Cross Reference to Related Application This application is a continuation-in-part of United States patent application serial no. 052,922, filed Kay 19, 1987, by Emerson L. Kumm and also entitled CONTINU- OUSLY. VARIABLE TRANSMISSION, now United States Patent 4,768,996

Field of the Invention This invention relates to the continuously variable transmission (CVT) art and, more particularly, to a hy- drauliσ/mechanical control system for establishing the speed ratio in a flat belt continuously variable trans- mission.

Background of the Invention Continuously variable transmissions of the class broadly characterizable as that in which a belt couples a pair of pulleys, each of which can-assume a more or less continuous range of effective diameters, generally fall into two categories; viz: a) those employing V-belts or variations thereof (such as link belts or chains) for transmitting power from one pulley to the other, and b) those systems employing flat, flexible belts between the variable diameter pulleys. Those skilled in the art have come to appreciate that CVT's employing• flat, flexible belts enjoy signifi- cant fundamental advantages over those systems employing V-belts. In the case of the latter, -the belts are co - posed of various compositions and have a trapezoidal cross section, the belt transmitting rotary motion at one speed from a source of power (such as an engine or motor) to an output shaft at another speed, the speed ratio be- ing varied in a continuous fashion from a minimum to a maximum as dependent on the geometry of the belt and the pulley system. The V-belt is compressed between smooth, conical sheave sections in each of the two pulleys by ex-

ternal axial forces acting on the sections to apply ten- sion or compression to the belt and friction between the sides of the V-belt in the sheave sections to prevent slippage. In operation, a force unbalance caused by changes in the axial loading of the sheave sections causes the V-belt to change its radial positions in the two pulleys uni_.il a force balance is achieved or a limit range stop is reached. For a large transmitted torque, the required axial forces exerted on the sheaves result in large compressive forces on the V-belt which requires that the belt have a substantial thickness to prevent its axial collapse or failure. This increase in thickness increases the belt's centrifugal force and causes higher belt tension load. in addition, as the belt thickness increases, the pulley size must be increased due to higher stress loads at a given design minimum pulley radius. Further, the typical V-belt must continuously "pull out" from the compressive sheave load on leaving each pulley which results in sig- nificant friction losses and belt fatigue which adversely affects the overall efficiency and operating life. Con- sequently, although variable speed pulley drives have successfully used V-belts in a wide range of applica- tions, they have been severely limited in their power ca- pabilities for more competitive smaller size equipment. As a result of these inherent drawbacks to the use of V-belts in continuously variable transmissions, a sec- ond category has developed which may broadly be desig- nated as flat belt drive continuously variable trans is- sions. As the name suggests, flat belts are employed be- tween driven and driving pulleys which are dynamically individually variable in diameter to obtain the sought- after ratio changes. No axial movement between the two pulley rims is necessary. On the other hand, it is nec- essary to somehow effect the diametric variations of the individual pulleys, and in one particularly effective system, this function is achieved by causing a circular array of drive elements in each pulley to translate radi-

ally inwardly or outwardly in concert as may be appropri- ate to obtain a given effective diameter of the pulley. Variable speed flat belt transmissions of this particular type, and their associated control systems, are disclosed in United States Patents 4,024,772; 4,295,836 and 4,591,351 as well as United States patent application Se- rial No. 871,254 filed June 6, 1986, and now United States Patent 4,714,452 _{/ a} n o Emerson L. Kumm. In all but the first patent enumerated above, the subject variable diameter pulley components have included a pair of pulley sheaves between which extend a series of belt engaging elements that are simultaneously moved both ra- dially and circumferentially. In one exemplary construc- tion, there is a series of twenty-four belt engaging ele- ments such that an angle of fifteen degrees extends be- tween runs of the belt coming off tangentially from one belt engaging element compared to that of an immediately adjacent belt engaging element. Each pulley sheave includes two pairs of two disks ^• ;'-i.esignated, respectively, the inner guideway disk and a outer guideway disk in each pair) which are parallel to each other with the inner and outer guideway disks of each pair being disposed immediately adjacent one an- other. Each of the guideway disks of an adjacent pair has a series of spiral grooves or guideways with the guideways of the pair oriented in the opposite sense such that the ends of the belt engaging elements are captured at the intersections of the spiral guideways. Thus, the radial adjustment to the belt engaging elements may be achieved by bringing about relative rotation- between the inner and outer guideway disks to change their angular relationship, this operation being, of course, carried out simultaneously and in coordination at both pairs of guideway disks of a pulley. Thus, it will be appreciated that the overall transmission ratio is dependent upon the respective angular relationships between the facing pairs of guideway disks on each of the two pulleys.

From the foregoing, it will be understood that the control system which establishes the instantaneous angu- lar relationship between the facing disks of each pair on each of the pulleys is a highly important system within the entire continuously variable transmission. A series of related disadvantages has been characteristic of the prior art control systems for establishing these angular relationships. More particularly, these disadvantages include the fact that the mechanical components of the prior art hydraulic/mechanical control systems have been physically large and heavy, contributing the majority of the overall size of these continuously variable transmis- sions and accounting for a large portion of their weight. Also, the prior art control systems have required sub- stantially larger hydraulic supply pressures to achieve the desired control response rates. As a result, the hy- draulic control systems have been, of necessity, corre- spondi ^' ngly complex, further contributing to the size, weight and cost of the prior art continuously variable trans issions. It is to provide a control system which overcomes these several related objections to the prior art control systems for flat belt continuously variable transmissions that my invention is directed.

objects of the Invention is therefore a broad object of my invention to provide an improved flat belt continuously variable transmission. It is another object of my invention to provide, in such a continuously variable transmission, an improved control system for establishing the effective transmis- sion ratio between an input shaft and an output shaft coupled by a flat belt. it is a still further object of my invention to pro- vide such a control system which is relatively small, lightweight, simple and inexpensive.

