AU2021240225A1 - A controller and method for controlling an ignition coil when starting a spark ignition internal combustion engine - Google Patents

Abstract

A CONTROLLER AND METHOD FOR CONTROLLING AN IGNITION COIL WHEN STARTING A SPARK IGNITION INTERNAL COMBUSTION ENGINE Abstract A controller and method of controlling an inductive ignition coil when starting an internal combustion engine whereby, in the event of failure to start, the stored energy of the ignition coil is transferred into and dissipated by the ignition controller to prevent a spark at the sparkplug. The starting mechanism, ignition coil, and ignition controller, are protected from damage due to electrical overstress on the electronics, and the mechanical components are protected from damage due to reverse torque. In particular, kick-back events on kick start motorcycles are prevented to eliminate physical damage to riders. The method also provides for protection of the high-tension ignition circuit, ignition coil, and ignition controller, in the event of a failure of a component in that high tension circuit. [Figure 4] 0 00 N4 CD u 0 Co U 0Q 0i ~N - ~ 0 0 t)-

Description

N4 CD

u

0

Co U 0Q 0i ~N - ~ 0 0 t)-

EDITORIAL NOTE 2021240225 There is only six pages of description.

A CONTROLLER AND METHOD FOR CONTROLLING AN IGNITION COIL WHEN STARTING A SPARK IGNITION INTERNAL COMBUSTION ENGINE Description The present invention relates to a controller and method for controlling an inductive ignition coil used in an ignition system of a spark ignition internal combustion engine. Inductive automotive ignition systems use the magnetic field created by current flowing through the primary winding of an ignition coil as a method of energy storage. This energy is built up over time as current through the coil increases over time: the rate being set by the coil inductance and battery voltage. Interruption of the current by, for example, contact breakers or an electronic ignition switching element, causes rapid collapse of the magnetic field. This rapidly changing magnetic field induces a very high voltage on the secondary winding of the ignition coil, that is electrically connected to the spark plug. When the voltage is high enough to jump the plug gap a spark is created that is intended to create combustion of the air fuel mixture and thus generate forward rotational momentum of the engine.

Sometimes, a spark ignition internal combustion engine will fail to start. This can occur for multiple reasons, for example, the temperature was inconsistent with the fuel delivery, and in particular the volatile compounds of the fuel at that temperature failed to present sufficient combustible fuel within the spark gap at the time of the spark. A similar situation can occur in which too much volatile fuel is presented and the required oxygen in the air is displaced, resulting in a non-combustible gas mixture.

A further reason causing an engine not to start is due to the rotational inertia of the engine being insufficient to carry the engine past the piston 'Top Dead Centre' (TDC) highest point. As the compression stroke progresses, a gas spring of air-fuel mixture is compressed by the rotation of the crank as, with both valves closed, the gas mixture is trapped in the cylinder at this time. If the gas spring force exceeds the rotational inertia of the engine plus any cranking force then the engine will not pass TDC but will instead bounce backwards. If a spark occurs under these circumstances the engine torque is significant and backwards. This effect is most common at the end of a person powered starter mechanism such as a rope pull-starter or motorcycle kick starter, or after the electric starter motor is de-energised. The same situation can occur significantly after the engine has come to rest, from seconds to tens of seconds and longer. The resistive force of TDC compression is the highest that a cranking force encounters. Failing to overcome TDC and resting a quite small angle before top dead centre (BTDC) is a common position in which an engine might come to rest, especially one with two or more cylinders whereby the valvetrain resistance on another cylinder imparts torque resistance against the reversing crank.

Most cranking systems, be they electric motors, kick starters, or pull starters, have a mechanism by which the torque they impart is in the forward rotational direction only. Usually, the engine starts, and some directional torque mechanism, such as a roller bearing clutch, a dog clutch or a Bendix gear, causes the starter mechanism to dis-engage if the engine rotates faster than the starter.

The challenge with an inductive ignition system when the engine fails to fire, whether crank rotation comes to a temporary stop or alonger resting position, is that the spark energy is already stored in the magnetic field around the ignition coil. Most ignition systems use an on-off type switching element that will create a spark when it is switched off.

If a spark occurs in either of the two conditions described previously then a combustion event can occur causing a high impulse of reverse torque. This is significant because, for many engines, for example motorbike roller bearing clutch electric starters, the starting mechanism is still engaged in these conditions. The reverse torque is transmitted through the starting mechanism. This can cause damage to the starting mechanism such as broken teeth. More seriously, physical damage to the rider of a kick-start motorcycle is well documented due to this 'kick back'.

A further problem exists with inductive ignition. In the event of a failure of the High-Tension (H.T.) circuit, either the coil, H.T. lead or sparkplug, the stored energy in the coil is unable to be dissipated across the spark gap. It must go somewhere, so it reflects back towards the switching element. Rather than spending, say, 8us at the applicable device clamp voltage, it spends 80us or more. This causes both high voltage stress and high thermal stress on the switching device. Whilst this is happening, the H.T. voltage is hanging at the maximum produced by the ignition coil, which might be kV. So, for example, a sparkplug failure might cause subsequent failure of H.T. lead, and/or ignition coil, and/or ignition module.

Therefore, there is a need to improve upon such ignition systems.

Preferred embodiments of the invention The description below refers to the use of an IGBT as the switching device, but a MOSFET might substitute, or a bipolar transistor. Where 'IGBT' is used, 'switching device' can be substituted.

