NC-controller strategy for the novel machine tool 'X-Cut'.
Puschitz, Falko ; Bleicher, Friedrich ; Mikats, Thomas 等
1. INTRODUCTION
The Laboratory for Production Engineering, part of the Institute for Production Engineering and High Power Laser Technologies at the Vienna University of Technology, has a long tradition in developing and building new machine tools. A main focus is set on machine tools with a parallel kinematic configuration. The first parallel kinematic machine tools were built with a degree of parallelism of one, e.g. Quickstep and Quickstep Neon (Bleicher, 2001). This implies that there are as many drives in the machine as there are degrees of freedom for the tool and, respectively, the tool centre point (TCP). The newly developed machine tool "X-Cut" has a degree of parallelism of two, incorporating four drives for the X-Y plane movement where only two drives are necessary. This circumstance results in the necessity of a new method in NC-Controller programming and parameterisation, which shall be drafted within this paper.
2. PARALLEL KINEMATICS
To begin with, a short overview of the definition of parallel kinematics is provided. Parallel kinematics refers to a set of closed kinematic chains which all act directly on the end effector. In this way, all kinematic chains work in parallel, and from this the name "parallel kinematics" is derived. By design and adjustment of the kinematic system, the machine can be matched to a specific function. It is possible to arrange the chains in such a way that they can only support unidirectional tensile or compressive forces. In this way the machine works like a framework, raising its ability for increased payload and stiffness. To achieve this, the kinematic chains are usually a joint-beam-joint combination called a strut. The joints of these struts are normally passive elements so that a drive is needed to create movement. Generally there are two different types: length-variable struts, in which the drive changes the length of the strut and requires the drive to be implemented within the strut itself (Bleicher, 2003), and fixed-length struts which are actuated by drives moving the whole strut. The variety of different kinematic concepts is completed by hybrid structures, where serial and parallel arrangements of mechanical elements are merged into one mechanism (Hennes, 2000). There are also differences in the exact specification of mechanical structures depending on the number of axes, degrees of freedom, different connecting elements and special layouts of the joint positions. In (Hebsacker, 1998; Neugebauer, 2000; Pritschow, 1997; Pritschow, 2000) different systems are compared and their advantages and disadvantages are identified.
3. PROJECT AIMS
A new approach in machine tool structure design should be found with the highly parallel system called "X-Cut". This structure exhibits two degrees of freedom, allowing movement in the X-Y-plane. This kind of movement could already be performed by a scissor-like kinematic structure (e.g. DynaM). In order to preload all kinematic transmission elements the scissor can be mirrored, changing the "V" into an "X," from which the name X-Cut is derived. In this way, a highly parallel mechanism is created. This principle allows that the negative structural characteristics of one kinematic adjustment near the singularity position in working space are compensated by the corresponding kinematic adjustment of the opposite chains. This leads to nearly constant mechanical characteristics over the whole working space (Fig.1).
[FIGURE 1 OMITTED]
Furthermore, due to the over-determined kinematic structure, it is possible to influence preloading of all kinematic transmission elements during processing, and therefore basic stiffness of the entire structure can be influenced in a broad manner. Preloading also effects structure damping characteristics whereby oscillations caused by acceleration and deceleration (e.g. during positioning of the main spindle) which are introduced into the structure fade away much faster. The third movement direction in Z, which is necessary for handling a great number of milling operations desired in industry, is realized by moving the work piece itself. Main goals during development of the "X-Cut" machine tool are the following:
* Unique machine structure capable of preloading of the kinematic chains for increased damping and stiffness characteristics
* Development of a controller software for real-time kinematic and inverse kinematic transformation
* Positioning accuracy of about 1um in length in all dimensions
* Controller algorithms for optimal preloading of the chains with respect to main spindle position
* Acceleration at the TCP of at least 2,5g
* Velocities of up to 150m/min
The resulting machine tool system is shown in Fig.2.
[FIGURE 2 OMITTED]
4. CONTROLLER STRATEGIES
The most sophisticated part in controller strategy is to handle the motion control loops of the four drives responsible for X-Y-plane movement. To find a solution the controller model had to be analyzed (Fig.3).
