Ultra precision machining on a CNC diamond turning machine

Transcription

1 Ultra precision machining on a CNC diamond turning machine J. Lamb Precision Engineering Laborator3;, A WE Aldennaston, United Kingdom Abstract This paper describes the features and operation of the Diamond Turning Facility at AWE Aldermaston. Examples are given of some of the sophisticated subsystems and constructional features of the machine along with the back-up equipment needed to run a facility of this type. The capability of the machine is discussed and Finally some typical work examples, along with results, are shown. demonstrating the precision and variety of work carried out on the machine. 1 Introduction AWE'S Diamond Turning Facility consists of the Ultra Precision Lathe or U.P.L., arguably one of the most accurate lathes of its type in the world, housed in an environmentally controlled room and supported by sophisticated setting, testing and inspection equipment, and of course the specialist personnel required to successfully run a facility of this type. The machine itself was delivered to Aldermaston from Cranfield Precision Engineering (the machine tool builders), approximately seven years ago after a lengthy procurement process. AWE engineers were heavily involved in the specification, design, build and test of the machine. Once delivered, the machine had to be assembled, the environmental room constructed around the machine, and then a demanding series of geometric and performance tests carried out to ensure that the machine met the design specifications. These tests themselves

2 210 Laser \fetrolocq~ atd,\laclilne Perfor-mrrnce were pushing the limits of what is achievable, and were far in excess of anything covered by the relevant IS0 standards. Some novel techniques were employed, which are described elsewhere[l]. One incident during the acceptance tests demonstrates the sensitivity of the machine to any form of disturbance. A 20 nanometre 'blip' was noticed in the output graph for one of the probes measuring the straightness of one of the axes. This 'blip' was repeated at regular intervals. After much investigation the cause was finally tracked down to a two chamber desiccant air drier. This was drying the instrument quality air before being fed to the laser system controlling axis movement. The drier was changing chamber every two minutes, causing a very slight pressure pulse in the system. This in turn was causing the axis control system to move the axis by the 20 nanometres to compensate, which produced the 'blip' on the graph. The simple solution adopted was to install an air reservoir in the line after the two chamber drier. This had the effect of dissipating the pressure pulse, and the 'blip' completely disappeared from the trace. This also shows the lengths that it is necessary to go to, in order to track down influences on a facility with such capability. 2 Features of facility There are many novel and sophisticated features of this facility which enable such a high class performance to be achieved. The following list picks out a representative sample. with brief descriptions of the important points of each feature. The polymer concrete bed consists of 8 tons of synthetic granite. This gives very good thermal stability, stiffness and vibration damping characteristics. Unusually in a diamond turning machine we have a hydraulic spindle. Again this gives good stiffness and increased load carrying capacity. The tool normal capability enables us to keep the cutting tool, under CNC control, normal to the profile of the surface being cut at all times. This has the advantage that if there are any imperfections in the tool nose radius profile, they do not get transferred to the job. The Laser Positioning system has a resolution of 1.25 nanometres and incorporates a wavelength tracker to compensate for any changes to the air in the room - (pressure, temperature, humidity, hydrocarbon content etc.).

3 The room itself is kept to 20 'C I 0.1 'C, and the temperature is independently monitored to "C in ten different locations. Atmospheric pressure and humidity is also carefully monitored. The CNC control is the CUPROC 3000 that utilises high resolution, fast feed forward techniques, thus reducing servo following error. These techniques also provide considerable real-time axis compensation as well as incorporating a 2 dimensional error compensation capability for straightness and orthogonality. 3 Operation There are some aspects of operating the facility that are very similar to a standard CNC lathe, and others that are totally different. The actual cutting tools tend to be different to enable the machine to achieve the form and surface finish accuracy. We tend to use tools with zero top rake, making setting of the tool more straightforward, and certainly for finishing cuts, tools with controlled radius forms, with form errors in the range 0.2 to 2 pm, are used. We primarily use some form of diamond for the tool material. Three different types are used; Natural, Synthetic monocrystal and PCD. Natural diamond produces the best quality tools. The quality of the natural stone determines the quality of the tool however, and imperfections can lead to chipping in use. The stone has to be ground in the correct orientation to produce the hardest wearing edge, but it is difficult to produce a really accurate form. A better form can be produced along a softer plane in the stone, but tool life will not be as good. Synthetic monocrystal diamonds overcome some of these problems in that there are no imperfections or variability which can occur in natural stones. These diamonds are grown under immense temperatures and pressure, but are not quite as hard as the natural material. Consequently tool life can be slightly reduced and we have found that component surface finish can also be slightly degraded. Poly Crystalline Diamond (PCD), is a completely different sort of material. It is made of diamond grit embedded in a metal matrix. Different sizes of grit and different matrix materials are used, but in general we have found tool wear mechanisms to be similar to those associated with the more common carbide tool materials. Surface finish and tool life are not as good as either of the two single crystal materials, but on some materials (for us, particularly Tantalum), PCD is the only form of diamond tool that can be used.

