Method of Torque Phase Shift Control for an Electronic Automatic Transmission System

1. Field of the Invention

The present invention relates to an automatic transmission primarily intended for motor vehicle use, and more particularly, to a method of controlling the torque phase of a shift for a transmission that is controlled electronically and hydraulically.

2. Description of Related Art

Generally speaking, land vehicles require three basic components. These components comprise a power plant (such as an internal combustion engine) a power train and wheels. The internal combustion engine produces force by the conversion of the chemical energy in a liquid fuel into the mechanical energy of motion (kinetic energy). The function of the power train is to transmit this resultant force to the wheels to provide movement of the vehicle.

The power train's main component is typically referred to as the "transmission". Engine torque and speed are converted in the transmission in accordance with the tractive-power demand of the vehicle. The vehicle's transmission is also capable of controlling the direction of rotation being applied to the wheels, so that the vehicle may be driven both forward and backward.

A conventional transmission includes a hydrodynamic torque converter to transfer engine torque from the engine crankshaft to a rotatable input member of the transmission through fluid-flow forces. The transmission also includes frictional units which couple the rotating input member to one or more members of a planetary gearset. Other frictional units, typically referred to as brakes, hold members of the planetary gearset stationary during flow of power. These frictional units are usually brake clutch assemblies or band brakes. The drive clutch assemblies can couple the rotating input member of the transmission to the desired elements of the planetary gearsets, while the brakes hold elements of these gearsets stationary. Such transmission systems also typically provide for one or more planetary gearsets in order to provide various ratios of torque and to ensure that the available torque and the respective tractive power demand are matched to each other.

Transmissions are generally referred to as manually actuated or automatic transmissions. Manual transmissions generally include mechanical mechanisms for coupling rotating gears to produce different ratio outputs to the drive wheels.

Automatic transmissions are designed to take automatic control of the frictional units, gear ratio selection and gear shifting. A thorough description of general automatic transmission design principals may be found in "Fundamentals of Automatic Transmissions and Transaxles," Chrysler Corporation Training Manual No. TM-508A. Additional descriptions of automatic transmissions may be found in U.S. Pat. No. 3,631,744, entitled "Hydromatic Transmission," issued Jan. 4, 1972 to Blomquist, et al., and U.S. Pat. No. 4,289,048, entitled "Lock-up System for Torque Converter," issued on Sept. 15, 1981 to Mikel, et al. Each of these patents is hereby incorporated by reference.

In general, the major components featured in such an automatic transmission are: a torque converter as above-mentioned; fluid pressure-operated multi-plate drive or brake clutches and/or brake bands which are connected to the individual elements of the planetary gearsets in order to perform gear shifts without interrupting the tractive power; one-way clutches in conjunction with the frictional units for optimization of power shifts; and transmission controls such as valves for applying and releasing elements to shift the gears (instant of shifting), for enabling power shifting, and for choosing the proper gear (shift point control), dependent on shift-program selection by the driver (selector lever), accelerator position, the engine condition and vehicle speed.

The control system of the automatic transmission is typically hydraulically operated through the use of several valves to direct and regulate the supply of pressure. This hydraulic pressure control will cause either the actuation or deactuation of the respective frictional units for effecting gear changes in the transmission. The valves used in the hydraulic control circuit typically comprise spring-biased spool valves, spring-biased accumulators and ball check valves. Since many of these valves rely upon springs to provide a predetermined amount of force, it will be appreciated that each transmission design represents a finely tuned arrangement of interdependent valve components. While this type of transmission control system has worked well over the years, it does have its limitations. For example, such hydraulically controlled transmissions are generally limited to one or a very small number of engines and vehicle designs. Therefore, considerable cost is incurred by an automobile manufacturer to design, test, build, inventory and repair several different transmission units in order to provide an acceptable broad model line for consumers.

Additionally, it should be appreciated that such hydraulically controlled transmission systems cannot readily adjust themselves in the field to compensate for varying conditions such as normal wear on the components, temperature swings and changes in engine performance over time. While each transmission is designed to operate most efficiently within certain specific tolerances, typical hydraulic control systems are incapable of taking self-corrective action on their own to maintain operation of the transmission at peak efficiency.

However, in recent years, a more advanced form of transmission control system has been proposed, which would offer the possibility of enabling the transmission to adapt itself to changing conditions. In this regard, U.S. Pat. No. 3,956,947, issued on May 18, 1976 to Leising, et al., which is hereby incorporated by reference, sets forth a fundamental development in this field. Specifically, this patent discloses an automatic transmission design which features an "adaptive" control system that includes electrically operated solenoid-actuated valves for controlling certain fluid pressures. In accordance with this electric/hydraulic control system, the automatic transmission would be "responsive" to an acceleration factor for controlling the output torque of the transmission during a shift from one ratio of rotation (between the input and output shafts of the transmission) to another. Specifically, the operation of the solenoid-actuated valves would cause a rotational speed versus time curve of a sensed rotational component of the transmission to substantially follow along a predetermined path during shifting.

3. Objects of the Present Invention

It is one of the principal objects of the present invention to provide a significantly advanced electronically controlled transmission which is fully adaptive. By fully adaptive, it is meant that substantially all shifts are made using closed-loop control (i.e. control based on feedback). In particular, the control is closed loop on speed, speed ratio, or slip speed of either N.sub.t (turbine of the torque converter) and N.sub.e (engine) or a combination of N.sub.t and N.sub.o (outPut) which will provide the speed ratio or slip speed. This transmission control is also capable of "learning" from past experience and making appropriate adjustments on that basis.

