Technical Field

This invention relates generally to a timing control for an internal combustion engine having a capacitor discharge ignition system and, more particularly, to a timing control which adjusts ignition timing to compensate for a delay between transformer energization and actual ignition.

Background Art

In the field of internal combustion engine controls it is a well recognized principle that accurate ignition timing is necessary to maintain stable and efficient engine operation. In spark ignited engines, ignition occurs upon firing of the spark plug in the cylinder. The "timing" of such an engine is defined as the time at which an ignition signal is delivered to the spark plug relative to the time at which the piston reaches the end of its stroke in the compression cycle. This position is commonly referred to as "top dead center" (TDC) , and timing is measured in crankshaft degrees with respect to TDC, for example 25 degrees before TDC (BTDC.)

In the past, ignition timing was controlled by magnetos and as such the timing angle could only be mechanically adjusted during servicing. Recently, electronics and in particular microprocessors have been adapted to control ignition timing. Electronic ignition controls have the advantage of being able to retard (closer to TDC) and advance (further from TDC) engine timing in response to sensed parameters, such as engine speed, engine load, air/fuel ratio and

altitude. Generally, the electronic controls set a timing angle which is used for all engine cylinders; however, more sophisticated controls develop separate timing angles for each cylinder. CDI's typically include a charge storage mechanism, such as a capacitor, and a step-up transformer with a secondary coil connected to a spark ignition device, such as a spark plug. The ignition timing controller is adapted to discharge the capacitor through the transformer primary coil at the desired timing angle. Discharge of the capacitor through the transformer primary coil induces a high voltage signal in the transformer secondary coil, which, if sufficiently high, causes a spark to arc across the spark plug gap. The voltage applied across a spark plug must be greater than or equal to a predetermined characteristic "spark ionization potential" (voltage) V_ _p in order to initiate the spark. Such ionization potentials are typically on the order of 10 Kv or more. The ionization potential V is dependent on factors such as spark plug gap, cylinder pressure, engine load, and air/fuel ratio. The time required for the transformer to energize to a level sufficient to cause sparking introduces a delay between the production of the ignition signal and actual ignition in the cylinder. This timing "error" can degrade overall engine operation and efficiency. To date, timing controls for engines having capacitor discharge ignitions have failed to address this problem. Partial compensation for the timing error could be achieved by introducing an predetermined timing offset. However, the delay between transformer energization and the spark plug firing is dependent on dynamic engine characteristics and performance factors, such as air/fuel ratio.

cylinder pressure, and spark plug gap. As such a predetermined timing offset would only achieve marginal improvements at best.

The subject invention is directed towards overcoming the above problems by providing timing control which adjusts ignition timing to compensate for a delay between transformer energization and actual ignition.

Disclosure of the Invention

An apparatus is provided for controlling ignition in an engine having an ignition system, a cylinder sensor, and a piston disposed in the cylinder and being movable to a top dead center (TDC) position. The cylinder sensor produces a first signal representing the position of the piston within the cylinder and having a frequency responsive to engine speed. The ignition system includes a transformer having a primary coil which is energized in response to a cylinder select signal to produce sparking across an associated spark plug gap. A first circuit is provided for sensing a delay between production of the cylinder select signal and ignition in the cylinder and responsively producing a timing error signal. A cylinder selector receives the first signal and the timing error signal and processes the first signal to producing a reference timing signal which represents the time between first and second reference points of • the first signal. The second reference point indicates the time at which the piston is at the _. TDC position. The cylinder selector further produces a desired timing signal representing the time between a third reference point of the first signal and the second reference point. The third reference point occurs prior to the second reference point and

indicates the time at which ignition desirably occurs. The cylinder selector further processes the timing error signal to produce a timing offset signal and produces a control delay signal responsive to the reference timing signal less the timing offset and desired timing signals. The selector means delivers the cylinder select signal at a time, as represented by the control delay signal, subsequent to the first point on the first signal. In this manner, the apparatus compensates for any timing error introduced by the time required for transformer energization.