1 It iε yet another object of my invention to provide

2 a continuously variable transmission in which the trans-

3 mission speed ratio may be more rapidly changed than in

4 the past.

5 In another aspect, it is a more specific object of

6 my invention to provide such a control system which is

7 self-energized from the energy transmitted -to the pulley.

8 Still more specifically, it is an object of my in-

9 'vention to provide such a control system in which the ef-

10 feet of centrifugal forces exerted by relatively heavy

11 belt-engaging elements are used to advantage and compen-

12 sated for as necessary. 13

14 Summary of the Invention

15 Briefly, these and other objects of my invention are

16 achieved by providing, in a flat belt continuously vari-

17 able ratio transmission, a control system which obtains

18 the energy necessary to effect pulley diameter changes

19 from the energy source driving the pulley and belt assem-

20 bly rather than from an external source. The inside and

21 outside guideway disks on both sides of eε-ch pulley are

22 connected together using differential gearing which iε

23 also coupled to a power consuming element. While it is

24 possible to use many different arrangements of differen-

25 tial gearing to obtain the desired operating torques be-

26 tween the inner and outer guideway disks, the presently 27 preferred embodiment of the invention employs a harmonic

. 28 gear drive to provide the differential geared relation-

• 29 ship between the inner and outer guideway disks and a

30 power consuming element. The harmonic drive consists of

31 an elliptically shaped wave generator and a flexible

32 splined gear sleeve (the flexspline) geared to two circu-

33 lar, internally splined rings designated the dynamic

34 spline and the circular spline, respectively. The circu-

35 lar spline typically has two more splines or teeth than 36 the dynamic spline and the flexspline (which have the 37 same number) resulting in a speed reduction between the 38 wave generator and the dynamic spline equal to one-half

of the number of splines on the dynamic spline. Hence, with the harmonic drive, it is possible to obtain speed reductions or -torque increases from the wave generator to the circular spline or dynamic spline exceeding 100 to 1 in a single stage of gearing. In the preferred embodiment of the continuously variable transmission disclosed in the above referenced parent patent .application serial no. 052,922 (now United States Patent 4,768,996 ) , for the driven pulley the circular spline is connected to the inner guideway disks and the dynamic spline is connected to the outer guideway disks. Conversely, for the driving pulley (which is con- nected to the external power source such as a motor or engine) , the dynamic spline is connected to the inner guideway disks and the circular spline is connected to the outer guideway disks. The orientation of the guide- way disks as related to the direction of rotation of the pulleys must be as specified in the previouεly referenced United States patent application Serial No. 871,254, now United States Patent 4,714,452 . With this arrange- ment, a drag torque provided by an output power consuming element connected to the wave generator of each pulley assembly gives a substantially proportional pulley actua- tor torque which, working through the harmonic gear drive, tends to move the belt engaging elements of the pulley radially outwardly. Hence, a fixed length belt can be tensioned around two rotating pulleys for trans- mitting torque using a drag torque on the output power consuming element in each pulley to achieve a self-ener- gized actuator drive. Since very high gear ratios may be used, correspondingly small torques (and hence small pow- ers) are transferred to the output power consuming ele- ment to thereby permit the generation of large actuator torques to be applied for maintaining a constant speed ratio in the CVT or, when desired, to change the CVT speed ratio by an appropriate transient application to dynamically adjust the angular relationships between the inner and outer guideway disks for each pulley.

While numerous arrangementε can be uεed for the con- trol torque abεorption on the output power consuming ele- ment (including an electrical power generator, a friction clutch, an air compressor, variable viscous fluid cou- pling, etc.), the employment of a positive displacement oil gear pump for this operation permits a wide range of drag torques to be effected simply by opening and closing an output valve to variably load the oil pump. In a re- finement of the control system, the oil pumps may selec- tively transiently operate as motors during rapid ratio changes. An oil supply subsystem is correspondingly transiently actuated to facilitate this temporary mode of operation. In the disclosure of the parent application, the use of relatively lightweight belt-engaging elements was as- sumed with such materialε as aluminum, ceramic and plas- tic contemplated. However,it is desirable to consider the use of more dense belt engaging elements (i.e., those made of steel, powdered steel alloys, etc.), which have fatigue strengths much higher than aluminum, ceramic or plastic, as the design pulley power is increaseε. When such relatively dense belt drive elements are employed, the centrifugal forces caused by pulley rotation and tending to counter rotate the two sets of discs in each pulley in the direction that moves the belt engaging ele- ments radially outwardly become dominant, and these cen- trifugal forces can be used to advantage. Thus, at high rotational pulley speeds, the torques generated by the centrifugal forces acting on the heavier belt engaging elements may be balanced by the torque of a hydraulic unit through the differential gearing of a harmonic drive if the specific connections between the dynamic spline or circular spline and the inner or outer guideway discε are reversed from those employed with the lighter weight belt engaging elements. In addition, for rotational speedε below those at which the centrifugal forces are domi- nant, the hydraulic units may be operated as motors, un- der the influence of a revised hydraulic control subsyε-

tern, to, in effect, return the CVT to the εa e mode of operation employed with the lighter weight belt engaging elementε.