Another object of the present invention is to provide an automatic transmission in which the shift quality is maintained approximately uniform regardless of the engine size, within engine performance variations or component condition (i.e. the transmission control system will adapt to changes in engine performance or in the condition of the various frictional units of the transmission).

It is a more specific object of the present invention to provide a method of matching torque build-up of an apply element to the torque fall-off of a release element for shifting an automatic transmission.

It is a further object of the present invention to achieve exceptionally smooth, yet quick kickdown shifts (i.e. second to first gear), and in so doing, make any power train feel more responsive without increasing harshness. Being adaptive, these controls will be capable of compensating for changes in engine or friction element torque, and provide consistent shift quality over the life of the transmission.

This application is one of several applications filed on the same date, all commonly assigned and having similar Specification and Drawings, these applications being identified below.

Commonly assigned application Ser. No. 07/187,772, filed Apr. 19, 1988 now U.S. Pat. No. 4,875,391 has been printed in its entirety. The Figures and the entire Specification of that application are specifically incorporated by reference. For a description of the above copending applications, reference is made to the above mentioned U.S. Pat. No. 4,875,391.

To achieve the foregoing objects, the present invention provides a comprehensive four-speed automatic transmission system. While this transmission system particularly features a fully adaptive electronic control system, numerous other important advances are incorporated into this unique transmission system, as will be described below in detail.

The transmission control system includes a microcomputer-based controller which receives input signals indicative of engine speed, turbine speed, output speed (vehicle speed), throttle angle position, brake application, predetermined hydraulic pressure, the driver selected gear or operating condition (PRNODDL), engine coolant temperature, and/or ambient temperature. This controller generates command or control signals for causing the actuation of a plurality of solenoid-actuated valves which regulate the application and release of pressure to and from the frictional units of the transmission system. Accordingly, the controller will execute predetermined shift schedules stored in the memory of the controller through appropriate command signals to the solenoid-actuated valves and the feedback which is provided by various input signals.

Another primary feature of the present invention is to provide an adaptive system based on closed-loop control. In other words, the adaptive control system performs its functions based on real-time feedback sensor information, i.e., the system takes an action which affects the output, reads the effect, and adjusts the action continuously in real-time. This is particularly advantageous because the control actuations can be corrected as opposed to an open loop control in which signals to various elements are processed in accordance with a predetermined program.

In addition to the advantages offered by the adaptive control system, the present invention provides a method of determining the apply rate of an apply element and adjusting the same to match the torque build-up of the apply element to the torque fall-off of the release element during shifting of the transmission.

Other objects, features and advantages of the present invention will become more fully apparent from the following detailed description of the preferred embodiment, the appended claims and in the accompanying drawings in which:

FIGS. 24A-L illustrate the shift methodology according to the present invention;

FIG. 24A is a shift graph tape for a third to first gear kickdown shift;

FIG. 24B is a graph of torque converter characteristics for the turbine torque methodology;

FIG. 24C is a partial shift tape graph of the learn methodology for kickdown shifts;

FIG. 24D is a shift tape graph for a first to second gear upshift;

FIG. 24E is a flow chart of the learn methodology;

FIG. 24F is a graph for the adaptive scheduling methodology for a fourth to third gear coastdown shift;

FIG. 24G is a phase plane graph for a second to first gear coastdown shift;

FIG. 24H is a partial shift tape graph for a second to first gear coastdown shift;

FIG. 24I is a flow chart of the release element logic for a kickdown or coastdown shift;

FIG. 24J is a flow chart of the apply element logic for a kickdown or coastdown shift;

FIG. 24K is a graph of pressure versus time for an element; and

FIG. 24L is a shift tape graph for a neutral to reverse gear garage shift.

The present provides fully adaptive electronic transmission controls. These adaptive controls perform their functions on real-time feedback sensor information, as is likewise performed by electronic antiskid brake controls. Additionally, the adaptive controls "learn" particular information by monitoring data such as the value for the fill time and apply rate of the applying element such as a clutch. This information is then stored in the memory of the transmission controller 3010 for future use.

The upshift methodology uses the learned values for the fill time and apply rate (torque build-up rate) of the ON-coming or applying element such as a clutch and for the release time of the OFF-going or releasing element such as a clutch. Learning apply element fill time permits timing the beginning-of-fill so as to compensate for orifice size or clutch clearance variations, etc. Learning the apply rate and release time allows compensation for variations in orifice size, clutch capacity, solenoid response, and torque requirement (included to some extend, different engines). Although learning is restricted to the HOT mode, some temperature compensation occurs between summer and winter.

In the power-ON upshift, the methodology adjusts the apply and release events so that release element slip occurs just before the apply element begins to develop torque. Slip must be allowed to occur so that the release and apply events can be identified by speed measurements. In simplified terms, release time is measured as the interval between initial-element-vent and beginning-of-slip; fill time is from beginning-of-fill to end-of-bump-along; and apply rate is end-of-bump-along to beginning-of-speedchange. Bump-along is a term coined to describe the bang-bang (fixed percent ON-OFF time) control period that the methodology uses to maintain a small amount of backward slip prior to the application of the apply element. The methodology delays either the beginning of the release vent or the beginning of apply fill so as to achieve approximately one cycle of bump-along.

