60/037,973, filed February 20, 1997, and titled "Apparatus and Method For Controlling Air/Fuel Ratio Using Ionization Measurements".

This invention relates generally to ignition systems in internal combustion engines and, more particularly, relates to an apparatus and method for utilizing ionization measurement for air/fuel ratio control to reduce engine emissions and increase engine efficiencies.

It is necessary to control the air/fuel ratio introduced into the cylinders of internal combustion engines for many reasons including emissions control, engine efficiency, catalytic converter efficiency, catalytic converter longevity and engine power. Numerous methods and apparatuses exist in the prior art to control the air/fuel ratio especially in light of governmental pressures to reduce certain emissions. Overall control of internal combustion engines is currently premised on the reading of various engine operating parameters such as engine speed, intake manifold pressure, coolant temperature, throttle position, and exhaust oxygen concentration. These parameters are used in conjunction with specific, predetermined base maps calibrated by a baseline engine to select the ignition timing, fuel injector duration, and exhaust gas recirculation ("EGR") of the engine so that the engine achieves maximum efficiency and minimum emissions as determined by the baseline engine.

Present engine control systems, and more specifically, air/fuel ratio control systems, do not adequately control internal combustion engines so that maximum efficiency and reduced emissions are achieved. For example, U.S. Pat. No. 4,543,934 provides a fuel-air mixture dilution control system by monitoring cycle-to-cycle fluctuations of the angular position of peak combustion pressure of each engine cylinder. This control system determines an air/fuel ratio at which engine stability changes between stable and unstable conditions. A controller attempts to continuously operate the engine at the engine stability point, leaning the fuel-air mixture until the engine becomes unstable, and enriching the fuel-air mixture until the engine becomes

stable again. This stability point is often beyond the point of maximum efficiency and is also often beyond the point of minimum emissions. Other control systems, such as the system disclosed in U.S. Pat. No. 4,736,724, control the air/fuel ratio by measuring the burn duration of each engine cylinder. The duration is compared to an adaptive engine map that determines the lean limit for the engine at a specific speed and load.

The engine is then controlled to operate at the most dilute point possible for a desired engine stability, but this point is often beyond the point of maximum efficiency, and is often beyond the point of minimum emissions. U.S. Pat. No. 4,621,603 discloses three different methods of controlling the level of fuel-air mixture dilution using pressure ratio management. The first system controls the amount of diluent at a specified value as a function of engine speed and load. The second system controls the amount of diluent to adjust the burn rate or combustion time. The third system controls the amount of diluent using cycle-to-cycle variability as both a method to balance fuel delivery to each combustion chamber, and as a method of stability control. Pressure ratio management allows for a simplified algorithm, but again does not supply the engine controller with enough information for complete engine control because taking pressure readings only at specific points allows the controller only to estimate engine stability, and therefore, this system suffers the same limitations of the previously mentioned systems. Alternatively, the system of U.S. Pat. No. 4,621,603 could be used at a specific air/fuel ratio that is calculated according to base maps, but even with an adaptive algorithm, the pressure ratio does not give enough information to allow the system to provide both maximum efficiency and minimum emissions. The system in U.S. Pat. No. 4,621,603, for example, would have extreme difficulty calculating the engine mean effective pressure if spark timing varies by large amounts. Such a calculation is necessary for an engine to achieve maximum efficiency at highly dilute mixtures and minimum emissions.

An important consideration in air/fuel ratio control methodology is catalytic converter performance. In order to optimize catalytic converter performance, a stoichiometric air/fuel ratio (about 14.7 to 1 for gasoline) is desirable. This is because with rich air/fuel ratios (i.e., less than 14.7 to 1) the fuel does not completely combust and the resulting emissions tend to clog the catalytic converter. A lean mixture (i.e., greater than 14.7 to 1), on the other side of stoichiometric, results in excess oxygen

("°2") in the emissions which in turn causes the operating temperature of the catalytic converter to rise and reduces or prevents the conversion of nitrogen-oxygen compounds ("NOx"). Exposure to elevated temperatures sharply reduces the operating life of the catalytic converter. In sum, catalytic converters are at their most efficient when a stoichiometric air/fuel ratio is used in the engine cylinders.

Most air/fuel ratio control methods use oxygen sensors in the exhaust system of the engine to measure the presence of oxygen which is indicative of whether the engine is running at stoichiometric mixtures. The 02 sensor measures the 02 in the exhaust of the engine in either the exhaust manifold or the exhaust pipe. One drawback to using an 02 sensor in the exhaust manifold or pipe is that the sensor reads a global air/fuel ratio for all engine cylinders. If one cylinder runs lean because, for example, a fuel injector is clogged, an air/fuel ratio controller that is based upon the 02 sensor will cause the other cylinders to run more richly thereby maintaining the desired global air/fuel ratio. Such a system achieves an average stoichiometric air/fuel ratio for all the cylinders, even though individual cylinders may be running at undesirably rich or lean mixtures.