Figure 1 shows low tension waveforms of an exemplary spark event using a semiconductor switching device and an inductive ignition coil. The most commonly used semiconductor switching device is an IGBT, and those designed for ignition systems include a voltage clamp to protect the device, achieved by in-built circuits that turn the device back on to limit the voltage across the device. After the IGBT is turned off by removal of the gate voltage at [102]. The collector voltage rises up to 450V at [103], at which point the clamp circuitry is engaged to partially switch on the device, [104], which limits the collector to a non-destructive voltage. At the time [103] of voltage limitation the current trace [105] shows that 5 amps is still flowing through the device resulting in an instantaneous power of 450 x 5= 2.25kW. The semiconductor die is around 4mmA2, so this energy density is several 100x greater than that required to cause an electric cooker ring to glow red-hot. This demonstrates the problem with relying on the switching device to absorb the energy stored in the ignition coil using its in-built voltage clamp. In the traces in Figure 1, the persistence of the high voltage from [103] results in a build-up of secondary voltage to a value at which the spark gap is jumped at [106]. A proportion of the energy of the coil is transferred by the H.T. wiring to the sparkplug. In this successful spark event the high voltage, high current in the switching device is present for only a few microseconds and the energy dissipation in the switching device is around 10mJ. However, to achieve a soft shut-down without a spark, the entirety of the stored energy in the ignition coil must be dissipated in the switching device and coil resistances. This could be up to ten times the energy documented in Figure 1, indicating the need for adherence to the SOA of the device.

In Figure 2, [200] shows an exemplary outline circuit of the invention. The Primary (inductance), Secondary (inductance), R_Primary (coilresistance), RSec. (secondary winding resistance) and LeakageInductance are bulk components that, electrically, model the behaviour of an ignition coil. The leakage inductance is significant because the fast rise in fly-back voltage is due to the leakage inductance more than the coupled primary and secondary inductances. The magnetic circuit of an ignition coil is usually not a closed loop so the behaviour of the coil is of a coupled inductor, not a transformer. The gate driver [202] amplifies the current from the Microcontroller [201] so that the IGBT

[203] can be turned on and off with the required switching times. The filter components R1 and C1 permit adjustment of the rise time of the collector voltage.

Description of the invention is most easily explained using the method [500] described in Figure 5, with references to the waveforms in Figures 3 and 4 at appropriate steps. Detection of crank rotation using information from the crank position sensor causes an ignition controller to begin scheduling spark events, and in the case of the invention method [500], monitoring for conditions requiring a soft shutdown at step [501]. In an inductive ignition system, the spark energy is stored in the magnetic field of an ignition coil. This is achieved by switching on an IGBT [203] at step [502] prior to the anticipated time at which a spark is required so that the current can build up through the primary inductance of the ignition coil. The energy stored in the magnetic field is proportional to the square of the current, which takes time, usually a few milliseconds, to build to the desired level. In step [503] an evaluation of the engine start is made: based upon the instantaneous crank position and rotation speed for this exemplary engine, will the rotational inertia carry it past 'Top Dead Centre' (TDC), the point of maximum compression?

If the answer is yes to [504] then at [505] the ignition controller switches off the IGBT at the predetermined angle for optimum starting of the engine based upon the prevailing conditions (rpm, temperature etc.). For a normal spark this is represented by the desired current peak [105] achieved at position [102], after which the current decays in a few microseconds to zero [107]. A proportion of the magnetic energy is transferred to the spark event at the sparkplug. If the engine starts then the ignition controller continues to schedule sparks based upon crank sensor position and coil dwell periods causing the engine to continue to run.

If the answer to question [504] is no then the soft shutdown algorithms are engaged at [507]. The intent of the soft shutdown process is to extract the stored energy from the ignition coil down to a level where a spark is impossible. The practical implementation of this means reducing the coil in the primary circuit to below alevel where the residual magnetic flux is capable of building up enough secondary voltage to cause a spark. In the'kick-back' problem described earlier, the compression is low and therefore the voltage required to create a spark might be only 2-3kV. The target current level can be single digit % of the normal current value at the end of the dwell period. The in-cylinder conditions, and often the electrical capacitance and resistance of the H.T. system, are unknown to the ignition controller so a conservative approach is required.

The method to reduce the coil current relies on transferring most of this energy into the IGBT, where it is dissipated as heat. This method results in some'IR losses', current x resistance in the ignition coil, which also dissipates some energy as heat. This second mechanism occurs both in the primary and the secondary: the H.T. circuit has capacitance, and charging and discharging this requires current flow through the resistance of the coil windings.

To reduce the stored energy in the coil, the IGBT is turned OFF at [509] for a duration insufficient to cause a spark. This is a finely calibrated duration involving microsecond or sub-microsecond periods depending upon the drive circuit, the IGBT, the ignition coil, and the instantaneous current applicable at the time of switching. Preferably, the internal clamp of the IGBT is not engaged. Use of the internal clamp is possible but places the IGBT under kW power stresses for much longer than usual and will often result in a spark at the sparkplug, defeating the invention.

After the OFF duration, the IGBT is switched ON at [509]. This has two effects: it quenches the build-up of both primary and secondary voltage, and it re-charges the energy in the coil. As the aim is to gradually reduce the stored energy, due to the latter effect it is necessary that the ON duration puts less energy back into the coil than has been removed by the dissipation during the OFF duration [509]. However, configuring the OFF peak at the collector to be 100+ volts (in this example) whereas the ON recharge is at 12V (in this example), means that a net reduction is easily achievable.

Figures 3 and 4 provide more information on this method.

In Figure 3, [300] shows waveforms from the initial phase of the energy extraction method. The IGBT gate driver [302] has been ON for 5ms in order to build up the coil current [301] to nearly 3.6A. The IGBT gate [302] is driven to OV for short durations [303] after 5.1ms to switch the device OFF. As a result, the IGBT collector voltage [304] increases to over 10OV, but the OFF duration is short enough to control the peak voltage. The IGBT is then switched back ON, which causes a quenching of the collector voltage. Using this calibrated OFF time and subsequent calibrated ON time, the collector voltage of the IGBT is limited to a value that does not stress the device or exceed its SOA. The energy dissipated in each OFF duration far exceeds the energy replaced by the ON duration. The net result is a clear reduction in current as shown during each OFF duration [305] without a corresponding increase during the ON duration [306]. By this method the energy stored in the ignition coil [205] is dissipated by the switching device [203] without risk of creating a spark.