Description of the Drawing- The subject matter of the invention is particularly pointed out and distinctly claimed in the concluding por- tion of the specification. The invention, however, both as to organization and method of operation, may best be understood by reference to the following description taken in conjunction with the subjoined claims and the accompanying drawing of which: Fig. 1 illustrates an edge on view of driving and driven pulleys coupled by a flat belt and representative of the class of continuously variable transmissions in which the present invention finds application; Fig. 2 is a cross sectional view, taken along the lines 2-2 of Fig. 1, of the pulley εyεtem illustrated in Fig. 1; Fig. 3 is a fragmentary perspective view, partially broken away, of a pulley particularly illustrating the relationships between inner and outer guideway disk com- ponents and belt engaging element components; Figs. 4a, 4b and 4c are illustrations showing the principle of operation of a harmonic drive, certain com- ponents being shown in an exaggerated elliptical shape in order to more clearly demonstrate the principle; Fig. 5 is a simplified cross sectional view of a flat belt continuously variable transmission illurt.rr.ting the fundamental aspectε of the mechanical componentε of the εubject control εystem for establishing the angular relationship between the inner and outer guideway diεkε of each pulley; Fig. 6 illustrates a hydraulic control subsystem for use in conjunction with the mechanical control subsystem illustrated in Fig. 5 to establiεh the angular relation- ship between inner and outer guideway disks according to the load and other demands;

Fig 7. is a view similar to Fig. 5 illustrating a variant embodiment of the continuously variable transmis- sion in which relatively heavy belt engaging elements are employed; and Fig. 8 is a view similar to Fig. 6 and illustrating a hydraulic control system ^' suitable in conjunction with the mechanical control subsystem shown in Fig. 7.

Detailed Description of the Invention Referring now to Figs. 1, 2 and 3, fundamental as- pects of the flat belt type of a continuously variable transmission, with which the subject control system is employed, are illustrated as embodied in a variable diam- eter pulley drive assembly 10 comprising variable diame- ter pulleys 11 and 12 connected by a flat drive belt 13. The pulley 11 will be considered as the driving pulley and the pulley 12 as the driven pulley in thiε discuε- εion, but it will be underεtood that the roles of these pulleys may be reversed without altering the concepts in- volved. The pulley 11 is appropriately mounted on a shaft 14, and the pulley 12 is similarly appropriately mounted on a εhaft 15. ε is well understood in the art. The pul- leys 11 .and 12 are similar to each other, and only one of them, namely 11, will be specifically described in this discussion. The belt 13 as shown in Fig. 3 corresponds to the position of the belt 13 of Fig. 2 in the dashed line position. The pulley 11 includes a pair of pulley sheaves 16 and 17 between which extend a series of belt engaging el- ements 18, the latter being engaged by the belt 13 for driving, or driven, conditions as will be underεtood. In one construction of the invention, there is a εerieε of twenty-four belt engaging elements 18 equally circuπfer- entially distributed whereby there is established an an- gle of fifteen degrees between runε of the belt 13 coming off tangentially from one belt engaging element 18 as compared to that of an _. immediately adjacent belt engaging

element 18. Each belt engaging element 18 includes a central shank 28, which engages the belt 13, and bearing regions 29 at each end. The pulley sheave 16 comprises a pair of pulley guideway diεkε 19 and 21 which are parallel to and lie immediately adjacent each other in juxtapoεition. Simi- larly the pulley sheave 17 comprises a pair of pulley guideway diεkε 22 and 23 which are parallel to and lie immediately adjacent each other in juxtapoεition. The longitudinal spacing between the pulley sheaveε 16 and 17 (i.e., the axial spacing between guideway disks 21 and 22) remains the same irrespective of the radial adjust- ment of the belt 13 for different driving or driven speeds. This spacing iε εufficient to accommodate with clearance the width of belt 13 which iε εelected to carry the load that the εyεte iε deεigned for as is well un- derεtood. The range of radial adjuεtment or poεition of the belt 13 on the pulley 11, as may be envisioned by the solid line and dashed line positions of belt 13 in Fig. 2, is achieved by altering the radial positions of the belt engaging elements 18. For example, in Fig. 2 the belt engaging elements 18 are close to the center of the shaft 14 in the solid line position of the belt 13 on pulley 11; conversely, the belt engaging elements are ra- dially farther out, namely adjacent the periphery, when the belt 13 is in its dashed line position which is also the position shown in Fig. 3. Variation in the radial positionε of the belt engag- ing elementε 18 is -achieved by relative -rotation to change the angular relationship between the outer guide- way disk 19 and the inner guideway ^* disk 21 of pulley sheave 11 and by identical relative rotation of the pul- ley guideway disks 22 and 23 of pulley sheave 17. As a practical matter, to inεure synchronous operation, the inner guideway disks 21 and 22 are physically locked to- gether, and the outer guideway disks 19 and 23 are also locked together. Power for εuch operation, not shown in

Figs 1, 2 or 3, has been achieved in the prior art typi- cally as disclosed in U.S. Patent No, 4,295,836 previ- ously referenced. The outer guideway disk 19 has a εerieε of logarith- mic spiral guideways 24 therein which progress outwardly from adjacent the center at an angle of forty-five da- grees with respect to the pulley radius. Similarly the inner guideway disk 21 has. a series of logarithmic spiral guideways 25 radiating outwardly at an angle of forty- five degrees with respect to the pulley radius, but in the opposite senεe to the guideways 24 of pulley disk 19. Since the guideways 24 and 25 radiate outwardly at angles of forty-five degrees with respect to the pulley radius, but in opposite senses, the intersections of these guide- ways exist at ninety degrees at all radial positions. This results in a subεtantially constant geometry at the intersections of the logarithmic spiral guideways 24 and 25 at all radial positions for receiving the bearing re- gion endε 29 of the belt engaging elements 18. Simi- larly, the inner guideway diεk 22 haε a series of loga- rithmic spiral guideways 26 radiating outwardly identi- cally to the guideways 25 of inner guideway diεk 21, and the outer guideway diεk 23 includes logarithmic spiral guideways 27 extending outwardly identically to the guideways 24 of outer guideway disk 19. Hence, the guideways 26 and 27 intersect at ninety degrees at all radial positions to give a constant interεection geometry identical to the logarithmic εpiral guidewayε 24 and 25 for receiving the other ends of the belt engaging ele- ments 18. While forty-five degree spirals have been shown and give ninety degree intersections, it will be understood that logarithmic spiralε of other angularities may be used as desired. Also, minor variations from a particu- lar angularity may be tolerated so long as the belt en- gaging element bearing ends supported at the guideway in- tersections will move appropriately when the sheaves are