There have been a number of attempts using 02 sensors to replace the above- described global emissions control with control of the air/fuel ratios in individual cylinders. The most common method of individually controlling the air/fuel ratio is to utilize fast acting 02 sensors to discern the exhaust 02 from each of the cylinders individually. The primary drawback with this implementation is that the 02 sensors are down-stream from the cylinders. The physical separation between the cylinder where combustion takes place and the sensor which measures the combustion characteristics introduces time delays, error and control difficulties. It is exceedingly difficult to calibrate this type of air/fuel ratio control system to account for the time delay and error at all engine speeds. Additionally, in some current production engines, four or more 02 sensors are required for this type of control thereby increasing the cost of implementation.

A relatively recent development allows certain in-cylinder combustion characteristics to be monitored. This monitoring technology revolves around electrically analyzing the gases in the cylinder before, during and after combustion.

These gases present in the cylinder include free ions which result from the combustion reaction.

The free ions present in the combustion gases are electrically conductive, and therefore measurable by applying a voltage across either an ionization probe or across the tip of a spark plug. The applied voltage induces a current in the ionized gases which can be measured to provide an ionization signal for analysis. For an example of ionization detection using the tip of a spark plug, see "Ignition System With Ionization Detection", U.S. Serial Number 08/595,558, filed February 1, 1996 which is commonly owned with the present invention and incorporated herein by reference.

There have been some attempts in the prior art to correlate an ionization signal to air/fuel ratios. The prior art strongly suggests, however, that feedback control of the air/fuel ratio in internal combustion engines based upon ionization signal data is impossible. See N. Callings et al., "Ignition Sensors for Feedback Control of Gasoline Engines", SAE Technical Paper Series No. 884711, 1988, pp. 43-47; R.L. Anderson, "In-Cylinder Measurement of Combustion Characteristics Using Ionization Sensors", SAE Technical Paper Series No. 860485, 1986, pp. 113-124.

In view of the foregoing, an object of the present invention to provide an improved control system and method for regulating the air/fuel ratio introduced into the cylinder of an internal combustion engine.

Another object of the present invention is to provide an improved control system and method of controlling the air/fuel ratio in an internal combustion engine based at least in part upon ionization detection.

Yet another object of the present invention is to provide a control system and method for controlling the air/fuel ratio in an internal combustion engine based upon an ionization signal derived from an ionization detection apparatus.

Still another object of the present invention is to provide a method for controlling the air/fuel ratio in an internal combustion engine that is inexpensive and efficient.

Summary of the Invention The foregoing objects are among those attained by the invention, which provides an air/fuel ratio control system for an internal combustion engine to reduce emissions

and increase engine efficiencies and includes in one aspect an ionization apparatus for measuring ionization within a combustion chamber of the engine and generating an ionization signal based upon the ionization measurements. Also included is an air/fuel ratio controller in electrical communication with the ionization apparatus. The controller receives the ionization signal and controls the air/fuel ratio in the engine based at least in part upon the ionization signal.

In another embodiment of the control system, the controller controls the air/fuel ratio based upon a first local peak in the ionization signal. In another embodiment, the controller controls the air/fuel ratio based upon maximizing the first local peak in the ionization signal. Another variation of the control system includes a processor for conditioning the ionization signal. The controller controls the air/fuel ratio based upon a the conditioned ionization signal.

In another embodiment the controller controls the air/fuel ratio to substantially maximize or minimize a second local peak in the ionization signal.

In still another preferred embodiment, the combustion chamber of the internal combustion engine includes a plurality of cylinders. Each cylinder is independently coupled to an ionization apparatus for detecting ionization within the cylinder and generating an ionization signal based upon the ionization measurements. The controller may independently control the air/fuel ratio two or more of the cylinders. The ionization measuring apparatus may further comprise a spark plug or an ionization probe in the cylinder for generating the ionization signal.

A method for reducing emissions and increasing engine efficiencies in an internal combustion engine is also disclosed. The method includes detecting ionization within a combustion cylinder of the engine with an ionization apparatus and generating an ionization signal with the ionization apparatus based upon the ionization detection.

The method further includes a step of adjusting an air/fuel mixture injected into the cylinder based upon the ionization signal.

The adjusting step of the method may be based on a number of features of the ionization signal, including a first local peak, maximizing the first local peak, a second local peak or maximizing and/or minimizing the second local peak. The method may further include a step of comparing the first local peak of the ionization signal of a first

cylinder with a first local peak of the ionization signal of a second cylinder. And may also be based upon maintaining the first local peaks of the first and second cylinder at substantially equal amplitudes.

Brief Description of the Drawings Fig. 1 is a graphical depiction of various emissions (specifically the gases CO, NO and HC) versus the excess air factor ("A"; defined below) for a typical internal combustion engine.

Fig. 2 is a schematic view depicting an air/fuel ratio control system of the present invention.

Fig. 3 is block diagram of the air/fuel ratio control system of the present invention.

Fig. 4 is a graphical presentation of experimental data showing ionization current versus engine piston crank angle for various engine load conditions.

Fig. 5 is a graphical presentation of experimental data showing cylinder pressure versus engine piston crank angle for various engine load conditions.