Figure 4 shows that the methods described by the detail of Figure 3 are effective over longer timescales. The upper trace is the current through the primary of the ignition coil and the lower trace is the voltage on the collector of the IGBT. Figure 3 is a significantly 'zoomed-in' view of the start of energy extraction at 5.1ms, at [401]. [402] shows the same trace as [304] over a much longer time duration. During this time, the reduction in stored energy results in alower fly-back collector voltage as seen at [403]. Because the current therefore stored energy is lower, the OFF/ON times can be revised without applying overstress to the IGBT or creating a collector voltage that risks a spark. These revised times are applied at 5.6ms creating a lower current discharge curve; and revised again at 6.3ms [405] creating a further reduced current discharge curve [404]. By the fourth revision of ON and OFF periods at about 7.Oms, the current falls sufficiently low that 200us later the IGBT can be switched off entirely without any risk of a spark [406]. This last step at [406] corresponds to a'yes' answer to question [510] and then actioning of [511].

The 5-step calibration of Figure 4 should be considered exemplary: it demonstrates the principle by a means that is visible in Figure 4 and can therefore be explained in steps. Implementation might be pre-determined in calibration using a few time-based break-points as per Figure 4. Thedurationsfor OFF and ON in steps [508] and [509] can be determined by a variety of methods. The calibration might be pre-determined based upon the implementation of circuitry. It might be a discrete function over time: this method is displayed in Figures 3 & 4. It might be a continuous, progressive function over time, leading to a faster extraction of energy at the same voltage, or a slower extraction at lower power and voltage. The exemplary methods shown in Figures 3 and 4 use a peak collector voltage of 110V. To create a spark, the secondary voltage must hang at the clamp level of 450V for a duration ranging from a few us to over 20us to charge the secondary circuit capacitance to the breakdown voltage. This difference in voltage and duration shows the margin that exists, which allows open-loop implementation of the invention through conservative calibration.

Two further preferred embodiments will now be described that enhance control through the addition of a feedback loop, and allow an alternative method of operation. A combination of these three preferred embodiments might be employed.

For Figure 6 the components of the invention [600] additional to the invention [200] will be described. A low value (10's milli-ohm) current sense resistor [601] is added to the emitter/source path of the switching element. A differential amplifier, shown diagrammatically as [602], converts this signal into a range suitable for reading by the Analog to Digital Converter (AD) of the Microcontroller (MCU). The Ignition Controller now has closed-loop instantaneous feedback of the current through the ignition coil primary. This allows tailoring of the energy extraction event in a number of ways. Variation of behaviour due to ignition component selection, but including the controller variation, can be accommodated through closed loop adjustment of PWM frequency and duty cycle. Where stepped changes in PWM frequency and duty cycle are used, such as per Figure 4, these are now determined by coil current and energy and are thus independent of the battery voltage that can change the charge curve of dwell current versus time. The starting energy for soft shut-down is known and the energy removal can operate in a closed loop to ensure successful energy removal to the point of final switch-off without a spark. The conservative time-only calibration can be superseded by a more aggressive shut-down. A strategy where the PWM aims for a constant fly-back peak is more consistent when based upon the instantaneous coil current. Therefore, a continuously variable ON and OFF PWM can also be applied to more aggressive shut-down.

If a non-standard ignition coil is in use, this can be identified by the charge curve during the dwell period, and the inductance and therefore energy storage identified so that a suitable soft shut-down PWM can be applied. Such a possibility is particularly relevant to an aftermarket ignition system.

For Figure 7 the components of the invention [700] additional to the invention [200] will be described. A resistive voltage divider formed by resistors [701] R2, R3, R4 and [702] R5 converts the collector voltage into the readable range of a high-speed ADC [703] in the Microcontroller. The use of multiple series resistors to reduce the high voltage is not essential but is the cheapest method to implement a surface mount resistor that can handle the 450V present with a running engine. The feedback of peak collector/drain voltage provides direct information by which the OFF-time can be adjusted to ensure the target window is being achieved of, high enough voltage to remove energy, and voltage too low to cause a spark. The target fly-back voltage can be more accurately set to match the SOA of the switching device in the Ignition Controller along with its mechanical heatsinking. Closed loop control of the collector peak voltage increases the accuracy of the method whereby a constant fly-back voltage is used to provide a more aggressive soft shut-down.

If a non-standard ignition coil is in use, this can be identified by a fly-back voltage outside the expected window. As the prime determinant that causes a spark is the duration and amplitude of the fly-back voltage, the ignition controller can adapt the PWM to bring the voltage back within the desired window. It can, without current feedback, continue to run closed-loop around the desired fly-back voltage, until the energy in the coil is no-longer sufficient and decays below the window, indicating that the switching device can be turned completely OFF [512]. Such a possibility is particularly relevant to an aftermarket ignition system.

