2004 International Continuously Variable and Hybrid Transmission Congress
September 23-25, 2004
Control and Operating Behavior of
Continuously Variable Chain Transmissions
Roland Mölle
Division Mobile Working Machinery
Prof. Dr.-Ing. Dr. h.c. K.-Th. Renius c/o
Institute of Automotive Engineering
Prof. Dr.-Ing. B. Heißing
Technische Universität München
 Introduction – Chain-CVT
 Clamping Systems
 Ratio Control Design during Range
Shifts in Autarkic Hybrid
 Expanded Control Layout for Universal Use
in Chain Variator Applications
 Summary
Presentation Outline
Mölle
2004
 PIV Chain Converter
Chain
Secondary Pulley
Primary Pulley
Hydro-Mechanical Torque Sensor
Introduction
Mölle
2004
Typical CVT chain of a
modern passenger car:
Audi multitronic
Torque Capacity up to 300 Nm
Nominal Power 162 kW (V6-3.0)
Pull type Chain in Audi/LuK CVT
Mölle
2004
Model
Engine
Maximum
Torque
Transmission ratio
(Variator, Overall)
Type of Variator
Audi
A4
4-Zyl., 2,0 l, 96 kW
195 Nm
?
Pull Type Chain
Audi
A4
6-Zyl., 3,0 l, 162 kW
300 Nm
?
Pull Type Chain
Audi
A6
4-Zyl., 1,8 l, 110 kW
210 Nm
2,4 - 0,4; 6,0
Pull Type Chain
Audi
A6
6-Zyl., 2,4 l, 121 kW
230 Nm
2,4 - 0,4; 6,0
Pull Type Chain
Audi
A6
6-Zyl., 2,8 l, 142 kW
280 Nm
2,4 - 0,4; 5,3
Pull Type Chain
Daewoo
Matiz
3-Zyl., 0,8 l, 38 kW
69 Nm
?
Push Belt
Daihatsu
Cuore
3-Zyl., 0,7 l, 43 kW
64 Nm
2,27 - 0,55; 6,77
Push Belt
Fiat
Punto
4-Zyl., 1,2 l, 59 kW
114 Nm
2,43 - 0,44; 5,25
Push Belt
Honda
Logo
4-Zyl., 1,3 l, 48 kW
108 Nm
2,47 - 0,45; 6,36
Push Belt
Honda
Insight
3-Zyl., 1,0 l, 50 kW
91 Nm
2,44 - 0,41; 5,69
Push Belt
Honda
Civic
4-Zyl., 1,4 l, 66 kW
130 Nm
2,47 - 0,45; 5,81
Push Belt
Honda
Civic
4-Zyl., 1,6 l, 81 kW
152 Nm
2,47 - 0,45; 5,81
Push Belt
Honda
Civic
4-Zyl., 1,7 l, 85 kW
149 Nm
2,47 - 0,45; 5,81
Push Belt
Honda
Capa
4-Zyl., 1,5 l, 72 kW
133 Nm
2,47 - 0,45; 6,34
Push Belt
Honda
HR-V
4-Zyl., 1,6 l, 77 kW
135 Nm
2,47 - 0,45; 6,88
Push Belt
Honda
HR-V
4-Zyl., 1,6 l, 92 kW
144 Nm
2,47 - 0,45; 6,88
Push Belt
Mazda
121
4-Zyl., 1,2 l, 55 kW
110 Nm
3,84 - 0,66; 3,84
Push Belt
MG
MGF
4-Zyl., 1,8 l, 88 kW
165 Nm
2,42 - 0,52; 4,05
Push Belt
Brand
CVT Passenger Cars (worldwide, 2001)
Mölle
2004
Brand
Model
Engine
Maximum
Torque
Transmission ratio
(Variator, Overall)
Type of Variator
Mini
Cooper
4-Zyl., 1,6 l, 85 kW
149 Nm
?
Push Belt
Mitsubishi
Lancer Cedia
4-Zyl., 1,8 l, 96 kW
177 Nm