This control methodology reduces the release element pressure to the minimum that will still support the input torque reaction, thereby establishing the optimum beginning conditions for the element exchange. The apply rate, then, is selected to develop the torque needed to begin speed change just as release element net-apply-pressure reaches zero. Thus, the duty-cycle-controlled apply rate matches the ballistic release rate of the OFF-going or releasing element. The purpose of the matched exchange, of course, is to minimize fight between the elements. Release time and apply rate are both learned relative to the throttle angle.

Once the speed change begins, the apply element pressure is controlled to provide the desired turbine acceleration alpha.sub.t. This speed change control is the heart of adaptive control since it adapts to changes in engine torque, friction coefficient, etc. and provides consistent control.

The acceptability of the power-OFF upshift must be verified with the appropriate engine control methodology. With carburetors, the engine speed N.sub.e drops quickly and may pull the turbine 128 through the speed change faster than desired. This can result in both elements going or staying off, which, if the throttle is opened, will result in a runaway bump as both elements race to apply. In order to prevent this, methodology was devised which uses elements to control the speed change and gradually accomplish the hand-off. With the electronic engine control, the engine may vary between power-ON and power-OFF. It may even provide the correct programmed rate past the target speed (N.sub.t =N.sub.j) without either element controlling, thus defeating the above approach. Methodology has been added which simply turns ON the apply element when the speed change is complete. Also, with this shift, it is desirable to release the OFF-going element quickly to avoid excessive engine braking torque.

The learned term for fill time is time remaining to nearly fill, T.sub.f. Using T.sub.f minimizes the possibility of a too aggressive element application and allows the use of a duty cycle to soften the initial application. T.sub.f is actually calculated from the learned clutch fill volume, V.sub.f. This term is stored in cubic inches so that different flow rates may be used to calculate T.sub.f. This allows the same learned volume to be used for different shifts which may have a different line pressure. The program or methodology continually tracks the fluid volume needed to apply each element.

The learned term for release time, above, is time to nearly release, T.sub.r, which is calculated as K.sub.s *T.sub.s -0.063. T.sub.s is a table value for the nominal observed release time. K.sub.S is the learned multiplier for that individual package. Since T.sub.s varies with THR (i.e. engine torque), a multiplier provides the best data match for the variables being accommodated. The 0.063 seconds, together with the T.sub.f differential, provides a margin to ensure that fight is minimized.

For good kickdown feel, it is essential that the shift occur quickly. The use of accumulators delays the clutch or element release so every effort is made to minimize the accumulator fill/vent times. The methodology turns OFF the release element at the start of the shift and does not apply it again until turbine acceleration exceeds a desired level by a small amount. A duty cycle (DC) nay then be initialized and updated to provide proportional control for the speed change. The primary element DC acceleration or proportional control (DC.sub.alpha, i.e. variable percent ON or OFF time) initialization level is calculated for N.sub.e and N.sub.t, the torque converter characteristics, and the element capacity; each DC.sub.alpha update is based on an anticipated turbine acceleration (alpha.sub.t) error.

As illustrated in FIG. 24A, a shift tape of the transmission characteristics for a third to first (3-1) kickdown shift is generally shown at 2200. Curve 2202 represents throttle angle. As throttle angle increases, engine speed N.sub.e shown in curve 2204 also increases. Simultaneously, the release element is released as shown in curve 2206 to drop its torque capacity. In other words, for a third to first (3-1) gear kickdown shift, the overdrive clutch 304 is released at the start of the shift. As shown by curve 2208, the fluid pressure of the overdrive clutch 304 vents down. When the torque capacity of the overdrive clutch 304 is low enough (at the fill volume), the turbine 128 will breakaway and a speed change will start as indicated by numeral 2210.

The present invention limits the rate at which the turbine 128 accelerates. This is accomplished by calculating and comparing an instantaneous turbine acceleration (alpha.sub.t) shown in curve 2212 against a desired acceleration (alpha.sub.desired or .alpha.*) level shown in curve 2214. Once the speed change begins at 2210, the controller 3010 attempts to match alpha.sub.t approximately equal with alpha.sub.desired.

When alpha.sub.t exceeds alpha.sub.desired, the release element is reapplied to control the rate at which the turbine 128 accelerates. The release element is reapplied under duty cycle acceleration or proportional control (DC.sub.alpha) to provide a controlled slope of speed change on the turbine 128.

As illustrated in FIG. 24A, curve 2212 of alpha.sub.t crosses curve 2214 of alpha.sub.desired at point 2216. At point 2216, the overdrive clutch 304 is reapplied by duty cycling the solenoid-actuated valve 632 as shown by part 2218 of curve 2206.

Simultaneously with speed change, the kickdown methodology adaptively applies the applying element (low/reverse clutch 310) as shown by curve 2220 based on the remaining turbine speed change which has to occur. As turbine speed N.sub.t increases in curve 2222, the methodology compares the actual turbine speed N.sub.t to a target speed N.sub.j (for a 3-1 shift, first gear ratio of first gear N.sub.j). Because the speed change is made at a known rate (because controlling release element at that rate), the methodology can predict how much time remains to fill the applying element. The methodology attempts to get the applying element filled after achieving the target speed N.sub.j for a predetermined time period such as 120 ms, which region 2224 of curve 2214 is called "hold-speed".

When N.sub.t exceeds the target speed N.sub.j at point 2226, i.e. enters the hold-speed region 2224, alpha.sub.desired is lowered again to a negative value at point 2228 on curve 2214 so that the release element will prevent further increases in N.sub.t. DC.sub.bb is again used for improved response before reentering DC.sub.alpha control. The release element hold-speed continues until the apply element is filled, begins to develop torque, and pulls N.sub.t down to the target level, N.sub.j. The methodology then turns OFF the release element when N.sub.t equals N.sub.j.

