High capacity aerodynamic air bearing (HCAB) for laser scanning applications

Sean M. Coleman Lincoln Laser Co., 234 E. Mohave, Phoenix, AZ 85004

ABSTRACT

A high capacity aerodynamic air bearing (HCAB) has been developed for the laser scanning market. The need for increasing accuracies in the prepress and print plate-making market is causing a shift from ball bearing to air bearing scanners. Aerostatic air bearings are a good option to meet this demand for better performance however, these bearings tend to be expensive and require an additional air supply, filtering and drying system. Commercially available aerodynamic bearings have been typically limited to small mirrors, on the order of 3.5” diameter and less than 0.5” thick. A large optical facet, hence a larger mirror, is required to generate the high number of pixels needed for this type of application. The larger optic necessitated the development of a high capacity ‘self-generating’ or aerodynamic air bearing that would meet the needs of the optical scanning market. Its capacity is rated up to 6.0” diameter and 1.0” thick optics. The performance of an aerodynamic air bearing is better than a ball bearing and similar to an aerostatic air bearing. It retains the low costs while eliminating the need for ancillary equipment required by an aerostatic bearing. Keywords: Scanner, spindle, polygon, airbearing, air bearing, aerodynamic, scanning, prepress, laser.

1. INTRODUCTION

Lincoln Laser Company has been producing air bearings for the scanning market since 1990. Until recently, the largest mirror used on these scanners was limited to 3” diameter by 0.4” thick. But now, with the inception of HCAB, it is possible to aerodynamically spin mirrors up to 6” diameter by 1” thick and up to 25,000rpm (limited by the windage or air drag of the optical configuration chosen).

2. BALL BEARING VS. AERODYNAMIC AIR BEARING

There are many problems associated with the use of precision ball bearings at high speeds and large payloads. These are related to issues such as lubrication, vibration, particle generation and shipment methods. Lubricants are the life blood of ball bearings. Much consideration is given to make sure that the right type and amount of lubricant is used in the ball bearing for a particular application. Lubricant sling, migration, and outgassing are critical problems associated with ball bearings. At high speeds, the lubricants have a tendency to ‘sling’ out of the bearing. Lubricants also tend to seep or ‘migrate’ along surfaces as well as evaporate, commonly referred to as outgassing. All three phenomena cause smudging of the polygon and optics close to the MPA. It is not possible to use seals to reduce the slinging for scanning applications because varying torque loads increase the stability error of the motor polygon assembly (MPA). Non-contacting shields are used to retain as much lubricant as possible without changing performance of the unit. A small gap is maintained between the shield and inner ring, which allows small amounts of lubricant to escape. However, as the high speed system continues to sling and migrate, the bearing eventually runs out of lubricant and fails. While outgassing rates are not considerably high, they produce hazing on the optics that is detrimental to performance of the optical scanning system.

