The trend is clear: More and more metalworking presses are going with servocontrol. “At trade shows, you increasingly see only hydraulic servos or presses running off cranks,” says Delta Computer Systems Inc. President Peter Nachtwey. “But customers tell us they use hydraulic servos because they make a variety of parts. Crank presses are limited.”
Delta Computer, in Vancouver, Wash., says makers of small presses (capacities below about 25 tons) often do the controls work themselves. But bigger or morec omp l i c a t e d unit s demand specialists. For example, “Powder- metal presses can be difficult because of their two concentric cylinders,” explains Nachtwey. “A lot of press makers are better at bending metal than devising controls, and they tend to customize machines extensively. Each is unique and that often gets them into trouble.”
Complicating the control-design process is a lack of information about hydraulic components. “You often don’t get useful specs from hydraulic-component makers,” says Nachtwey. “Big OEMs just can’t get valve manufacturers to give them transfer functions or equations that describe how components work as a function of input, so they’ll work on it until they know more about the valve than the manufacturer does.”
Another complicating factor is that servos for hydraulic presses aren’t built in the same way as ordinary positioning systems. Typical positioning servos have a velocity- feedback loop running within a position-feedback loop. But hydraulic servos usually don’t employ velocity feedback. “You only need a velocity loop if you are controlling torque,” says Nachtwey. “It is also easy to end up with overlapping gains between the two loops. This can, in turn, produce more poles in the transfer function, depending on how that inner loop is implemented.”
Most press systems use different loops depending on where the cylinder is in the cycle. A position loop typically manages the cylinder until it nears the end of the stroke. At that point, a second loop takes over using pressure as the feedback parameter.
During the part of the stroke controlled by the position loop, the goal is to give the moving press the right amount of kinetic energy that will generate the right amount of work for the forming action. The goal is somewhat different in the case of a die-casting press, however. There, designers generally are unconcerned with balancing the kinetic energy involved. They usually opt for generating as much force as possible during the stroke, then control gets handed off to the pressure loop.
“If you move too slowly under the control of the position loop, pressure builds up as you are forming the part. Move too fast and pressure will spike too high because you can’t remove the excess kinetic energy through the servovalve,” says Nachtwey.
The change from position to pressure/force control can take place by various means. One is to use some criteria for when the changeover should take place. The triggering event can be a pressure/force setpoint, a specific rate-of-change, or a combination of factors.
The transition from control via the position loop to the force loop can be tricky because of instability in the process. So some controls run position and pressure loops simultaneously. This creates smooth transitions between position and pressure/force modes because the controller is always comparing the control signal from the position and pressure/force. It uses the lowest of the two to control the output. This lets the actuator go to position or force mode, depending on which setpoint is reached first. The scheme works because the control output goes to zero when either the position or the pressure/force reach their respective setpoints.
An example of this approach comes from machine builder EUROelectronics srl in Italy. The firm devised a control system for a high-speed die casting press that has a hydraulic cylinder moving at up to 7.5 m/sec. Controls monitor cylinder position and compare it against operator- entered values.
The high speeds and pressures involved dictate that the position loop controlling the cylinder close at a minimum 1-kHz rate. Similarly, the system closes a loop around a pressure sensor to manage the cylinder as it nears the end of the stroke. The pressure sensor loop closes every few milliseconds.
In a 1-msec scanning cycle, the controller measures valve position, calculates cylinder speed, compares both to the setpoint, then corrects cylinder movement using a PID (proportional, integral, and derivative) algorithm. Controls watch pressure values in the front and back of the cylinder to avoid instantaneous pressure peaks and keep the hydraulic circuit balanced.
The hydraulic servovalve has a distinctive nonlinear behavior, so its PID tuning is via a linearization table of valve-response values. The resulting algorithm is a way to schedule PID gain so the system gets accurate responses at low velocities (from 0.05 to 0.30 m/sec when the cylinder starts moving) and at the 7.5 m/sec maximum velocity. Designers also applied feedforward and smoothing to the PID command signal to ensure the cylinder would smoothly transition into its rapid-movement mode at the right point.
The position-loop controller is a CompactRIO device from National Instruments Corp. An FPGA on the controller synthesizes an interface for the cylinder position encoders, basically linear magnetic stripe sensors. The FPGA accepts signals from these sensors via two high-speed digital inputs with no intermediate amplification or processing. All programming was done in NI LabView and used the real-time version of LabView.
Pressure control starts when the piston reaches the end of its movement and the die cavity is full of metal. At this point, a conventional PLC takes control of the machine and drives a separate servovalve different from the one controlling piston movement. The CompactRIO controller communicates with the PLC through an Ethernet port.
A supervising application developed using NI LabWindows/CVI also runs on the CompactRIO. This operator interface lets supervisory software define the cylinder’s injection profile in two ways, either by inputting numerical values, or by drawing profiles interactively using a graphical procedure. The operator also can set various machinecycle parameters, including position, velocity, pressure, and time.
A PCI-6025E data-acquisition board feeds diagnostic signals to the supervisory software such as position, pressure, and temperature profiles for each injection. Machine-monitoring software plots machine operation and calculates various control values such as mean and peak velocities, times, pressures, and temperatures.
EUROelectronics engineers say National Instruments software and hardware made it possible to go from prototyping to the final machine setup in only three weeks.