Brief Description of the Drawings

Fig. 1A is an illustrative block diagram of a capacitive discharge ignition system which can be adapted for use with the immediate invention; Fig. IB is a circuit diagram of the capacitive discharge ignition system of Fig. 1; Figs. 2A-E contain waveform diagrams illustrating timing parameters involved in the present invention;

Figs. 3A-E are graphic illustrations of certain waveforms associated with the circuits of Figs. 1A, IB and 4; Fig. 4 is circuit diagram of an embodiment of the ignition system of Figs. 1A and IB which incorporates the immediate invention;

Fig. 5A is a software flowchart of a Timing Error Subroutine which can be used in programming a microprocessor to measure timing errors;

Figs. 5B and 5C are software flowcharts illustrating alternate embodiments of a Timing Error Subroutine which can be used in programming a microprocessor to adjust timing in response to a sensed timing error; and

Fig. 6 is a graph of actual firing time versus measured timing error.

Best Mode for Carrying Out the Invention Referring now to the drawings, the immediate timing controller 8 will be described in connection with a capacitor discharge ignition system 10. The timing controller 8 can be adapted for use with numerous capacitor discharge ignition systems. However, the timing controller 8 will be described in connection with an ignition system as disclosed in U.S. Patent Application Serial No. 07/630,578, filed on December 20, 1990, entitled "Spark Duration Control For A Capacitor Discharge Ignition System," assigned to the assignee herein, and the disclosure of which is specifically incorporated by reference.

The ignition system 10 is shown generally in Figs. 1A-B. Fig. 4 illustrates the ignition system 10 incorporating the immediate timing controller 8. The timing controller 8 and the ignition system 10 will work with an internal combustion engine having any number of cylinders provided electrical components are sized properly. Currently, the timing controller 8 and the ignition systems 10 are being developed for use with a series 3500 SI engine as manufactured by Caterpillar, Inc. of Peoria, Illinois. The series 3500 SI engine has 16 cylinders; however, for simplification Fig. 1A is described in connection with a six cylinder engine, and Figs. IB and 4 are illustrated in connection with a single engine cylinder.

The ignition system 10 includes a power source 12, such as a battery, connected to a DC-to-DC power converter 14. The power converter 14 is a continuously operating, high speed charging circuit

and it is electrically connected to first and second terminals 16a, 16b of an ignition capacitor 18. The power converter 14 is provided for rapidly charging the ignition capacitor 18 and continuously supplying power to the capacitor 18 to maintain the capacitor first terminal 16a at a predetermined electrical potential above the capacitor second terminal 16b. More particularly, the capacitor second terminal 16b is connected to system ground and the first terminal 16a is maintained a preselected potential V above system ground. In the preferred embodiment, the preselected potential V is on the order of 200 volts. Power converters of this type are common in the art and, therefore, it will not be explained in greater detail. One such circuit is generally disclosed in

U.S. patent 3,677,253 which issued on July 18, 1972 to Oishi et al.

Each engine cylinder (not shown) includes a spark plug (not shown) having an associated spark gap 22a-f. Step-up transformers 24a-f are provided for each cylinder to control operation of an associated spark plug. Each transformer 24a-f has a primary coil 26a-f and a secondary coil 28a-f. The transformer primary coils 26a-f each include first and second terminals 30a-f, 32a-f. The transformer secondary coils 28a-f are electrically connected in series with spark gaps 22a-f in associated engine cylinders. Selector switches 34a-f are connected between the ignition capacitor first terminal 16a and an associated one of the primary coil first terminals 30a-f. Numerous electrical switching devices, such as transistors, can be adapted to perform the functions of the selector switches 34a-f and, therefore, the selector switches 34a-f will not be described in great detail. The selector switches 34a-f are normally

biased open and are adapted to close in response to receiving a cylinder select signals (see Fig. 3A) from a cylinder selector means 36 (i.e. timing controller.) When a selector switch 34 is biased closed, the ignition capacitor first terminal 16a and the primary coil first terminal 30a-f, of an associated transformer 24a-f, are electrically connected, thereby establishing a current path through the primary coil 26a-f. The cylinder selector means 36 is provided for operating the selector switches 34a-f in a timed sequence corresponding to a desired ignition sequence for the engine. The cylinder selector means 36 may be implemented with any suitable hardware including analog or digital circuits; however, the cylinder selector means 36 is preferably embodied in a microcontroller (MCϋ) 38 operating under software control. A number of commercially available devices are adequate for performing the control functions of the MCU 38, such as the MC6800 series component manufactured by Motorola Semiconductor Products, Inc. , Austin, Texas.