Lubricant sling, migration, and outgassing are problematic for clean room environments because they increase the particle count. While outgassing can be controlled by using vacuum lubricants, there is no way of preventing sling and migration. Another particular problem of ball bearing scanners is related to the aerodynamic effect of a spinning flat plate optic or polygon. The air wants to adhere to the spinning surface and is thrown out of the unit by its own momentum. This creates a low pressure area in the housing, which can exacerbate the sling problem by trying to flow air though the ball bearing and dispersing lubricant into the environment like a fan. The bearing lubricant fill amount needs to be controlled in order to lessen the problems mentioned earlier. There must be enough lubricant to keep the ball retainer wetted as well as free lube in the races to prevent metal-to-metal contact. However, care should be taken to ensure that there is no excess lubricant, which would accelerate sling and migration problems. The correct fill is usually obtained empirically. The major advantage of HCAB is that it does not need lubricants other than air. All materials used in the HCAB are either non-outgassing or low outgassing, therefore no materials get expelled that can harm optics or other surfaces in the system. Ball bearings support a rotor by rolling element contact with the bearing races. Errors in the geometry of balls and raceways are transmitted to the unit housing and rotor as vibration, noise, and displacement. These errors result in ‘ball pass frequencies’. When a ball runs over a disparity, it creates a displacement. This is usually cyclical because the same disparity is encountered again and again; hence, the term ‘ball pass frequency’. There are four resulting frequencies - fundamental train, outer race, inner race, and ball defect. These frequencies are approximately calculable and proportional to the rotational speed of the rotor. These frequencies can excite optical mounts if the resonant frequencies are similar. The HCAB aerodynamic bearing ‘floats’ on air and transmits only an unbalance vibration, once per revolution frequency. This unbalance frequency is common among both the ball bearing and aerodynamic bearing. Precision ball bearings are delicate and require additional protection during shipping and handling. These bearings use a different raceway curvature than their standard grade counterparts. Standard grade bearings are meant to carry more load. The raceway curvature matches the ball curvature closely. The larger contact area results in higher load capability, but increased start and running torques. The increased torques are not acceptable for precision applications that require ultra smooth rotation. Precision ball bearings have less contact area, to reduce torque, leading to high susceptibility to impact damage. Heavy optics coupled with shipping shock loads can easily brinell precision bearings. Therefore, softer and thicker than normal shipping foam is required to reduce possible shock amplitudes to an acceptable level. The aerodynamic bearing is impervious to impact damage within the range associated with shipping. The large conforming surfaces transmit shock without causing stress risers such as a ball on its raceway. Balance disturbance, although rare, is the most common failure due to mishandling of aerodynamic units during shipping.

3. DESIGN PARAMETERS

There were two designs investigated for the HCAB – the first one did not include an encoder, while the second included an encoder. The design parameters for the first version of HCAB were: aerodynamic, capable of spinning a 6” diameter by 1” thick mirror, bearing stability up to 25krpm, bi-directional, retrofitable for pre-existing ball bearing units, impervious to airborne contamination, capable of horizontal and vertical operation, and have a life of at least 20k start-stop cycles. The second version was designed with same parameters, but in addition, it included encoder capabilities.

Figure 1: Motor polygon assembly (MPA) cut-away view.

4. DESIGN

There are several commonly used aerodynamic bearing configurations: cylindrical, biconic, cylindrical with thrust, and spherical with thrust. For cost concerns and ease of manufacturing, a cylindrical bearing (fig 1) was chosen. The cylindrical air bearing design does not inherently provide axial support due to lack of a thrust bearing, so a magnetic axial support system is employed. The polygon rotor is levitated by the interaction of a stationary, high energy, permanent magnet situated in the housing and a ferrous coupling band on the rotor. The geometries are chosen to maximize the axial stiffness and minimize any radial pull. Radial pull will side-load the bearing and cause premature failure due to sleeve contact. The cylindrical bearing design is ideal for polygon mirror applications since axial mirror position is not critical. For stabilization of the bearing we chose the patented Lincoln Laser Speed Bump design. The speed bump consists of an eccentric grind on the center surface of the bearing support shaft (fig 2). This creates a ‘wedge effect’ that is used to preload, or slightly offset, the spin axis from the center of the support shaft. Without the speed bump, the bearing will become unstable as the rotational speed increases.

Figure 2: Stationary shaft showing speed bump.

If we look at the air gap of the bearing as a flat surface, it can be modeled as parallel plates (fig 3).

Figure 3: Sheared fluid between parallel plates.

If the clearance between the plates remains constant, there is no pressure generation to support the bearing. The shearing of the fluid is generated solely by the velocity gradient and no other forces are at work to change the profile from one end to the other1. Hence, no internal pressures are created. The clearance needs to be varied to create a pressure wedge action.

Figure 4: Sheared fluid between wedged plates.

