Alexander P. Semenov*, Vladimir E. Patrikeev, Anatoly V. Samuylov, Yury A. Sharov

* Correspondence: e-mail: bcrubin@glasnet.ru; Telefone: 007-095-552-15-47; Fax 007-095-552-17-90

The large technical-scientific achievements made to date in the field of astronomy, space and laser technology have been feasible to a great extent due to forthcoming of the new high-precision optical ground and space-based systems. In this connection, the requirements to the quality of optical surface have enhanced; a range of their overall dimensions has increased; components with off-axial aspherical surfaces, and with an arbitrary shape of outer perimeter of the components and their holes have been used often. Alongside the traditional materials used in the optical manufacture (optical glass, Zerodur, Sitall, ULE etc.), non-traditional materials (silicone carbide, beryllium, and etc.) find ever-growing use.

To create ground and space-based telescope mirror systems JSC LZOS (Lytkarino Optical Glass Plant), Russia, when manufacturing mirrors, uses widely glass ceramic material Sitall C0-115M which is analogous to Zerodure (Schott, Germany) on optical and physical-mechanical properties. The multiyear experience of figuring mirrors made of Sitall have showed its reliability and applicability for fabricating large-size optical instruments with monolithic, lightweighted and thin optical components.

The classical methods of polishing can not ensure to the full extent the finishing of such complicated optical components with the required accuracy. To solve successfully the task of complicated optical system fabrication, it is necessary to improve the manufacturing methods of the system components developing computer-controlled methods of surface polishing, which allow to run controlled processes of surface finishing up to the high quality.

On the basis of the production-engineering system, consisting of thermostatic optical halls, vacuum test stands, test equipment, automatic computer-controlled machines, the great experience in the manufacturing of the new types of lightweighted mirrors (Fig. 1) and high-precision computer-controlled figuring of optical surfaces on lightweighted and thin substrates of an arbitrary shape of an outer perimeter and holes has been gathered at LZOS. The development of such systems speeds up essentially the high-precision optics manufacturing process and allows to predict assuredly the surface shape and a time of optical surface figuring.

The developed manufacturing process is founded on the programmed control of the optical surface polishing system (Fig.2) using computer-based processing of surface data, with calculation, redetermination and prediction of technological process conditions and controlled of movement of a small-size polishing tool.

The computer-controlled system for large-size optical components grinding and polishing includes:

• automatic machines of AD series, which are controlled by a computer, to polish optical workpieces of 100mm to 4000mm using small oscillating tools moving over a surface in the Cartesian system of coordinate;
• a set of interferometers to test a surface shape during all manufacturing stages, that includes an automatic photoelectric system for recording and processing surface interferograms;
• vibration-isolated test stands and technological supporting mounts of membrane-pneumatic type to support and stabilize the workpiece surface shape on the stages of its machining and testing, including automatic maintenance of adjustment parameters when the external conditions (atmospheric pressure, temperature) are changed;
• a system of technological software KCPM and AD2 used for processing wave-front interferograms of the workpiece to be tested on a real time basis, calculating technological parameters of computer-controlled polishing, automatic correction of the manufacturing process according to the results of a computerized treatment session, prodiction a surface shape to be reached.

When machining the workpiece, it is placed in a specially designed technological mount of membrane-pneumatic type or in a supporting fixture, which insures its stable position during the testing to provide a successful session of computer-controlled polishing. The mount is equipped with a system of automatic stabilization of a mirror position at varying external conditions (atmospheric pressure, humidity) when testing a surface, that ensures constant surface shape conditions with the required accuracy. To avoid air turbulence the testing of the workpiece in vacuum in a special test set-up UVC-6/70 (Fig. 3) is possible.

A computer-controlled polishing cycle begins with the testing of a working surface shape. Interferometers such as Fizeau, Twyman-Green and so on are used for the testing of a surface shape. In order to test aspherical surfaces we use interferometer IKAP-2 with a wavefront corrector, which converts a plane wavefront into an aspherical one.

To test a surface shape on a real time basis we use a computer-based interferogram processing system, which is a television system of interferogram image signal processing. It includes a unit of photoelectric recording fulfilled on a matrix photoelectric converter on CCD of 256x256 pixels. The time of accumulation is 3msec. More perfect photoelectric recording unit is made on CCD camera of 1024x1024 pixels in size with the time of accumulation of 0.4 msec. The small time of accumulation and the possibility to average a large amount of surface topographies allow to get the necessary information about the surface shape under workshop conditions with the required accuracy to calculate computer-controlled polishing process.

With this system we can conduct processing of interferograms of 3000 up to 7000 points to get more detailed surface topography needed to make satisfactory computer-controlled polishing.

To make an analysis of an optical surface shape according to a wave-front interferogram the normal practice is to use approximation of Zernike's polynomials, power polynomials, spline etc. But the experience of optical surface computer-controlled polishing has shown that when the workpiece is machined with a small tool of a size in the order of 1/5 to 1/40 of the workpiece diameter these methods of a surface shape description is not acceptable, since they do not describe local errors of the surface with the required accuracy, especially on the outer edge of the workpiece.

As a result of comparison of errors of different methods, considering our experience in of the computer-controlled polishing and increasing an accuracy of interference fringe coordinate determination due-to using a ÑÑD camera, to plot deviation topography, we use the method of local interpolation into the required knots of a grid, when plotting the topography of surface shape deviations from the nearest comparison surface.

When testing aspherical surfaces with a wavefront compensator an adjusting coma and distortion deviations, introduced into the interferogram image, are subtracted mathematically.