The cylinder selector means 36 receives a first signal corresponding to engine speed and cylinder position from a sensor means 48. Preferably this function is performed using a single sensor such as that disclosed in U. S. Patent No. 4,972,323 which issued on November 20, 1990 to Luebbering et al., is assigned to the assignee herein, and the disclosure of which is specifically incorporated herein. However, it is foreseeable to use separate sensors for engine speed and cylinder position, respectively. The cylinder sensor means 48 is in the form of a toothed timing wheel or gear 49 and a magnetic pickup unit (MPU) 50 such as a Hall effect device. The timing

wheel 49 includes a series of circumferentially spaced teeth 51. In addition, the wheel is mounted on a shaft (not shown) which is in turn coupled to a crankshaft or camshaft of the engine. The wheel 49 thus rotates as the engine is running, causing the teeth to pass beneath the MPϋ 50. In response to the passage of the teeth, the MPϋ 50 develops the first signal in the form pulse train, as illustrated in the waveform of Fig. 2A. The positions of the pistons in the engine cylinders are referenced to particular pulses on the first signal and the frequency of the first signal is responsive to engine speed.

A variety of other parameters can also be input to the cylinder selector means 36, such as engine load and air/fuel ratio. The cylinder selector means 36 processes these signals to produce cylinder select signals for controlling operation of the select switches 34a-f. The cylinder selector means 36 produces the cylinder select signals for a period of time T corresponding to the desired spark duration in an associated cylinder, (see Fig. 3A.) The selector switch 34a-f to which the selector signal is delivered remains closed while the selector signal is produced. The desired spark duration can be a constant period of time or it can be adjusted in response to sensed engine parameters, as would be apparent to one skilled in the art. In as much as timing controls of this type are well known in the art, no further description of the selector means 36 will be provided. A modulation switch 52 is connected between the primary coil second terminals 32a-f and system ground for completing a current path for the primary coils 26a-f. When a cylinder select switch 34a-f and the modulation switch 52 are closed current begins to flow from the ignition capacitor 18 through the

associated primary coil 26a-f. Numerous electrical switching devices, such as an n-channel MOSFET, can be adapted to perform the functions of the modulation switch 52 and, therefore, the modulation switch 52 will not be described in greater detail.

A current sensing means 62 senses the current flowing through any of the transformer primary coils 26a-f and responsively produces a primary current signal, (see Fig. 3B.) The current sensing means 62 includes a first current sensing resistor 64 connected between the selector switches 34a-f and the ignition capacitor first terminal 16a.. A current mirror circuit 66 is connected to the first current sensing resistor 64 such that the current flowing through the resistor 64 is an input to the current mirror circuit 66. The current mirror circuit 66 delivers an output current signal which has a magnitude responsive to the magnitude of the current flowing through any of the primary coils 26a-f. Only one current mirror circuit 66 is required since only one of the cylinder select switches 34a-f is closed at any given instance in time.

The current mirror circuit 66 includes first and second pnp transistors 68, 70 wherein both transistors 68, 70 have bases connected to the other and to the collector of the first transistor 68. The collectors of the transistors 68, 70 are further connected to system ground through first and second resistors 72, 74, respectively. The emitter of the first pnp transistor 68 is connected to the ignition capacitor first terminal 16a through the first current sensing resistor 64. The emitter of the second pnp transistor 70 is connected to the ignition capacitor first terminal 16a through a second current sensing resistor 76. As would be apparent to one skilled in

