Excellence in Motion Control Systems

High performance, low cost, easily integrated MOTION CONTROLLERS
NEE Controls, reducing job cycle times
 
Search for
 
reducing job cycle times

reducing job cycle times

Reverse engraved and filled example
Reducing job cycle times can be achieved by the best motion controllers

Advances in computer performance, together with the availability of powerful computer-aided design (CAD) software packages, enable very complex multidimensional parts to be accurately and rapidly designed.

Chris Williams reports.

In most motion control applications the traditional approach is over conservative in two ways - the designer specifies the complete motion profile of all axes to follow for any given part, although much of the motion profile is non-critical and the designer creates a 'tracking controller' that closely follows or tracks the reference (desired motion profile).

This over conservatism cost significantly by increasing production cycle times, reduced part quality and in some cases, machine reliability. The fact that non-critical parts of a motion profile are specified can mean that substantial control effort is expended on achieving a useless end. Here we focus primarily on the feature of vector blending that permits job cycle times to be reduced by between 10 and 60 percent, while improving motion smoothness, machined part quality and increased machine reliability at the expense of part accuracy.

The mechanical accuracy and repeatability of each contouring axis effects the accuracy of the component, how well the control loop is configured to track the reference position, tool wear and cutting forces when controlling cutters in routing applications and kerf specification error when cutting with plasma.

Machine controller designers at the forefront of technology have been working on these issues for some time. Machine controllers are now available that can produce accurate parts with exceptional finishes, in shorter times, and at the same time, increase the reliability of the machine producing the parts.

Modern CAD packages produce large data files to define complex multidimensional shapes. The data is comprised of line, arcs, spline and perhaps bezier curve elements. The length or machine movement associated with each element is generally very short and on multidimensional shapes the direction of each consecutive element varies considerably.

From the machine perspective and from the plant managers perspective the best results are achieved by operating the machine smoothly, and rapidly, provided that accuracy can be maintained. So the problem from the machine controller designers perspective is to be able to run the machine smoothly, rapidly and accurately from complex and large multidimensional data files.

Early machine controllers ran each data element separately. When each element was completed the machine stopped while the controller calculated the next move, the machine then executed the move. The next generation of controller calculated the next move while executing the current move, minimising the time between moves. Many currently available machine controllers calculate or 'look ahead' a number of elements or moves, (typically 2 to 10), of the currently executing block and attempt to 'join' moves of differing lengths and directions together where possible.

The ability of the controller to, a. join moves together, and b. look at sufficient distance ahead to operate the machine at the required speed, mostly determines the performance of the machine.

Obtaining optimum machine performance at the junction of two movements is a complex operation. What is required at the junction of two moves, where the directions of the move differ, are instantaneous speed changes on the machine moving axes. Unsophisticated controllers let the servo amplifiers take care of the motor machine between moves for larger direction changes.

In order to achieve this the 'motion planner' must look sufficiently far ahead through the upcoming data to run the machine at the required speed.

'Blending' between moves to reduce job cycle times

Best machine operation is achieved by 'blending' between moves of differing directions by, in effect, joining them with an arc. To achieve this the controller 'motion planner' must generate a velocity profile for each axis, keeping it within its acceleration, deceleration and velocity limits. To simplify this for processing the different elements, the input data is converted into a common format, generally short vectors.

Modern motion controllers must be capable of smoothly accelerating or decelerating at the intersection of two vectors when the direction changes. Therefore tight control over accelerations needs be exercised at all intersects to limit against the possibility of any axes exceeding the pre-defined acceleration limits. Due to there being no control of deceleration at the intersect of each vector, the deceleration will be completed during one sample, therefore yielding a declaration rate that exceeds the predefined limit for the given axis.

Obviously motion controllers do not allow this to happen, as they will ensure in the case of a simple square, one axis will complete its deceleration and stop before the next axis starts to accelerate.

However by taking this over conservative approach of limiting axis velocities based on the knowledge that there is a large change in direction, this has a severe negative impact on the part production time. A solution needs to be found that will keep each axis within its physical specification (acceleration and velocity limits) while providing a smooth velocity profile without instantaneous changes in velocity that will increase the productivity of your machine.

If positional accuracy can be relaxed to within a predefined tolerance then this provides the motion controller with a measurable physical distance to adjust the velocity profile to provide optimal performance.

The instantaneous change in velocity at the intersect determines the time and distance required to accelerate or decelerate from the end velocity of the current vector to the start velocity of the next. To illustrate how the part production job cycle times can be radically reduced, if we take a practical job, say a 16 sided polygon that is to be nested 2000 times on a sheet of metal and is to be cut using plasma. If we compare the resultant torch speed of one polygon executed without vector blending against another we can see the immediate improvement.

As the direction changes are large at the vector intersects the axes are having to decelerate to a stop. However when using vector blending the resultant torch speed never fully decelerates to a stop and in this example never falls below 130ipm. Due to the blend at each intersect the acceleration rate is never exceeded and therefore there will be no undue stresses put onto the axis due to high rates of acceleration or deceleration.

The time taken to execute the non-blended polygon is 4.92 seconds compared against the blended polygon taking 3.01 seconds, which equates to a 39 percent reduction of the production cycle. Multiplying up by the number of parts required gives us a saving of over 11/2 hours for a run of 2000 parts.

The complexity and amount of CAD data required to define complex multidimensional shapes places huge demands on manufacturing machine controllers. Unless the controller has the computing power and suitable control algorithms, part quality, machine productivity and machine reliability are all impaired.

To conclude there are many sources of errors that will effect the accuracy of the final component. Therefore in majority of applications it is acceptable to deviate from the programmed path providing the deviation does not exceed the allowable tolerance of the component being machined. If your application permits motion profiles that do not exactly follow the programmed path then vector blending can significantly improve the output of your machine.

More Information

Home |  Applications |  Contact Us |  Top of the Page

NEE Controls. Offices in UK and US. Distributors around the globe