Most people familiar with large-scale industrial laser processing have seen a high-throughput laser CNC machine cutting large steel plates and tubes at dizzying speeds. Those of us in laser micromachining, where part quality depends on micron-level machining accuracy, wondered if we could achieve such high machine throughput and still produce highly accurate parts. The answer is yes - and then the question becomes "how?" This article explores the basic considerations in machine design and control that one must be familiar with to achieve maximum throughput from a precision laser microprocessor.
In the manufacturing process, the criteria for determining acceptable parts are often non-negotiable. Part tolerances are defined by the requirements for normal or safe operation of the part. They define the allowable error budget for the manufacturing process. The error budget is then "depleted" by different sources of error arising from machine design, controller functionality, and laser material interactions during machining. The key to maximizing throughput when manufacturing high precision parts is to leave as much error budget as possible for dynamic tracking errors. Following sound system and structural design principles and selecting a powerful motion controller - one that takes maximum advantage of the dynamic tracking error budget - will maximize throughput and therefore the economic rationale for laser micromachining systems.
The structural design of the manufacturing system is fundamental to improving the ability of the manufacturing system to operate at high throughput. In order for the control system to reject and minimize errors, the sensors used to "see" the motion within the system must be able to observe the relative motion between the tool and the part. In most systems, these sensors do not directly observe the motion of the tool tip, i.e., the laser spot; instead, they derive their information from an optical readhead that views an encoder scale (effectively a ruler) embedded in the motion system mechanism. Therefore, in order to save as much error budget as possible for the dynamic tracking budget in the controller, the designer must minimize unobservable errors due to bending or vibration within the frame. The key to minimizing unobservable error is to maximize the stiffness of the structure. One way to achieve maximum stiffness is to minimize the length of the machine's structural loops. A structural loop is a path of forces generated by the motion of a machine that matches or is equal or opposite to the forces generated by the corresponding structural elements. Imagine that the materials that make up the structural elements of a machine are formed by thousands of tiny springs connected in series. Adding more springs to a tandem chain actually reduces the chain's stiffness. Therefore, designers should shorten the structural "chain" of spring elements to stiffen the machine. In addition, adding spring elements in parallel makes the chain more rigid. To maximize stiffness, designers should add redundant structural elements to the frame of the machine to support inertial forces. The stiffer the machine, the more energy is injected into the structure without causing unwanted motion. This allows the user to push motion control elements faster, with more acceleration and energy, while minimizing unobservable processing errors. Figure 1 below depicts the series and parallel connection of the machine's structural loops and spring elements.
Figure 1 shows. Adding springs in series makes the spring chain less stiff, while adding springs in parallel makes the spring chain stiffer. This principle can be used to maximize the stiffness of a machine's structural circuit.
A stiffer machine that allows more energy to be injected without bending, saving more of the error budget for elsewhere, is an immediate improvement. This paves the way for the next area of focus in improving throughput: machine dynamics principles. As the stiffness of motion platforms and racks increases, so does their intrinsic frequency. As their intrinsic frequency increases, so does their controllability and production speed.
Each motion trajectory - the path required for a laser spot to create a part - has spectral content for each axis involved in generating the motion. Each axis command has a certain sinusoidal frequency band that needs to be represented in a mathematical series or summation to represent it. Figure 2 below shows an example of a step function and its sinusoidal approximation using a finite bandwidth.
Figure 2. approximation of a step function using a sine wave in terms of levels and sums. The more sine wave frequencies or bandwidths used in the approximation, the closer the approximation is to the step function. The step function requires an infinite number of steps of sinusoids to represent it perfectly, but the smooth function can be represented by a finite number of steps or bandwidths.
In this example of a step function, an infinite bandwidth is needed to perfectly approximate the step, which makes it impossible to implement in a real machine. This is one of the main reasons why motion programmers try to avoid discontinuities in the commands sent to the machine. The principle demonstrated in Figure 2 applies to every command signal. When the motion profile is multidimensional and involves multiple axes of motion, the rate at which the machine traverses that profile changes the bandwidth of the commands sent to each relevant axis. A simple example of this relationship is using two axes to create a circle. In basic trigonometry, two axes travel through a circle, experiencing a sinusoidal wave in position, velocity, and acceleration. The frequency of the sine wave that each axis is asked to perform is proportional to the speed at which the circle passes. The faster the machine is required to travel a circle, the higher the frequency of the sine wave for each axis involved must be able to perform in position, velocity and acceleration. For any axis of motion to execute the command profile provided, the bandwidth of that profile must be within the bandwidth of the motion system. That's right, every motion system has a bandwidth.
The control system relies on feedback signals, servo control loops, and powerful motors to react to commands and match the actual results to the desired results. The responsiveness of the control system depends on how quickly the controller can make decisions and effect changes when the actual motion does not exactly match the commanded motion. This "control system responsiveness" is almost entirely dependent on the specifications and design of the control product used. Specifications such as the rate of trajectory generation, the rate of current closure (the rate at which the current generated by a given motor drive can be changed), and the peak force generated by the device motor will determine the control system's response rate. Therefore, it is a somewhat obvious conclusion that choosing a powerful control product and a powerful motor will benefit the designer. However, the response rate of the control system is only one part of the overall motion system's ability to respond to commands, i.e., the motion system bandwidth. The combination of the physical stiffness of the motion platform and the bandwidth of the control system determines the dynamic capability of the entire system. Given the same control system and motor, the higher the intrinsic frequency of the mechanical system, i.e., the stiffer it is, the greater the frequency bandwidth at which the system can successfully respond.
In general, the most important signal in motion control is the acceleration command. Acceleration is the primary signal of interest to the machine operator because it is most closely related to what the machine controller is actually controlling, the current to the motors. The current fed to each axis motor is proportional to the force generated by each motor. The force generated by each motor is proportional to the acceleration experienced by that degree of freedom as the machine moves. Tracking error, or error injected into the production process due to the motion system's inability to perfectly follow the commanded trajectory, is proportional to the portion of the commanded acceleration bandwidth that exceeds the motion system's bandwidth. A car based on suspension, engine, and driver can only cross a race track at a certain speed; if it is forced to turn at a speed that exceeds its limits, it will run off the road. This is the same for laser processing machines. By understanding the bandwidth of the acceleration commands sent to the machine in the motion profile, as well as the bandwidth of the machine's responsiveness or dynamics, we have a solid foundation for ensuring high quality parts are produced at maximum throughput. Some advanced motion controllers actually offer features that allow the programmer to automatically take into account the bandwidth of the motion system and self-limit the acceleration commands sent to the machine components to prevent too many errors from occurring.
Combining these concepts creates a meaningful message for the machine designer. The more rigid the frame structure, the less machine bending and vibration will affect machining results, leaving more error budget for dynamic tracking errors. The more rigid the mechanical design of the motion system, the higher the bandwidth of the motion system. The higher the performance of the control products used, the higher the bandwidth of the motion system. The higher the bandwidth of the motion system, the greater the bandwidth of acceleration commands it can respond to without creating the same level of part error. The higher the bandwidth of acceleration commands allowed without creating a bad part, the faster the machine can be commanded to traverse the desired contour during part production. Therefore, machine designers should consider every possible way to maximize machine stiffness and control system bandwidth to maximize process throughput without compromising part quality.