Two further advantages accrue from the voltage feedback of [700]. One is a direct ability to detect a hardware fault in the H.T. circuit from the ignition coil through to the spark gap: it could be as simple as an H.T. lead not properly attached to a sparkplug. This fault condition applies the maximum secondary voltage to the ignition system, which might be 30kV or more. Failures of insulation in ignition coils, leads, caps or plugs are a common result. The traces that result from such a fault condition are depicted in Figure 8. In Figure 8 the ignition components and switch-off parameters are analogous to Figure 1, so comparisons can be made with the non-faulted operation of a spark in Figure 1. Note that the time duration of Figure 8 is significantly longer than Figure 1. In this simulation a MOSFET is used as the switching element. The gate voltage is depicted by trace [802], and shows the same characteristic whereby the switching device is partially turned back on to limit the drain voltage, as is shown by trace [804]. It can be seen that, without the spark event in Figure 1, the Drain voltage hangs at 450V for 80us rather than the 4us of Figure 1. Rather than the current decay from 6A to zero taking 7.5us as in Figure 1, the same current decay takes 80us in trace [803]. The result is a peak power dissipation in [801] of 2.3kW and an average power dissipation of the switching device of over 1kW during this 80us decay time. All the stored energy is dissipated in the switching device. This creates both significant voltage stress and significant thermal stress in the switching device. The voltage stress can lead to over-stress failures of the semiconductor structures in the device, and the thermal stress can lead to over-temperature failure whereby the silicon turns from being a semiconductor to a conductor, leading to catastrophic thermal runaway and device failure (identifiable by escaping smoke). It is imperative to maintain semiconductor die temperatures below about 170 degrees Celsius. This requires a rate of power input whereby the heatsinking path from die, through device lead-frame, through PCB (if applicable), through insulating thermal washer, into the Ignition Controller heatsink, that prevents catastrophic temperatures in the die. This fault condition is worst-case in a multi-cylinder engine because the working cylinder(s) will cause the energy dump to recur as fast as every 10ms milliseconds.

Invention [700] makes this a survivable condition even in multi-cylinder engines by detecting the High-Tension fault and extracting the energy using a soft shut-down method. The more aggressive soft shut-down approaches enabled by feedback permit this protection to operate at high engine rpm. Note that both Ignition Controller and Ignition Coil are protected by soft shut-down. Any of the soft shut-down mechanisms declared in this disclosure are sufficient to provide protection.

The indication of a secondary wiring H.T. fault is that the collector voltage [804] read by the invention

[700] hangs high long after a spark should have occurred in a working ignition system. At some point after this spark event would be expected but significantly less than the full maximum power energy dump shown in Figure 8, for example, after 15us, the soft shut down mechanism is triggered. The advantage of the soft shut-down is that the mechanism intentionally takes longer to dissipate the same energy at a much lower average power, and does so at alower switching voltage. This reduces both voltage stress and temperature stress on both the device and the ignition coil.

The second advantage enabled by switching voltage feedback is depicted in Figure 9, again using a MOSFET switching device. In this method of soft shut down, the PWM is switched ON and OFF using voltage feedback of the invention [700]. The switching device is switched OFF when the drain (or collector) voltage exceeds a calibrated value. In Figure 9, 120V is used. The switching device is switched back ON when the drain (or collector) voltage drops below a calibrated value. The specific value depends upon factors such as the switching device capacitance, the drive capability of the current amplifier [202], and the low pass-filter formed by resistor R1 and capacitor C1 in [200]. The effect of these filter components plus gate capacitance is to create a Digital to Analog Converter (DAC) whereby the digital PWM from the Microcontroller [201] is converted into an analogue voltage [901] on the gate of the switching device that controls the switching device in a substantially linear mode of operation. A result is depicted in Figure 9. The MOSFET gate voltage [902] is constant and substantially linear and varies such that the'fly-back' drain voltage is controlled within a narrow range just above 120V for the duration of the controlled discharge. The drop in current [904] is also substantially linear. Extracting the stored coil energy at this lower voltage over a longer duration, about 200us in this example, is safe for both the switching device and the ignition coil. The peak power [903] is under 400W, rather than 2.3kW, and the average power is around 200W.

As there are no high and fast voltage transients, electromagnetic noise and radiation is significantly reduced. The PWM switching frequency can be adjusted by tuning the switching voltages, and the DAC conversion into an analogue voltage can be tuned by varying the values of resistor R1and capacitor C1 in [200] to achieve the desired effect.

Changing the collector voltage becomes an easy method to trade off the duration of energy removal versus the peak and average power required. The total energy for the soft shut-down is reduced because the switching device is never switched fully back on, so there is no re-charging of stored coil energy.

By this mechanism the rate of energy dissipation can be tailored to match the capabilities of both the switching device, its thermal heatsinking, and switching device peak and average power dissipation.

It should be noted that all three embodiments [200], [600] and [700] are capable of all four methods of soft shut-down: time-based PWM calibration, coil current based PWM calibration, peak voltage based PWM calibration, and voltage based PWM DAC calibration. Calibration for the invention [200] must be developed for open-loop implementation. The preferred closed-loop feedback enhancements of [600] and [700] increase the operational envelope of soft shut-down (e.g., engine RPM), improve the precision of the methods described, and allow them to accommodate non-standard ignition components such as a replacement ignition coil.

EDITORIAL NOTE 2021240225 There is only six pages of claims.

A CONTROLLER AND METHOD FOR CONTROLLING AN IGNITION COIL WHEN STARTING A SPARK IGNITION INTERNAL COMBUSTION ENGINE

Preferred embodiments of the invention and claims The description below refers to the use of an IGBT as the switching device, but a MOSFET might substitute, or a bipolar transistor. Where 'IGBT' is used, 'switching device' can be substituted.

