SPACE LUBRICATION AND PERFORMANCE OF HARMONIC DRIVE GEARS I Schafer*, P Bourlier*, F Hantschack*, E W Roberts **, S.D.

Lewis**, D J Forster** and C John ** * Harmonic Drive AG, Limburg Germany Phone +49 6431 5008 851 Fax: +49 6431 5008 59 E-mail: [email protected] **ESTL, AEA Technology plc, Darwin House, 414 The Quadrant, Birchwood Park, Warrington, WA3 6AT U.K. Phone +44 870 190 2315 Fax: +44 870 190 2304 E-mail: [email protected]

ABSTRACT Harmonic Drive gears are commonly used in space mechanisms. However, lubrication is a critical aspect for proper, effective performance. This paper describes the principle of operation, explains the key parameters that define gear performance and discusses the nature of the various tribological interfaces to be found within HD gears and which call for effective lubrication. Test data are presented on a compact type of harmonic Drive, type CSD 20, that has a reduction ratio of 160:1. Two such HD gears have been lubricated with two space greases namely Maplub SH050a (a grease containing Nye 2001 oil and PTFE) and Maplub PF100a (a grease containing Z25 oil and PTFE). These units have been subjected to thermal-vacuum testing during which measurements were made of input torque, efficiency, break-out torque and no-load backdriving torque. The dependency of these parameters on output load, rotational speed and temperature have been quantified. The work identifies the interface between the wave generators outer race and the inside surface of the flexspline as the most tribologically demanding and highlights the importance of achieving effective lubrication in this area 1. A BRIEF HISTORY OF THE HARMONIC DRIVE GEAR From its origins in aerospace the harmonic drive gear has now established itself in this field as the ideal solution in a wide range of different uses. The first major space application was in 1971 as the mechanical transmission element within the individual wheel drives of the Lunar Rover Vehicle on the Apollo 15, 16 and 17 mission. Soon afterwards Harmonic Drive gears were applied as part of the telescope drive actuator for the imaging photopolarimeter that flew on NASAs Pioneer 10 and Pioneer 11 planetary probes, launched in March 1972. At that time the hermetic sealing possible with the harmonic drive gear was an important additional attribute, given the high vapour pressure and migrating tendencies of the silicone-based lubricants then used. In the meantime this gear has become the preferred solution for compact, powerful positioning drives with references including the Hubble space telescope, the Mars Exploration Rovers and many scientific and commercial satellites. 2. PRINCIPLE OF OPERATION The harmonic drive gear is unique in transmitting high torque through an elastically deformable component. The gear has just three concentric elements: The Circular Spline (CS) is a solid cylindrical ring with internal gear teeth. The Flexspline (FS) is a non-rigid, thin cylindrical cup with external teeth at the open end of the cup. The closed end of the cup is provided with a flange connection to following machine elements. The Wave Generator (WG) comprises a thin-raced ball bearing fitted onto an elliptical plug, serving as a high efficiency torque converter. These three basic components function in the following way: The Flexspline is slightly smaller in diameter than the circular spline and usually has two fewer teeth than the CS. The elliptical shape of the Wave Generator causes the teeth of the FS to engage the CS at two regions at opposite ends of the major axis of the ellipse.

The performance requirements of space vehicles are increasing steadily. This would not be possible without continuous improvements from the applied gear and actuator technology. Precision gears and actuators must fulfill a complex set of requirements they must provide high positioning accuracy and repeatability, high torque capacity and high torsional stiffness, a compact and light design at a competitive price. These requirements have led to significant further development of the harmonic drive gear. This gear type, also known as strain wave gearing is a standard transmission component in a wide range of application areas, from industrial robots and machine tools to printing machines, medical equipment and machines for semi-conductor manufacturing.

As the WG (input) rotates, the zone of tooth engagement travels with the major axis of the ellipse. For each 180o clockwise movement of the WG, the FS (output) moves counterclockwise by one tooth