To reduce the energy (and provide a faster shift), learning is used to limit the hold-speed period to the minimum that will accomplish the "apply" identification and improve "shift feel". To know whether to turn ON the apply element (i.e. cause the solenoid to apply), the methodology starts which a "hold-speed" time allowance and adds to that a time-to-complete-speed-change, which is calculated by (N.sub.j -N.sub.t /alpha.sub.desired. This "time available (from now until the element should apply)", is continuously compared to the "time required (element volume divided by fill rate)" and the solenoid-actuated valve is turned ON or OFF as required. Since the methodology tracks element volume during solenoid OFF and ON times, there is little error that can develop if alpha.sub.t is lower than alpha.sub.desired. When alpha.sub.t is low and the actual N.sub.t becomes lower than projected, the methodology simply turns OFF the element and waits for N.sub.t to catch up to the projected level. If alpha.sub.t is higher than alpha.sub.desired, the controls have no means to catch up, but since the initial release element vent time and the alpha.sub.desired "feather" control cause alpha.sub.t to be lower than alpha.sub.desired normal, there is no opportunity for significant "fall-behind" error to develop.

To achieve 120 ms. of hold-speed, the present invention utilizes "adaptive" kickdown start logic which is based on a "learned" fill volume of the applying element. The equation for the kickdown start logic may be defined as follows:

S is the kickdown start value (r.p.m. of turbine remaining) which equals t.sub.f multiplied by alpha.sub.desired. As illustrated in FIG. 24A, curve 2232 represents the kickdown start value S. t.sub.f is the time needed to fill the applying element to the level that will provide the correct amount of bump-along time or kickdown hold-speed time (i.e. 120 ms). It is continuously updated and includes compensation for any expected duty cycle use during the remaining fill time. t.sub.f is calculated as follows: ##EQU1## K.sub.f =DC COMPENSATION FACTOR: Corrects for the reduced fill rate when DC use is expected. K.sub.f =1 for kickdown shift

V.sub.f =fill volume of the applying element

Q.sub.f =flow rate of the applying element

M=correction factor for line pressure

V=instantaneous element volume

Since N.sub.j is the ratio multiplied by N.sub.i, N.sub.t can be controlled at a desired slope by the release element so that N.sub.t goes to N.sub.j during t.sub.f having 120 ms of hold-speed to completely fill the apply element. t.sub.f is continuously calculated to give the kickdown start value S. Whenever N.sub.t crosses S (i.e. N.sub.t >N.sub.j -S), the apply element is turned ON which reduces S because the apply element is filling. If N.sub.j -S>N.sub.t (i.e. N.sub.t falls below S), the apply element is turned OFF. This results in an irregular or variable DC on the apply element. In other words, once the kickdown start value S is calculated, the transmission controller 3010 compares N.sub.t to S. If N.sub.t is greater than N.sub.j minus S, the methodology instructs the transmission controller 3010 to turn ON the applying element to reduce S to zero. Hence, the methodology drives S to equal zero just as N.sub.t crosses or equals N.sub.j at point 2226. This allows 120 ms. of time remaining to complete the fill (hold-speed), resulting in better shift quality. Otherwise, the shift quality would be "jerky" if the apply element were applied just as N.sub.t crossed N.sub.j.

Referring to FIG. 24A, until alpha.sub.t crosses alpha.sub.desired for the first time at point 2216, the release element is held completely OFF so that any initial speed change is not delayed. Once the speed change is complete at point 2228, it is desired not to overshoot alpha.sub.desired. Therefore, a duty cycle is calculated that will hold or maintain alpha.sub.desired. The turbine torque calculation is used to calculate the initial percent ON time, indicated at 2216, for the duty cycle (DC) of the release element.

The initial percent ON time of the release element for either a downshift or garage shift is calculated as follows:

whereby,

DC.sub.o =Zero torque DC estimate

I.sub.t =Equivalent turbine inertia

K.sub.t =Gain, DC to turbine torque (T.sub.t)

In the above equation, DC.sub.o is the duty cycle needed to maintain fill pressure on the release element, which is predetermined value. I.sub.t .times..alpha..sub.desired is the net torque to maintain desired acceleration which is also a predetermined value. K.sub.t is the gain from the DC to the turbine torque which is a predetermined value. DC.sub.o, I.sub.t and K.sub.t vary for the shift involved, i.e. fourth to third gear, fourth to second gear, etc. The equation for turbine torque (T.sub.t) is defined below: ##EQU2##

As illustrated in FIG. 24B, the equation for the turbine torque (T.sub.t) is derived by the graph of turbine torque T.sub.t divided by engine speed N.sub.e squared (which is the same as impeller speed squared) versus speed ratio of turbine speed N.sub.t divided by engine speed N.sub.e which is curve 2280. For turbine speed N.sub.t less than a predetermined constant K.sub.3 times engine speed N.sub.e, the equation for turbine torque T.sub.t is indicated by part 2282 of curve 2280. For turbine speed N.sub.t equal to or greater than K.sub.2 multiplied by N.sub.e, the equation for turbine torque T.sub.t is indicated by part 2284 of curve 2280.