rotated relative to each other to change the angular re- lationship between the inner and outer guideway disks. It will be clear that the belt 13, as it passes around the pulley 11 or ^* 12, engages the central portion of the belt engaging elements 18 and cauεes one pulley to drive and the other pulley to be driven in the obvious fashion. The foregoing description of the baεic drive εystem, the pulleys 11 and 12, the belt 13 and the belt engaging elements 18 iε εet forth in greater detail in United States patent No. 4,295,836, dated October 20, 1981, pre- viously referred to and doeε not form a εpecific part of the invention described in this application, but forms the environment in which the invention functions. While it is possible to use many different differen- tial gearing arrangements to obtain the desired operating torqueε between the inner and outer guideway diεks of the pulleyε (which, as will be seen below, is a necessary op- eration for practicing the present invention) , the presently preferred embodiment employs a so-called "harmonic" gear drive to provide the differential geared relationship between the inner and outer guideway diskε and certain power abεorbing elements. Referring to Figs. 4a, 4b and 4c, and particularly to the somewhat enlarged Fig. 4a, the basic principles of a harmonic drive gear reduction apparatus are presented. In this most ele en- tary form, a harmonic drive employε three concentric com- ponents to produce high mechanical advantage and speed reduction. The use of nonrigid body mechanics . allows a continuous elliptical deflection wave to be induced in a nonrigid external gear, thereby providing a continuous rolling mesh with a rigid internal gear. Thus, referring to Fig. 4a, an elliptical wave generator 30 deflects a flexspline 31 which carries outside teeth and therefore meshes with the inside teeth of a rigid circular spline 32. The elliptical shape of the flexspline and the amount of flexspline deflection is shown greatly exagger- ated in Figs. 4a, 4b. and 4c in order to demonstrate the

principle. The actual deflection is very much smaller than shown and is well within the material fatigue li - its. Since the teeth on the non-rigid flexspline 31 and the rigid circular spline 32 are in continuous engagement and since the flexspline 31 typically has two teeth fewer than the circular spline 32, one revolution of the wave generator 30 causes relative motion between the flex- spline and the circular spline equal to two teeth. Thus, with the circular spline 32 rotationally fixed, the flexspline 31 will rotate in the opposite direction to the wave generator (the system input in the example) at a reduction ratio equal to the number of teeth ^' on the flexspline divided by two. This relative motion may be visualized by examining the motion of a εingle flexspline tooth 34 over one-half of an input revolution in the direction shown by the ar- row 35. Since the input to the wave generator 30, in the example, causeε clockwise rotation of the wave generator, the flexspline rotates counterclockwiεe. Thuε, referring ^" to Fig. 4b, it will be seen that the tooth 34, after one- quarter revolution of the wave generator 30, has moved counterclockwise one-half of one flexspline tooth posi- tion. It will also be noted that when the wave generator 30 axis has rotated 90°, the tooth 34 is fully disen- gaged. Full reengagement occurs in the adjacent circular spline tooth space when the major axiε of the wave gener- ator 30 iε rotated to 180° aε εhown in Fig. 4c, and the tooth 34 haε now advanced one full tooth position. This motion repeats as the major axiε rotateε another 180° back to zero, thereby producing the two tooth advancement per input revolution to the wave generator 30. Conventional tabulationε of harmonic drive gear re- duction ratioε aεεu e the flexεpline is the output member with the circular spline rotationally fixed. However, any of the drive elements may function as the input, out- p t or fixed member depending upon whether the gearing is

1 used for εpeed reduction, εpeed increasing or differen-

2 tial operation.

3 The harmonic drive principle can be extended by the

4 addition of a fourth element designated the dynamic

5 spline. The dynamic spline is an internal gear that ro-

6 tates at the same εpeed and in the εarae direction as the

7 flexspline. Unlike the circular spline (to which it is

8 parallel, also engaging the flexspline) , _. the dynamic

9 spline has the same number of teeth as the flexspline. 0 Flexspline εhape rotation reεultε in tooth engage- 1 ment/diεengagement within the εa e tooth εpace of the dy- 2 namic spline such that the ratio between the two is one 3 to one. The system, therefore, iε a flexspline output 4 with the same characteristics as the three element har- 5 onic drive model; i.e., gear reduction ratio tabulated 6 with the direction of rotation opposite to the input. 7 Ultra high dual ratio capability can be obtained by using ° two circular splines in mesh with the flexspline with 9 each developing a different single-stage ratio. Merely C by way of example, the compounding of single-stage ^' ratios 1 of 160:1 and 159:1 resultε in a total reduction ratio of 2 12,720:1. Harmonic driveε εuitable for uεe in the pre- 3 εent invention may be obtained from the Harmonic Drive 4 Division of E hart Machinery Group in Wakefield, Masε. 5 The subject invention, in the presently preferred 6 embodiment, employs a four element harmonic drive in 7 which the fourth element is a dynamic spline. Referring 8 now to Fig. 5, a slightly simplified representation of a 9 flat belt continuously variable transmission according to 0 the present invention is εhown. Preliminarily, it may be noted that, with the proper connection of the dynamic spline or the circular spline to the inner guideway disks or the outer guideway disks on a given rotat -.g pulley using the guideway disk orientation as related "LO the di- rection of pulley rotation and direction of power flow as 6 given in the previously referenced United States patent application Serial No. 871,254 (now United States Patent 4.714.452 ) ι a drag torque on a power consuming element