2,32 - 0,45; 5,22
Push Belt
Nissan
Micra/March
4-Zyl., 1,0 l, 44 kW
80 Nm
2,43 - 0,44; 6,3
Push Belt
Nissan
Micra/March
4-Zyl., 1,4 l, 60 kW
108 Nm
2,43 - 0,44; 5,25
Push Belt
Nissan
Cube
4-Zyl., 1,3 l, 63 kW
120 Nm
2,43 - 0,44; 5,24
Push Belt
Nissan
Bluebird
4-Zyl., 2,0 l, 110 kW
200 Nm
?
Push Belt
Nissan
Almera Tino
4-Zyl., 2,0 l, 100 kW
175 Nm
2,33 - 0,43; 5,47
Push Belt
Nissan
Primera
4-Zyl., 2,0 l, 103 kW
181 Nm
2,33 - 0,43; 4,18
Push Belt
Nissan
Primera '02
4-Zyl., 2,0 l, 110 kW
200 Nm
2,33 - 0,43; 5,47
Push Belt
Nissan
Primera '02
4-Zyl., 2,0 l, 125 kW
245 Nm
2,1 - 0,43; 5,47
Push Belt
Nissan
Prairie/Liberty
4-Zyl., 2,0 l, 103 kW
186 Nm
2,33 - 0,43; 5,47
Push Belt
Nissan
Serena
4-Zyl., 2,0 l, 107 kW
186 Nm
2,33 - 0,43; 5,74
Push Belt
Nissan
Cedric/Gloria
6-Zyl., 3,0 l, 206 kW
388 Nm
2,86 - 0,66; 3,69
Half Toroid
Rover
R45
4-Zyl., 1,8 l, 86 kW
160 Nm
2,42 - 0,44; 5,76
Push Belt
Subaru
Pleo
4-Zyl., 0,7 l, 33 kW
56 Nm
2,43 - 0,44; 4,67
Push Belt
Suzuki
Alto/Kei
3-Zyl., 0,7 l, 34 kW
57 Nm
2,42 - 0,55; 6,77
Push Belt
Toyota
Opa
4-Zyl., 2,0 l, 112 kW
200 Nm
2,4 - 0,43; 5,18
Push Belt
CVT Passenger Cars (worldwide, 2001)
Mölle
2004
 Introduction – Chain-CVT
 Clamping Systems
 Control Design for Range Shifts in
Autarkic Hybrid
 Expanded Control Layout for Universal Use
in Chain Variator Applications
 Summary
Presentation Outline
Mölle
2004
nCVT1
Pressure
Transducer
Pulley 1
p
nCVT2
Pulley 2
U
p
U
Main Advantage:
 Oil flow on demand
Disadvantages:
 Torque information
supplied by engine
controller: Poor
Directional
dynamics and limited
Control Valve
accuracy
 Need for high over
clamping for security
reasons or additional
measures
Constant Pressure
Oil Supply
 Oil flow always at
maximum pressure level
Constant Pressure System
Mölle
2004
Advantages:
Torque
Sensor
Four Edges
Spool Valve
 Clamping pressures are
automatically achieved
Pulley 1
without superior control
Md
Line Pressure
Valve
Md
 High dynamically set
clamping pressures due
Pulley 2
to the “pump function”
Actuator
Speed Ratio Control
Pressure Differential Hydraulic
Valve
Control Unit
Constant Flow
Oil supply
 Clamping pressure and
speed ratio control
independent
Main Disadvantage:
 Permanent, constant
oil flow required
Constant Oil Flow System (PIV)
Mölle
2004
Movable sensor plate
A
A-A
AF
Fp
bF
A
T
AF
Fax
rMd
jF
sF
Fu
sF (axial movement of sensorplate)
Characteristics:

