Air bearings are beautifully simple in concept, yet their suitability for high-value applications has resulted in them becoming extremely highly developed.
A typical air bearing (or aerostatic bearing) consists of a plain shaft running inside a sleeve, separated by a gap around 5 to 50microns wide. Clean dry air is fed under pressure through small orifices in the sleeve, from where it flows along the gap and out of the ends of the bearing. When the assembly experiences a radial load, the gap on the loaded side of the bearing increases, which causes the pressure to fall. On the other side, the gap decreases and the air pressure rises. The resultant pressure differential across the assembly therefore supports the applied load without any metal-to-metal contact between the shaft and sleeve, whether or not the shaft is rotating.
Compared with rolling element bearings, lubricated plain bearings or hydrodynamic bearings, air bearings have low friction, allow high speeds to be achieved with minimal thermal effects (water cooling can virtually eliminate thermal effects), have high stiffness, are extremely accurate (error motions of less than 0.05micron total indicated runout are possible), cause virtually no vibration, and can be used in clean room environments. They also need no maintenance, though the air preparation equipment must be maintained in order to prevent contamination of the bearing.
Most applications for air bearings are in metrology and ultra-high-precision machining, such as in the microelectronic and optical industries. In particular, air bearing spindles in their many forms are evolving continually to meet the unrelenting demands for improved performance, where low motion errors, high static and dynamic stiffness and low thermal growth are important. To maximise the benefits, in many cases customised designs are created. This process involves tailoring spindle layouts to suit individual requirements, typically by integrating auxiliary components and, hence, optimising the overall system. In addition, improvements are continually being made to bearing design and the manufacturing process.
Realising these improvements is both an incremental and an ongoing dynamic process, with new developments occurring all the time. Recently these have focussed on four key areas: bearings; spindle stiffness and damping; motion errors; and thermal distortion. All four areas are yielding gains that substantially enhance spindle performance.
Air bearing developments
Although a number of different bearing geometries are used in high-precision machining, in air bearing spindles it is mostly cylindrical journal and plain thrust bearings that are used, as these can be manufactured to the high standards of precision and finish that are essential for low motion errors. In addition, these types of bearings also exhibit low thermal distortion, meaning that they are suitable for both high-speed and low-speed operation. This facility for high speeds must be put into context, however, as the speeds in the vast majority of high-precision machining applications are modest and the emphasis is usually to achieve maximum static stiffness for the bearing across the required speed range.
The stiffness of an air bearing is inversely proportional to its internal clearance, so maximum stiffness is achieved by minimising clearance. However, there are limits to how far the clearance can be reduced; one is thermal distortion the other is the geometrical accuracy to which the bearings can be manufactured.
In allowing for thermal distortion, bearing clearances cannot be reduced to the extent that opposing surfaces come into contact under operational conditions and, in practice, the bearing is designed to take maximum load at a safe working gap. The larger the surface geometry errors, the larger this gap and the bearing design clearance. Happily, over the last two decades, a combination of improved methods of manufacture, better understanding of thermal distortion and improved internal design with regard to air flows has succeeded in reducing the bearing design clearances by approximately 40percent, resulting in much improved stiffness that benefits the overall performance of the spindle.
Spindle stiffness and damping
The role of bearings in the overall radial stiffness of an air bearing spindle, while important, is not the only major factor. Stiffness also depends on the rigidity of other components such as the shaft and workholding arrangement. Recently, computer models for determining the elastic deformation of a bearing/shaft system have enabled better optimisation of the spindle layout, thereby minimising the cantilevering effect of an overhung work position.
Such models have also proved to be vital for improving the dynamic behaviour of spindles. The data they provide enables critical speeds in the operating speed range of the spindle to be avoided and dynamic stiffness at the work position to be optimised by judicial choice of system parameters.