connected to the wave generator can be caused to give a largely proportional force which tends to move the belt engaging elements radially outwardly. As a result, a fixed length belt can ^* be tensioned around two rotating pulleys for transmitting torque using a drag torque on the energy consuming element in each pulley to give a self-energized actuator drive. Since very high ratios (greater than 100:1) may be used, very small torques (and hence small powers) transmitted to the power absorbing element achieves the generation of large actuator torques to be applied in positioning the belt engaging elements of the continuously variable transmiεεion pulleys. In Fig. 5, the pulley assembly 40 may be deemed the driving pulley (which receives torque via the input shaft 42 from an external power source not shown such as an engine or motor) and the pulley assembly 41 may be deemed the driven pulley which receives power via the flat belt 43 which is applied to an output shaft 44. As may be ap- preciated from the manner in which Fig. 5 is cross hatched, the inner guideway diskε 45 of the pulley asse - bly 40 are physically connected together to effect an in- ner guideway disk structure. Similarly, the outer guide- way disks 46 of the pulley asεembly 40 are also connected together to effect an outer guideway disk structure. The inner guideway diskε 47 and the outer guideway diεks 48 of the driven pulley assembly 41 are similarly connected. Briefly comparing Fig. 5 to Figs. 2 and 3, the inner guideways diskε 45 correspond to the inner guideway disks 21, 22; the outer guideway diskε 46 correspond to the outer guideway diεkε 19,- 23; and the directions of rota- tion are as indicated by the arrows in each Fig. to es- tabliεh "the correct relationship between the components, as discussed more fully in the previously referenced United States patent application Serial No. 871,254 (now United States Patent 4,714,452 ) _f which contribute to the correct operation of the subject invention. A four element harmonic drive 50 differentially in- terconnects the outer guideway disks 46, the inner guide-

way diskε 45 (which are fastened to the εhaft 42) , and an output drive 54 of the pulley assembly 40 to a power con- suming element εuch aε an oil pump 55. Similarly, as to the pulley assembly 41, a four element harmonic drive generator 51 differentially connects the outer guideway diskε 48, the inner guideway disks 47 (which are fastened to the shaft 44) , and an output drive 56 which is coupled to a second power consuming element εuch as an oil pump 57. In the driving pulley assembly 40, the inner guide- way diskε 45 are connected to the dynamic εplinε 60 of the harmonic drive 50, and the outer guideway diεkε 46 are connected to the circular εpline 61. The inner guideway diεkε 45 are connected to the shaft 42 by a col- lar 52 over one element of the outer guideway diskε 46. The output drive 54 between the oil pump 55 and the har- manic drive 50 iε coupled to the wave generator 58. Th.e driven pulley aεsembly 41 is similarly config- ured, but the positions of the dynamic spline and the circular spline are reversed. Thuε, the circular εpline 62 iε connected to the inner guideway diεkε 47, and the dynamic εpline 63 iε connected to the outer guideway diεkε 48. The inner guideway diεkε 47 are connected to the εhaft 44 by collar 53 over one element of the outer guideway diεkε 48. The output drive 56 to the oil pump 57 iε coupled to the wave generator 59 of the harmonic drive 51. Any constant radial position for the belt engaging elements 64 (driving pulley assembly 40) or 65 (driven pulley assembly 41) results in all components of the har- onic drive for that pulley assembly (i.e., the wave gen- erator, the flexspline, the dynamic spline, and the cir- cular εpline) rotating at the same speed as the pulley shaft. Hence, the power consuming element (the oil pumps 55 or 57 in Fig. 5) rotates at a εpeed proportional to the εhaft speed producing a hydraulic oil flow whose pressure output (against which it works) can be changed by a control valve. It can be shown that the drag torque

of the positive displacement oil pump used as the power consuming element is substantially proportional to the generated oil pressure. Thus, the actuator torque in the pulley can be maintained constant at any belt drive ra- dius and pulley εpeed by holding the poεitive displace- ment pump discharge pressure constant. Consequently, the pulley speed ratio may be changed by increasing the actu- ator torque on one pulley versus the other. For example, if a higher output speed is desired for a given input speed, increasing the discharge pressure on the oil pump 55 would cause the wave generator 58 to transiently rotate more slowly relative to the εhaft 42 causing the circular spline 61 connected to the outer guideway diskε 46 to rotate more slowly relative to the inner guideway diskε 45 connected to the shaft 42. The movement of the inner guideway disks relative to the outer guideway diskε cauεeε the interεections of the guideways, and hence the belt engaging elements 64, to move radially outwardly in the driving pulley asεembly 40. ^■ With a fixed length belt, thiε can only happen if the belt drive radiuε in the driven pulley assembly 41 simultaneouεly decreases. An increase in belt tension due to the increase in actuator torque in driving pulley assembly 40 will result in increasing the torque on the power consuming unit, oil pump 57, in the driven pulley asεembly 41 subsequently increasing the rotational speed of the oil pump 57 as the belt drive radius of driven pulley asεembly 41 is decreasing. Torque can be transmitted in either direction through the harmonic drives 50, 51 although the drive ef- ficiency iε a few percentage pointε lower when transmit- ting power from the circular spline or dynamic spline to the wave generator as compared to the reverse case. This is not a substantial concern in the subject system since operation at any constant speed ratio gives very low power losses in the transmisεion of power to the output power consuming elements (oil pumps 55, 57) since, again, there is no change in. the relative position of any ele-