[FIGURE 3 OMITTED]
Based on this analysis the following approach was specified: all four drives of the X-Y-plane should have a full NC positioning control loop with slight differences between the two master drives on the left and the two slave drives on the right side of the machine tool (Fig.2). The master drives should operate with a conventional positioning controller. An additional feed forward torque provided for limiting lag-error due to the weight of the kinematic structure was implemented as well. The slave drives on the right side had to be handled quite delicately. The positioning control loop for these drives has a very small proportional gain and therefore reacts quite slowly to positioning deviations. The feed forward momentum is designed to react much quicker to any error between its programmed value and its actual measured value; therefore it describes the dominant parameter for these drives. The nominal value is provided by the application program and stored in a table. The values have been calculated based on FEM modelling of the CAD-Model.
Furthermore a kinematic transformation, direct as well as inverse, also had to be implemented in the NC-controller. The transformation is necessary to calculate the nominal positioning values for the drive in relation to the position of the TCP. The kinematic problem is quite simple since all motion occurs exclusively in the X-Y-plane. The solution can even be found in a fully analytically explicit manner using sine and cosine functions.
The next step is much more difficult: implementing the algorithm within the NC-Control. Traditionally NC-Control-Vendors implement algorithms for their customers, of course for a price and without teaching how the algorithms work. This was not acceptable for this project. The Institute wanted to be involved with all aspects of the project and additionally wanted the flexibility to alter the algorithm whenever necessary without relying on NC-control vendors. Fortunately, a close cooperation with Bernecker & Rainer (B&R) was established. B&R was developing an all new NC-Controller for machine tools, and IFT had the chance to be involved from a very early stage. This new NC-Controller enables implementation of all necessities, like transformation, calibration etc., with conventional high-level-language programming. For that reason all knowledge regarding the operation of the X-Cut could be kept within the laboratory, allowing for further improvement without any dependencies on NC-Controller vendors.
5. CONCLUSION
The IFT developed a totally new machine structure with an over-determined, highly parallel kinematic structure, allowing for increased payload and stiffness of the kinematic chains. This machine tool made it necessary to develop new methods in NC-Controller programming and parameterization. Thanks to close cooperation with B&R, an adequate controller could be found allowing implementation of these methods in a unique form, providing the opportunity for new development and further improvement. Future work will be to implement calibration strategies and methods for increased machine accuracy.
6. REFERENCES
Bleicher, F. (2001). Entwicklung einer parallelkinematischen Bohr--und Fraseinheit fur den flexiblen Einsatz in automatisierten Anlagen, TU Wien, Habil.-Schrift, Vienna
Bleicher, F. (2003). Parellelkinematische Werkzeugmaschinen, Neuer Wissenschaftlicher Verlag GmbH Nfg KG, Auflage: 1., ISBN-10: 3708301188, Vienna
Hebsacker, M. (1998). Die Auslegung der Kinematik des Hexaglide--Methodik fur die Auslegung paralleler Werkzeugmaschinen, In: VDI-Berichte 1427, Neue Maschinenkonzepte mit parallelen Strukturen fur Handhabung und Produktion, VDI, pp 51-66, VDI-Verlag Dusseldorf, ISBN: 3-18-091427-0, Dusseldorf
Hennes, H. (2000). Ecospeed--Ein hybridkinematisches Maschinenkonzept zur 5-Achsen Hochleistungszerspanung groBer Strukturbauteile im Flugzeugbau, In: Beitrag zum Fertigungstechnischen Kolloquium, Gesellschaft fur Fertigungstechnik, pp 342-360, Springer-Verlag Berlin Heidelberg New York, ISBN: 3-540-67984-7, Stuttgart
Neugebauer, R.; Stoll, A.;Kirchner, J.; Ihlenfeldt, S. (2000). Gestaltung, Bewertung und Einsatzerfahrungen von Parallelkinematiken, In: Beitrag zum Fertigungstechnischen Kolloquium, Gesellschaft fur Fertigungstechnik, p.316-333, Springer-Verlag Berlin Heidelberg New York, ISBN: 3540-67984-7, Stuttgart
Pritschow, G., Wurst, K.H. (1997). Linapod--ein Baukastensystem fur Stabkinematikmaschinen, wt Werkstattstechnik 87, pp 437-440, ISSN: 0340-4544
Pritschow, G. (2000). Neue Wege fur Parallelkinematikkonzepte, In: Beitrag zum Fertigungstechnischen Kolloquium, Gesellschaft fur Fertigungstechnik, pp 96-114, Springer Verlag Berlin Heidelberg New York, ISBN: 3-540-67984-7, Stuttgart