An example showing the radial dynamic response of a lens turning spindle measured at the work position is featured in Fig.1. The spindle operates at speeds of up to 10000rpm and the magnitude of the response at 610Hz defines the dynamic stiffness of the spindle, which in this case is 5.3N/micron and compares to a static stiffness of 17.2N/micron. The significance of these figures is that the dynamic stiffness of an air spindle is predominantly a function of bearing damping and the typical response (Fig.1) shows this to be high compared to the level of damping found in most mechanical structures.
In addition to their role in providing spindle stiffness and damping, bearings also have an impact on motion error. In air bearing spindles this is often characterised by eccentricity, higher-order synchronous motion and asynchronous motion errors. In general these are primarily a function of such factors as geometrical errors within the bearings, magnetic imbalances in the spindle drive motor and mechanical imbalance.
Many ultra-precision machining applications now require peak-to-peak motion errors better than 0.05microns, so it is critical to address all of these sources. Basic eccentricities in motion error are normally addressed by paying close attention to bearing alignments and balancing, while reducing synchronous and asynchronous motion requires a combination of attention to air flows and bearing geometry.
One of the major factors in ensuring low asynchronous motion error is an air supply that is free from pressure pulses and contamination. Minimising air turbulence within the bearing and random pressure fluctuations in the exhaust air also help in this respect. In contrast, synchronous motion errors arise if the bearing geometry departs from perfect round, (or flat, in the case of a thrust bearing). The air film in an air bearing averages the effect of bearing geometry errors to a low level and ratios of motion error/geometry error as low as 0.05 have been observed in practice. However, sub-micron precision of bearing surfaces is still required to meet most ultra-precision motion specifications. Despite these problems, improvements in manufacturing methods have had an important effect on motion error and enabled errors to be reduced from about 0.2microns (typically) on commercial machining spindles 20years ago to the present level of 0.05microns peak-to-peak.
Whereas motion errors in air bearings contain both synchronous and asynchronous components, spindle drive motors are usually dominated by the former. Over the last 10years much development work has been undertaken to reduce magnetic imbalance on both framed and frameless motors. In addition, motor design and manufacturing methods have improved significantly over the period. The net result of this work is that both types of motors are now available with out-of-balance magnetic forces of less than 1N.
In contrast to the relative complexity of the development work to reduce motion errors, the initiatives to limit heat build-up and the resulting thermal growth are comparatively straightforward. In standard high-precision machining spindles, thermal growth is limited by locating the thrust bearing closer to the work position and placing the main heat source – the drive motor – towards the spindle rear. However, in more critical applications, forced cooling is required to control thermal growth. The type and degree of cooling is normally matched to the application requirements. Chilled water-cooling affords the best results, allowing changes in spindle dimensions to be kept as low as 1 micron.
The effectiveness of the recent developments in air bearing spindle technology is evident from the new designs that have emerged as a result of the work. These include high-precision spindles for turning contact lenses (Fig.2), cup grinding and diamond turning. All are particularly notable for the levels of precision they provide. Nevertheless, even among spindles that achieve surface finishes better than 0.1micronRa, the cup grinding spindle stands out.
One such spindle is fitted to Cranfield University’s TetraformC, a grinding machine that is required to achieve optical surface finishes when grinding both brittle materials and hard metallic surfaces at relatively high material removal rates. In terms of spindle performance, these requirements demand high static and dynamic stiffness and low motion errors over a wide speed range up to 6000rpm. When grinding materials such as glass, silicon and M50 tool steel, the machine is achieving surface finishes consistently better than 10nmRa and as low as 1nmRa.
Grooved hybrid bearings
Driven by the demands of the machine tool industry and the requirements for micro-machining in the electronics and optical industries, ultra-high speed aerostatic bearings are now achieving speeds in excess of 3x106DN, which is equivalent to shaft surface speeds of more than 150m/s. However, with bearing clearances in the range of 5-20microns, air films are subject to very high shear rates at these speeds; the resulting viscous forces can have a detrimental affect on both the thermal and dynamic characteristics of the bearing, but Loadpoint has addressed these problems with a new design of grooved hybrid aerostatic bearing. The benefit of this design is an improvement in high-speed performance that may be exploited in terms of increased limiting speed, higher stiffness or reduced running temperature.