Fig. 6 is a graphical presentation of experimental data showing a correlation between the excess air factor (X) and ionization for numerous engine load conditions.

Fig. 7 is a graphical presentation of experimental data showing ionization versus engine load for various values of the excess air factor (A).

Description of Preferred Embodiments Referring initially to Fig. 1, a graph depicting various emissions gases versus an excess-air factor ("X") for a typical engine under typical operating conditions is shown.

Fig. 1 is derived from the Bosch Automotive Handbook, 1986, page 439. As used herein, the excess-air factor (X) is simply a factor indicating the amount that the air/fuel ratio is above or below a stoichiometric mixture (e.g., 14.7 to 1 for gasoline). Thus, for example, A=1 corresponds to an air/fuel ratio equal to stoichiometric, A=1.2 corresponds to an air/fuel ratio that is 120% of stoichiometric, A=0.8 corresponds to an air/fuel ratio that is 80% of stoichiometric, and A=2 corresponds to an air/fuel ratio twice stoichiometric (e.g., 29.4 to 1 for gasoline).

It is seen in Fig. 1 that the concentration of NO peaks at a value slightly leaner (R > 1) than a stoichiometric air/fuel ratio. The presence of NO is a sample representation of the presence of all NOx gases.

As explained above, ionization detection and measurement is known in the art.

One type of ionization detection apparatus for detecting and measuring ionization includes a spark plug which utilizes a spark gap across which a voltage is applied. The voltage across the spark gap induces a current (across the spark gap) in the ionization gases during and after combustion. The current is detected by a circuit and analyzed to determine combustion characteristics. See, for example, "Ignition System With Ionization Detection", U.S. Serial Number 08/595,558, filed February 1, 1996, incorporated herein by reference. Another ionization detection apparatus employs a probe, similar to the spark plug, except its primary function is to detect ionization gases.

Turning now to Fig. 2, a control system 10 according to the present invention is shown. An internal combustion engine (not shown) includes a cylinder 12, a piston 14, an intake valve 16 and an exhaust valve 18. An intake manifold 20 is in communication with the cylinder 12 through the intake valve 16. An exhaust manifold 22 receives exhaust gases from the cylinder 12 via the exhaust valve 18. A spark plug 20 with a spark gap 22 ignites the air and fuel in cylinder 12.

A conventional engine controller 30 typically controls various engine operating parameters and components including fuel injector 32 and idle air valve 34. The engine controller 30 also receives position data from a throttle position sensor (not shown) coupled to a throttle valve 36 and manifold pressure data from a manifold pressure sensor 38. The throttle valve 36 provided in the intake manifold 20 controls air flow to the cylinder 12. The engine controller 30 also typically receives data from an O2 sensor 40 located in the exhaust manifold 22 or elsewhere downstream from the exhaust valve 18.

An ionization detection apparatus 50 includes an ionization detector which, as shown in Fig. 2, comprises spark plug 20 located partially in the cylinder to detect ionization in the cylinder 12. The ionization detected by the spark plug or ionization detector 20 is communicated to the ionization apparatus 50. The ionization apparatus 50 receives ionization data from the ionization detector (either the spark plug 20, an

ionization probe or any another conventional device for detecting ionization) and communicates an ionization signal 52 to the engine controller 30.

The engine controller 30 controls the fuel injector 32 and may control the throttle valve 36 to deliver air and fuel, at a desired ratio, to the cylinder 12. The engine controller 30 may be any conventional controller adapted to receive feedback, in the form of ionization signal 52, from the ionization apparatus 50 to adjust the air/fuel ratio. The use of the ionization signal 52 by the engine controller is described more fully below.

In Fig. 3, there is shown a block diagram of the control system 10 in accordance with the present invention. Engine 11 includes the spark plug 20 which, in this embodiment, provides ionization detection (other ionization detection apparatus may also be used such as an ionization probe). The ionization apparatus 50 receives ionization detection data from the spark plug 20 and converts it into an ionization signal 52. The ionization signal 52 is processed and analyzed, which may include a statistical analysis (explained further below), in processor 50b. Processed ionization signals 52a and 52b are transmitted to the engine controller 30 (also commonly referred to as an engine control unit ("ECU")) which in turn provides the ionization apparatus 50 with other engine data including engine speed, ignition timing and ignition duration via signal 56. The engine controller 30 also receives data from other engine sensors such as engine speed and °2 sensor data. Among other operating parameters, the engine controller 30 controls the fuel introduced into the engine 11 via the fuel injector 32 and fuel pump 33. The engine controller may also control the air introduced to the engine (not shown in Fig. 3). The engine controller 30 (or ECU) may thereby control the air/fuel ratio based at least in part on the ionization signal 52.

The ionization apparatus 50 includes an ionization circuit 50a and may also include a processor 50b. The processor may include analysis software, including statistical analysis routines for analyzing the ionization signal 52. The ionization apparatus may further include conventional buffers and memory for storing the ionization signal 52 and the processed signals 52a, 52b.

