Like so many of the myths and misconceptions surrounding hydraulic fluid power, the idea that valves are used to turn on and shut off flow is widespread. But, alas, in many cases it is wrong. The reason is simple: In spool valves, a physical clearance exists between the spool and its mating bore that causes internal leakage.
Internal leakage, in itself, can cause a connected cylinder to creep, the rate of which is a complex function of the amount of clearance in that valve specimen, the size of the cylinder, and the magnitude of load. However, and I am guilty of this, it is often stated that “the valve centers, the flow stops, and the cylinder stops.” It is easy to visualize this, and it serves the immediate purpose.
However, this rationale leads to a misunderstanding of the exact mechanisms at work in precision electrohydraulic motion control and servomechanisms. It also yields to misleading conclusions about where the errors come from in a positional servomechanism. More importantly, we need to see how a design methodology can be developed that allows building systems that meet a specified positional accuracy criterion.
Why the piston stops
For example, a performance specification for an electrohydraulic motion control system might be: The system must move a load of 10,000 lb at a maximum speed of 14 in./sec, provide a productivity rate of 8 cycles/min, dwell for 2 sec at the end of cylinder extension, and move into position with no more than 0.010 in. of error. Systems routinely are designed to meet load and speed requirements, cycle rates, and dwell times. The requirement for achieving position with no more than a specified positioning error is probably an unfamiliar concept. However, we can predict the conditions necessary to reach such a goal and even design for it. But to do so, we need to understand how the valve, cylinder, and load all interact to bring the cylinder to a stop.
This raises the question: What mechanisms are at work to bring the servo cylinder to a stop if the valve cannot shut off flow? Isaac Newton provides us with the answer: The cylinder and load stop because the valve brings them all into force equilibrium, so the flow stops because the piston stops. The actual process has the valve causing force balance, which results in the flow being blocked by the piston, not the other way around.
This raises another question: What is it in the valve that leads to a force balance condition in the load? The short answer is found in the valve’s pressure metering characteristics, or, as it is expressed more simply in technical data from valve manufacturers, in the pressure gain of the valve.
The pressure metering test
Valve flow metering characteristics are commonly known and understood. Pressure metering, however, may be alien to those not directly concerned with precision positioning systems. And yet, it is not a difficult concept to comprehend. We can begin with a look at a simplified cut-away diagram of a typical four-way directional control valve, Figure 1.
|Figure 1. Cutaway view of a simplified 4-way directional-control valve (spool actuation means is not shown). |
Note that the means for shifting the spool in Figure 1 are not shown. It is not important to the immediate discussion, although the shifting means is very important to the operation of the real valve in the application. Further note that the hydraulic power source produces constant pressure, such as a conventional pressure-compensated pump. Also, the spool is drawn with some spoolto- bore clearance, and the work ports are terminated with pressure gauges, or, perhaps, pressure transducers. The valve is drawn as having, more or less, zero lap. Although pressure metering is most often associated with zero lapped (servo) valves, over and under lapped valves can be tested as well.
Figure 1 is the typical setup for measuring and recording the pressure metering characteristics of the valve. It shows a dead-head situation — that is, a zero flow output test. This is deliberate, as well as necessary, because it tells us what the valve and its load will experience when there is no work-port flow.
When hydraulic power is turned on, high pressure develops at the P port. Because of the internal clearances, there will be leakage from P to A and from P to B. However, there can be no flow out the work ports, so we can conclude that all the P-to-A leakage must pass through the B-to-T clearance, while the P-to-B leakage must pass through the A-to-T clearance.
Because of the internal leakage, we can also conclude that the deadhead pressures in the two work ports are not zero. In fact, if the valve were “perfectly manufactured,” the work port pressures would be exactly onehalf the supply pressure when the spool is perfectly centered. This is a consequence of the inability of the valve to shut off.