Figure 1 shows low tension waveforms of an exemplary spark event using a semiconductor switching device and an inductive ignition coil. The most commonly used semiconductor switching device is an IGBT, and those designed for ignition systems include a voltage clamp to protect the device, achieved by in-built circuits that turn the device back on to limit the voltage across the device. After the IGBT is turned off by removal of the gate voltage at [102]. The collector voltage rises up to 450V at [103], at which point the clamp circuitry is engaged to partially switch on the device, [104], which limits the collector to a non-destructive voltage. At the time [103] of voltage limitation the current trace [105] shows that 5 amps is still flowing through the device resulting in an instantaneous power of 450 x 5= 2.25kW. The semiconductor die is around 4mmA2, so this energy density is several 100x greater than that required to cause an electric cooker ring to glow red-hot. This demonstrates the problem with relying on the switching device to absorb the energy stored in the ignition coil using its in-built voltage clamp. In the traces in Figure 1, the persistence of the high voltage from [103] results in a build-up of secondary voltage to a value at which the spark gap is jumped at [106]. A proportion of the energy of the coil is transferred by the H.T. wiring to the sparkplug. In this successful spark event the high voltage, high current in the switching device is present for only a few microseconds and the energy dissipation in the switching device is around 10mJ. However, to achieve a soft shut-down without a spark, the entirety of the stored energy in the ignition coil must be dissipated in the switching device and coil resistances. This could be up to ten times the energy documented in Figure 1, indicating the need for adherence to the SOA of the device.

In Figure 2, [200] shows an exemplary outline circuit of the invention. The Primary (inductance), Secondary (inductance), RPrimary (coilresistance), RSec. (secondary winding resistance) and LeakageInductance are bulk components that, electrically, model the behaviour of an ignition coil. The leakage inductance is significant because the fast rise in fly-back voltage is due to the leakage inductance more than the coupled primary and secondary inductances. The magnetic circuit of an ignition coil is usually not a closed loop so the behaviour of the coil is of a coupled inductor, not a transformer. The gate driver [202] amplifies the current from the Microcontroller [201] so that the IGBT

[203] can be turned on and off with the required switching times. The filter components R1 and C1 permit adjustment of the rise time of the collector voltage.

Description of the invention is most easily explained using the method [500] described in Figure 5, with references to the waveforms in Figures 3 and 4 at appropriate steps. Detection of crank rotation using information from the crank position sensor causes an ignition controller to begin scheduling spark events, and in the case of the invention method [500], monitoring for conditions requiring a soft shutdown at step [501]. In an inductive ignition system, the spark energy is stored in the magnetic field of an ignition coil. This is achieved by switching on an IGBT [203] at step [502] prior to the anticipated time at which a spark is required so that the current can build up through the primary inductance of the ignition coil. The energy stored in the magnetic field is proportional to the square of the current, which takes time, usually a few milliseconds, to build to the desired level. In step [503] an evaluation of the engine start is made: based upon the instantaneous crank position and rotation speed for this exemplary engine, will the rotational inertia carry it past'Top Dead Centre' (TDC), the point of maximum compression? If the answer is yes to [504] then at [505] the ignition controller switches off the IGBT at the predetermined angle for optimum starting of the engine based upon the prevailing conditions (rpm, temperature etc.). For a normal spark this is represented by the desired current peak [105] achieved at position [102], after which the current decays in a few microseconds to zero [107]. A proportion of the magnetic energy is transferred to the spark event at the sparkplug. If the engine starts then the ignition controller continues to schedule sparks based upon crank sensor position and coil dwell periods causing the engine to continue to run.

If the answer to question [504] is no then the soft shutdown algorithms are engaged at [507]. The intent of the soft shutdown process is to extract the stored energy from the ignition coil down to a level where a spark is impossible. The practical implementation of this means reducing the coil in the primary circuit to below alevel where the residual magnetic flux is capable of building up enough secondary voltage to cause a spark. In the'kick-back' problem described earlier, the compression is low and therefore the voltage required to create a spark might be only 2-3kV. The target current level can be single digit % of the normal current value at the end of the dwell period. The in-cylinder conditions, and often the electrical capacitance and resistance of the H.T. system, are unknown to the ignition controller so a conservative approach is required.

The method to reduce the coil current relies on transferring most of this energy into the IGBT, where it is dissipated as heat. This method results in some'IR losses', current x resistance in the ignition coil, which also dissipates some energy as heat. This second mechanism occurs both in the primary and the secondary: the H.T. circuit has capacitance, and charging and discharging this requires current flow through the resistance of the coil windings.

To reduce the stored energy in the coil, the IGBT is turned OFF at [509] for a duration insufficient to cause a spark. This is a finely calibrated duration involving microsecond or sub-microsecond periods depending upon the drive circuit, the IGBT, the ignition coil, and the instantaneous current applicable at the time of switching. Preferably, the internal clamp of the IGBT is not engaged. Use of the internal clamp is possible but places the IGBT under kW power stresses for much longer than usual and will often result in a spark at the sparkplug, defeating the invention.

After the OFF duration, the IGBT is switched ON at [509]. This has two effects: it quenches the build-up of both primary and secondary voltage, and it re-charges the energy in the coil. As the aim is to gradually reduce the stored energy, due to the latter effect it is necessary that the ON duration puts less energy back into the coil than has been removed by the dissipation during the OFF duration [509]. However, configuring the OFF peak at the collector to be 100+ volts (in this example) whereas the ON recharge is at 12V (in this example), means that a net reduction is easily achievable.

Figures 3 and 4 provide more information on this method.

In Figure 3, [300] shows waveforms from the initial phase of the energy extraction method. The IGBT gate driver [302] has been ON for 5ms in order to build up the coil current [301] to nearly 3.6A. The IGBT gate [302] is driven to OV for short durations [303] after 5.1ms to switch the device OFF. As a result, the IGBT collector voltage [304] increases to over 10OV, but the OFF duration is short enough to control the peak voltage. The IGBT is then switched back ON, which causes a quenching of the collector voltage. Using this calibrated OFF time and subsequent calibrated ON time, the collector voltage of the IGBT is limited to a value that does not stress the device or exceed its SOA. The energy dissipated in each OFF duration far exceeds the energy replaced by the ON duration. The net result is a clear reduction in current as shown during each OFF duration [305] without a corresponding increase during the ON duration [306]. By this method the energy stored in the ignition coil [205] is dissipated by the switching device [203] without risk of creating a spark.