the art, selection of the ohmic values of the first and second current resistors 64, 76 controls the relationship between the input and output of the current mirror circuit 66. The output of the current sensing means 62 is delivered to a control logic means 78 which produces control signals in response to the current mirror output signal. The control signals are applied to the modulation switch 52 to respectively open and close the modulation switch 52. The control logic means 78 operates the modulation switch 52 while a selector switch 34a-f is closed such that the current flowing in an associated primary coil initially rises to a first current threshold II which is normally sufficient to cause a spark to arc an associated spark gap 22a-f, (see Fig. 3B.) Thereafter, the spark is maintained by modulating the current in the primary coil 26a-f between the first current threshold II and a second current threshold 12 which is lower than the first current threshold II. It should be noted that the primary current could be modulated at other levels to further minimize the current draw on the capacitor 18, as would be apparent to those skilled in the art. The time required to reach the first current threshold provides an indication of the secondary load because it is a function of the voltage required to initiate a spark across the spark plug gap, (i.e. the characteristic ionization potential V _gp , (see Fig. 3C.) The subject invention measures this time and processes it to determine a timing offset for adjusting ignition timing, as explained below.

The control logic means 78 includes a first comparator 80 having an inverting input terminal adapted to receive the current mirror output signal. The first comparator 80 is an open-collector type

comparator having its inverting input terminal connected to the junction of the second pnp transistor 70 and the second resistor 74 through an R-C network 82. The current output from the current mirror circuit 66 establishes a voltage across the second resistor 74 which is applied to the first comparator inverting input terminal. As should be apparent, this voltage is proportional to the current flowing through the first current sensing resistor 64 and thus to the current in the primary coil 26a-f. The R-C network 82 includes a third resistor 84 serially connected between the junction of the second transistor's emitter and the second resistor 74 and first comparator inverting input terminal. The R-C network 82 further includes a first capacitor 86 connected between the junction of the third resistor 84 and the first comparator inverting input terminal and system ground.

The non-inverting input terminal of the comparator 80 is connected to a voltage divider network 87 for controlling the voltage level applied thereto. More particularly, the non-inverting input terminal is connected to a preselected reference potential V _RE „ through a pull-up resistor 88 and to system ground through a fourth resistor 90. The non-inverting input terminal is further connected to the output terminal of the first comparator 80 through a seventh resistor 92. The output terminal of the first comparator 80 switches between logic "low" and logic "high" in response to the primary current signal rising above and falling below the the first and second current thresholds II, 12, respectively.

When the first comparator output terminal is pulled "high," the voltage divider network 87 applies a third voltage potential to the first comparator

non-inverting input terminal. The third voltage potential corresponds to a primary current having magnitude equal to the first current threshold II. The first comparator output terminal is pulled "low" when the voltage applied to its inverting input terminal rises to the third voltage potential, thereby indicating that the primary current has reached the first current threshold II. When the first comparator output terminal is pulled "low," the voltage divider network 87 applies a fourth voltage potential, which is lower than the third voltage potential, to the first comparator non-inverting input terminal. The fourth voltage potential corresponds to a primary current equal to the second current threshold 12. The output from the first comparator 80 is delivered to the modulation switch 52 to control operation of the switch. The modulation switch 52 is biased open and closed when the first comparator output is pulled "low" and "high," respectively. Referring now to Figs. 2A-E, operation of the present invention will be described in greater detail. Figs. 2A-E are waveform diagrams illustrating timing parameters involved in the present invention. More specifically. Fig. 2A illustrates the first signal as produced by the sensor means 48. Figs. 2B and 2C respectively illustrate production of a cylinder select signal and ignition in a respective cylinder of an engine without the present invention. Figs. 2D and 2E respectively illustrate production of a select signal and ignition in a respective cylinder in an engine equipped with the present invention.

The selector means 36 stores a value REFTIM in memory which is a reference timing value representing the time between a first point T. of the first signal and a second point T, of the first signal

which corresponds to the time at which the piston reaches TDC. In the preferred embodiment, the time period represented by REFTIM commences upon the rising edge of a particular pulse of the signal developed by the MPU 50 at the first point time T. and ends at the second point T_. Also stored is a value or a series of values DESTIM representing the time between a third point T_ corresponding to the desired ignition timing and the second point T_. The DESTIM values are developed as a function of engine parameters such as engine speed, engine load and air/fuel ratio. A single DESTIM value can be used for all engine cylinders; however, it is also foreseeable that different DESTIM values can be developed for each cylinder.