By ‘squeezing’ the air through a wedged area, a film pressure is produced (fig 4). It is this pressure that is used to create bearing stability. In a plain cylindrical bearing with no stabilization attribute, a wedge is inherently created as the rotor becomes eccentric to the support shaft (fig 5). The pressure profile that is created forces the rotor back to the center. However, because of the momentum built up in the rotor, it over travels center. This will lead to dynamic instabilities as rotational speeds increase. This situation is much akin to a ball bearing system that has no preload. A stabilization system is needed for high stiffness and stability.

Figure 5: Cylindrical journal bearing offset to create a wedge.

As the bearing becomes unstable, it tends to result in a ‘coning’ action of the rotor. Coning is described as the deviation of the rotor’s principal axis rotating about the true rotational axis, as defined by the support shaft (fig 6). A coning rotor will look similar to a dynamically unbalanced rotor, but with a much more severe and lower frequency compared to the rotational speed. The instability usually increases toward a whirl speed of half the rotational speed of the rotor or ‘half speed whirl’. Coning is detectable by optical systems even at low ratios of whirl to rotational frequencies. By the time the rotor encounters half speed whirl, the bearing is already in the process of experiencing contact failure. A contact failure of the bearing will occur at the bottom and top edge of the rotor due to the tilt. The failure occurs very quickly after the onset of coning because of the forces and surface velocities involved.

Figure 6: Coning or whirl. The speed bump stabilization system works by creating a physical wedge geometry, which in turn creates inherent wedges on opposing side of the shaft due to the offset. The geometry of the speed bump area creates a wedge as air is sheared over the land (fig 7). The speed bump land creates a high pressure area causing the axis of rotation to shift in the radial direction (fig 8). The upper and lower sections of the cylindrical bearing surfaces react like another wedge is occurring. In essence, the top and bottom surfaces are preloaded against the land creating a very stable bearing.

Figure 7: Pressure profile caused by

speed bump land. Figure 8: Rotor displacement caused by speed bump

and resulting pressure profiles To be a possible replacement for ball bearing scanners, the HCAB must be capable of horizontal and vertical spin axis orientations. The Lincoln Laser Company speed bump design lends itself very well to both orientations and tilt attitudes between. The speed bump works well in the vertical position by preloading the bearing in the radial direction. The same preloading is used to “float” the rotor in the horizontal position. The wedge effect created by the speed bump is so effective that in some payload configurations, lower lift-off numbers are seen in the horizontal position. This was unexpected because the weight of the rotor was directly on the bearing surfaces instead of the magnetic axial support system. One caveat of the horizontal (or off vertical) orientation is that the speed bump land must be facing up to avoid high lift off numbers and eventual premature failure. The speed bump must work against gravity, not with it, to float the bearing during starts and stops. Any off-vertical angular orientation can be chosen so long as the speed bump land is on the upper side of the shaft (fig 9).

Figure 9: Support shaft showing correct orientation of speed bump.

The orientation of the speed bump cannot be seen in an assembled unit because it is inside the bearing. This requires an attribute on the assembly, such as a window or cable exit, to be chosen to insert the support shaft in a particular angular orientation. For example, if a unit is to be used horizontally in a cable up orientation, the speed bump land will be aligned to the cable exit upon assembly. The relation is kept so that the final orientation of the unit (assembly) is compatible for horizontal use. Before physical prototypes where built, different speed bump designs were simulated and optimized to a few choices using Lincoln Laser proprietary software. The simulations were conducted to ensure smooth operation through the entire speed range of 4 – 25 krpm. The software is capable of running dynamic simulations outputting rotor displacement, attitude, and pressure profiles. The displacement is displayed in x,y,z coordinates and the attitude or tilt is displayed in radians. Bearing stiffness can be calculated from the

