much as 40% of spool travel — just to hold position!
There is no oil flow when the spool is in the deadband region. Thus, the response measured by the position or pressure transducers is effectively zero during the time (potentially several milliseconds) it takes to shift the spool through the deadband.
Valve manufacturers rate their valves by how quickly they shift their spool, but in many applications, that isn't as important as how well they control the flow of oil around the zero point. This behavior around zero, Figure 2 (a), is certainly not the linear response the control algorithm wants. Because of the negative impact on response and the extreme non-linearity around zero, valves with deadbands should be avoided in servo applications — use zero overlap valves instead, Figure 2 (b).
The only time that valves with deadbands can be controlled reliably with precision is in applications where the spool is shifted to one side and only operated in that region. Velocity control applications are an example of control in this manner — the velocity of the hydraulic actuator is controlled by how far the spool is shifted to one side. Valves with deadband can also work in some pressure control applications if an orifice is placed across the valve's A and B ports. In this case, the valve must be shifted past the deadband, so the flow is offset by the leakage from the orifice which maintains pressure.
Counterbalance blocking valves
Counterbalance valves are a type of safety valve used to keep loads from dropping when a loss of hydraulic pressure occurs. However, in servo systems, use of these valves often creates major problems. As a general rule, there should be only one active valve that controls flow in a servo system. Any valve other than the servovalve that changes the flow will interfere with the control of the system.
When a counterbalance valve is used in a closed-loop system, it must be applied properly. In a typical application involving a vertical cylinder, a counterbalance valve might be installed between the rod end of the cylinder and the servovalve. This means the servovalve and the counterbalance valve must both be open in order for the actuator to move down. The counterbalance valve is controlled by pressure — whenever the pressure is above a threshold, pilot pressure opens the counterbalance valve. For plumbing convenience, the pilot port is sometimes connected to the cap end of the cylinder, Figure 3.
With this configuration, it is possible to have a condition where the servovalve is trying to extend the piston rod, but the flow is blocked because the pressure on the cap end of the cylinder is not high enough to open the counterbalance valve.
When the pressure is finally high enough to open the counterbalance valve, the actuator moves down too rapidly, due to the excess pressure/force on the cap end of the piston and the load weight hanging from the vertical cylinder. The controller then throttles back the flow to the servovalve, in order to slow the motion. This causes the pressure on the top of the piston to drop below the pilot set point so the counterbalance valve shuts abruptly — jerking the axis to a stop. (This pressure decrease occurs because the cap end of the cylinder requires more oil than the rod end, so it cannot get enough oil fast enough to hold pressure.) Now the controller again increases the control signal to cause the cylinder to move down which increases the pressure again and the cycle repeats.
In this situation, the cylinder will chatter as it extends. This effect can be minimized by reducing the pressure setpoint so the counterbalance valve opens sooner. However, this technique is merely minimizing a flawed control situation. It would be better if the counterbalance valve pilot port was connected to the supply pressure so it is always open during normal operation and interferes as little as possible with the servovalve.
Solenoid-actuated blocking valves
When the application requires servo control, often a much better approach is to use normally-closed blocking valves that are energized open during normal operation by the logic associated with the emergency stop control. The open blocking valves have no effect on the motion, nor do they interfere with flow during normal operation so the motion controller performance will not be adversely affected, Figure 4. When an error does occur or power is lost, the valve shuts and hydraulically locks the actuator in position.
For more information, visit www.deltamotion.com or contact the author at peter@deltamotion.com.