3. KEY FEATURES Compared to conventional gearing the harmonic drive gear offers the user a number of significant advantages. The key feature is the exceptionally high positioning accuracy and repeatability. This is the result of the high transmission accuracy, which is better than 30 arcsec for standard series gears and better than 20 arcsec. Due to the very low hysteresis losses a repeatability of +/- 5 arc seconds is achieved for standard series gears and better than +/- 3 arc seconds for selected gears is possible. These values are assisted by the fact that the harmonic drive gear can operate with zero backlash. Due to natural radial pre-loading in the region of tooth engagement, the gear operates without backlash. Since power is transmitted through multiple tooth engagement, harmonic drive gears offer a very high torque capacity, equal to conventional drives twice its size and three times its weight. Harmonic drive gears exhibit very high torsional stiffness with an almost linear stiffness characteristic. The hysteresis losses are also extremely low, which reflects the very low internal friction within the gear assembly. The high efficiency as a speed reducer means that the gear is reversible. In an emergency situation it is possible to back-drive the gear. For space applications the compact, lightweight design are particularly important, as well as the design flexibility offered by this principle. 4. LUBRICATION OF HARMONIC DRIVES Lubrication is a key element for space mechanisms and has a huge influence on performance and life characteristics. The lubricant has to be understood as an integral part of the mechanism. Lubricants heritage A variety of different lubricants have been used successfully over the years with Harmonic Drive gears in different space applications. This heritage influences the users of Harmonic Drives in the choice of the lubricant but does not give any information about the characteristics of the lubrication and how close to the limits the application was. Still the most commonly used lubricant is Braycote 601 EF. The phenolic resin bearing retainer is impregnated with the base oil in a vacuum environment. Other lubricants which have successfully been used are the Maplub SH100 and the Maplub PF 100a. Gold-plating has been used to support the grease lubrication of gear teeth. Similarly, silver ion plating of the inner and outer bearing races and also of the wave generator balls has been undertaken. For dry lubrication, initial experiments at ESTL with sputtered and spray-bonded MoS2 were not successful, but tests are continuing at both Japanese

i=

n fs ncs n fs

relative to the CS (fixed). where i = reduction ratio (input speed/output speed) ncs = number of CS teeth nfs = number of FS teeth

Each complete clockwise rotation of the WG results in the FS moving counterclockwise by two teeth from its previous position relative to the CS. The reduction ratio is therefore not a function of the relative sizes of the toothed components, as is the case for spur gears or planetary gears, but simply of the number of teeth.

Fig 1 Principle of operation

and German research facilities to research appropriate dry lubricants for harmonic drive gears. Tests with Dicronite WS2 are running currently at Harmonic Drive AG. Specific movement between surface in contact

Fig 4 shows the sliding speed in the tooth area at an input speed of 2000 rpm. It can be seen that the tooth is approaching at a relative high speed which is decreasing when reaching the point of maximum engagement. The figure shows as well that there is a difference in sliding speed for the different longitudinal positions of the tooth length.

Fig 4: Sliding speed in teeth system (CSD 20) Fig 2: Lubricating surfaces For space gears, three areas of lubrication can be identified for the Harmonic Drive gear: - The teeth system FS/CS. - The Wave generator bearing. - The contact surface between the Wave Generator output race and the honing surface of the FS This paragraph describes the characteristics of the different lubricated areas. The Tooth System Different to conventional gears with involute tooth profiles, the movement of the teeth in the Harmonic Drive is primarily slide with a low amount of roll as can be seen in Fig 3 shows. The tooth engagement movement has been optimized by developing the special IH-tooth profile which takes this into account. Wave Generator Due to the elliptical shape of the wave generator, the speed of the balls in the bearing is not constant. This has an effect for highly dynamic applications and high input speeds. The balls situated on the main elliptical axis of the wave generator where the teeth are fully matching and load is transmitted, are the ones which see the highest load. Due to the nature of the Wave Generator, they are rolling with negligible spin motion but with some micro-slip. The contact stresses in the WG bearing are higher than in the teeth system.

Fig 5: Load distribution on the wave generator Measurements carried out at a size 40 Harmonic Drive gear shows the load distribution over the circumference of the WG plug (see Fig 5). It can be seen that the distribution changes with the applied load and that the new N-shape geometry of the plug reduces the peak loads.

Fig 3: Sliding movement of the teeth

Contact between Wave Generator bearing and Flexspline The most difficult movement to conceptualize is the sliding movement between the outer side of the Wave Generator bearing and the inner surface of the Flexspline. This movement depends on the input speed, on the temperature and on the applied load. This movement is typically 15 25 times slower than the Flexspline, i.e. outer bearing ring makes relative to the flexspline one rotation for every 15 25 flexspline revolution. During the rotation, the point of maximum contact stress is varying in the longitudinal direction due to the coning effect of the Flexspline, which deforms it three-dimensionally. This results in a sinusoidal pattern of this point, which explains the often-to-be-observed traces on the bearing ring at the extremes of this pattern.