To achieve ignition at the third point T_, the selector means 36 calculates a control delay CNTDLY corresponding to the length of time between between the reference pulse at first point T. and the desired ignition timing the third point T_. This is achieved by subtracting the desired timing value DESTIM from the reference timing value REFTIM to determine the control delay value CNTDLY. However, in prior systems a timing error e _T - _rM arises due to a delay between production of the cylinder select signal at the third point T_ and actual ignition at T4 (see Fig. 3C.) The timing error e _TjM occurs as a result of the time required for the transformer to energize to a level sufficient to cause sparking. The timing error ^e τM ^{can det} ? ^rade overall engine operation and efficiency.

The immediate invention constitutes an improvement over prior timing controls for capacitor discharge ignition systems because it actively compensates for the timing error e _τ _ _M by measuring the

time between production of a cylinder select signal and ignition in an associated cylinder. This measured timing error e _n ,. _™ , is converted into a timing offset TIMOFFS which is used to adjust ignition timing. This is achieved by providing a combination of hardware and software for approximately sensing the timing error e_, _TM .. The timing offset TIMOFFS and desired timing value DESTIM are subtracted from the reference timing value REFTIM to arrive at the control delay value CNTDLY (see Fig. 2D) . The selector means 36 produces the selector signal at a time T ₅ , corresponding to the control delay value CNTDLY, following the first point T-, thereby causing ignition to occur at the desired time T_. Preferably, individual timing errors e _ττ . _M are measured each cylinders and used to calculated individual timing offsets TIMOFFS for each cylinder. However, due to limitations on the available processing capability of the MCU, the selector means 36 currently uses the average of the individual timing errors e _TIM to calculate a timing offset TIMOFFS which is used for all of the cylinders.

Referring now to Fig. 4, the electrical hardware required to practice the subject timing controller 8 will be discussed. The timing controller 8 measures the time between production of a cylinder select signal and ignition in a respective cylinder, as evidenced by the time at which the primary current signal reaches the first current threshold II. The invention is based on the premise that the step-up transformers 24a-f have a mutual inductance between their primary and secondary coils. Our research shows that changes in transformer output loads (i.e. the characteristic spark ionization potential V S_i_r) can accurately be determined by sensing changes in the primary inductance. Because the voltage provided by

the ignition capacitor 18 is maintained at essentially a constant magnitude by the power converter 14, an accurate indication of primary inductance and thus the timing error e _TIM can be obtained by measuring the time required for the primary current to reach a fixed current level.

The timing controller 8 includes a first means 98 which receives the cylinder select signals, senses a time delay between reception of a cylinder select signal and sparking in an associated cylinder. It should be noted that the timing error e _T -- _M is not an absolute measure because there is no absolute indication of when sparking occurs. Rather, what is measured is the time required for the primary current to reach the first current threshold II, and this time is a function of the time required for sparking to occur. The first means 98 responsively produces a timing error signal which is indicative of the sensed delay. The first means 98 includes a timer means 100 which measures a time delay between the production of a cylinder select signal and the time at which the current in an associated cylinder reaches the first preselected current threshold II. Preferably the timer means 100 includes a free-running clock which is internal to the MCU 38; however, it is foreseeable that the timer means 100 could be embodied in additional hardware circuitry. Production of a cylinder select signal causes a begin time (TB) to be stored in memory. The begin time (TB) corresponds to the time indicated by the free-running clock when the cylinder select signal is produced.