4 A CNC part programme still has to be written, in order to produce the required component shape, although the language is a form of basic, making programming less structured but giving more flexibility over machine control. A little more care and attention is also paid to tool setting. We use a microscope (usually at X) to measure the tool radius and to give us an indication of the nose radius form. Depending on the nature of the job, errors in the value of tool nose radius, may or may not have a large effect on the accuracy of the component form. On some jobs however, errors in tool radius form will be directly transferred to the component. Once the tool has been mounted on the machine, accurate LVDT's are used to establish the tool height and position in the X-axis (diameter). Z-axis position (length), is set by touching the tool on a location face. (See later). For the more accurate work we will cut a partial test sphere and examine this on the laser form interferometer. This will show the errors in the component and allow us to make the final setting adjustments to produce components down to the 200 nanometre region. We have the option of operating the machine remotely. Again for the most accurate work, the presence of the operator adjacent to the machine will generate heat which will affect the component and machine structure. We have a remote control console in another room together with viewing screens showing the image from two cameras. One of the cameras mentioned above is a miniature system, which can be positioned very close to the cutting tool itself. This enables us to datum on a machined surface by inching the tool onto the surface in small increments and watching for the first generation of swarf. It is easily possible to pick up the surface to better than 100 nanometres in this fashion. One other task which has to be undertaken before any really precision work can be carried out is to balance the spindle. An accelerometer is used to measure the runout and tell us where to add weight to improve the balance. Once this has been done we allow the spindle to coast through its speed range to pick the quietest spindle speed. Diamond is very forgiving of cutting speed variations enabling us to pick a quiet speed for the spindle, almost irrespective of the component material and size. Also as a rule we tend to use constant spindle speed rather than surface speed, as we do not want to introduce disturbance (speed increase) to the spindle during cutting. Once all of the above has been carefully carried out, in the majority of cases, the biggest influence on component accuracy is the fixturing. Providing secure location and holding of the component with absolute minimal distortion can be

5 P P P P P P Transactions on Engineering Sciences vol 34, 2001 WIT Press, ISSN Luser- \/err-otog? arzd Zltrchrne Per-for-mmce 21 3 very difficult, especially as the majority of our components have thin sections and are often out of materials with low rigidity 4 Capabilities SPEClFICATZON OF U.P.L * STRAIGITJYESS 500 NM FULL STROKE * ORTHOGONALITY 0.5 ARC SECONDS * SPINDLE ERRORS + 50 NM * 100 NM FORM ACCURACY * < 5NM Ra SURFACE ROUGHNESS AS GOOD AS ANY LATHE IN THE WORLD! When Diamond Turning, there are two parameters of primary importance, namely Form Accuracy, and Surface Finish. The potential for the machine in terms of form accuracy is 100 nanometres. We have come very close to achieving this on real components. The machine specification said that we ought to be able to achieve 5 nanometres Ra Surface Roughness. We have in fact achieved better than 1 nanometre Ra on a flat copper disc. The first part of the above chart shows three typical values obtained from the geometric acceptance tests carried out when the machine was delivered. Typically we will take cuts from 100 nanometres to 100 pm, and we can produce step heights which can be identified down to 10 nanometres. 5 Support equipment In order to make sure that we have produced what we were supposed to, some sophisticated inspection equipment is required. Remembering that in general it is form and surface roughness that are the features of primary importance, we have specialised equipment for examining these features.