Loss mechanisms which are covered in the engineering model are: Losses in the tooth system WG bearing losses Viscosity effects of the lubricant Size effects of the gear Gear reduction ratio Gear type (tooth geometry, FS shape, WG shape) The basic equation for the efficiency model compares the output torque with the necessary input torque (multiplied by the gear ratio). The discussion here considers only a component set, so other parasitic losses are not covered. Basically, the losses can be separated into the losses in the tooth system and the bearing losses:

Tout Tout + NLRT R R teeth

No load running torque Gear ratio Gear output torque Efficiency of the teeth meshing

Fig 6: Coning of the Flexspline As a result of this, the lubricant grease or oil migrates in the gear such as shown on the Fig 7. Therefore it is difficult to separate the different lubrication areas from each other.

with: NLRT R: Tout teeth

The No Load Running Torque is the input torque level required to run the gear with zero output torque, i. e. all torque applied is consumed up by internal gear losses. The NLRT is influenced by: No-load starting torque: the NLRT at zero speed. Viscosity loss of the lubricant, which is speed dependent. Gear size Gear type (HFUC, HFUS, CSD, etc.) Gear ratio.

Fig 7: Migration of lubrication Fig.7 Efficiency of the HD gear Harmonic Drive AG has developed an engineering model which describes the losses within the gear due to various mechanisms. For the temperatures above 10C, the model has been shown to provide very good results compared to measurements. Tests are currently being undertaken in order to fill the gap of knowledge for lower temperatures,.

The viscosity loss of the lubricant is calculated based on the base oil viscosity used for lubrication. The problem at low temperatures is that the efficiency is significantly influenced by the fact that the gear losses result in a local temperature increase, dependent on thermal conductivity and grease loading. This makes the calculation under these conditions difficult.

Fig 8: Input torque and efficiency-calculation tool This model, although it has its obvious limitations due to the simplicity of the calculation, has shown acceptable accuracy for conceptual layout work and preliminary estimates of the gear performance.

measurement of break-out torque, efficiency, hysteresis loss, lost motion (akin to backlash), torsional stiffness and no-load back-driving torque. Initially each unit was characterised in vacuum over a range of temperature (from temperatures close to the pour point of the grease up to +80 deg.C). Subsequently the harmonic drives were subjected to life testing under thermal-vacuum conditions in which the soak conditions cycled between 40 deg.C and + 80 deg.C. In these tests the aim was to achieve 125,000 revs of the output shaft (equivalent to 20 million rotations of the input shaft). In the event the life tests were curtailed due to fall-offs in efficiencies.

Test items The test items were 2-off, type CSD-20-160-2A-GRSP harmonic drive units, each having a reduction ratio of 1:160. Each unit was manufactured from the following materials:

Fig 9. Harmonic drive CSD component set

Circular spline (CS) and wave generator plug: SUS 630 H1150 (equivalent to 17-4PH) Flexspline (FS): 15-5PH H1075 Wave generator (WG) bearing races and balls: SUS 440C (equivalent to AISI 440C) The retainer is cotton-reinforced phenolic material (impregnated using the base-oil of the test grease). These were supplied dry to ESTL. Retainer stopper: SUS 304L (equivalent to AISI 304L) Fasteners are in SUS XM7 (equivalent to AISI 302HQ (1.4567))

The nominal rated output torque of the unit is 28Nm. 5. TESTS AT ESTL ON CSD COMPONENT SETS Lubrication The availability of a compact, low mass, high torque capacity gearbox for space applications is highly attractive. Harmonic Drive produce a number of ranges, the latest of which, type CSD (Fig.9) is much more compact than previous units and, for space use, has the advantage of much higher specific torque output. The units have a reduction ratio of 1:160. Tribologically the most critical interface is that between the wave-generator bearings outer race and the inside surface of the flexspline. The ability to lubricate this interface effectively is therefore expected to determine the useful life of the gearbox. Two such Harmonic Drive units (type CSD-20-1602A-GR-SP) have been lubricated with two space greases namely Maplub PF100a a grease containing Z25 oil and PTFE, and Maplub SH050a a grease containing Nye 2001a oil and PTFE. Each gearbox has been tested using an ESTL-designed test rig that allows The harmonic drives were lubricated with either Maplub SH050a or Maplub PF100a both of which are greases formulated for use in space. In each case the phenolic cage of the wave generator bearing was vacuum-impregnated with the base oil of the test grease. Lubrication details are summarised in Table 1. Table 1 Lubrication details for CSD-20-160-2A-GR-SP harmonic drive gearboxes Unit No. WG bearing; CS & FS gear teeth; FS/WG interface Maplub SH050a Maplub PF100a Phenolic cage

1 2

Nye 2001a Z25

Creep barrier (FC732) was applied on appropriate surfaces within the harmonic drive to minimise oil loss.