Figure 4 shows that the methods described by the detail of Figure 3 are effective over longer timescales. The upper trace is the current through the primary of the ignition coil and the lower trace is the voltage on the collector of the IGBT. Figure 3 is a significantly 'zoomed-in'view of the start of energy extraction at 5.1ms, at [401]. [402] shows the same trace as [304] over a much longer time duration. During this time, the reduction in stored energy results in alower fly-back collector voltage as seen at [403]. Because the current therefore stored energy is lower, the OFF/ON times can be revised without applying overstress to the IGBT or creating a collector voltage that risks a spark. These revised times are applied at 5.6ms creating alower current discharge curve; and revised again at 6.3ms [405] creating a further reduced current discharge curve [404]. By the fourth revision of ON and OFF periods at about 7.Oms, the current falls sufficiently low that 200us later the IGBT can be switched off entirely without any risk of a spark [406]. This last step at [406] corresponds to a 'yes' answer to question [510] and then actioning of [511]. The 5-step calibration of Figure 4 should be considered exemplary: it demonstrates the principle by a means that is visible in Figure 4 and can therefore be explained in steps. Implementation might be pre-determined in calibration using a few time-based break-points as per Figure 4. Thedurationsfor OFF and ON in steps [508] and [509] can be determined by a variety of methods. The calibration might be pre-determined based upon the implementation of circuitry. It might be a discrete function over time: this method is displayed in Figures 3 & 4. It might be a continuous, progressive function over time, leading to a faster extraction of energy at the same voltage, or a slower extraction at lower power and voltage. The exemplary methods shown in Figures 3 and 4 use a peak collector voltage of 110V. To create a spark, the secondary voltage must hang at the clamp level of 450V for a duration ranging from a few us to over 20us to charge the secondary circuit capacitance to the breakdown voltage. This difference in voltage and duration shows the margin that exists, which allows open-loop implementation of the invention through conservative calibration.

Two further preferred embodiments will now be described that enhance control through the addition of a feedback loop, and allow an alternative method of operation. A combination of these three preferred embodiments might be employed.

For Figure 6 the components of the invention [600] additional to the invention [200] will be described. A low value (10's milli-ohm) current sense resistor [601] is added to the emitter/source path of the switching element. A differential amplifier, shown diagrammatically as [602], converts this signal into a range suitable for reading by the Analog to Digital Converter (ADC) of the Microcontroller (MCU). The Ignition Controller now has closed-loop instantaneous feedback of the current through the ignition coil primary. This allows tailoring of the energy extraction event in a number of ways. Variation of behaviour due to ignition component selection, but including the controller variation, can be accommodated through closed loop adjustment of PWM frequency and duty cycle. Where stepped changes in PWM frequency and duty cycle are used, such as per Figure 4, these are now determined by coil current and energy and are thus independent of the battery voltage that can change the charge curve of dwell current versus time. The starting energy for soft shut-down is known and the energy removal can operate in a closed loop to ensure successful energy removal to the point of final switch-off without a spark. The conservative time-only calibration can be superseded by a more aggressive shut-down. A strategy where the PWM aims for a constant fly-back peak is more consistent when based upon the instantaneous coil current. Therefore, a continuously variable ON and OFF PWM can also be applied to more aggressive shut-down.

If a non-standard ignition coil is in use, this can be identified by the charge curve during the dwell period, and the inductance and therefore energy storage identified so that a suitable soft shut-down PWM can be applied. Such a possibility is particularly relevant to an aftermarket ignition system.

For Figure 7 the components of the invention [700] additional to the invention [200] will be described. A resistive voltage divider formed by resistors [701] R2, R3, R4 and [702] R5 converts the collector voltage into the readable range of a high-speed ADC [703] in the Microcontroller. The use of multiple series resistors to reduce the high voltage is not essential but is the cheapest method to implement a surface mount resistor that can handle the 450V present with a running engine. The feedback of peak collector/drain voltage provides direct information by which the OFF-time can be adjusted to ensure the target window is being achieved of, high enough voltage to remove energy, and voltage too low to cause a spark. The target fly-back voltage can be more accurately set to match the SOA of the switching device in the Ignition Controller along with its mechanical heatsinking. Closed loop control of the collector peak voltage increases the accuracy of the method whereby a constant fly-back voltage is used to provide a more aggressive soft shut-down.

If a non-standard ignition coil is in use, this can be identified by a fly-back voltage outside the expected window. As the prime determinant that causes a spark is the duration and amplitude of the fly-back voltage, the ignition controller can adapt the PWM to bring the voltage back within the desired window. It can, without current feedback, continue to run closed-loop around the desired fly-back voltage, until the energy in the coil is no-longer sufficient and decays below the window, indicating that the switching device can be turned completely OFF [512]. Such a possibility is particularly relevant to an aftermarket ignition system.