In Fig. 4 there are shown experimental data that include a statistical average of 100 combustion cycles of ionization data at five different load levels on a particular engine. The curves in Fig. 4 are labeled 1, 2, 3, 4 and 5 and represent the ionization

signal (as a current in milliamperes) as a function of piston crank angle (in degrees; wherein 360 degrees is top dead center) for different and increasing engine loads, respectively.

In general, chemi-ionization in the flame zone is primarily responsible for the measured ionization data. However, there are two local peaks 11, 12 seen in these curves. The first local peak 11 primarily relates to flame speed in the engine cylinder.

Clearly, when the air and fuel combust, the chemical reaction sharply increases the number of ions present in the cylinder chamber, and hence ionization detection increases.

The second local peak 12 seen in some of the curves of Fig. 4 relates to temperature and pressure-based ionization and concentration. The second local peak is primarily related to the presence of NOx molecules or NOx emissions developed during the combustion process. When the temperature and pressure in the cylinder increase immediately after combustion occurs, the concentration and production of NOx correspondingly increases. It is seen that the curves 1, 2 corresponding to lower load levels do not have a second local peak. This is because the load level is too low to generate sufficient temperature and pressure to increase the quantity and concentration of NOx and cause a second local peak in the ionization signal. In curves 3, 4 and 5, the increase in load and resulting increase in pressure from the combustion process increases the temperature and the NO emissions, thereby producing increased ionization (and increased concentration of the ions) in the cylinder and resulting in an ionization curve with a second local peak at 12.

As seen in Fig. 5, the second local peak 12 accurately locates (in the combustion cycle) the peak pressure in the cylinder. The curves in Fig. 5 are labeled la, 2a, 3a, 4a and 5a and represent relative average pressure over 100 combustion cycles as a function of piston crank angle (in degrees; wherein 360 degrees is top dead center) for different and increasing engine loads, respectively. These curves directly correspond to and are measurements from the same test as the curves shown in Fig. 4. In Fig. 5, it is seen that the peak pressure in the cylinder occurs at approximately 395 degrees. This is approximately the same location as the second local peak 12 of curves 3, 4 and 5 shown in Fig. 4. Thus, by determining the location of the second local peak 12 from

the ionization data, the location of the peak pressure can be derived from the ionization data.

The ionization information in Fig. 4 can be statistically processed and analyzed to provide data that is averaged over numerous combustion cycles and has noise from cycle to cycle variations filtered out. Statistical processing and analysis may use any of a number of conventional statistical methods on the overall ionization data, and these are especially useful in the analysis of the first local peak 11 (the flame propagation portion) as well as the maximum intensity and location of the second local peak 12 (the pressure and temperature portion).

Turning now to Fig. 6, experimental data measuring the first local peak of the ionization signal as a function of A is shown. The measured ionization was converted into an ionization signal in volts. The data shown as curve 6a is the first local peak (the flame ionization portion) of the ionization signal versus A (i.e., various air/fuel ratio conditions). The curve 6a roughly drawn through the data points reaches a maximum between approximately A=0.90 and A=0.95.

A similar curve, curve 6b, represents the second local peak of the ionization signal as a function of A. This curve 6b reaches its maximum at approximately A=1 .00 to 1.10.

Thus, as air/fuel ratio is varied (rather than as a function of piston crank angle as in Figs. 4 and 5) over numerous engine cycles, the first local peak of the ionization signal will reach a maximum in the range of A = 0.90 to 0.95. The second local peak of the ionization signal will reach a maximum in the range of A=1.00 to 1.10. As discussed above, in order for there to be a second local peak, the load on the engine must be sufficiently high to raise the temperature and pressure in the cylinder to promote creation and concentration of NOx molecules. This effect must be great enough so that the second local peak has a sufficient magnitude to be detected.

For the reason that the second local peak is more difficult to measure, the first local peak in the ionization signal is the more reliable of the two local peaks to be used for air/fuel ratio control. Based on the data depicted in Figs. 4 and 6, it is clear that the magnitude of the first local peak 11 in the ionization curves 1, 2, 3, 4 and 5, can change as a function of both A and load. It is therefore important to insure that minimum load variation when compiling statistical averages to analyze the air/fuel ratio

and optimize the air/fuel ratio. This can be accomplished by insuring that ignition timing, mass air flow and engine revolutions per minute ("rpm") are held constant during the change in air/fuel ratio that is associated with the optimization process. It is also possible to make the changes to only one cylinder at a time, in order to determine the statistical information for that cylinder, without affecting the load of the overall engine.

Fig. 7 shows a graph of the first local peak of the ionization signal versus load for three different air/fuel ratios. The topmost curve 7 is for A=1. The other curves 8, 9 are for A=1 .2 and A=0.7, respectively. It is apparent from Fig. 7 that over a certain range of cylinder loading conditions, the ionization level for stoichiometric air/fuel mixtures is higher (and measurably so) than that for air/fuel mixtures corresponding to A=1.2 and 0.7.