The areas treated included that area on the internal surface of the FS not in contact with the WG bearing.
100.0

Output Torque/Efficiency Curves In Air Testing, MAPLUB SH050a Grease

Test procedure In outline the test procedure was as follows:

Efficiency %

90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0

50rpm 100rpm 500rpm 250 rpm

a) Installation of the test harmonic drive. Measurement of its stiffness/lost motion as determined from an evaluation of the drive hysteresis (with the input shaft locked). b) Run-in to distribute the grease. Operation of the HD in air with an input shaft speed of 50rpm, while monitoring the torque behaviour for anomalies during a period of 0.5 hours clockwise followed by 0.5 hours anticlockwise. c) In-air characterisation of HD. Measurement of the no-load back-driving torque, lost motion, hysteresis loss and stiffness. Operation of test unit at speeds of 50, 100, 250 and 500 rpm. At each speed, measurement of efficiency as function of torque load. Measurements of the break-out torque at 0, 5 and 14 Nm load. d) In-vacuum characterisation of HD. Characterisation of the HD in vacuum, in terms of its break-out torque and efficiency as a function of temperature at 250 rpm i/p speed. The temperature range was from ~10 deg.C above the test greases pour point to +80 deg.C. The no load back-driving torque was also measured. e) Thermally-cycled, life testing. Subjection of the HD to the following thermal sequence in vacuum: 25,000 o/p revs. at 40 deg.C 25,000 o/p revs. at +80 deg.C 25,000 o/p revs. at 40 deg.C 25,000 o/p revs. at +80 deg.C 25,000 o/p revs. at room temperature The output load was 14Nm and the i/p shaft speed nominally 500 rpm. Input torque was monitored throughout. It was decided that the life-test would be stopped and the parts examined if the performance degraded to the point where the efficiency decreased by 10% or more (under a given test condition). Test results Unit 1 (test grease SH050a) Efficiency measurements made in air and vacuum are shown in Figs. 10 and 11 below. As expected, the efficiency increases with increasing output load, decreases with increasing rotational speed and decreases with decreasing temperature.

0.0 0 2 4 6 8 10 12 14 16 Output Torque Nm

Fig. 10 Efficiency of Unit 1 HD (Maplub SH050a lubricant) as a function of output load, in air at 21 deg.C
Efficiency/Output Torque Graphs @ 250rpm MAPLUB SH050a Grease 100 90 80 70 Efficiency % 60 50 40 30 20 10 0 0 2 4 6 8 10 12 14 16 Output Torque Nm
Vac @ -45C Vac @ -35C Vac @ -25C Vac @ -15C Vac @ -5C Vac @ +5C Vac @ +15C Vac @ +20C Vac @ +40C Vac @ 60C Vac @ 80C

Fig. 11 Efficiency of Unit 1 HD (Maplub SH050a lubricant) as a function of output load, in vacuum in temperature range 45 deg.C to +80 deg.C.
350 300

Input torque (mNm)

0 Nm o/p tq. 5 Nm o/p tq. 14 Nm o/p tq.

250 200 150 100 50 0 -70 -50 -30 -10 10 30

50

70

90

Temperature (deg.C)

Fig.12 Input (running) torque as a function of temperature obtained at 250 rpm (in vacuum). The influence of temperature is further illustrated in Fig.12 which plots the steady-state input torque as a

function of temperature for the no-load condition and with applied output loads of 5 and 14 Nm. The break-out (starting torque) torque was only significant at output loads of 0 and 5Nm, being typically 9% and 3% (respectively) higher than the running (steady-state) torque. At an output load of 14 Nm the break-out torque did not exceed the running torque. The torque required to back-drive the unit increased with decreasing temperature as illustrated in Fig.13.
9 8 7 6 5 4 3 2 1 0 -60 -40 -20 0 20 40 60 80 100 Temperature (deg.C)

an indication that lubrication at this interface had not been fully effective.

Test results Unit 2

We present below the equivalent data for Unit 2. See Figs. 14 17. The general trends in terms of efficiency, input torque and back-driving torque are similar to those seen for the unit lubricated with Maplub SH050a grease. Generally, the efficiency of the PF100a-lubricated gearbox was lower than that lubricated with SH050a although the difference became negligible below 0 deg.C.
70 60
Efficiency

Back-driving torque (Nm)

50 40 30 20 10 0 0 5 10 15
50 rpm 100 rpm 250 rpm 500 rpm

Fig. 13 No-load back-driving torque of Unit 1 (Maplub SH050a) as function of temperature (in vacuum)

Output torque (Nm)

Fig. 14 Efficiency of Unit 2 HD (Maplub PF100a lubricant) as a function of output load, in air at 21 deg.C