program’s output by varying the inputs of side load or unbalance. The change in rotor displacement can then be plotted against the side load, or unbalance force, to obtain bearing stiffness. Airborne contamination is a problem for aerodynamic bearings, especially since they do not have a pressurized supply of clean air like aerostatic bearings. The Lincoln Laser speed bump stabilization system does not cause flow of outside air through the bearing. It uses air that already exists in the gap making it less susceptible to contamination. On prior aerodynamic bearings of similar configuration, experiments were conducted with powdered substances to simulate contamination conditions. No bearing failures resulted from these experiments. It appeared that the small air gap coupled with the shearing forces repels common contaminants such as printer toner, baby powder and chalk. Pumping-style stabilization systems such as grooved herring bone, double chevron and some thrust bearings are susceptible to contamination. Care must be taken with these systems by providing a clean environment or adding a filtering media to the intake side of the bearing. Tribology is of major concern to aerodynamic bearings. The only lubricant that is used is air molecules. Common wearing materials are: ceramic on ceramic, stainless steel on ceramic, nickel plate on hard anodize and tool steel on tool steel. Lincoln Laser has had the best results with a stainless steel sleeve rotor on a stationary ceramic post. However, initial simulations showed that increasing the bearing surface area was not able to overcome the increase in rotor weight thereby leading to instability or severely decreased load capacity. This demonstrated the need for a different material dynamic to create a lighter weight rotor. Initially, ceramic was considered a good choice since it was less than half the density of steel. It was decided to reverse the materials of the two wearing surfaces. However, cost concerns arose at the aspect of machining complex rotor shapes from ceramic. At the same time, a very robust ceramic coating was developed and applied to the inside wearing surface of the rotor. This allowed the rotor to be changed to aluminum for ease of machinability and weight reduction. The support shaft was changed to stainless steel from solid ceramic further reducing cost and increasing manufacturability. The ceramic coating on the inside of the aluminum rotor and the stainless steel shaft retained the same tried and true tribology and allowed a much larger payload than expected. It was decided to use an external motor supplier to produce a brushless direct current motor capable of the speed and torque required to overcome the windage of the optic. The first iteration motor tested was supposed to perform the function of drive motor and magnetic axial support of the rotor. It was discovered that the high strength magnet required for axial support compromised the upper speed due to magnetic viscous losses and high back emf. More notably, the increased axial support resulted in proportionately higher unwanted radial pull. This resulted in high liftoff speeds and short bearing life. The usual methods of centering the motor stator did not help. Small differences in the strength of the individual magnet segments could not be easily corrected. The second iteration motor was designed only for rotational power. Axial support was provided by the axial support magnet and ring mentioned before. With the axial support taken care of, the magnetic coupling of the motor magnet and stator were decreased significantly. Thus allowing higher speeds and longer bearing life due to reduced side pull.

5. DESIGN VARIATION WITH ENCODER A second variation of the HCAB was designed with an internal optical encoder. Cylindrical air bearings are difficult to pair up with an optical encoder because of the axial movement of the rotor. Most optical encoders consist of a light source and read head that straddles a transparent disk with printed or etched lines. The internal clearances are usually very small, on the order of thousandths of an inch. This is too small for the axial displacements seen with shipping impacts or even axial flutter while the unit is in operation.

The encoder chosen was a newly developed large gap reflective optical encoder. This allowed a stop to be used in only one axial direction of the rotor and facilitated easy assembly. The one sided nature of the reflective encoder allowed the rotor to be inserted or removed without disassembly of the encoder. The straddling style encoder needed rotor stops in both directions and needed to be removed if the rotor was to be pulled out of the housing. Also, the encoder position would have to be reset upon re-assembly. The large gap of the reflective encoder tolerated considerable amounts of axial flutter with little error in position reporting. Rotor position to encoder signal error was typically under 3 arc seconds. The modular nature of the encoder required that the grating disk be manually centered at Lincoln Laser Company. A special fixture incorporating a static air bearing spindle was used to support the HCAB rotor for rotation while the disk was centered with an optical scope. After centering, the disk was affixed with a special adhesive. Radial run out of the grating track can be controlled to .0002 inches in this manner for high accuracy.