mentε in the pulleys or harmonic drives during operation at a constant transmission ratio. When a speed change occurs, the temporary increase in power loss iε inversely dependent upon the time that it takes for the speed ratio change to be completed. Typically, the angular movement of the outer guideway diskε relative to the inner guide- way diεkε of a given pulley for maximum radiuε ratio change (εpeed ratio change) iε about 100° of angular εhift. Hence, if thiε change occurs in one second (typical of a maximum speed change) , this corresponds to a 16.7 RPM speed of the inner guideway diskε relative to the outer guideway diεkε for that period. Given an exe - plary 100:1 harmonic drive, a temporary change ¹ in the power consuming unit of 1667 RPM is brought about. In- creasing the rotational rate of the drive to the oil pump 57 by 1667 RPM for one second will cause the output pul- ley aeεembly ^" 41 to increase in speed and require a de- crease in the oil pump 55 in the driving pulley 40 by 1667 RPM for the same one second period using pulleys of the same size. The discharge oil pressure from the oil pump 55 in the driving pulley asεembly 40 may be increaεed by re- εtricting its output control valve flow area to effect such a decrease in the oil pump 55 speed and provide the higher actuator torque to move the belt engaging elements 64 radially outwardly. The εimultaneouε increase in the speed of the oil pump 57 of the driven pulley 41 can si- multaneously be aided by opening ^' itε output control valve flow area. In an automotive application for the -continuously variable transmission, the output shaft 44 of the driven pulley asεembly 41 iε geared directly to the vehicle wheelε εo that the rotational rate of the driven pulley assembly is directly proportional to the vehicle speed. The inertia of a vehicle does not permit the absolute value of the driven pulley asεembly'ε 41 εpeed to in- creaεe very rapidly (i.e., in a εecond or two). However, the more critical operation for an automotive continu-

ousl variable transmission installation consists of ob- taining maximum vehicle acceleration at any drive speed. This correspondε to rapidly accelerating the engine con- nected to the driving pulley assembly 40 to a higher εpeed to give more power to accelerate the vehicle. In εuch a caεe, the discharge oil pressure from the oil pump 57 of the driven pulley assembly 41 would be increased by restricting its output control valve flow area to effect a decrease in the oil pump 57 speed and give higher actu- ator torque to move the belt engaging elements 65 radi- ally outwardly. The simultaneouε increaεe in the εpeed of the oil pump 55 of the driving pulley assembly 40 can also be aided by opening its output control valve flow area. The overall resulting' output torque from the con- tinuouεly variable transmission versus rime depends on many factors, but chiefly the inertia of the engine . _.. and the other components and the controlled output pressures of the oil pumps 55, 57. The simultaneous rate of increase in the speed of the oil pump 55 or 57 can be 'aided by operating the ac- celerating oil pump as a motor for a brief time duration (e.g., a second to a few εecondε for any one εpeed change) . Thiε feature can be incorporated by transiently supplying oil to the oil pump functioning as a motor from an independent boost pump, under control of an appropri- ate solenoid valve, for the very brief duration required to accommodate the rapid engine input speed acceleration or deceleration. The boost pump volume flow must be ade- quate for the maximum flow requirement, but the boost presεure can be relatively low "thuε keeping the motor power requirement to about 5-10% of the minimum oil pump power. The control system must always maintain adequate belt tension to prevent slippage during εpeed ratio changes. One or more circumferential springs (represented schematically by the springs 63, 69 in Fig. 5) may also be incorporated in each actuator drive oriented to give a torque between the inner guideway disks and outer guide-

way diskε whose direction would cause the belt engaging elements to be moved radially inwardly in the driving pulley asεembly 40 and radially outwardly in the output or driven pulley aεεembly 41. This arrangement permits the continuously variable transmisεion to εtart operation at a maximum output torque to input torque ratio with the belt under εome tenεion at all times to avoid initial slippage. Such springε alεo are helpful in obtaining a maximum εpeed increaεe of the driving pulley aεεe bly 40 relative to the driven pulley asεembly 41 as desired dur- ing the acceleration of the vehicle; i.e., if the oil presεure on the oil pump 55 iε reduced εufficiently, the circumferential spring torque must help drive the belt engaging elements 64 radially inwardly permitting very rapid engine acceleration. However, the specific desired changeε aε applied to the automotive continuouεly vari- able transmiεεion application normally require that there iε no net loεs in output torque during operator-demanded vehicle acceleration that has a duration of more than about 0.1 second. As a result, the rate and magnitude of output presεure changeε in the oil pumps 55, 57 have the major effect. Consider now the exemplary hydraulic control εubsys- tern which determines the input and output preεεureε of the oil pu pε 55, 57 for the reaεons previously diεcuεεed to control the εpeed ratio between the driving pulley as- sembly 40 and the driven pulley assembly 41 via the belt 43. all as shown in Fig. 6. Oil from a reservoir 70 is supplied, through conduit 71 and check valves 72, 73, to the suction εideε 74, 75, respectively, of oil pumps 55, 57. The presεure sides 76, 77, respectively, of the oil pumps 55, 57 are connected, by reεpective conduitε 78, 79 _r back to the reservoir 70 after the oil hae passed through a cooler 80. The conduit 78 includes an inline pressure sensor 81 and, downstream of the pressure sensor 81, a flow rate control valve 82. Similarly, the conduit 79 includes an inline pressure sensor S3 and a flow rate control valve 84 downstream from the pressure sensor S3.