In the thermal-cycling life test the unit successfully completed the first cold soak phase (25,000 o/p revs.) with no performance degradation. Upon completion of the cold soak the temperature of the gearbox was increased, as planned, to 80 deg.C. On operating the gearbox at 250 rpm the efficiency was measured at 76%, but after 10 mins of operation at 250 rpm (i/p shaft) the input torque increased significantly and the efficiency reduced to 62%. In addition, the operation of the gearbox became noticeably noisy (audible noise). The unit was stopped and left stationary for 48 hours; the temperature was maintained at +80 deg.C. On restarting the gearbox, it again operated noisily at the reduced efficiency (i.e. 62%). Given the noisy performance and decreased efficiency the test was stopped and the gearbox removed for examination. A strip examination of the gearbox revealed that the gear teeth on the flexspline and circular spline were in a good condition with no signs of significant or excessive wear. Likewise the internal parts of the wave-generator bearing appeared in a good condition indicating that lubrication had been effective. However, the surfaces of the bearings outer race and the inside surface of the flexi-spline showed distinct signs of wear. There was evidence that metal had transferred from the flexspline to bearing counterface

80 70
Efficiency (%)

-50 deg.C' -40 deg.C' -30 deg.C -20 deg.C -10 deg.C 0 deg.C 20 deg.C 40 deg.C 60 deg.C 80 deg.C

60 50 40 30 20 10 0 0 5 10 15

Output torque (Nm)

Fig. 15 Efficiency of Unit 2 HD (Maplub PF100a lubricant) as a function of output load, in vacuum in temperature range 45 deg.C to +80 deg.C.

400 350

and Maplub PF100a greases. Measurements made under thermal-vacuum conditions have shown that:
0 Nm o/p tq. 5 Nm o/p tq. 14 Nm o/p tq.

Input torque (mNm)

300 250 200 150 100 50 0 -70 -50 -30 -10 10 30

the driving torque increases (and efficiency decreases) as the temperature of operation is decreased. The most rapid rise in driving torque occurs below 0 deg.C as the temperature approaches the pour point of the grease. the efficiency of the PF100a-lubricated gearbox was generally lower than the unit lubricated with SH050a, although the difference became negligible below 0 deg.C. the no-load back-driving torque proved sensitive to temperature; its value increasing with decreasing temperature the life testing of both units was stopped during the life tests due to decreases in efficiency of 10% or greater. At the time of stoppage the first unit (SH050a) had completed 4 million i/p revs. at 40 deg.C and had undertaken a small number of revs. at +80 deg.C; whilst the second unit (PF100a) had completed 10,000 i/p revs. only at 40 deg.C. Further work is required to establish whether this is due to failure of the lubricant or due to loss (displacement) of lubricant from the FS/WG interface due to the coning action of the flexspline. 6. CONCLUSIONS

50 70 90

Temperature (deg.C)

Fig.16 Input (running) torque as a function of temperature obtained at 250 rpm (in vacuum). 16

Back-driving torque (Nm)

14 12 10 8 6 4 2 0
-60 -40 -20 0 20 40 60 80 100

Temperature (deg.C)

Fig.17 No-load back-driving torque of Unit 2 (Maplub PF100a) as function of temperature (in vacuum) The break-out torque was low, being up to 5 % higher than the running (steady-state) torque. As previously the back-driving torque increased with decreasing temperature. The proposed thermal-cycling life test was again stopped prematurely. During the initial operation in the cold phase (-40 deg.C) the initial efficiency was ~29% - consistent with previous measurements. However, after the completion of some 10,000 input revs. the input torque increased significantly, corresponding to a drop in efficiency of 10%. The test was stopped and the unit dis-assembled for examination. The gear teeth and internal surfaces of the bearing were in good condition. The internal surface of the flexi-spline (that had been in contact with the wave-generator bearing) showed clear signs of wear as did the counterface on the WG bearing. The lubrication in this area had thus not been fully effective. Discussion/conclusions of test results Two compact, CSD type harmonic drive gearboxes have been lubricated with respectively Maplub SH050a

The working principles of Harmonic Drive gears have been described and a detailed description of the various tribological interfaces presented. Test data obtained under thermal-vacuum conditions have been obtained on two compact (CSD) HD gears lubricated with greases formulated for use in space. The variation of input torque, efficiency, break-out torque and no-load back-driving torque with load, speed and temperature have been determined. The WG/FS interface is confirmed as being, tribologically, the most demanding of all the loaded contact zones within the gear system. Further work to optimise the lubrication of this area is required and to ensure long life under thermal-vacuum conditions.