Valve null explained
The process of nulling the valve involves positioning its spool to the “hydraulic center” of its travel. To null the valve, the electronic driver (the amplifier) is disconnected from the coil, and the spool is moved mechanically until the two work ports have the same pressure. Then the amplifier is reconnected and the valve has been electronically nulled.
Back pressure metering
The pressure metering test involves recording the dead-head work port pressures as the spool is moved across its “center region,” usually referred to as the null zone in technical literature. Without need to resort to complex concepts, we can reason out what is expected with the pressure metering test.
First, we have found that when the spool is centered, internal leakage will cause the work port pressures to be approximately one-half the supply pressure. But what would we expect when the spool is shifted away from null? It follows directly that if the spool is fully shifted — say in the direction of opening from P to A — the A-port pressure will rise to the supply pressure (because there is no work port flow), and the B-port will go to tank pressure. Conversely, if the spool is shifted P to B, the B-port will rise to the supply pressure, and the A-port pressure drop to tank pressure.
Emphasizing that at no time is there any work port flow, it should be clear that a relatively small shift away from the hydraulic center will result in one work port or the other reaching supply pressure while the opposite work port reaches tank port pressure. In fact, in zero-lapped (servo) valves, all of the pressure metering takes place in less than 10% of the spool travel, and often in the center 5%. A rule of thumb is that pressure metering takes place in the over-lapped zone plus about 5% or 10% of total spool travel.
|Figure 2. Pressure metering curves control the conditions that will stop the cylinder. |
A closer look
Figure 2 shows typical pressure metering curves for a nominally zero- lapped valve. The vertical axis is deadhead pressure and the horizontal axis is control. The two curves are the deadhead work port pressures, PA and PB, respectively, recorded as the control moves slowly from a negative value to a positive value. Control could be spool position, command current, or command voltage. The total range of the control variable is, as stated above, the center 5% to 10% of maximum control input. Outside the pressure metering null zone, the work port pressures rise to the supply pressure and tank pressure, alternately and respectively. The slopes of the metering curves, MA and MB, are the respective “per-port pressure gains.”
An important aspect of the pressure metering graph is that the two work port pressures, PA and PB, are equal at a point where the control is not zero. This is labeled as the force equilibrium point. The non-zero control means that the valve is shown as being mis-nulled — not necessarily a disastrous, or even undesirable condition. Also, at that point (where the two pressures are equal, just to the left of zero control) their equalized value, PN, is each less than half of the supply pressure.
This means that either of two conditions exist within the valve — either the return lands of the valve might be slightly underlapped, or the powered lands may be slightly overlapped. We cannot be sure which condition exists without further testing. However, as long as the amount comes close to one-half the supply pressure, it is not necessarily disastrous or a cause to reject the valve after its manufacture. With the non-zero, or mis-nulled condition, when the control is zero, the two work port pressures are not zero, labeled as P0A and P0B, respectively. To help illustrate the force balance condition, Figure 2 implies that the valve is “connected” to a single-rod cylinder. This indicates that the PA side of the valve is connected to the cap end of the cylinder, and the PB side is connected to the rod end.
A case in point
Consider a rather common industrial application situation where no load is connected to the cylinder. We now ask, “What control input is necessary to cause the cylinder to stop?” To find an answer, let’s look at some what ifs. What if the control input (let’s assume it is valve input current, I) is set to zero? The two cylinder pressures would be P0A and P0B. This results in cap-end pressure being higher than rod-end pressure. We know from basic hydraulic theory that high pressure on the cap end area does not constitute force equilibrium in an unloaded cylinder. The cylinder would creep — and we know it will creep outward at a rate determined by the internal leakages of the valve and cylinder areas.
Now, what if the control current is set to a value that results in equal pressures? Again, we know that equal pressures on rod and cap ends will not result in force balance because the cap-end pressure is too high. Therefore, the cylinder will again creep outward. To find the current that results in force equilibrium, we need to know more about the cylinder.