FIG. 24B is based on the characteristics of a particular model of torque converter. This can be used at any time that the lockup clutch is disengaged to calculate an input torque to the transmission 100. For a particular element involved (knowing what its capacity is), the transmission controller 3010 can calculate the DC necessary to provide the appropriate level of element torque (i.e. initial DC). After the initial percent ON time for the DC, the DC adaptively adjusts to maintain alpha.sub.desired.

The only learned quantity used for making downshifts is the fill time of the applying element or clutch. As previously mentioned, the element volumes are actually learned and stored. Fill times are calculated by using the learned element volume and an appropriate flow rate from a look-up table and graph of flow rate characteristics for each element for example. The learned volume information for a given element is shared between different shifts, both upshifts and downshifts. The flow rate used accounts for the individual hydraulic flow rates and compensates for line pressure differences which exist between different shifts (i.e. for element fill rates, not vent rates).

With a coastdown shift, however, the pump 200 will not, under all conditions, have the capacity to maintain the regulated line pressure. To compensate for the resulting low line pressure, a learned fill rate is used for coastdown shifts only. This fill rate is set at the regulated line pressure level with each start-up (because with cold fluid, the pump 200 will maintain the regulated pressure) and it will learn any reduction in the fill rate with each subsequent shift.

Learning fill time with downshifts is similar to upshifts in that the beginning of apply (end of fill time for the apply element) is identified by the ending of a "hold-speed" control maintained by the release element in power-ON shifts. Implicit with this is the necessity of establishing some "hold-speed" control rather than timing an exact application to be described herein. It is also necessary to handle OFF and ON times correctly since the fill event is seldom a continuous ON; the flow rates, mentioned above, provide this capability.

The learn logic for kickdown shifts tracts the instantaneous volume of the apply element and compares that value with the current fill volume such that the apply element is completely filled at the end of the hold-speed region.

As illustrated in FIG. 24C, curve 2250 represents a desired acceleration (.alpha.*) of the turbine 128. Curve 2252 represents turbine speed N.sub.t and curve 2254 represents a target speed (N.sub.j) of the turbine 128. Curve 2256 represents an instantaneous fill volume (V.sub.I) of the apply element and curve 2258 represents the current fill volume (V.sub.f) of the apply element. As N.sub.t approaches N.sub.j, N.sub.t comes within a predetermined range 2260 of N.sub.j. At point 2263 when N.sub.t reaches the lower limit of the predetermined range 2260, the learned volume (V.sub.L) of the apply element is latched at that volume of the instantaneous fill volume (V.sub.I). Once N.sub.t leaves the upper limit of the predetermined range 2260 at point 2264, the learned volume again tracks the instantaneous fill volume until N.sub.t enters the predetermined region 2260 at point 2266. At point 2266, the learned volume of the apply element is latched at that value of the instantaneous fill volume. At the end of the shift (EOS), the transmission controller 3010 takes a step out of current fill volume (V.sub.f) which is a percentage of the difference between V.sub.f and V.sub.L at point 2266.

The fill volume (V.sub.f) of the apply element is also "learned" and adaptively adjusted based on bump-along (i.e. element slip). As illustrated in FIG. 24D, a shift tape of the transmission characteristics is shown for a first to second (1-2) upshift. Curve 2270 represents the stored or previously learned current fill volume (V.sub.f) of the apply element. Curve 2272 represents the instantaneous volume (V.sub.I) of the apply element (i.e. two/four shift clutch 308). Curve 2274 represents the learned volume (V.sub.L).

While a shift is in progress, the learned volume (V.sub.L) is set equal to the instantaneous fill volume (V.sub.I) whenever (t.sub.f >0) or (t.sub.f =0 and N.sub.f >N.sub.j +30). As shown in FIG. 24D, V.sub.L tracts V.sub.I until point 2274 because t.sub.f was greater than 0. At point 2276, t.sub.f equals zero and V.sub.L stops tracking V and is set equal to the value of V.sub.I at point 2276. When t.sub.f =0, the apply element is filling in the hold-speed region. If N.sub.t is greater than N.sub.i plus a predetermined value such as 30 (i.e. slip occurs), called bump-along, V.sub.L is updated to the value of V.sub.I at point 2278. At point 2278, V.sub.L again tracks V.sub.I until N.sub.t is not greater than N.sub.i plus the predetermined value at point 2280. At point 2280, V.sub.L is set equal to the value of V.sub.I and stops tracking. This methodology is repeated whenever N.sub.t is greater than N.sub.i plus the predetermined value. At the end of the shift, the transmission controller 3010 compares V.sub.L to V.sub.f. If V.sub.L is greater than V.sub.f, as shown in FIG. 24D, V.sub.f is adjusted or increased a percentage of difference between V.sub.L and V.sub.f. If V.sub.L equals V.sub.f, no adjustment is made. Otherwise, if V.sub.L is less than V.sub.f, V.sub.f is decreased.

Referring to FIG. 24E, a flow chart of the learn methodology is shown. At the start of the methodology in bubble 2290, the methodology advances to block 2292. At block 2292, the methodology intercepts or determines the time to bump-along, time to speed change, and instantaneous volume during bump-along of the element. The methodology then advances to diamond 2294 and determined whether the shift has been completed. If no, the methodology loops back to block 2292. If the shift has been completed, the methodology advances to block 2296 and learns the fill volume if the conditions are valid, learns K.sub.s (release time multiplier), if conditions are valid and learns DC.sub.t (adjustment) if conditions are valid. From block 2186, the methodology returns.