A preferred method of achieving a stoichiometric mixture in each cylinder utilizes a single °2 sensor and air/fuel ratio control based upon the ionization signal in each individual cylinder. At least one °2 sensor in the exhaust system of the engine is probably required in engines with a catalytic converter. A global determination (rather than cylinder-by-cylinder) of exhaust gases may be necessary because there is usually just one catalytic converter in the exhaust system of the engine. The O, sensor in the exhaust is used to determine the total or global stoichiometric mixture of the engine.

The engine controller then utilizes methodology for equalizing the amplitude or the location (or both) of first local peak of the ionization signal in each individual cylinder. When statistical equality in the individual cylinders is achieved with an air/fuel mixture at stoichiometry based on the O2 sensor, and knowing the slope of the first local peak of the ionization signal relative the stoichiometric mixture, the engine will be in balance. In this type of system, the ionization is used as a balancing mechanism for improving catalyst efficiency by maintaining a mixture closer to stoichiometric in all cylinders, as compared to current production systems that utilize multiple exhaust oxygen seniors, in order to get sensitivity to the individual cylinders, as well as to the global engine air/fuel ratio.

One preferred method for controlling a stoichiometric mixture for each cylinder is to approximately equalize the statistical first local peak of the ionization signal amongst all cylinders for a given engine operating condition. Because of the slope of

the ionization curve, perturbations of the air/fuel ratio from rich to lean of stoichiometric will be readily detected. The lean cylinders will have significantly different first local peak (of the ionization signal) amplitudes as compared to the rich cylinders. This will give a clear indication of which cylinders are running rich, and which are running lean, thereby allowing the system to achieve a better balance of the overall air/fuel ratio from each cylinder. Then the air/fuel ratios in individual cylinders can be controllably adjusted to achieve relative equality of individual first local peaks of the ionization signals among the cylinders. This adjustment would be performed relatively slowly, at fairly steady engine operating conditions, so that statistical information can be gathered and analyzed by the engine controller. The controller would then determine the offset value of each fuel injector (and hence the quantity of fuel) in order to achieve approximate equality between the different cylinders. These offsets would then be used during the entire engine operating range, in order to maintain or evenly balance air/fuel ratio amongst the cylinders under all operating conditions.

Engine modeling can be utilized to determine the off-set peak ionization relative to the stoichiometric air/fuel ratio of the particular engine. This methodology can be accomplished in each cylinder separately so that individual cylinder air/fuel ratio control can be optimized to a stoichiometric mixture. Each cylinder off-set from the base engine map can be determined and then utilized to maintain that particular cylinder's stoichiometric air/fuel ratio.

Due to manufacturing imperfections and other operating variables, the amount of air and fuel delivered to each cylinder is at least slightly different. Using the air/fuel ratio control system as depicted in Figs. 2 and 3, we can calibrate for the appropriate injection time for each cylinder's stoichiometric air/fuel ratio. The calibration of an engine is very important to the emissions level achieved in the engine. One of the things that is most difficult parameters to calibrate in an engine is the amount of air allowed into each cylinder during each cycle. This has a lot to do with intake manifold design, valve timing, cam profiles, as well as conditions of back pressure that change the EGR inherent in the engine. These differences in air admitted into the cylinder in each cycle, as well as the air admitted into each cylinder versus its neighboring

cylinders, makes it difficult for conventional systems to accurately determine a stoichiometric mixture for each cylinder.

With the ability to adaptively control around the stoichiometric mixture using ionization signal data, the engine control system can achieve an accurate off-set in fuel control to accommodate the differences in each cylinder's air intake. This methodology can also accommodate for changes over the life of the engine, like clogging of fuel injectors or other wearing conditions that may change the air and fuel conditions or delivery thereof for each particular cylinder.

Certain engines, such as lawn mower engines and small utility engines, do not have the same emission standards or requirements for catalytic converters that current automotive production engines require. For these engines, an ionization methodology for air/fuel ratio control is even more valuable than it is in some automotive applications. In these engines, an ignition system is required, however, an oxygen sensor is not the optimum methodology for air/fuel ratio control given the fact that these engines in most cases meet the emission standards without a catalytic converter.

These engines require accurate control of the air/fuel ratio to prevent running too rich and producing too much pollution, as well as not running too lean and overheating the engine.

In has been determined that these smaller utility engines have an optimum operating range in the vicinity of X = 0.90 to 0.95, a level at which they operate efficiently and produce reasonably low levels of hydrocarbon and carbon monoxide emissions. The control strategy for these engines is ideal for ionization detection methodology because it simply entails the maximization of the first local peak of ionization signal during almost all operating conditions of the engine. A very simple control system can be employed with an ignition system (that includes an ionization apparatus), to achieve a low-cost, accurate and efficient air/fuel ratio control system.

