2
N II 1 Diamond turning won't do everything! Given the recent impressive gains in diamond turning pre- cision optics it is fair to ask why we have virtually given up trying to use diamond turning methods to make mirror segments for the proposed US National 15m Telescope. The scale of this giant, ground-besed optical instrument is such that its primary mirror must be sub-divided into smaller segments; one of the leading design concepts we are now considering would use up to 100 precisely-made 1-2m mirror elements. When nested together in a mosaic and put under appropriate position servo control, these mirrors would function as a single, nearly-parabolic, optical sur- face. Manufacturing these elements would seem to be a natural application for the budding capabilities of diamond turning. Regrettably, this is not the case, even though the repetitive precision of diamond turning would be an enor- mous advantage in making identical mirror segments. The first problem is that all but the central element will be off-axis, aspheric mirrors, individually not rotat- ionally symmetric, and the outer segments will be more than 7m off-axis. This suggests the need to generate the optical surface with the workpiece offset on a huge spindle (impractical); or the ability to work on axis while rapidly and precisely controlling the cutting tool motion in and out of the workpiece surface at least once/revolution; or the ability to pre-bend the workpiece such that, after machining, presumably to some manageable on-axis con- figuration, and the workpiece is released, it will spring into the desired off-axis shape. We are currently attempt- ing to use the latter technique to polish a 2m diameter low-expansion glass mirror using a conventional optical polishing machine and have appropriately dubbed it the 'bend-and-polish' method. However, the technology for using any of these approaches on a diamond turning mach- ine is presently well-hidden from my view, if it exists at all. Secondly, the size of the workpiece (1-2m) mandates excellent workpiece support on the turning spindle if the required 25-50nm (1-2/~in) surface accuracy is not to be lost. Telescope designers have grappled for many years with the flexible nature of large mirrors and although our methods work well and might adapt to diamond turning, we have not worried much about centrifugal forces or distortions due to machining stresses. Today 'optical quality' mirrors up to 0.5m can be diamond turned, but a new philosophy for workpiece support must be developed to do as well at 2m. Mirror surface flexure problems worsen more or less as the third or fourth power of mirror radius and it seems unlikely that present holding methods can be improved 64- fold or more by any simple modifications. (My apologies to the structural engineers for the cavalier interpretation of plate bending theory implicit in this statement.) The point is that workpieces above 0.5m deserve as much careful analysis as the diamond turning machine itself and I see no widespread recognition of this problem. A third problem is how to measure the optical quality of the finished surface. A 15m telescope mirror segment will have a basic surface radius (ie radius of the 'best-fit' spherical approximation to the actual surface) in the range 50-60m. The centre of curvature is the point where optical mirror testing is most often performed. In this case, the optical centre of curvature could also be 7m off-axis. The dimensions of the testing set-up would thus be formidable. Conceivably, one could perform interferometric tests or a Hartmann test with the workpiece in situ on the turning machine, but I suspect that air turbulence will force testing to be done with the optical axis vertical (to minimize thermal layering effects). Facilities and technical expertise of the type required for this form of testing are uncommon: a lesson we are learning painfully from our 'bend-and-polish' experiment. Finally, as if achieving the final product were not sufficiently difficult, the long-term dimensional stability of diamond-turned metal mirrors will have to be conclu- sively demonstrated to sceptical astronomers. Cast, nickel coated, aluminium mirrors were in vogue for telescopes in the 1955-65 period (for economy), but have lost favour since due to optical surface changes brought on by thermal expansion differentials and stress relaxation in the substrate. Since telescopes are almost always built one-at-a-time from non-renewable funds and without a back-up, astronomers will be understandably reluctant to opt for metal mirrors again without positive proof that these problems have been solved. There is little evidence to show that the marvellous accuracies achieved from diamond turning are matched by long-term stability of the workpiece. It is not easy to demonstrate. The stability of the test method comes into question and some methods of storing data (eg holograms) have dubious long-term value. Mirrors may, in fact, be the appropriate test vehicle because they can be re-tested with confidence, given an appropriate method. If an invariant wavelength of light is used for testing, then any changes can be blamed on the mirror substrate. However, this work will require, at least, several years to complete even after the proper approach is selected. As a mechanical engineer, I was fascinated several years ago with the prospect of using diamond turning to make telescope mirrors, but my early enthusiasm has been replaced by the gloomy realisation that it may require ten years or more for solutions to emerge for the problems cited above. We would rather not wait that long before deciding how to build our telescope. Kitt Peak National Observatory, Tucson, Arizona, USA PRECISION ENGINEERING 187

Diamond turning won't do everything

  • Upload
    ld-barr

  • View
    213

  • Download
    0

Embed Size (px)

Citation preview

N II 1 Diamond turning won't do everything!