The shift schedule (bubble 810 of FIG. 12) has logic which compares engine speed N.sub.e and target speed N.sub.j and delays any coastdown shift that would go from power-ON to power-OFF since these shifts involve crossing drivetrain backlash and may result in a "clunk". The 2-1 and 2-1 shifts are power-ON coastdowns (a 3-2 power-ON coastdown shift is not made); the 4-3 is typically a power-OFF shift (it may be power-ON if the shift is inhibited by the below "backlash" logic).

As illustrated in FIG. 24F, a graph of speed (r.p.m.) versus time is shown at 2300 for an adaptive fourth to third (4-3) gear coastdown shift. Curve 2302 represents the output speed N.sub.o or target speed N.sub.j for third gear. Curve 2302 represents the engine speed N.sub.e. Curve 2306 represents turbine speed N.sub.t.

If a shift is scheduled by the transmission controller 3010 when N.sub.e is less than N.sub.j, the start of the 4-3 shift will occur at point 2308. As the shift occurs, N.sub.t will increase and cross over N.sub.e, as indicated by point 2310, from positive to negative torque, resulting in a "clunk" of the drivetrain.

The present invention provides the feature of delaying or inhibiting the start of the shift by the transmission controller 3010 until N.sub.e is at least equal to or greater than N.sub.j, as indicated by point 2312. This is accomplished by delaying the actuation and/or deactuation (i.e. turning ON and/or OFF) of the appropriate solenoid-actuated valves. By inhibiting the shift, N.sub.t will remain less than N.sub.e during the entire shift, resulting in only positive torque and preventing any "clunk" of the drivetrain.

As illustrated in FIG. 24G, a phase plane graph of turbine acceleration (alpha.sub.t) versus turbine speed N.sub.t minus target N.sub.j (first gear) for a second to first (2-1) gear coastdown shift is shown at 2320. The solid line curve 2322 represents the desired acceleration (alpha.sub.desired or .alpha.*) which is a function of slip. Alpha.sub.desired goes to a negative value in the hold-speed region of the downshift.

The present invention provides methodology for controlling alpha.sub.t at point 2324 which is approximately 25 r.p.m. This is accomplished by using proportional control (DC.sub.alpha or DC.alpha.). DC.sub.alpha is used during coastdown shifts because real tight control hold-speed is needed and is lacking otherwise.

Referring to FIG. 24G, curve 2326 represents the vent release element (VRE) which is identified during a coastdown shift by alpha.sub.desired minus a predetermined value such as 1000. VRE is used where the applying element may be ON, or it is desired to vent the release element faster than normal DC.sub.alpha would (rather than backing off the release element's duty cycle by DC.sub.alpha, which would eventually release the element). If actual alpha.sub.t is below VRE curve 2326, as indicated by the arrow, the release element is turned OFF. This would result in actual alpha.sub.t coming back above the VRE curve 2326 if the apply element was not ON. Once alpha.sub.t was above the VRE curve 2326, the methodology would instruct the transmission controller 3010 to turn the release element ON. If the apply element was ON, alpha.sub.t would not come back above the VRE curve 2326.

Referring to FIG. 24G, curve 2328 represents hold the apply pressure (HAP). HAP is used where there is too much negative alpha.sub.t. In other words, HAP is used where alpha.sub.t is less than a predetermined value such as -1700. HAP prevents the apply element from applying hard quickly by duty cycling the apply element to maintain it at a predetermined pressure. This prevents the apply element from building up torque any faster in the hold-speed region, causing alpha.sub.t to come back above the HAP curve 2328.

As illustrated in FIG. 24H, a plot of actual turbine acceleration (alpha.sub.t) represented by curve 2330 and desired acceleration (alpha.sub.desired or .alpha.*) represented by curve 2332 is shown for a second to first (2-1) gear coastdown shift. A logic curve 2234 represents VRE and logic curve 2236 represents HAP. A plot of turbine speed N.sub.t represented by curve 2338, target speed N.sub.j represented by curve 2340, and output speed N.sub.o represented by curve 2342 is shown from the start to the end of the second to first gear coastdown shift. Logic curves 2344 and 2346 show the element logic for the release element (two/four shift clutch 308) and the apply element (low/reverse clutch 310), respectively.

Referring to FIG. 24H, the release element is ON until the start of shift at point 2348. At that time, the methodology turns the release element OFF. Simultaneously, the apply element which has been previously OFF is maintained OFF. Also, curve 2338 of N.sub.t is less than curve 2340 of N.sub.j.

After the start of shift at point 2348, alpha.sub.t starts to rise or increase. When alpha.sub.t crosses alpha.sub.desired at point A (wait until slip), the release element is turned ON or reapplied using duty cycle bang-bang (DC.sub.bb). DC.sub.bb is used until alpha.sub.t again crosses alpha.sub.desired at point B. Also, N.sub.t crosses N.sub.j at point B. At point B, the release element switches from DC.sub.bb to proportional control (DC.sub.alpha or DC.alpha.).

Referring to FIG. 24H, the apply element comes on before point B to be ready at the right time into hold-speed region (starts at point C). At point C, alpha.sub.desired enters the hold-speed region. The release element against switches to DC.sub.bb while the apply element is under DC.sub.alpha. If alpha.sub.t goes too far below alpha.sub.desired, VRE is applied as previously described. Alternatively, if alpha.sub.t is below the HAP value, HAP will be applied as previously described. Thus, N.sub.t is matched to N.sub.j and alpha.sub.t is matched to alpha.sub.desired at the end of the shift by using DC.sub.bb, DC.sub.alpha, VRE and/or HAP.