For our purposes, let’s assume the cylinder area ratio is 2 to 1. Now, with no external load on the cylinder, we know the pressure on the cap end must be one-half the pressure on the rod end. That is, we must find a point on the metering curves where PB is twice that of PA. There is only one point on the pressure metering curves where the two pressures meet that criterion. It is labeled as the I0 point on the control axis. When no load is on the cylinder, and if we set the control current to I0, the unloaded cylinder will stop because a condition of equilibrium is established in the face of internal valve leakage.
Carrying the case further
Continuing with another what if, what if the external load on the cylinder was increased — say, in a direction that attempts to retract the cylinder? If we set the control input current to I0 (the current that stops the unloaded cylinder), the cap-end pressure will be too low to sustain the external load, and, indeed, the cylinder will creep inward. So what must we do with the control current to stop the externally loaded cylinder? Clearly, we need to move the current to the right, to a value that results in a higher PA and a lower PB. The new input current must be more positive than the zero load equilibrium current. We need to know much more about the specific values to identify the exact point of the curves — an exercise that is beyond the scope of this basic explanation.
|Figure 3. Combination cutaway and schematic diagram of the closed-loop positional mechanism. |
Now, what if the valve is connected to a cylinder that has a position transducer connected, and the system is further connected to a servo-proportional amplifier in a closed positional servo loop? A simplification of such a system is shown in Figure 3. Note that the valve is now outfitted with some unspecified electromagnetic actuation device, such as a proportional solenoid or torque motor. The specific means of valve actuation is unimportant to our discussion.
Further note that the cylinder is outfitted with a position transducer, labeled H in the figure. The transducer output is fed negatively into an electronic summing junction as VX, where it is compared to (subtracted from) the command input voltage, labeled C. The error voltage, E (the difference between the command and the feedback signal), is fed into the amplifier, A. The output current from the amplifier goes into the valve coil to shift the valve and, thus, cause the cylinder to follow the command voltage. The valve’s pressure metering characteristics are shown between the valve and cylinder in order to establish the stopping conditions.
To explain the operation of the circuit in Figure 3, an error voltage, E, exists if the command voltage is increased above the existing value of the feedback voltage. This voltage is amplified and fed into the valve coil, causing the valve to shift away from null, routing flow into the cylinder. This, in turn, causes the cylinder to move in a direction that increases the feedback voltage.
At some point the feedback and command voltages become equal, the error goes to zero, the valve control current goes to zero, and the spool centers, shuts off the flow, so the cylinder stops. This entire last sentence is in error for reasons already expounded upon. It should really read: The cylinder moves in a direction to increase the feedback voltage, and at some point the error voltage reaches a value that causes a force equilibrium between the cylinder and any external load. This causes the cylinder and load to stop, forcing the valve output flow to be zero. In this condition, neither the error voltage nor the valve current are likely to be zero!
Wrapping it up
We see that the valve will stop the load only when the spool position produces a force balance between the cylinder and the load. The pressure metering characteristics of the valve are vital in reaching this condition. Zero output flow is a consequence of this action, not the cause.
If the load on the cylinder changes, the control current must change to establish new balancing pressures. Therefore, the cylinder must also move. This causes variable errors in the positional servo mechanism, and system designers must take such unfortunate realities into account or be prepared to accept failure in the application.
A special offer for H&P readers
• everything from fundamental hydraulic circuit analysis to dynamic testing of motion control systems,
• more than 500 pages of useful formulas and dozens of fully worked examples, and
• 20 chapters covering hundreds of topics, including components, systems, circuit analysis, modeling, servomechanisms and control, simulation, and testing.
The book normally sells for $99.95, plus shipping and handling. However, readers who order it directly from IDAS Engineering will also receive a free copy of the newly revised and expanded Lexicon of Electrohydraulic Motion Control — a $19.95 value at NO additional cost to you. The Lexicon contains more than 700 terms and their definitions, covering all aspects of electronic components and circuits, analog and digital electronics, transducers, hydraulic principles related to electronic control, and much more.