6 214 Lmer \letrolog md Ilnchlne Perjormnnce Our laser form interferometer is, as the name suggests, for examination of component form. A large diameter laser beam is reflected off the component, back into the instrument. Here it interferes w~th the reference beam producing interference fringes. These fringes are viewed by a camera, digitised, analysed by computer and the results displayed in the format requested by the user. There are however some severe limitations. Firstly you need a reflective surface on the component, and as already mentioned, not all materials produce a reflective surface even when diamond turned. Secondly you have to create a laser beam wavefront the same shape as the component being examined. It is very difficult and extremely expensive to produce accurate aspheric lenses, so this effectively limts this instrument to examination of flats and spherics. Having said that the instrument is still extremely useful. It enables us to find all sorts of errors on components. Tool setting errors, clamping points, fixturing problems etc., all become obvious. We can also tell a lot about the health of the machine. For examining component surface roughness we have a filtered white light surface profiler. As with the form interferometer, piezo electrics are used to drive the lens through a fixed distance, taking a number of views as it does so. This machine is however examining a much smaller area of the component, (typically 0.25 mm square), and can give a height resolution of better than 0.1 pm. By using a two wavelength option, step heights of up to 10 pm can be accommodated and rougher surfaces examined. Again you can tell a great deal about your cutting process by careful examination of the surface produced. 6 Example l - laser target One of our major customers is AWE'S high powered experimental laser group. The components they require are typically very small with dimensions measured in micrometres, and well defined precision features measured in nanometres, usually in materials which lend themselves well to diamond turning. The following sketch represents a typical target: When we first saw the specification for this target, we thought that it would not be possible to machine an acceptable quality sine wave of these dimensions. We were concerned that the noise of the machine axis movements would ruin the form. However, after much trial and error, not least on fixturing, we machined a set of targets. The results were better than we dared hope. The sme wave was exactly the correct pitch at 50pnh and the peak to valley dimension was 195 nanometres rather than 200. This type of component could not be economically produced by

7 any other method, and it gives our laser group a capability it did not have previously. It is an example of the amazing detail and precision that can be obtained from the machine TYPICAL WORK EXAMPLE * LASER TARGETS 200 NM P + V SINE WAVE 300 FM DIAM 7 Example 2 - precision shell At the other end of the size range, we manufacture thin walled, essentially hemispherical components out of difficult materials, in this case Tantalum. These are typically 100 to 200 mm diameter and about 1 mm wall thickness. They form part of some Hydro-dynamic experiments, the results of which require much extrapolation, so that small dimensional errors in the components can lead to large uncertainties in the experimental results. The major problems with this type of work are concerned with the nature of the material and the configuration of the component. This in turn leads to fixturing and tooling problems, and as a consequence errors in the final component tend to be larger than one would expect from other materials and stiffer components.

8 INNER AND OUTER PROFILE ERRORS -601 POLE Angle [Degrees] EQUATOR THICKNESS RUNOUT g 20 E I5 P I0 W 5 +PREVIOUS RUN-OUT POLE Angle [Degrees] EQUATOR The graphs above show the plots of the inner and outer profiles and the thickness runout, (the variation in wall thickness around halos of the component), with the UPL results compared with conventional lathes. In both graphs the Y axis represents error in Micromenes, and the UPL plots are the ones nearest to the X axis, demonstrating close to an order of magnitude improvement in each case. The source of some of the errors, particularly the profile errors, has been located and the next set of shells should be an improvement on these results. Although this may not be a typical application of a diamond turning machine, being able to manufacture to these levels of precision has enabled our customer to dramatically

9 increase the effectiveness of their process. These results could not be obtained on conventional machines. 8 Summary This has been a very brief top level description of what is a technically very sophisticated facility performing near unique work. We are getting world class performance from the facility thanks to the dedication and professionalism of the engineering team running and supporting the facility. I should stress the importance of the people. This is not a technology that one can buy, plug-in, press a button and produce good results. Each job presents unique problems which have to be overcome and the solutions implemented. The people involved not only have to have the right technical background, they also need the right temperament and a 'feel' for the process and technology. There is still an element of 'black magic' in getting the facility to perform to its limits. Hopefully this paper has given an appreciation of the facility and its capabilities, together with an idea of some of the work currently being undertaken. Subject to availability, we are keen to expand the variety of work on both the precision machining and inspection sides, and believe there are applications in a variety of industries. The author is keen to establish links with anybody in this area of technology whether or not they may have work for the facility. References [l] Gull MS, Final acceptance tests for the Nanocentre 250. Lamdamap Conference Proceedings Copyright 0 British Crown Copyright 2001 /MOD Reproduced with the permission of the controller of Her Britannic Majesty's Stationery Office.