Referring to FIG. 24I, the methodology for the release element used during a coastdown or kickdown shift is generally shown at 2400. The methodology enters through bubble 2402 and advances to diamond 2404. At diamond 2404, the methodology determines whether the conditions are present indicating that the apply element is applying. In other words, are conditions present for VRE (i.e. THR<5.degree. and alpha.sub.t <alpha.sub.desired -1000). If that criteria is true, the methodology advances to block 2406 and vents the release element (applies VRE). The methodology then returns. If that criteria is not true, the methodology advances to block 2408 and establishes the phase of the shift: phase 1 equals the start; phase 2 equals the feather start (reduction in desired acceleration); and phase 3 equals target speed (hold-speed). This is accomplished by performing speed calculations and setting a flag for each phase of the shift. The methodology then advances to block 2410 and performs a pre-DC.sub.alpha flag check by setting the flag with slip and alpha.sub.t is HI or the release element is below fill volume, and clearing the flag with a change in the phase of the shift. The methodology then advances to block 2412 and performs a duty cycle.sub.alpha flag check. The methodology sets the DC.sub.alpha flag when the pre-DC.sub.alpha flag has been set and alpha is LOW (i.e. alpha.sub.t, high-to-low crossover) and it cleared with the change in phase of the shift. The methodology then advances to diamond 2414 and determines whether the DC.sub.alpha flag has been set. If the flag has been set, the methodology advances to block 2416 and uses DC.sub.alpha control or DC.sub.alpha on release element. DC.sub.alpha control is when the total period is fixed and the ON and OFF time is calculated and adjusted (i.e. variable ON and OFF time). The methodology then returns. If the flag has not been set, the methodology advances to diamond 2418 and determines whether alpha.sub.t is HI. If that criteria is true, the methodology advances to block 2418 and performs DC.sub.bang-bang control or DC.sub.bb on the release element and returns. DC.sub.bb control is when the total period is fixed and the ON and OFF time is fixed (e.g. at 60% ON). If that criteria is not true, the methodology advances to block 2420 and vents the release element and returns.

Referring to FIG. 24J, the methodology for the apply element is generally shown at 2450 for a coastdown or kickdown shift. The methodology enters through bubble 2452 and advances to diamond 2454. At diamond 2454, the methodology determines whether the phase of the shift is equal to one or two and N.sub.t is less than N.sub.j. If any of this criteria is true, the methodology advances to diamond 2456 and determines whether N.sub.t is above the speed associated with the correct apply timing (i.e. will element be late). In other words, the methodology determines whether N.sub.t is greater than S (kickdown start valve previously described). If that criteria is true, the methodology advances to block 2458 and applies the apply element and returns. If that criteria is not true, the methodology advances to block 2460 and vents the apply element and returns.

At diamond 2454, if any of that criteria is not true, the methodology advances to diamond 2462 and determines whether the apply element will apply within 120 ms if run at a predetermined duty cycle by looking at the fill volume (V.sub.f). If that criteria is not true, the methodology advances to block 2464 and applies the apply element and returns. If that criteria is true, the methodology advances to diamond 2466 and determines whether vehicle speed or N.sub.o is greater than a predetermined speed such as 8 mph and less than 300 r.p.m. of run away for the turbine 128. If that criteria is true, the methodology advances to block 2468 and applies the apply element and returns. If that criteria is not true, the methodology advances to diamond 2468 and determines whether conditions are present indicating apply element should "hold" (for a coastdown, alpha.sub.t very negative). In other words, the methodology determines whether the conditions are present to apply HAP (i.e. THR<5.degree. and .alpha..sub.t <-1700). If that criteria is true, the methodology advances to block 2470 and performs DC.sub.HAP on the apply element and returns. If that criteria is not true, the methodology advances to block 2472 and performs DC.sub.alpha2 (secondary element DC acceleration control) on the apply element and returns.

Another feature of the present invention used during a coastdown shift is a methodology called "wait-for-slip". At the beginning of the coastdown shift, the release element is vented. Whenever slip is present (i.e. N.sub.t .noteq.N.sub.j) and V.sub.I <V.sub.f for the release element and V<V.sub.f for the apply element, and THR.gtoreq.5.degree. or 2-1 or 3-2 or 3-1 shift is occurring, the methodology controls the release element at a low limit percent ON for its DC.sub.alpha. The methodology attempts to keep the release element from further venting because the release element may be needed to apply again. Once, the above conditions are no longer present, the release element continues to vent.

As illustrated in FIGS. 5A-L, the hydraulic system 600 includes accumulators 630,640, 642, 644 for the clutch assemblies 302,304, 308 and 310, respectively. The accumulators provide mechanical cushion so that extreme changes in pressure are not realized as the solenoid-actuated valves are turned ON or OFF. These accumulators help reduce the axial length of the transmission 100 and give more flexibility to the hydraulic system. This is advantageous over prior systems which used large cushion springs built in the clutch packs, increasing the axial length of the transmission.

As illustrated in FIG. 24K, a curve 2480 of pressure versus time for applying and venting (releasing) of an element or clutch is shown. The accumulator control zone, represented by part 2482 of the curve 2480, provides compliance or softness so that it takes time to develop a large change in pressure. Otherwise, if no accumulator was used, the slope of this part of the curve would be steeper and a small change in ON time would result in a large change in pressure, making torque capacity and shift quality unbearable.

In other words, control is performed in the accumulator control zone to prevent large excursions in the output torque (T.sub.o) which would create jerkiness or harshness in shift quality. For example, turning the release element ON during slip or bump-along without an accumulator would produce a steeper slope in the output torque, resulting in an inability to limit slip without harsh control.