In other industrial engine applications, misfire detection can be employed to determine the lean operating limit of a particular engine. The lean operating limit can be determined with the misfire detection capability of the ionization signal. Engine misfire is detected when there is little or no amplitude in the ionization signal across the entire combustion duration time frame. A control strategy that leans the air/fuel ratio just short of engine misfire, can be utilized to maximize fuel efficiency in an

engine that employs an ionization detection circuitry. The control strategy utilized would be one that incrementally makes the air/fuel ratio leaner and leaner, until a misfire is detected in one of the cylinders, in a global strategy, or in each individual cylinder to determine each individual cylinder's lean misfire limit, and then backing off a certain factor from that misfiring air/fuel ratio in order to operate at a stable condition with some margin of assurance that a misfire is not going to occur. In certain small engine applications two strategies may be advantageously used. One is a maximization strategy that would be utilized at certain high speed and load conditions and the other is the lean operating limit strategy described above. The two strategies would be employed under conditions of engine operation in order to achieve the best balance between emissions and proper operation of the engine during high load conditions.

In certain engine applications the control system tuning capability makes it possible to achieve a desired air/fuel ratio simply by maximizing the ionization signal, the first or second peak of the ionization signal, or an integral of the ionization signal (or a combination thereof). This significantly simplifies the algorithm needed for achieving a desired air/fuel ratio in each cylinder.

Using the above described ionization detection and analysis and the correlation between ionization and air/fuel ratio, feedback may be provided to an air/fuel ratio control system. Each cylinder can be optimized for either a stoichiometric air/fuel ratio, or an appropriate air/fuel ratio for the operating condition desired by the engine controller.

The use of ionization sensing for cylinder-to-cylinder air/fuel ratio control supplements other potential uses of the ionization signal. See, e.g., in Appendix A attached hereto and incorporated herein in full by reference a pre-print of a paper to be published by the Society of Automotive Engineers as SAE Technical Paper Series 980166, by Eric N. Balles, Edward A. VanDyne, Alexandre M. Wahl Kenneth Ratton and Ming-Chia Lai, "In-Cylinder Air/Fuel Ratio Approximation Using Spark Gap Ionization Sensing". The ionization signal can deliver multiple pieces of information regarding the events and conditions in the combustion chamber. As an example, the ionization signal can determine misfire, knocking conditions, as well as variations in the cylinder pressure of an engine. Additionally, the ionization signal can be utilized to control the exhaust gas re-circulation ("EGR") system. Sensitivity of the ionization

signal sensitivity to NOx in the vicinity of the second local peak can be used by the EGR system to reduce the NOx emissions. This EGR control system can utilize comparative ionization values to reduce NOx levels without the presence of misfire.

The combination of magnitude of the second local peak of the ionization signal and the statistical magnitude of the misfire occurrence can be utilized together to control the maximum tolerable EGR achievable in the engine at each running condition.

It has been shown that because NOx is the most conductive of the gases resulting from combustion, the second peak of the ionization signal increases as a function of the NOx molecules available. This correlation between ionization signal and the presence of NOx molecules follows the load on the engine, whereby higher NOx emissions are indicated by higher ionization signal measurements.

The use of ionization detection and analysis can be used to minimize NOx emissions because of the direct correlation between the second local peak in the ionization signal and NOx emissions. Therefore, based upon the second local peak of the ionization signal, information about the concentration and amount of NOx, present in the combustion chamber can be determined. Over a range of air/fuel ratios, NOx emissions increase as the air/fuel ratio is increased from a rich mixture to a stoichiometric mixture. NOx emissions peak at a air/fuel ratio that is slightly higher than stoichiometric, and then fall again after about a 16 to 1 air/fuel ratio (for gasoline).

This air/fuel ratio (R between approximately 1.00 to 1.10) is typically the where NOx emissions are at their highest. Again, see Fig. 1.

Utilizing this concept, that NOX emissions peak slightly above stoichiometric and this peak corresponds to the second local peak in the ionization signal, the air/fuel ratio can be adaptively controlled based on the ionization signal. Using the relative increase in ionization signal amplitude together with the sensitivity to other information within the ionization signal, air/fuel ratio can be optimized for each cylinder. In conjunction with an oxygen sensor measuring the overall oxygen level of the entire engine. the ionization signal within each cylinder can be used to provide valuable feedback control for modifying the air/fuel ratio in individual cylinders thereby providing balance to all cylinders.

It should be understood that the preceding is merely a detailed description of certain preferred embodiments. It therefore should be apparent to those skilled in the art that various modifications and equivalents can be made without departing from the spirit or scope of the invention. APPENDIX A In-Cylinder Air/Fuel Ratio Approximation Using Spark Gap Ionization Sensing Eric N. Balles, Edward A. VanDyne, and Alexandre M. Wahl Adrenaline Research, Inc.

Kenneth Ratton and Ming-Chia Lai Wayne State University ABSTRACT Experiments were conducted on a single cylinder engine to measure the ionization current across the spark plug electrodes as a function of key operating parameters including air/fuei ratio. A unique ignition circuit was adapted to measure the ion current as early as 300 microseconds after the initiation of spark discharge.

A strong relationship between air/fuel ratio and features of the measured ion current was observed. This relationship can be exploited via relatively simple algorithms in a wide range of electronic engine control strategies. Measurements of spark plug ion current for approximating air/fuel ratio may be especially useful for use with low cost mixture control in small engine applications. Cyiinder-to-cylinder mixture balancing in conjunction with a global exhaust gas oxygen sensor is another promising application of spark plug ion current measurement.