In order to test large plane mirrors the most acceptable test set-up is Ritchey-Common set-up. Surface shape reconstruction methods, when testing at the Common test set-up at two different angles of incidence are stated in papers. However they possess a disadvantage essential for compute-controlled polishing, that when approximating of a surface shape with power polynomials or Zernike's polynomials, it is impossible to receive the detailed description of the surface shape and its local errors comparable with the size of a tool used for polishing.

In the method we developed testing of large-size planes is conducted at two different angles of incidence of a beam to a surface normal. Generally the first angle is about 30^o and the second one about 55^o to 60^o. According to the results of a surface astigmatism topography determination at the two positions, a surface sphericity value is defined with the iteration method, and it complies with given topographies. At that the astigmatic surface topographies for this sphericity are calculated analytically for each mirror positions. After determination of sphericity, the topography of a real plane surface to be machined is plotted.

Checking of the method was made during testing small planes with a non-round outside shape at the Ritchey-Common set-up and at Fizeau interferometer. Differences in shape surface deviations made up in the order of 5%. The practical machining of a large amount of planes up to 1m in diameter have proved reliability of the designed test method.

At the heart of the polishing technique, the method of small-size tool controlling, when a tool is moved along a prescribed path over a surface in the three-dimensional Cartesian coordinate system, was set. Besides an influence of a number of factors producing an effect on the polishing process, when a toll operates on the workpiece, is minimized. A polishing velocity remains constant thanks to small-size tool oscillation with a fixed eccentricity. A pressure applied to the tool is constant. A controlling factor is a time of polisher’s stay within an elementary area, this is accomplished by means of slow movement of a tool center of rotation within the given area. At the simulation of computer-controlled polishing , the calculation of topography of optical surface normal deviations with the required accuracy has essential importance.

A principle of the calculation is based on an iterative process with gradual increasing of the designed removal depth and at the same time with controlling rms deviation changes of a surface from theoretical. This way allows to choose easily tooling to be used for polishing, to calculate an optimal removal from the workpiece surface and an optimal duration of a machining session.

The calculation of removal in each area of a surface is fulfilled for several sets of tools of different diameters by means of redistribution of removal so as to ensure a smooth shape of a surface to be obtained and to minimize an effect of fluctuations on material removal.

During the first machining sessions the optical surface, as a rule, has regular errors of such as zonal and astigmatic ones. A way of redetermination of a technological coefficient is based on the fact that during the polishing session some wide areas remain unpolished and serve as base surfaces, from which a value of an absolute removal is determined according to topographies before and after the polishing process. The necessary condition to make proper determination of an absolute removal is maintaining of a unpolished area shape when testing, that is stability of the surface shape on this stage of a technological cycle.

According to the results of the analysis of absolute removal distribution over a surface, a quality of the removal can be examined in time on the areas with different surface errors, and a mathematical removal model can be redeterminated. With unpolished areas of a surface we can evaluate errors and a level of stability of a supporting system of a technological mount.

Based on the results of material removal calculation, a tooling movement path and a control program for machines are calculated for each of subsessions.

By now using the computer-controlled method, plane, spherical and aspherical surfaces of 100 to 3200mm in diameter have been manufactured at LZOS. The achieved rms error from the required surface is equal to 0.01 to 0.03 (=632.8nm), a peak-to-valley errors is about 0.1 (=632.8nm). For a series of optical components the designed polishing method is an only possibility to achieve the required accuracy.

The final polishing of the workpiece surface up to 2m in diameter from p-t-v error in the order of 8 to 10 and rms error about 1 to 2 up to rms errors in the order 0.01 to 0.02 takes about 1.5 months.

The mathematical software we have developed for computer-controlled polishing allows to study features of substrate material behavior, deformation-thermal effects, arising from the polishing process and in the process of handling and installation at a test set-up.

Among the fabricated mirrors are lightweighted (weight reduction up to 80%) and thin optical mirrors both axial and off-axial with arbitrary shape of the outer perimeter. A series of light-weighted mirrors of diameter up to 1600mm with aspherical surfaces, a series of telescope optics for some European countries of diameter up to 700mm, a plane mirror of 1060mm in diameter for Carl Zeiss, and the secondary mirror for Kottamia Telescope in Egypt were manufactured at LZOS.

For example, we will consider the manufacturing of hyperbolic surfaces of lightweighted mirrors of 1540mm in diameter from Sitall CO-115M. A weight of the mirror after lightweighting is 377kg. A lightweiting factor is 1.8. A maximum deviation from the nearest sphere is 8.3m. The mirror was based on a technological mount installed on a polishing machine AD-2000. In order to avoid air turbulence a protective cover was used. The Polished surface has rms error about 0.015. An interferogram of the mirror and the point spread function is shown in Fig.4.

Our plant work in close contact with Carl Zeiss Jena (Germany) in the field of manufacturing of large-size optics.

During 1997-98 we manufactured for Carl Zeiss a hyperbolic mirror for MPI Heildelberg of 1250mm in diameter and two mirrors for Royal Greenwich Observatory (RGO) of 2050mm in diameter. On the RGO mirrors the enclosed energy of 80% in a diameter of 0.2 arc seconds was reached and that is close to the theoretical diffraction limit.

At the moment we are manufacturing for Carl Zeiss Jena a mirror of the 2280mm telescope for the National Observatory in Athens (NOA), Greece, and the optics of VST telescope (VLT Survey Telescope) with a 2650mm primary for Kapodimonte Astronomical Observatory, Naples, Italy.