Loadpoint's new design seeks to overcome the two major factors that have previously prevented higher speeds being achieved with aerostatic bearings: high friction losses due to shearing of the air within the bearing gap; and half-speed whirl, a destructive instability that adversely affects the stiffness of the bearings and can be so severe as to cause seizure.
Combining aerostatic and aerodynamic principles, Loadpoint’s grooved hybrid bearings – journal type air bearings with helical grooves, and thrust bearings with spiral grooves – are designed to pump air into the centre of the bearing when they rotate (Fig.3). At high speeds, the pumping action of the grooves significantly changes the pressure distribution within the bearing, generally improving load carrying capacity and stiffness. The grooves also change the velocity gradients within the bearing's air film, which, in turn, reduces the bearing’s viscous friction loss and improves its whirl response.
Whirl response for the grooved hybrid bearings is generally more complex than for a conventional aerostatic bearing; it is highly dependent on groove design, where, in general, the higher the groove volume, the more the response departs from that of the conventional bearing.
The response takes a similar form to that shown in Fig.4, with maximum loss of aerodynamic lift occurring at a particular speed. However, unlike conventional air bearings, this speed does not always occur at half shaft rotational speed, and the maximum loss is not always the total aerodynamic component, which would otherwise result in total loss of load capacity.
As viscous friction loss in aerostatic bearings is proportional to area and inversely proportional to gap, the grooved hybrid bearings also produce lower levels of friction than conventional air bearings operating with the same minimum clearance. The benefits become more pronounced at high speeds, as the ratio of groove clearance to ridge clearance in the hybrid bearing increases.
A typical comparison between grooved hybrid and conventional aerostatic bearing performance illustrates this point (Fig.5). Each curve shows the trade-off between bearing stiffness, taking into account whirl response and viscous friction loss, as shaft speed increases from 0 to 160000rpm. As speed is increased, the bearing’s minimum clearance reduces from 28 to 16microns due to shaft growth. At low and moderate speeds the results are similar to the conventional air bearing, exhibiting a marginally better trade-off between stiffness and power loss. However, at speeds above 120000rpm, the grooved bearing shows clear benefits and has the potential to reach significantly higher speeds.
Comparing the effects of the grooved hybrid design across annular thrust bearings and journal bearings shows that the benefits are more pronounced in the former types. Taking the example of a machining spindle arrangement, the thrust bearings do not experience whirl, so the trade-off is usually stiffness against power loss. This is an important consideration, as thrust bearings are invariably larger in diameter than journal bearings, so surface speeds are higher.
Importantly, grooved hybrid thrust bearings can be designed to give the same static stiffness as conventional aerostatic thrust bearings; however, because of the grooving, they exhibit significantly lower power consumption throughout the speed range. Furthermore, their stiffness increases with speed due to aerodynamic effects, whereas conventional aerostatic thrust bearings do not generate any aerodynamic lift at all.
An early application of Loadpoint’s grooved hybrid technology is the rotor of a bore-grinding spindle. This uses four short aerostatic journal bearings to achieve high stiffness and to support a high-powered overhung motor. In this application grooved hybrid bearings have been compared with conventional aerostatic bearings and found to increase whirl speed by approximately 25percent; power consumption, measured in terms of drive current, is reduced by 10percent.
Clearly further developments of air bearings will lead to ever-higher performance, but it is difficult to predict where the technological limitations will be met. What is certain, however, is that developments with air bearings are not yet at their peak, and that ongoing improvements in some of the key areas mentioned above are consistently redefining the boundaries of what can be achieved and broadening the scope for their use.
The author wishes to thank Dr Frank Wardle, technical director of Loadpoint Bearings, for his assistance with this article. "