INTRODUCTION Sensors perform a critical role in the control of modern, highly developed internal combustion engines. A wide variety of information ranging from ambient conditions, to engine operating conditions (e.g., speed, throttle position, intake manifold pressure, etc.), to chemical composition (e.g., exhaust oxygen concentration) are interpreted by the vehicle's engine management system which in turn controls functions such as spark timing and fuel injection. The sophistication of today's engine systems allow passenger cars to operate with low pollutant emissions and high thermal efficiency. On the other hand, small utility engines, which have seen little or no regulation in the past, pose extra challenges for meeting future emission levels due to the size, weight, and cost constraints imposed by their unique applications.

Because of these constraints, sensors and control strategies that are appropriate for passenger cars frequently do not translate well to small engines.

Continued development of new and advanced sensors, corresponding control algorithms, and actuators is required to meet the increasingly stringent worldwide emission standards and fuel economy mandates for passenger cars and other internal combustion engine applications.

The ability to obtain useful information about the combustion event via the measurement of incylinder ionization current has been known for decades and continues to be studied for use with modern engine control strategies.1234 There are several examples of the use of ionization current sensing in production engine applications. Saab implements a version of spark plug ionization sensing for misfire detection to meet on-board diagnostic requirements. Caterpillar, in one of its lean burn natural gas engines, uses an ionization probe (non- firing spark plug) in the combustion chamber opposite the ignition spark plug. The ionization probe detects the flame arrival time which is then used to maintain a target overall flame propagation time by controlling the air/fuel ratio. This strategy works well because of the strong correlation of airifuel with flame speed.

To further study the use of spark plug ionization current sensing as a method for air/fuel ratio approximation, an experimental program was conducted at Wayne State University on a small single cylinder overhead valve production engine. A Dual Energy Ignition system. developed by Adrenaline Research, Inc., and patented by the Massachusetts Institute of Technology,s was used in these experiments. This ignition circuit was adapted to sense the ion current at the spark plug gap and patented separateiy by Adrenaline 6 Results described in this paper focus on the relationship between air/fuel ratio and the ion current measurements.

IGNITION AND ION CURRENT SENSING SYSTEM CIRCUIT DESIGN OVERVIEW - The ignition and ion current sensing system used in this experimental program was based on Adrenaline's Dual Energy Ignition (DEI) circuit. This circuit is the combination of a capacitive discharge ignition circuit and a strobe light circuit, which allows for both high voltage and high current discharges across the spark plug gap for improved ignition.7.8 After the Dual Energy circuit is discharged, the secondary capacitor (seen in Figure 1) is recharged to present a high voltage across the spark gap electrodes for ionization sensing. A zener diode and parallel resistor between the secondary capacitor and ground allow this portion of the circuit to deliver the high current for ignition and to perform the ion current sensing function. The zener diode provides the circuit path for charging and discharging of the high voltage capacitor for ignition purposes. The sensing resistor provides a <BR> <BR> measurable voltage 0: praportional to the ion current flowing across the spark gap. Prlmary capacitor I I t r .1 Triggering 30kV Circuitry r l Diode Spark Gap 3 0+ e teconcapa7 Source Ionization Analysis e ~ Circuitry Secondary Battery power Source 1 (loop to -s200v) Figure 1. Schematic diagram of the basic Dual Energy Ignition circuit combined with an ionization sensing circuit.

SYSTEM DESIGN FEATURES WHICH ENHANCE IONIZATION SENSING - Implementation of this combined approach has several significant advantages for spark gap ionization sensing.

1. Short duration spark discharges (typical of CD ignition systems) allow for ionization sensing to begin within 300 lls after the initial spark has occurred.

This is a significantly shorter time delay period than an inductive ignition can provide which has typical spark duration of 1.5 to 3 ms. Beginning ionization sensing within a few hundred microseconds after spark provides data during the chemi-ionization phase (primarily flame propagation in the vicinity of the spark gap). This information would otherwise be missed with longer duration ignition systems.

2. The large secondary capacitor used in the DEI circuit for high current spark discharge is also used to hold the voltage across the spark plug electrodes for ion current measurement. This large capacitor retains most of its charge during the ionization portion of the combustion process - the ionization current is approximately 30 to150 itA out of a capacitor that can sustain 100 A for 3 ps when discharging in the ignition phase. This feature therefore provides essentially a constant voltage throughout the ion current measurement phase.

3. Another benefit is the low resistance of the capacitive discharge coil's secondary winding that allows for a higher current flow through the secondary windings during the ionization phase. The ability to present a high, negative voltage across the gap also improves the ionization signal-to-noise ratio.

4. A major cost benefit is derived from using the same circuit to produce both the high-energy ignition that assists combustion, especially in the lean burn mode, as well as the diagnostic ionization feedback.

The combination of improved lean burn ignition plus ionization feedback for misfire purposes is especially useful in cold starting engines and operating just inside the misfire limit for leanest possible cold starts in a closed loop fashion. Another benefit of the combined circuits is that it can be used to detect pre- ignition as well as diagnose ignition problems like improper discharge of the ignition. primary or secondary, as well as improper charging of the capacitor itself, giving us a complete diagnostic tool for the total system.