6. TESTING

An air bearing does not experience any wear when operating above its lift-off speed. The shearing action of air in the gap is not enough to cause erosion of bearing materials. Theoretically, an air bearing can run forever. However, an aerodynamic bearing has a weakness much like an airplane wing; it needs a certain velocity to lift-off. When the bearing is at rest there is physical contact between the wearing materials. After powering up a unit, a sliding contact continues until the lift-off speed is reached. At this point the rotor should remain airborne through its entire speed range until it is powered off and slows back down to the lift off speed and begins a sliding contact again to a stop. Therefore, starting and stopping is the only time that the bearing experiences wear. Why not continually run the MPA? Typically, in optical systems, there are power requirements necessitating certain duty cycles and foremost, dusty environments cause impingement (sand blasting) on the optical facet. A sealed, glass windowed MPA housing can alleviate the sand blasting, but it creates heating induced balance and wobble problems at higher speeds. The life span of aerodynamic air bearings is measured in hours of run time and number of start-stop cycles. As stated before, the run time affects the optic more than the bearing and is dependent on the environment it is used in. The key performance attribute is designing a bearing that survives thousands of start-stop cycles. The prototype HCAB bearings were placed on an automatic cycling machine that brought the units up to an appropriate speed, powered off and allowed to coast to a stop, then restarted and so on. This was done in blocks of 10,000 cycles. All prototype bearings passed a 30k cycle limit for testing but not all bearings were brought to failure due to time constraints. However, early prototypes allowed to run typically failed in the 100k+ cycle range; well beyond the 20k start stop cycle life requirement. The stiffness of the bearing was theoretically calculated using Lincoln Laser Company bearing design software. Unbalance was added statically and dynamically to the rotor (fig 10). The static unbalance was used to calculate the radial stiffness and the dynamic unbalance was used to calculate tilt stiffness. Axial stiffness was provided by the magnetic axial support and was not as important for polygon scanning. Hence, axial stiffness was not controlled by the bearing and was not a consideration in the air bearing software. When the rotor assembly dynamics were simulated, a displacement or eccentricity was seen with the unbalance. The unbalance contributes a force to the rotor and can be calculated from the mass and angular speed of the rotor. The force of the static simulation is divided by the distance (eccentricity) to obtain the radial stiffness. The torque of the dynamic simulation is divided by the angular displacement of the rotor to obtain tilt stiffness. The stiffness was found to be 0.26 lb/µin in the radial direction and 0.51 lb·in/µrad in the tilt direction.

Figure 10: Static and dynamic unbalance.

Axial stiffness is provided by the magnetic axial support system. This is much ‘softer’ than the bearing stiffness. Axial position sensitive systems, such as drum scanning, can not be accomplished with this arrangement. Because of the soft system, ‘sag’ of the optical unit can be seen with gravity. ‘Sag’ refers to the change in waterline height of the optic due to gravity and must be accounted for when designing a unit. The axial stiffness was empirically found to be 66 lbs/in with a break away force of 5 lbs. Windage refers to the amount of torque that is required to spin the polygon rotor assembly against aerodynamic forces wherein motor is the limiting speed factor. The bearing is stable with all payload configurations, but the motor/controller combination does not have enough power to make 25krpm with all configurations. Since windage is the major cause of power consumption, it is also the largest contributor of heating. More torque could be supplied by a different controller, but heating becomes a concern. For example: an eight faceted 5.5” diameter, 0.60” thick mirror at 18 krpm requires 16.9 oz·in of torque. This equates to 225 watts. If the unit requires a glass window, the heat is contained and a substantial temperature rise of the unit will occur, again, leading to balance and track problems.

7. CONCLUSION

The HCAB development process has resulted in a precision rotary solution that eliminated many of the problems associated with traditional ball bearing scanners. In addition, HCAB provides many advantages of an aerostatic air bearing without requiring any additional equipment.

8. REFERENCES

1. Bernard J. Hamrock, Fundamentals of Fluid Film Lubrication, McGraw Hill, 1994