The first means 98 further includes a second comparator 102 having an inverting input terminal connected to the output of the first comparator 80 through a second R-C network 104. The second R-C

network 104 is provided to filter high frequencies caused by ignition noise. The second comparator 102 also has a non-inverting input terminal connected to a voltage divider network 106. The voltage divider network 106 includes sixth and seventh resistors 108, 110 serially connected between a reference voltage V-,- _™ and system ground. The second comparator non-inverting input terminal is connected between the resistors 108, 110, thereby maintaining the non-inverting input terminal at a preselected voltage potential. Preferably the preselected voltage potential is one-half the switching voltage of the comparator 102 to ensure proper switching of the comparator 102. The output terminal of the second comparator 102 is held high by a pull-up resistor 112 as long as the inverting input terminal has a higher potential than the non-inverting input terminal. More specifically, the second comparator 102 outputs a square wave signal which tracks the output signal from the first comparator 80.

A monostable multivibrator 114 is adapted to receive the primary current signal and produce a stop time signal in response to the primary current signal reaching the first current threshold II. The multivibrator 114 has an inverted clock pin (CLK') connected to the second comparator's output terminal and being adapted to sense the comparator's output signal. An inverted reset pin (RES') connected to the junction of the cylinder selector means 36 and the selector switch for receiving the selector signals. A second R-C network 116 is connected between the multivibrator 114 and the cylinder selector means 36 for filtering noise from the selector signal.

The multivibrator 114 also has an output terminal connected to an input terminal on the MCϋ 38.

The multivibrator output terminal is adapted to produce the stop time signal when the primary current reaches the first potential. More particularly, when the current in a primary coil reaches the first current potential, the second comparator output goes low. This low potential is received by the multivibrator inverted clock pin (CLK'), thereby turning the multivibrator 114 "on", (i.e. causing its output terminal (Q) to go high.) A timing circuit 118 is connected to input pins on the multivibrator to lock the multivibrator 114 "on" for a predetermined period. The timing circuit 118 is connected between the multivibrator external timer pin RX/CX and a reference voltage V -. The timing circuit includes an eighth resistor 120 and a second capacitor 122 which are connected between the reference potential and the external timing pin RX/CX. The components of the timing circuit are selected to keep the multivibrator 114 "on" for a preselected time, as is common in the art.

When the leading edge of the stop time signal is sensed by the MCU 38, the MCU 38 sets a stop time (TS) variable in memory in response to the time at which the stop time signal was received. This stop time corresponds to the time at which the primary current reaches the first current threshold II. The MCU 38 calculates the timing error e _τ _ _M by subtracting the begin time (TB) from the stop time (TB) .

Referring now to Figs. 5A-C, embodiments of software for controlling the MCU 38 in accordance with certain aspects of the immediate timing controller 8 is explained. Figs. 5A-C are flowcharts illustrating computer software subroutines for implementing the preferred embodiment of the present invention. The subroutines depicted in these flowcharts are

particularly well adapted for use with the MCU 38 and associated components described above, although any suitable microprocessor may be utilized in practicing an embodiment of the present invention. These flowcharts constitute a complete and workable design of the preferred software program, and have been reduced to practice on the series 6800 microprocessor system. The software subroutines may be readily coded from these detailed flowcharts using the instruction set associated with this system, or may be coded with the instructions of any other suitable conventional microprocessor. The process of writing software code from flowcharts such as these is a mere mechanical step for one skilled in the art. Fig. 5A corresponds to a Timing Error

Subroutine which is performed each time a cylinder select signal is produced to update a timing error in memory with timing errors e _TIM ^for individual cylinders. Figs. 5B and 5C represent alternate embodiments of a Timing Adjustment Subroutine which is executed each time a Main Control Routine (not shown) executes. The Timing Adjustment Subroutine retrieves timing errors e _TIM . from the timing error table and uses the timing errors e _TIM to achieve more accurate ignition timing. Fig. 5B illustrates an embodiment wherein timing is adjusted on a individual cylinder basis. In this embodiment, the timing error table contains a plurality of timing errors e__ for each cylinder. This embodiment could also be incorporated using a time averaged or filtered value of the timing errors ©„,-.„. Preferably, the table contains timing errors e _TI M ^{for tne} last five firings of each cylinder and the table is updated on a first-in, last-out basis. Fig. 5C illustrates an embodiment where an average of the timing errors e__ _M for all the engine

cylinders is used to adjust ignition timing for all of the cylinders. In the second embodiment, the timing error table only contains one timing error e. _™ for each cylinder. Referring now specifically to Fig. 5A, the

