Study On Techniques In Computer-controlled Grinding And Polishing For Large And Medium Aspheric Surfaces

Aspheric optics are being used more and more widely in modern optical systems, due to their ability of correcting aberrations, enhancing the image quality, enlarging the field of view and extending the range of effect, while reducing the weight and volume of the system. With the ever-increasing demands on optical system performances, requirements for aspheric optical components are more and more critical, which involve aperture, relative aperture, accuracy, lightweight extent, manufacturing efficiency and cost. Computer controlled optical surfacing (CCOS) technique is used widely in machining process of the large and medium aspheric surfaces, because of its high accuracy, simple process conditions, low cost and other merits. However, there are some problems at present in CCOS technique, such as low convergence rate, small scale manufacturing errors and edge effect. These problems have influenced machining accuracy and efficiency of optical components seriously. This thesis is dedicated to solve the key problems in CCOS technique, in order to consummate CCOS technique and improve capability for manufacturing large and medium aspheric surfaces. The major research efforts include the following points.1. Aspheric optical compound machining tool (AOCMT) is introduced, which has milling, grinding and polishing, and contact measuring functions. A dual rotors grinding and polishing tool is designed, based on the timing belt drive. The required material removal function can be obtained by adjusting rotational speeds of the dual rotors, eccentricity of the two rotating centers, working pressure of the cylinder and other parameters.2. The material removal function model of the dual rotors grinding and polishing tool is built. Systematic experiments are done to study the grinding and polishing parameters. The influences of parameters on material removal rate, surface quality and subsurface damage (SSD) depth are gained. The methods for selecting process parameters in each fabrication stage are given. The SSD depth values in grinding stage are gained experimentally, which provides basis for determining the material removal quantity in the subsequent process. With optimized parameters, the SSD depth of K9 optical glass in grinding stage is controlled within about 2.2μm. Consequently the polishing time is considerably decreased and the machining efficiency is improved.3. The figuring ability of the removal functions of the dual rotors grinding and polishing tool is analyzed. The optimized parameters are gained, with the eccentricity ratio being 0.8 and the rotate speed ratio -3. The transfer relation of removal function size, spatial error wavelength, and extra material removal quantity to error convergence ratio is analyzed. Based on CCOS convolution model, the generating rules of residual errors due to convolution effect are gained by computer simulations. The optimizing control method in the shape error convergence process is brought forward.4. Four reasons for generation of small scale manufacturing errors are recognized. Based on maximum entropy principle, a new method for expressing the polishing effect and optimizing processing parameters is presented. The typical method for controlling small scale manufacturing errors is to polish the whole surface uniformly with a large tool. Aiming at this method, the principle for selecting main processing parameters including kinematic velocity and size of polishing tool is put forward. Moreover, in order to increase working efficiency, a new method for controlling small scale manufacturing errors is brought forward, which suggests correcting errors in definite areas.5. Finally, a CCOS control strategy is proposed to control full aperture errors and full band of frequency errors of the large and medium aspheric surfaces and to increase working efficiency. As an application, a 500-mm diameter, f/3, parabolic mirror was successfully fabricated on AOCMT within 233 hours. The finished mirror has a shape accuracy of 9.4nm rms (λ/67 rms,λ=632.8nm), surface roughness of about 1.5nm, and curvature error of 1.2mm (0.4‰). The magnitude of the surface errors from 100mm to 2mm scale is 3.6nm rms. The results meet expected requirements.