Two further advantages accrue from the voltage feedback of [700]. One is a direct ability to detect a hardware fault in the H.T. circuit from the ignition coil through to the spark gap: it could be as simple as an H.T. lead not properly attached to a sparkplug. This fault condition applies the maximum secondary voltage to the ignition system, which might be 30kV or more. Failures of insulation in ignition coils, leads, caps or plugs are a common result. The traces that result from such a fault condition are depicted in Figure 8. In Figure 8 the ignition components and switch-off parameters are analogous to Figure 1, so comparisons can be made with the non-faulted operation of a spark in Figure 1. Note that the time duration of Figure 8 is significantly longer than Figure 1. In this simulation a MOSFET is used as the switching element. The gate voltage is depicted by trace [802], and shows the same characteristic whereby the switching device is partially turned back on to limit the drain voltage, as is shown by trace [804]. It can be seen that, without the spark event in Figure 1, the Drain voltage hangs at 450V for 80us rather than the 4us of Figure 1. Rather than the current decay from 6A to zero taking 7.5us as in Figure 1, the same current decay takes 80us in trace [803]. The result is a peak power dissipation in [801] of 2.3kW and an average power dissipation of the switching device of over 1kW during this 80us decay time. All the stored energy is dissipated in the switching device. This creates both significant voltage stress and significant thermal stress in the switching device. The voltage stress can lead to over-stress failures of the semiconductor structures in the device, and the thermal stress can lead to over-temperature failure whereby the silicon turns from being a semiconductor to a conductor, leading to catastrophic thermal runaway and device failure (identifiable by escaping smoke). It is imperative to maintain semiconductor die temperatures below about 170 degrees Celsius. This requires a rate of power input whereby the heatsinking path from die, through device lead-frame, through PCB (if applicable), through insulating thermal washer, into the Ignition Controller heatsink, that prevents catastrophic temperatures in the die. This fault condition is worst-case in a multi-cylinder engine because the working cylinder(s) will cause the energy dump to recur as fast as every 10ms milliseconds.

Invention [700] makes this a survivable condition even in multi-cylinder engines by detecting the High-Tension fault and extracting the energy using a soft shut-down method. The more aggressive soft shut-down approaches enabled by feedback permit this protection to operate at high engine rpm. Note that both Ignition Controller and Ignition Coil are protected by soft shut-down. Any of the soft shut-down mechanisms declared in this disclosure are sufficient to provide protection.

The indication of a secondary wiring H.T. fault is that the collector voltage [804] read by the invention

[700] hangs high long after a spark should have occurred in a working ignition system. At some point after this spark event would be expected but significantly less than the full maximum power energy dump shown in Figure 8, for example, after 15us, the soft shut down mechanism is triggered. The advantage of the soft shut-down is that the mechanism intentionally takes longer to dissipate the same energy at a much lower average power, and does so at alower switching voltage. This reduces both voltage stress and temperature stress on both the device and the ignition coil.

The second advantage enabled by switching voltage feedback is depicted in Figure 9, again using a MOSFET switching device. In this method of soft shut down, the PWM is switched ON and OFF using voltage feedback of the invention [700]. The switching device is switched OFF when the drain (or collector) voltage exceeds a calibrated value. In Figure 9, 120V is used. The switching device is switched back ON when the drain (or collector) voltage drops below a calibrated value. The specific value depends upon factors such as the switching device capacitance, the drive capability of the current amplifier [202], and the low pass-filter formed by resistor R1 and capacitor C1 in [200]. The effect of these filter components plus gate capacitance is to create a Digital to Analog Converter (DAC) whereby the digital PWM from the Microcontroller [201] is converted into an analogue voltage [901] on the gate of the switching device that controls the switching device in a substantially linear mode of operation. A result is depicted in Figure 9. The MOSFET gate voltage [902] is constant and substantially linear and varies such that the 'fly-back' drain voltage is controlled within a narrow range just above 120V for the duration of the controlled discharge. The drop in current [904] is also substantially linear. Extracting the stored coil energy at this lower voltage over a longer duration, about 200us in this example, is safe for both the switching device and the ignition coil. The peak power [903] is under 400W, rather than 2.3kW, and the average power is around 200W. As there are no high and fast voltage transients, electromagnetic noise and radiation is significantly reduced. The PWM switching frequency can be adjusted by tuning the switching voltages, and the DAC conversion into an analogue voltage can be tuned by varying the values of resistor R1 and capacitor C1 in [200] to achieve the desired effect.

Changing the collector voltage becomes an easy method to trade off the duration of energy removal versus the peak and average power required. The total energy for the soft shut-down is reduced because the switching device is never switched fully back on, so there is no re-charging of stored coil energy. By this mechanism the rate of energy dissipation can be tailored to match the capabilities of both the switching device, its thermal heatsinking, and switching device peak and average power dissipation.

It should be noted that all three embodiments [200], [600] and [700] are capable of all four methods of soft shut-down: time-based PWM calibration, coil current based PWM calibration, peak voltage based PWM calibration, and voltage based PWM DAC calibration. Calibration for the invention [200] must be developed for open-loop implementation. The preferred closed-loop feedback enhancements of [600] and [700] increase the operational envelope of soft shut-down (e.g., engine RPM), improve the precision of the methods described, and allow them to accommodate non-standard ignition components such as a replacement ignition coil.

Claims 1. An ignition controller for an inductive ignition system using a semiconductor switching device whereby, when desirable during engine starting, the ignition controller is capable of extracting sufficient stored energy from an ignition coil to prevent a spark at a sparkplug 2. An ignition controller as per claim 1 whereby spark prevention is desirable to prevent kick-back of a motorcycle with a kick-starter 3. An ignition controller as per claim 2 in which the decision on desirability to prevent spark is a function of the recent crank speed and current crank position 4. An ignition controller as per any of claims 1 to 3 whereby stored energy is extracted from an ignition coil using a PWM waveform from a Microcontroller to control a semiconductor switching device 5. An ignition controller as per claim 4 whereby the semiconductor switching device is a MOSFET or IGBT 6. An ignition controller as per claim 5 whereby the off-time of a semiconductor switching device is sufficiently brief such that the secondary circuit of an ignition coil is unable to generate sufficient voltage to create a spark at a spark plug 7. An ignition controller as per claim 5 whereby the off-time of a semiconductor switching device is sufficiently brief such that the collector or drain terminal remains significantly below the collector or drain voltage clamp threshold of the device