Timing Error Subroutine will be discussed. The Timing Error Subroutine is triggered by an interrupt operating in real-time which causes the subroutine to be executed each time a cylinder select signal is produced. Initially, in the block 200, the begin time (TB) , as indicated by the free-running clock, is stored in memory. Control is then passed to the block 205, where the routine checks to see if a stop time signal has been received from the multivibrator 114. When a stop time signal is detected in the block 205, control is passed to the block 210, thereby causing the stop time (TS) to be recorded in memory. If a stop time signal has not been received, control is passed to the block 215. In the block 215, the time elapsed since production of the cylinder select signal, as indicated by the free-running clock, is compared to a maximum time limit. The maximum time limit is empirically determined and it corresponds to a time which is significantly longer than a timing error e _TTM for normal ignition. In the preferred embodiment, the maximum time limit is on the order of 300 microseconds; however, this value will vary in dependance on the particular engine on which the system 8 is installed. If the elapsed time exceeds the maximum time limit, control is passed to the block 220. Otherwise control is returned to the block 205.

Control continues to loop between the blocks 205 to 215 until the maximum time limit is exceeded. Thereafter, control is passed to the block 220 where

memory is examined to see if a stop time (TS) was received and recorded in memory. If a stop time (TS) was recorded, control is passed to the block 225 where the timing error e _TTM is determined by subtracting the begin time (TB) from the stop time (TS) . The timing error e _TIM is then stored in a timing error table in memory. The format of the timing error table depends on which embodiment of the Timing Adjustment Subroutine is being utilized. Referring now to Fig. 5B a first embodiment of the Timing Adjustment Subroutine will be described. Initially in the block 300, a software pointer is set to point to the first timing error e _TIM in the timing - error table. The first embodiment calculates separate timing offsets TIMOFFS(n) for each cylinder in response to a plurality of past timing errors ^e _{ΦT .r} . _v -i f° ^{r a} respective cylinder, where n represents the cylinder and x represents the timing error. Control is then passed to the block 305 where the first timing error ^e _TIM , _{n x\} ^for this cylinder is retrieved from the error table. Thereafter, control is passed to the block 305 where the timing error ^e _r τ _n™ iM- _/ (n-,,x) _\ is examined to determine if it falls within an acceptable range of normal error times. As is illustrated in the Fig. 6, the timing errors e__ approximately follow an exponential curve. Fig. 6 is a graph of measured actual firing delay versus the measured timing error e _τ - _rM - Timing errors e _τ _ which fall outside the normal range are assumed to correspond to a faulty ignition conditions, such as short or open circuit conditions in the secondary coil.

If the timing error e _TIM , _{n χj} is not within the normal range, control is passed to the block 320 where the pointer x is incremented. Control is then

passed to the block 325 where it is determined if all the timing errors e _TIM for this cylinder have been retrieved. If they have not, control is returned to the block 305 causing the next timing error e„ TI- ₁ M _{ut /} ( _^ n,x). to be retrieved and checked to determine if it falls within the normal range. If it does, control is passed to the block 310 where the timing error ^e TIM(n,x) ^{is added to a timing error sum e} SUM(n)' Subsequently in the block 315 a counter is incremented to indicate the number of timing errors e _ττ „ which have been added to the timing error sura es_ _T u _T M _lur/ (n)..

After all of the timing errors e _TIM for a cylinder have been checked, control is passed to the block 330 where an average timing error ■2 _{AVG n} - _l is calculated by dividing the timing error sum e _s - _rτMi' n_ ^ ^v the value of the counter. After the average is calculated, the counter is reset to zero and control is then passed to the block 335. In the blocks 355-360, the average timing error e _{AVG n\} i examined to determine if it falls within one of three ranges within the normal range. For each range, a curve is used to linearly approximate a timing offset TIMOFFS(n) from the average timing error e- _VG . .. The curves are empirically determined under lab conditions by sensing secondary voltage to determine when the spark plug actually fires. Direct sensing secondary voltage on production engines is impractical because it requires an extra "tap" on the transformers and additional circuits to condition the high voltage signal from the transformer. These additional components would add unnecessary expense and complexity to the engine. Therefore, the subject invention relies on the measured timing errors e,--... to approximate when ignition actually occurs. It has been found that there is an approximately exponential

relationship between the measured timing error e_ and the time at which the plug actually fires. The exponential curve has been further subdivided into three ranges where the relationship between measured timing error _TIM and actual firing time can be linearized. If the delay time average does not fall within any of the three ranges, the timing offset TIMOFFS(n) is set to zero in the block 360.