The learned term for apply rate is torque phase duty cycle, DC.sub.t, The purpose of the torque phase duty cycle is to make the hand-off smooth between the release element letting go of torque and the apply element taking over torque. This is accomplished by timing the apply element to have sufficient capacity to start the speed change just as the release element capacity reaches zero. In other words, the methodology attempts to build-up apply element torque capacity to match torque fall-off capacity of the release element.

The torque phase duty cycle is adaptively adjusted to match torque build-up of the apply element to torque fall-off of the release element according to the following equation: ##EQU3## Where: THR=throttle angle

B=slip (40 r.p.m.)

The above equation is based on a table value, DC.sub.tt or nominal DC.sub.t values (fixed % ON time) based on throttle angle, plus a learned adjustment, DC.sub.ta. Since the intent is to have the speed change begin as the release element net-apply-pressure reaches zero, the methodology selects a DC.sub.t which will achieve the start of speed change at an interval after the start of venting of the release clutch. This interval is equal to the learned time to release at zero degrees throttle angle plus an allowance for one bump-along cycle. The transmission controller 3010 does this by achieving and maintaining t.sub.f equal to zero until slip occurs, then DC.sub.t is allowed to proceed.

Referring to FIG. 24D, curve 2500 represents the logic state of the release element. Curve 2502 represents slip in the transmission 100. At point 2504 on curve 2500, the release element is turned OFF or starts to vent. The interval between the start of vent at point 2504 until the start of speed change, which is point 2506 on curve 2502, is known as t* which is a predetermined value different for each upshift. Curve 2508 represents the logic state of the apply element. At point 2510 on curve 2508, the apply element is initially turned OFF or vented. At point 2510, t.sub.f is equal to zero and DC.sub.t starts for the apply element.

The slope of DC.sub.t is tailored so that it matches the build-up in apply element torque capacity. For throttle angles greater than 10.degree., the apply element is given a 10% boost in its duty cycle so that the actual turbine acceleration (alpha.sub.t or .alpha..sub.t) will achieve the desired acceleration (alpha.sub.desired or .alpha.*).

As illustrated in FIG. 24D, curve 2512 represents the desired acceleration (alpha.sub.desired) and curve 2514 represents the actual turbine acceleration (alpha.sub.t). At point 2506 on curve 2502, the speed change begins. Alpha.sub.t is greater than alpha.sub.desired. Therefore, DC.sub.ta adds 10% boost in ON time to DC.sub.t for the apply element such that alpha.sub.t will be momentarily equal to alpha.sub.desired at or near the end of DC.sub.t.

As shown and described above, DC.sub.ta is the learned adjustment to DC.sub.t. DC.sub.ta is used so that the start of the speed change from the initial release occurs within a predetermined time period called time to start speed change (t.sub.n). This time is when it is desired to have the speed change begin because the release element pressure will have decayed down to the fill pressure such that no torque capacity is on the element. Otherwise, if the speed change begins earlier or prior to this time, fight will occur because both the apply and release element have capacity. t.sub.n is defined as follows:

where:

t.sub.t =value of time `t` with N.sub.t .gtoreq.N.sub.i -B or previous value of t.sub.t with N.sub.t <N.sub.i -B

t.sub.v =value of time `t` at initial venting of release element or last occurrence of V.gtoreq.V.sub.f +V.sub.a for release element

Initially, DC.sub.ta is equal to zero (i.e. battery disconnect). Then, DC.sub.ta is defined as follows: ##EQU4## In the above equation, t.sup.*a is an adjusted value of t* (a predetermined table value) based on a learned value of K.sub.s. K.sub.s is used to predict where the first cycle of bump-along occurs because of changes in temperature. K.sub.s is used to adjust t* based on temperature so that start of DC.sub.t for the apply element occurs just prior to the first bump-along cycle.

Referring to the equation for DC.sub.t, a delta term is used when the transmission system has not learned out properly the above variables. If t* is less than the start of speed change at point 2506 on curve 2502, the % ON time for DC.sub.t is increased or incremented until the start of speed change begins at the end of .sub.t *. Thus, the delta term provides added protection by reacting immediately.

Referring to FIG. 24L, a shift tape representation of various characteristics of the transmission 100 is shown. Curve 2502 represents the logic state of the apply element and curve 2504 represents the logic state of the release element. Curve 2506 represents the desired acceleration (alpha.sub.desired) and curve 2508 represents the actual turbine acceleration (alpha.sub.t). Curve 2510 represents the pressure of low/reverse element and curve 2512 represents the pressure of the reverse element.

When the manual valve 604 is shifted to reverse R, the low/reverse element starts to vent. The low/reverse clutch solenoid-actuated valve 636 is turned OFF as indicated by point 2514 on curve 2502. The pressure in the low/reverse element starts to decrease or decay as shown by part 2516 of curve 2510. During this time, the reverse element is filling and the pressure starts to increase as shown by part 2518 of curve 2512. When the pressure in the low/reverse element has decayed to a fairly low level as indicated by point 2520 on curve 2510, the low/reverse element is reapplied under DC control at point 2520 on curve 2504.

The present invention has been described in an illustrative manner. It is to be understood that the terminology which has been used is intended to be in the nature of words of description rather than of limitation.

Obviously, many modifications and variations are possible in light of the above teachings. Therefore, the subject invention may be practiced otherwise than as specifically described.