The main specification of the proposed system is its simple applicability to existing motorcycle independently on the different ignition control system.
The torque applied to the rear wheel can be controlled reducing the gasoline injected closing the butterfly valve or reducing the electrical current to the sparking plug. The former cannot be easily obtained without changing the injection controller, while the latter has been simply obtained, as it will be shown in this section.
Figure 6 shows a general scheme that gives the electrical current to the sparking plug. Every motorcycle has a manual switch used to turn off the engine; the switch simply bypasses to ground the electrical current flowing in the sparking coil. The proposed system modifies the sparking scheme inserting an additional switch in parallel to the manual turn off switch. This switch is controlled by a microcontroller on the basis of the output of some additional sensors, as shown in Figure 7.
The effect of the traction controller is shown in Figure 8. In the top of the figure, the voltage applied to the spark plug during a normal sparking is reported, in the bottom, a cut in the pulse, introduced by the traction controller, can be seen.
The cut on the electrical current of the spark plug is obtained in two ways: defining the cut_off delay between the start of the ignition spark and the intervention of the traction control, as shown in Figure 8, and defining the number of consecutive ignition sparks for which the traction control takes action, an example is shown in Figure 9. The ignition spark cut off, imposed by the traction controller, modifies the torque applied to the wheel.
The complete architecture of the traction control system is reported in Figure 10.
The core of the control algorithm is implemented in a 40 MHz Microchip PIC18F6527 microcontroller.
Six sensors are added to give information on the situation of the motorcycle to the controller: front wheel speed sensor, real wheel speed sensor, sparking signal, rolling angle sensor, butterfly valve opening sensor, and engine r.p.m. sensor. The driver can modify the parameters of the controller pushing few buttons even driving the motorcycle. The microcontroller stores information on a 1 Mbyte flash memory with the purpose of monitoring the performances of the control system. A display is placed in the motorcycle to allow the driver to verify in real time the status of the controller. An RS232 interface is used to communicate with an external PC when the motorcycle is parking.
3.1. Wheel Speed Sensors
The wheel speed is measured using hall effect gear tooth sensors. The sensor output voltage is 5 V when the sensor is in proximity to a ferromagnetic material, otherwise the output voltage is 0.2 V. The hall sensor is placed close to the 4 ferromagnetic bolts of the wheel of our motorcycle, as shown in Figure 11. During wheel rotation, the sensor sends a pulse when the bolt is close to the sensor. The time interval between two pulses is inversely proportional to the angular speed of the wheel. The wheel speed is estimated knowing the time interval between the pulses, the effective diameter of the tyre, and the number of bolts in each wheel. Figure 11 shows the sensor applied to the front and rear wheels in our prototype.
The same principle is applied in the BMW K1200R.
3.2. Rolling Angle Sensors
The wheel speed estimation depends on the rolling angle of the motorcycle and other parameters like tyre, pressure, and temperature. Figure 12 shows the effect of the rolling angle on the effective tyre diameter. An increment in the rolling angle causes a reduction of the effective diameter of the wheel and therefore a reduction in the wheel speed.
The difference between the nominal and effective tyre diameter in bend can be higher than 10% considering that the motorcycle in bend can have a rolling angle higher than 60° and the distortion of the part of the wheel that touches the ground during the bend. Furthermore, this difference depends on the tyre pressure, temperature, and consumption. The error on the wheel speed estimation due to this effect is not negligible, therefore, a rolling sensor of the VTI technologies has been inserted and the information are used in the algorithm for the wheel speed estimation.
3.3. Butterfly Opening Sensor and Engine r.p.m. Sensor
The opening of the butterfly has been measured with a precision potentiometer of Vishay company connected to the accelerator cable. The r.p.m. of the engine is indispensable to measure the torque and the power of the engine.
3.4. Electrical Current Switch
Different solutions have been studied to obtain a cut in the electric current of the sparking plug. An electromechanic relay of multicomp was not fast enough to obtain a shape similar to the one shown in Figure 8. The solution chosen is an IGBT.