Finally, in the block 370, the timing offset is used to calculate the control delay value CNTDLY(n) in accordance with the following equation:

CNTDLY(n) = REFTIM - DESTIM - TIMOFFS(n)

The control delay value CNTDLY(n) is subsequently used to control production of the cylinder select signal for a respective cylinder. Control is then passed to the block 380 where it is determined if control delay values CNTDLY have been calculated for each cylinder. If they have, control is returned to the Main Control Loop. Otherwise, control is passed to the block causing the pointer to be incremented to the next cylinder in the timing error table. The steps 305-370 are repeated until control delay values CNTDLY have been determined for each cylinder.

Referring now to Fig. 5C a second embodiment of the immediate timing control will be described. Fig. 5C generally parallels Fig. 5B with the exception that the routine is only executed once to derive a single control delay value CNTDLY which is used to adjust timing in all of the engine cylinders. The second embodiment can be used where processing time and capability are a concern. As indicated above, the timing error table for this embodiment only contains

one timing error e _™ , , per cylinder. Blocks 405 to 425 are repeated until all of the "normal" timing errors e_* _IM are retrieved from the timing error table and summed to arrive at a timing error sum e _s -_,. An average timing error e _AVG is subsequently calculated in the block 430 by dividing the timing error sum e _gUM by the number of "normal" timing errors. The average timing error e _AVG is converted into a timing offset TIMOFFS using one of three linear curves in the blocks 430 to 465. Finally, in the block 470 the control delay value CNTDLY is calculated in response to the average timing offset TIMOFFS. The control delay value CNTDLY is used to effect efficient timing in all of the cylinders.

Industrial Applicability

In operation of the engine a cylinder select means 36 produces cylinder select or ignition signals in a predetermined sequence and in response to sensed engine parameter to effect ignition in individual cylinder in accordance with a desired engine firing order. In response to production of the cylinder select signal, the Timing Error Subroutine is executed. Initially, a begin time (TB) is recorded in the block 200. The selector signal biases a respective selector switch 34 closed, thereby allowing current to flow through an associated primary coil 30. The current sensing means 62 senses the current flowing through the primary coil and responsively produces a primary current signal. The monostable multivibrator 114 responsively produces a stop time signal when the primary current reaches the first current threshold II. The stop time signal is sensed by the MCU 38 causing a stop time (TS) to be recorded

The Timing Error Subroutine continues to loop between the blocks 205 to 215 until the maximum time limit is exceeded. If a stop time was recoded, a timing error £„,_-. is calculated and stored in the timing error table.

The Timing Adjustment Subroutine executes each time the Main Control Loop is executed. The Timing Adjustment Loop retrieves the timing errors e _ττ - _M from the error table and calculates a timing error sum e _s __- in response to the sum of the "normal" timing errors. Subsequently, an average timing error e _AV _ is calculated by dividing the timing error sum ^e _sτ - _rM/n -. by the number of "normal" timing errors. One of three linear equations it then used to convert the average timing error e _A ττ _G into a timing offset TIMOFFS(n) .

The timing offset TIMOFFS(n) is then used to calculate the control delay value CNTDLY(n) which is used to control ignition timing in the engine cylinder n. The Timing Adjustment Subroutine is repeated until control delay values values CNTDLY(n) have been calculated for each cylinder. By calculating the control delay value CNTDLY in accordance with the present invention, it is possible to actively compensate ignition timing for the delay introduced by transformer energization and more precisely effect ignition at the desired point in time.