Given the recent impressive gains in diamond turning pre- cision optics it is fair to ask why we have virtually given up trying to use diamond turning methods to make mirror segments for the proposed US National 15m Telescope. The scale of this giant, ground-besed optical instrument is such that its primary mirror must be sub-divided into smaller segments; one of the leading design concepts we are now considering would use up to 100 precisely-made 1-2m mirror elements. When nested together in a mosaic and put under appropriate position servo control, these mirrors would function as a single, nearly-parabolic, optical sur- face. Manufacturing these elements would seem to be a natural application for the budding capabilities of diamond turning. Regrettably, this is not the case, even though the repetitive precision of diamond turning would be an enor- mous advantage in making identical mirror segments.

The first problem is that all but the central element will be off-axis, aspheric mirrors, individually not rotat- ionally symmetric, and the outer segments wil l be more than 7m off-axis. This suggests the need to generate the optical surface with the workpiece offset on a huge spindle (impractical); or the abil ity to work on axis while rapidly and precisely controlling the cutting tool motion in and out of the workpiece surface at least once/revolution; or the abil i ty to pre-bend the workpiece such that, after machining, presumably to some manageable on-axis con- figuration, and the workpiece is released, it will spring into the desired off-axis shape. We are currently attempt- ing to use the latter technique to polish a 2m diameter low-expansion glass mirror using a conventional optical polishing machine and have appropriately dubbed it the 'bend-and-polish' method. However, the technology for using any of these approaches on a diamond turning mach- ine is presently well-hidden from my view, if it exists at all.

Secondly, the size of the workpiece (1-2m) mandates excellent workpiece support on the turning spindle if the required 25-50nm (1-2/~in) surface accuracy is not to be lost. Telescope designers have grappled for many years with the flexible nature of large mirrors and although our methods work well and might adapt to diamond turning, we have not worried much about centrifugal forces or distortions due to machining stresses. Today 'optical quality' mirrors up to 0.5m can be diamond turned, but a new philosophy for workpiece support must be developed to do as well at 2m. Mirror surface flexure problems worsen more or less as the third or fourth power of mirror radius and it seems unlikely that present holding methods can be improved 64- fold or more by any simple modifications. (My apologies to the structural engineers for the cavalier interpretation of plate bending theory implicit in this statement.) The point is that workpieces above 0.5m deserve as much careful analysis as the diamond turning machine itself and I see no widespread recognition of this problem.

A third problem is how to measure the optical quality of the finished surface. A 15m telescope mirror segment wil l have a basic surface radius (ie radius of the 'best-fit' spherical approximation to the actual surface) in the range 50-60m. The centre of curvature is the point where optical mirror testing is most often performed. In this case, the optical centre of curvature could also be 7m off-axis. The dimensions of the testing set-up would thus be formidable. Conceivably, one could perform interferometric tests or a Hartmann test with the workpiece in situ on the turning machine, but I suspect that air turbulence wil l force testing to be done with the optical axis vertical (to minimize thermal layering effects). Facilities and technical expertise of the type required for this form of testing are uncommon: a lesson we are learning painfully from our 'bend-and-polish' experiment.

Finally, as i f achieving the final product were not sufficiently diff icult, the long-term dimensional stability of diamond-turned metal mirrors wil l have to be conclu- sively demonstrated to sceptical astronomers. Cast, nickel coated, aluminium mirrors were in vogue for telescopes in the 1955-65 period (for economy), but have lost favour since due to optical surface changes brought on by thermal expansion differentials and stress relaxation in the substrate. Since telescopes are almost always built one-at-a-time from non-renewable funds and without a back-up, astronomers will be understandably reluctant to opt for metal mirrors again without positive proof that these problems have been solved. There is little evidence to show that the marvellous accuracies achieved from diamond turning are matched by long-term stability of the workpiece. It is not easy to demonstrate. The stability of the test method comes into question and some methods of storing data (eg holograms) have dubious long-term value. Mirrors may, in fact, be the appropriate test vehicle because they can be re-tested with confidence, given an appropriate method. If an invariant wavelength of light is used for testing, then any changes can be blamed on the mirror substrate. However, this work will require, at least, several years to complete even after the proper approach is selected.

As a mechanical engineer, I was fascinated several years ago with the prospect of using diamond turning to make telescope mirrors, but my early enthusiasm has been replaced by the gloomy realisation that it may require ten years or more for solutions to emerge for the problems cited above. We would rather not wait that long before deciding how to build our telescope.