3.5. Microcontroller and Control Algorithm
The control algorithm used is shown in Figure 13. The parameter settings are stored in the memory, but they can be modified by the driver during the race using push-buttons to increment or decrement the values of
and
. The parameter
is the minimum increment or decrement of the cut_off delay represented in Figure 8. When the ignition cut
is higher than a fixed value, the ignition spark is completely eliminated and the width of the successive ignition spark is reduced, as shown in Figure 9. The complete elimination up to three successive sparks does not have effect on the driving, as it has been verified by experimental results in a real track.
The parameter
defines the value of difference between front wheel speed and rear wheel speed for which the traction control system takes action. This parameter depends on the driver style of drive and on the ground conditions (asphalt or ground, wet or dry). The front wheel speed and rear wheel speed are estimated, as reported in subsections 3.1 and 3.2, and their difference is used to calculate the width of the cut on the electrical current of the spark plug. When the microcontroller receives an interrupt from the wheel speed sensors, it evaluates if the ignition spark cut must be done and the cut_off delay. When it receives the interrupt from the sparking signal that indicates that the ignition spark started, it eventually waits for a time equivalent to the cut_off delay and operates the ignition spark cut. The algorithm has been translated in assembly code and implemented in the Microchip PIC18F6527 microcontroller with 40 MHz clock, 10 MHz bus clock, and 100 nanoseconds instruction time. The maximum time required to generate the cut has been estimated.
(i)Front wheel speed estimation 4.7 microseconds
(ii)Rear wheel speed estimation 4.7 microseconds.
(iii)Rolling angle estimation 38.0 microseconds.
(iv)Butterfly opening estimation 10.0 microseconds.
(v)Engine r.p.m. estimation 4.7 microseconds.
(vi)Err calculus 118.8 microseconds.
(vii)Total time 176.7 microseconds.
The time required for the traction control by the microcontroller is, therefore, negligible compared with the minimum time between two consecutive ignition sparks (180 milliseconds for 20000 r.p.m.) and with the average time interval between two pulses coming from the hall sensor used to estimate the wheel speed (about 20 milliseconds for a speed of 20 m/s). Therefore, the digital controller implemented is able to control in real time the traction of the motorcycle.
The critical aspect of the traction control system is the number of bolts in each wheel and not the computation time of the microcontroller. The distance
covered by the wheel in the time interval
between two pulses of the hall effect sensor is
where n is the number of bolts, and
is the effective diameter of the wheel, considering the rolling angle effect. As an example, let us consider the case of a constant acceleration a starting from an initial speed
. In this case, the following relationship is valid:
Therefore, using (1) and (2), it results
The value of
in (3) is the sampling time of the speed estimate and it is the delay with which the control system knows the wheel speed.
Figure 14 reports the value of
in milliseconds as a function of the acceleration and of the velocity
for different values of the number of bolts of the wheel for a wheel diameter of 60 cm.
It can be seen that the delay of the control system (177 microseconds) is three-order of magnitude less than
.
Conversely
is of the same order of magnitude of the time interval between two consecutive ignition sparks, which depends on the engine r.p.m.
A reduction of
can be obtained increasing the number of bolts, but this solution is expensive. The BMW K1200R motorcycle uses 100 of pick-up points for wheel speed estimation, while the Ducati motoGP uses 8 pick-up points (the bolts).
The speed estimation
performed by the microcontroller is
Using (3), we obtain
The maximum relative error between the speed estimation
and the effective speed is
where
The error is reduced by increasing the number of bolts (4 is the minimum acceptable), it increases for strong acceleration and low values of speed. Figure 15 reports the value of
(%) as a function of the acceleration and of the velocity
for different values of the number of bolts of the wheel for a wheel diameter of 60 cm. This error depends on speed and acceleration; therefore it reduces the efficiency of the traction control system since it can be only partially compensated by an appropriate tuning of the control parameters.
To verify the error on speed estimation in practical cases, a simple numerical simulator has been developed. Figure 16 reports the effective and estimated wheel speed in a simulation example of a curve with 8 bolts for wheel.
Figure 17 reports the time interval between two spikes of the wheel speed sensor for different number of bolts in the same example, while Figure 18 shows the absolute value of the relative error between estimated and effective speed for different number of bolts.
When the motorcycle is exiting from the curve, the speed is low and the acceleration is high and the error due to a reduced number of bolts is critical.