The status of the flow rate control valves 82, 84 iε determined by outputs from a control module 86 which re- ceives input information from the pressure sensors 81, 83 and from external sources εuch aε engine εpeed and throt- tie poεition. When the control module senses that the pressures existing in the conduits 78 and 79, from the sensors 81, 83 (and hence the transmission ratio between the pulley assemblies 40, 41), are incorrect for sensed engine speed and throttle position conditions (as, for example, during rapid acceleration) , the control module responds by increasing the flow area through the flow rate control valve 82 and decreasing the flow area through the flow rate control valve 84 which momentarily slows the oil pump 57 and speeds the oil pump 55 until the effective radius of the driving pulley assembly 40 is decreased and that of the driven pulley assembly 41 is increased to ^" reach a new and correct system balance point. Also, when decelerating and using the engine as a partial vehicle brake, the control module 86 causes the flow rate through" the flow rate control valve 84 to de- crease and that through the flow rate control valve 82 to increase, momentarily slowing the oil pump 57 and accel- erating the oil pump 55 until a syεtem balance iε again achieved. As previously discuεεed, under certain conditions, the response speed of the oil pump 55 or 57 whose speed is being increased can be facilitated by permitting it to operate briefly as a motor. This ^' feature is accomplished with the addition of a low pressure pump 88 driven by a motor 89. The pump 88 is supplied from the reservoir 70 and is connected, via conduits 90, 91, 92 to the suction sides 74, 75 of the oil pumps 55, 57, respectively. A solenoid operated valve 93 in line in the conduit 91, when opened under the influence of the control module 86, supplies an extra volume of oil to the suction side 74 of the pump 55 to permit its transient operation as a motor. Similarly, solenoid operated valve 94, in line in the conduit 92, when opened under the influence of the con-

trol module 86, permits the pump 88 to supply an extra volume of oil to the suction side 75 of the pump 57 to permit itε tranεient operation aε a motor. The pump 88/motor 89 and the εolenoid operated valves 93, 94 need only be operated when rapid tranεmiεεion ratio changes are undertaken and only facilitate that rapid change. Thus, it will be appreciated that the portion of the hy- draulic circuit including the pump 88, the conduits 90, 91, 92 and the solenoid operated valves 93, 94 is op- tional and may not be required for all systems. The preceding description of the operation and spe- cific connectionε between the inner guideway diεcε and dynamic spline or circular εpline and alεo between the outer guideway diεcs and circular spline or dynamic spline for the input and output power pulleys assume the use of relatively light weight belt engaging elements. Such belt engaging elements may be made typically of alu- minum, ceramic or plaεtic for low weight. The belt en- gaging ele entε are acted upon by centrifugal force cauεed by pulley rotation and, due- to their poεitionε in the guideway discs, give torques operating to counter ro- tate the two sets of discε in each pulley in the direc- tion that moves the belt engaging elementε radially out- wardly. The effect of the centrifugal force acting on the relatively light weight belt engaging elementε of each pulley is usually inεufficient to require provision for compensation in the control εyεtera. However, it iε deεirable to conεider the uεe of more denεe belt engaging elements (i.e., those made of steel, powdered steel alloys, etc.), which have fatigue strengths much higher than aluminum, ceramic or plastic, as the design pulley power is increased. At high rota- tional pulley speeds, the torque generated by the cen- trifugal force of the heavier belt engaging elements may be balanced by the torque of a hydraulic pump through the differential gearing of a harmonic drive provided only that the specific connections between the dynamic spline or circular εpline and the inner or outer guideway discs

are reversed from those shown in the preceding descrip- tion in which the uεe of relatively light weight belt en- gaging elementε was assumed. Thus, referring now to Fig. 7 which illustrates a flat belt continuously variable transmission employing belt engaging elements 164, 165 of significant weight, the driving pulley assembly 140 has its outer guideway discs 146 connected to the dynamic spline 160, and the inner guideway discε 145 are connected to the circular εpline 161. Similarly, the driven pulley assembly 141 haε itε outer guideway discs 148 connected to the circu- lar spline 162, and the inner guideway discε 147 are con- nected to the dynamic spline 163. It will again'be noted that these relationships are reversed from those shown and discussed with reference to the embodiment illus- trated in Fig. 5. As the operating speedε of the pulley assemblies 140, 141 decrease, the centrifugal forces acting on the belt engaging elementε 164, 165 decrease with the square of the speed change. . Thus, at lower pulley assembly speeds, there is lesε torque on the guideway discs from this source and correspondingly less tendency to cause the belt engaging elements to move radially outwardly. Therefore, at lower speedε,. it' iε necessary to operate the hydraulic units 155, 157 connected to the wave gener- ators 158, 159 of the harmonic driveε aε motors rather than aε pu pε in order to ^' provide the necessary torque tending to move the belt engaging elements outwardly to obtain the required belt tension for transferring power between the pulleys without slippage. In a given contin- uously variable transmission employing the heavier belt engaging elements, the pulley asεembly εpeed below which it is necessary to operate each of the hydraulic units 155, 157 as a motor rather than as a pur.p is dependent chiefly on the weight of the belt engaging elements 164, 165, the pulley torque and the overall effective friction factor between the belt 143 and the belt engaging ele- ents. However, εince the hydraulic flow rate to a fixed