Torque sensor pressure – proportional to torque at the shaft

Additional “pump function” at high torque gradients
Conventional Torque Sensor
(System PIV)
Mölle
2004
40
p CYL1
Pcyl1
p TORQUE
pTorque
400
p CYL2
Pcyl2
p PUMP
Pump
Nm
20
200
shift speed ds/dt
100
10
0
-1,5
0
-1
-0,5
0
0,5
mm
1,5
Slide valve travel
Characteristic Curve of Actuator in
Conventional PIV Clamping Systems
Mölle
2004
Torque
Pressure
bar
 Introduction – Chain-CVT
 Clamping Systems
 Control Design for Range Shifts in
Autarkic Hybrid
 Expanded Control Layout for Universal Use
in Chain Variator Applications
 Summary
Presentation Outline
Mölle
2004
B
C
L2
The Autarkic Hybrid
E
K1
D
L1
Opel Astra Caravan

60 kW Diesel engine

10 kW electric motor (120V)

i2-CVT gearbox
Range shift in Autarkic Hybrid
raised the need for improved speed
ratio control:
K2
A
F

A: input shaft
B: shaft of CVT
C: shaft of CVT
D: output shaft
E: differential gear
F: electric engine
Extremely high torque gradients
during range shift
(CVT engaged vs. disengaged)
Error signal <0,002 required
Driveline of the Autarkic Hybrid
Mölle
2004
System deviation
Error signal limit positive high
Error signal limit positive low
Selection of control parameters:
u Absolute value of deviation
u Algebraic sign of deviation
0
Variation of param. (gain scheduling):
u Value of control variable
Error signal limit negative low
Error signal limit negative high
PI-Control
PD-Control
setpoint
Linear
Controller
CVT
CVT Controller, Variable in Structure
Mölle
2004
FAb=pAb.Az
Ratio of
clamping forces
pAb
z=
Az
Main disturbance
variables:
• Torque
• Speed Ratio
... lead to a change in
required z-ratio for
steady state operation
FAn
FAb
FAn=pAn.Az
Az
pAn
Influence of Disturbance Variables
Mölle
2004
Problem:
Improved control system is needed for speed
ratio control at SYN (i=0,458) during range shift.
Solution:
Disturbance feedforward (torque)
setpoint
Linear
Controller
CVT
Extension of the Speed Ratio Controller
Mölle
2004
20
%
1000 1/min
1500 1/min
2000 1/min
0
-10
Slide valve travel
Slide valve travel
20
2500 1/min
3000 1/min
-20
V2
0
-10
-20
-30
-40
V1
-50
-30
0
5
bar
10
15
20
Torque sensor pressure
30
5
10 15 20 bar
Torque sensor pressure
0
Disturbance
feedforward
T, n
setpoint
Linear
Controller
CVT
Extension of the Speed Ratio Controller
Mölle
2004
30
The taken measures resulted in a significant
improvement of the quality of speed ratio control
and reliability of range shifts.
Apply same principles to the CVT controller for universal use:
 Regard to further disturbance variables
 Improved control over the whole spreading range
(improvements in quality, efficiency etc.)
 Enable different control strategies:
ratio based strategies (e.g. ground speed pto) vs.
di/dt control (passenger car / transportation work)
Results and further Aims
Mölle
2004
 Introduction – Chain-CVT
 Clamping Systems
 Control Design for Range Shifts in
Autarkic Hybrid
 Expanded Control Layout for Universal Use
in Chain Variator Applications
 Summary
Presentation Outline
Mölle
2004
Main disturbance variables torque
and speed ratio lead to:
Pulley Misalignment, shaft
deflection, pulley distortion, …
… change in clamping force ratio
Characteristic z-map
Further disturbance variables:
• Speed (rotating hydraulic cylinder)
• Spring (basic clamping force)
Algebraic compensation
Disturbance Variables
Mölle
2004
Disturbance
Variables
z-map
Distrubance
feedforward
E=mc
Mathematic
2
Compensation
setpoint
Linear
Controller
CVT
Extension of the Control Structure
actual
value
Mölle
2004
Question:
Prerequisites
Where to
get for
the
z-map from ?
Output
of Linear
adaptation:
Controller supposed to
Zero instate
steady state!
beSteady
(T, n, manipulated var.)
 …
setpoint
background task
(duration ?)
Linear
Controller
Adaptation
Disturbance
Variables
z-map
Distrubance
feedforward
E=mc
Mathematic
2
Compensation
CVT
constant task time
(e.g. 5ms)
Adaptation of z-map
actual
value
Mölle
2004
Adaptation of the
sampling points:
Value of the
manipulated variable
from linear controller
x weighting factor.
Weighting functions:
• Gauss
• Cone
• ...
Adaptation Law
Mölle
2004
START
Visualization and Discussion of the
Adaptation Process
Mölle
2004
T1
J1
 Power demand leads
to desired engine speed.
ω1
i
di/dt
ω2
J2
T2
 New engine speed is
achieved by changing
the CVT’s speed ratio i.
 Change in speed ratio
di/dt affects the available
torque at the wheel T2!
 Controlling the rate of speed ratio change is favorable
CVT in Drive Train Configuration
Mölle
2004
Modification of the
control structure:
Adaptation
 Delete Feedback Loop
Disturbance
Variables
 Stop Adaptation Process
 Replace Controller
z-map
Distrubance
feedforward
E=mc
Mathematic
2
Compensation
setpoint
setpoint
speed ratio
di/dt
pdyn
f(di/dt,n,
Linear
Controller
geometry)
Control of the Rate of
Speed Ratio Change di/dt
CVT
di/dt speed
ratio
Mölle
2004
pdyn = ds/dt / ( ACYL·D )
 Axial pulley speed
ds/dt = f ( di/dt, geometry )
 Damping coefficient
D = f ( speed… )
* ü = 1/i
Control of the Rate of
Speed Ratio Change di/dt
Mölle
2004
Measured Results of the Control of
Speed Ratio Change ds/dt
Mölle
2004
 Introduction – Chain-CVT
 Clamping Systems
 Control Design for Range Shifts in
Autarkic Hybrid
 Expanded Control Layout for Universal Use
in Chain Variator Applications
 Summary
Presentation Outline
Mölle
2004
 Quality of speed ratio control was significantly improved
 The control structure was implemented using a RCPsystem running under Matlab/Simulink (xPCTarget)
and is currently running on a test rig
 For use in tractor applications it was also implemented
on a typical electronic control unit (C167) both manually
and using code generation (dSpaceTargetLink 2.0)
 Gathered z-maps can be used for different purposes
(scientific work, onboard diagnostic purposes etc.)
 Further optimization possible (improved di/dt, z-max)
Summary
Mölle
2004
2004 International Continuously Variable and Hybrid Transmission Congress
September 23-25, 2004
Control and Operating Behavior of
Continuously Variable Chain Transmissions
Roland Mölle
Division Mobile Working Machinery
Prof. Dr.-Ing. Dr. h.c. K.-Th. Renius c/o
Institute of Automotive Engineering
Prof. Dr.-Ing. B. Heißing
Technische Universität München
Mölle
2004