Kitt Peak National Observatory, Tucson, Arizona, USA

PRECISION E N G I N E E R I N G 187

Notes for authors PRECISION ENGINEERING covers the full spectrum of advanced technologies and takes an integrated approach to all subjects related to the development, design, manu- facture and application of precision machines and com- ponents. This quarterly journal is specifically aimed at research and development workers, designers and engineers in industry, institutions, and the academic world.

Contributions on all aspects of the subject will be welcomed. All types of contribution should emphasise the results of the work and their application. Specifically, it is a condition of acceptance for original research papers that they note the practical implications of the work in the introduction. The following categories may guide authors in planning their papers: • Articles (surveys, reviews etc), usually planned in

consultation with the Editor • Original papers on development research work • Technical papers on new techniques and devices • Brief research reports describing current projects

and giving the trend of results • Conference reports, book reviews etc planned in

consultation with the Editor • Letters containing comments on any aspect of

precision engineering and technology, discussion of previously published papers, or advanced informal reports of research work Authors wishing to make contributions should con-

tact the Editor. Only papers not previously published will normally be accepted: if any tables or illustrations have been published elsewhere, the Editor must be informed so that the necessary permission to reproduce can be obtained.

Copyr igh t Before publication authors are requested to assign Copy- right to IPC. This allows IPC to sanction reprints and photocopies and to authorise the reprint of cemplete issues of volumes according to demand. Authors' tradi- tional rights will not be jeopardised by assigning Copyright in this manner as they will retain the right to re-use and a veto over third party publication.

The appropriate Copyright Transfer Form will be sent to authors when their paper has been approved for publication.

Refereeing All major contributions are refereed before publication, and may require modification in the light of the referees' comments. On acceptance, contributions are subject to editorial amendment.

Style Authors are urged to write concisely. Spelling should follow that recommended in the Concise Oxford Dictionary. Scripts should be in English. Contributions in other languages can be translated, but there will inevitably be a delay and the author may be asked to pay the cost of translation.

Derivations of equations should be relegated to an appendix. Reference to other publications should be made, rather than arguments repeated in full. Articles should be arranged in the following order:

Full title:

Names of authors: Abstract of the paper, including the purpose, scope, methods and conclusions of the work: Text, suitably sub-divided under headings: Acknowledgements (if any): References: Ta bl es: Captions to illustrations

Scripts Two copies should be submitted, typed on one side of the paper only in double spacing and with wide margins.

Unless a typewriter for typing mathematics is available, mathematics should be written as clearly as possible by hand with Greek letters identified, the first time used, by their name in words. SI units should always be used; the fps equivalent may be given in brackets after- wards. Ensure that symbols and abbreviations used for units follow the latest editions of British Standard 1991, parts 1-6. In particular, note the following points:

Always put a zero before decimal quantities less than unity, ie 0.376 not .376. Use full stops as decimal points, not as multiplication signs. Use fractional indices instead of root signs, eg (A + 2b) 1/2 is preferred to ~/(A + 2b)

References References should be indicated throughout the text by superior Arabic numerals. The references themselves should be listed in numerical order at the end of the script. References should include the authors, tit le of article, publication, year of publication, volume number, issue number and page numbers, eg: Tallian T.E. On Competing Failure Modes in Roller Contact. ASLE Trans, 1967, 10(4) 418-423.

Standard abbreviations for the title of the publication should be used, transliterated where necessary into the Latin alphabet.

I l lus t ra t ions Authors need not prepare finished versions of their draw- ings as these will normally have to be redrawn to suit the printing process. Pencil sketches are good enough, provided that they are clear. Photographs are encouraged but authors should note that it is impossible to reproduce a photo- graph from the printed version in another journal. Full plate (250mm x 200mm) black and white glossy prints as rich in contrast as possible should be sent. Authors are urged to pack scripts carefully to avoid damage to illustrations.

Proofs Galley proofs are sent to authors for checking, but authors should confine themselves to correcting typesetter's mis- takes. Alterations made to the text at the proof stage are expensive and may necessitate the resetting of whole para- graphs for the insertion or deletion of a single word. We reserve the right to charge authors for changes at proof stage to the text originally submitted.

Repr ints Authors are sent 25 run-on offprints of their article free of charge. Additional reprints (minimum quantity usually 100) can be ordered from the Reprint Dept, IPC Science and Technology Press Ltd.

Submission of papers Send your paper to: The Editor, PRECISION ENGINEER- ING, IPC Science and Technology Press Ltd, PO Box 63, Westbury House, Bury Street, Guildford, Surrey GU2 5BH, UK. Tel: Guildford (0483) 31261 Telex: 859556 SCITEC G

188 PRECISION E N G I N E E R I N G