8. An ignition controller as per claim 5 whereby the on-times of a semiconductor switching device are sufficiently long to quench the voltage built up on the collector or drain terminal 9. An ignition controller as per claim 5 whereby the on-times of a semiconductor switching device are sufficiently brief such that the increases in stored energy in the ignition coil due to increased current are significantly lower than the decreases in stored energy caused by the off-times of a semiconductor switching device 10. An ignition controller as per claim 5 whereby the on-times and off-times of a semiconductor switching device are configured to ensure that the switching device remains within its safe operating parameters as per its data sheet for voltage, current, rate of change of current d/dt, pulse power dissipation and RMS power dissipation 11. An ignition controller as per claim 5 that uses a multitude of pre-determined durations of on-times and off-times for a multitude of predetermined durations 12. An ignition controller as per claim 5 whereby the durations of on-times and off-times and the total duration are a function of time since the first PWM edge started 13. An ignition controller as per claim 5 that includes a current measurement device in series with the ignition coil circuit and adapts the PWM on- and off-times of the semiconductor switching device and the duration of PWM based upon feedback of measured current 14. An ignition controller as per claim 5 that includes a voltage measurement feedback of the collector or drain voltage and adapts the PWM on- and off-times of the semiconductor switching device and the duration of PWM based upon feedback of measured voltage 15. An ignition controller as per claim 5 that includes both a current measurement device in series with the switching device and voltage measurement feedback of the collector or drain voltage of the switching device 16. An ignition controller as per claim 5 whereby the durations of on-times and off-times of a PWM originating from a Microcontroller and configured to control the gate of a switching device are sufficiently fast that the switching device behaves in a substantially linear manner as if responding to a substantially stable analogue voltage 17. An ignition controller as per claims 14 or 15 or 16 whereby the method for transfer of energy from the ignition coil to the switching device is engaged based upon the time and voltage detected at the collector or drain of the switching device 18. An ignition controller as per claim 5 that includes a voltage measurement feedback of the gate voltage and detects the partial switching on of the switching device by the voltage clamp circuitry internal to said device and uses this to determine the on- and off-times of the switching device 19. An ignition controller as per claim 5 that implements a method that prevents a spark in an inductive ignition in which current is flowing through an ignition coil primary by repeatedly switching on and off a semiconductor switching element in series with said ignition coil primary

ÂÁÇ ÂÁÈ ÂÁÆ ÂÁÄ ÂÁÂ ÂÁÅ ÂÁÃ )LJXUH ÃÁÁ ÃÁÆ ÃÁÇ ÃÁÅ ÃÁÄ ÃÁÂ ÃÁÃ )LJXUHOHDNDJHLQGXFWDQFHPRGHOIRU IO\EDFNVLPXODWLRQ ÄÁÅ ÄÁÁ ÄÁÂ ÄÁÆ ÄÁÇ ÄÁÃ ÄÁÄ )LJXUHIO\EDFNYROWDJHOLPLWHGE\RIIWLPH ÅÁÂ ÅÁÁ ÅÁÃ ÅÁÅ ÅÁÄ ÅÁÆ ÅÁÇ

QOÅIE<s]NEkggI[j<[GI[IgOs

/j<gjh]Nj hPkjG]q[Z][Qj]g ÆÁÂ

EjQp<jIhqQjEPQ[O ÆÁÁ IYIZI[j ÆÁÃ

[<YshIEg<[Xd]hQjQ][ hI[h]gQ[N]gZ<jQ][ ÆÁÄ )LJ0HWKRG 9Ih h<[I[OQ[Ihj<gj d]hhQDYI‹ ÆÁÅ ÆÁÆ "] I<EjQp<jIhqQjEPQ[O I<EjQp<jIhqQjEPQ[OIYIZI[j<[GNQgI [O<OIh]NjhPkjG]q[ hd<gX<jdgIGIjIgZQ[IGEg<[X<[OYI hd<gX<jdgIGIjI <YO]gQjPZh ÆÁÈ

"] 0kg[hqQjEPQ[OIYIZI[j$N]g< ÆÁÉ ÆÁÇ Gkg<jQ][Q[hkNNQEQI[jj]EgI<jI<hd<gX QGjPII[OQ[I hj<gj‹

0kg[hqQjEPQ[OIYIZI[j$"N]g<Gkg<jQ][ 9Ih hkNNQEQI[jj]fkI[EPjPIDkQYGŸkd]NPQOP ÆÁÊ p]Yj<OIQ[jPIPQOPjI[hQ][EQgEkQj .k[I[OQ[I "]

hjPIE]QYEkggI[jDIY]q< dgIŸGIjIgZQ[IGjPgIhP]YG‹ ÆÂÁ

9Ih

0kg[hqQjEPQ[OIYIZI[j$ ÆÂÂ

/j]d ÆÂÃ

ÇÁÁ ÇÁÂ ÇÁÄ ÇÁÃ )LJ&XUUHQWIHHGEDFNIRUFORVHGORRS HQHUJ\H[WUDFWLRQ ÈÁÁ ÈÁÄ ÈÁÂ ÈÁÃ

)LJ9ROWDJHIHHGEDFNIRUFORVHGORRS HQHUJ\H[WUDFWLRQ

ÉÁÁ ÉÁÂ ÉÁÃ ÉÁÄ ÉÁÅ

)LJZDYHIRUPVGXULQJDQ+7IDXOW

ÊÁÁ ÊÁÂ ÊÁÄ ÊÁÃ ÊÁÅ

)LJ9ROWDJHIHHGEDFNXVHGWRFRQWUROWKH VZLWFKLQJYROWDJHWRDVDIHFRQWLQXRXV SHDNYROWDJHRYHUGLVFKDUJH