displacement hydraulic motor varies directly with its εpeed, the required auxiliary supply pump flow and power required are minimized by operating only in the lower speed regime of the pulleys. The hydraulic control subsystem in this arrangement for using heavier, stronger belt engaging elements there- fore requires an auxiliary pump to operate continuously at the lower pulley asεembly εpeedε to εupply the re- quired hydraulic flows to the hydraulic units connected to the wave generators in the harmonic drives on the pul- leys. (It may be noted that hydraulic flow requirements for a constant pulley torque will decrease as the speed decreases.) An exemplary hydraulic control subsystem for per- forming this function is illustrated in Fig. 8. While this embodiment of the hydraulic control subsystem bears a substantial resemblance to that illustrated in Fig. 6, many of the signals and subsystem responses are reversed as a consequence of the differences in the mechanical configurations illustrated in Figs. 5 and 7 as discussed immediately above. However, the primary functions of the hydraulic control subsystem illustrated in Fig. 8 remains the determination of the input and output pressures of the hydraulic units 155, 157 to control the speed ratio between the driving pulley assembly 140 and the driven pulley assembly 141 via the belt 143. Thus, oil from a reservoir 170 is supplied, through conduit 171 and check valves 172, 173, to the suction sides 174, 175, respectively, of hydraulic units 155, 157. The pressure sides 176, 177, respectively, of the hydraulic units 155, 157 are connected, by respective conduits 178, 179, back to the reservoir 170 after the oil has passed through a cooler 180. The conduit 178 in- eludes an inline pressure sensor 181 and, downstream of the presεure sensor 181, a flow rate control valve 182. Similarly, the conduit 179 includes an inline preεsure sensor 183 and a flow rate control valve 184 downstream from the pressure sensor 183. A low presεure pump 188,

driven by a motor 189, is supplied from the reservoir 170 and is connected, via conduits 190, 191, 192 to the suc- tion sides 174, 175 of the hydraulic units 155, 157, re- spectively. A εolenoid operated valve 193 in line in the conduit 191, when opened under the influence of the control mod- ule 186, εupplieε an extra volume of oil to the εuction εide 174 of the hydraulic unit 155 to permit itε opera- tion as a motor. Similarly, solenoid operated valve 194, in line in the conduit 192, when opened under the influ¬ ence of the control module 186, permits the pump 188 to supply an extra volume of oil to the suction side 175 of the hydraulic unit 157 to permit its operation as a o- tor. Sensorε 195, 196 (for which there are no equivalent components in the hydraulic control subsyεte illuεtrated in Fig. 5) are operatively disposed, respectively, be- tween the solenoid operated valve 193 and the suction side 174 of the hydraulic unit 155 and between the solenoid operated valve 194 and the suction side 175 of the hydraulic unit 157. The pump 188/motor 189 and the solenoid operated valves 193, 194 need only be operated wherPthe speed of the pulley assemblies 140, 141 fall be- low a rotational rate at which the centrifugal forces ex- erted by the belt engaging elements 164, 165 (Fig. 7) fall below a value at which they are the dominant forces. The status of the flow rate control valves 182, 184, 193, 194 is determined by outputs from a control module 186 which receives input information from the pressure sensors 181, 183, 195, 196 and from external sources such as engine speed and throttle position. At operating speeds at which centrifugal force becomes a substantial factor, when the control module 186 senses that the pres- sures existing across the hydraulic units 155, 157 (and hence the transmisεion ratio between the pulley asse - blieε 140, 141), are incorrect for sensed engine and throttle position conditions (as, for example, during rapid acceleration) , the control module responds by de- creasing the flow area through the flow rate control

valve 182 and increasing the flow area through the flow rate control valve 184 which momentarily slows the hy- draulic unit 155 and speeds up the hydraulic unit 157 un- til the effective radius of the driving pulley aεεembly 140 is decreased and that of the driven pulley asεembly 141 iε increaεed to reach a new and correct system bal- ance point. Also, when decelerating and using the engine as a partial vehicle brake, the control module 186 causes the flow rate through the flow rate control valve 184 to increase and that through the flow rate control valve 182 to decrease, momentarily speeding the hydraulic unit 157 and εlowing the hydraulic unit 155 until a εyεte balance iε again achieved. Aε previouεly mentioned, when the rotational speeds of the pulley assemblies 140, 141 fall below a value at which the centrifugal forces exerted by the belt engaging elementε 164, 165 are εubεtantial, the hydraulic control εyεtem illuεtrated in Fig. 8 must revert to operation similar to that of the hydraulic control syεtem illuε- trated in Fig. 6. Thiε iε achieved by energizing the o- tor 188/pump 189 and additionally governing the flow ar- eaε through the flow rate control valveε 193, 194, to provide preεεureε appearing on the εuction εideε 174, 175 of the hydraulic units 155, 157 (as senεed by the sensors 195, 196) such that the hydraulic units operate aε motors rather than pumps in thiε lower εpeed range. The result is that, in this lower speed range, the hydraulic torque input (albeit reversed from the drags diεcussed in con- junction with the embodiment illustrated in Figs. 5 and 6, exerted by the hydraulic unitε 155, 157) become the sole forces determining the transient adjustment of the relative positionε of the inner and outer guideway diεcs in the pulley assemblieε 140, 141 and hence the transfer speed ratio between them. This operation may be summarized as follows: When the rotational speed of a pulley assembly is above a value at which the centrifugal force exerted by the belt engaging eler.entε it carrieε becomes a dominant force,

the control subsystem operates the corresponding hy- draulic unit to effect a change in the angular relation- ship between the firεt and second guideway diskε to cause a tendency for change in the radial positions of the belt engaging elementε opposite to the tendency for change in the radial positionε due to centrifugal force. . Con- versely, when the rotational speed of a pulley assembly is below a value at which the centrifugal force exerted by the belt engaging elements it carries becomes a dominant force, the control subεyεtem operateε the corre- εponding hydraulic unit to effect a change in the angular relationship between the firεt and εecond guideway diεkε to cause a tendency for change in the radial positions of the belt engaging elements which are in the same direc- tion as the tendency for change in the radial positions due to centrifugal force. While the principles of the invention have now been made clear in illustrative embodiments, there will be i - mediately obvious to those skilled in the art many modi- ficationε of εtructure, arrangementε, proportionε, the elements, materials, and components, used in the practice of the invention which are particularly adapted for a specific environment and operating requirements without departing from those principles.