CRITICAL CIRCUIT DESIGN ISSUES - The most important issue for reliability of an ionization circuit is the conductivity of the contacts made between the coil and the spark plug in order to maintain a reliable path for the ion current which is only between 30 and 150 pA. It is also important to have a 12-bit analog-to<ligital converter for measuring the voltage drop across the sensing resistor in order to get sufficient resolution of the signal to calculate misfire, knock, and other information. In order not to leave a high voltage at the gap after the ignition has been tumed off, a bleed resistor is incorporated in the charge circuit to prevent the secondary capacitor from maintaining a high voltage charge. This bleed resistor results in a zero offset during ion current measurement. This needs to be taken into account when analyzing the signal and determining the zero ion current level which can be affected by the secondary capacitor voltage.

EXPERIMENT DESCRIPTION ENGINE AND DYNAMOMETEK SETUP - Experiments to measure spark plug ion current and to correlate with engine operating conditions were conducted on a small production engine in a controlled laboratory environment.

The test engine was a 219 cc VanGuard 9 hp Briggs and Stratton engine (model 185432). This overhead valve. air-cooled engine was originally supplied with a gasoline carburetor which was converted to propane fuel for these experiments. A surge tank was added to the intake system to improve the air mass flow measurement accuracy. The exhaust system was modified to allow for a lambda sensor but retained the factory muffler which exhausted to atmospheric pressure.

Table 1 Test Engine Specifications displacement 219 cc bore x stroke 59 mm x 80 mm compression ratio j 8.1 rated power (gasoline) 6.7 kW (9 hp) peak torque (gasoline) 7.5 Nm (13 ft-lbs) peak torque (propane 6.6 Nm (11.5 ft-lbs) conversion) The dynamometer and engine control, as well as the data acquisition was carried out using a PC-based LabVIEW system.9 The configuration is shown schematically in Figure 2.

KEY INSTRUMENTATION - This experiment used a heated Bosch lambda sensor, LSM11, for exhaust air/fuel ratio determination. The sensor was connected to a Bosch LA2 amplifier and heater control unit. This was the primary measurement used to correlate with measured spark plug ion current.

The ionization sensing portion of Adrenaline's ignition system was configured by running an empirical set of experiments. Configuration consisted basically of setting the voltage applied across the spark gap for ion current measurement and the time delay between the initiation of the spark discharge (ignition) and the recharge of the secondary capacitor (see Fig 1). The best voltage and delay time settings, determined empirically, translated to values of approximately 200 volts and 300 i£s delay (i.e., ion current data within approximately five crank degrees after ignition at an engine speed of 2800 rev/min). T- I 1 I SC.205 J I -L ' scoao cs Adanw ~ SC MO ~ c~d l sh | AT10 l6E I | P | ~ NllBosd | prosne u ~ tnj£cLion Ventun' AasDIer END into ratt-t Nl hd e W Nl 1 ì ì J L CS At T Sststs dac mnn,IPM I Snaa ha 7- Enmrr54 Li 15HP d + Dc X Sz56 1r1 Figure 2. Schematic diagram of dynamometer, engine control, and data acquisition systems.

RESULTS Spark plug gap ion current data was collected over a range of air/fuel ratio and torque set points. in these experiments speed was held constant at 2800 revtmin while air/fuel ratio traverses (A = 0.85 to A = 1.10) were conducted at discrete torque set points (approximately 40% to 60% of full ioad). Throttle angle was adjusted to maintain the torque set point as air/fuel ratio was varied.

Spark timing was fixed for all experiment test points (spark set point at approximately 29 degrees BTC).

Characteristic features of the ion current changed in a well-behaved manner as air/fuel ratio changed. Figures 3a through 3d show the spark plug gap ion current as a function of crank angle. The ionization curves shown are 40 cycle averages and each curve represents a distinct air/fuel ratio. Each graph represents a different load condition. Overall, the characteristic shape of these ionization curves indicate that chemi-ionization (from species in the reacting flame zone) is primarily responsible for the measured ion current across the spark gap under these test conditions. Other researchers working with production automotive engines have shown that the chemi-ionization portion of the ion current signal is a strong function of air/fuel ratio.2 The results presented here for a small utility engine also show that certain features of the ion current may be used in conjunction with the appropriate algorithms or neural network to approximate overall airlfuel ratio.

Severai straighfforward characteristics of the ionization waveform were analyzed as a function of air/fuei ratio.

Figure 4 shows how the peak spark gap ion current is influenced by air/fuel ratio. The peak current decreases significantly as a function of airlfuel ratio as the mixture is leaned out past A=0.95 (approximately). Ion current appears to peak in the range of A = 0.90 to 0.95 for these conditions. This was confirmed by other data from this engine (partial data set taken earlier in the experimental program) which showed a dramatic fall-off in ion current as the mixture was made very rich (e.g., A = 0.7).9 Figure 5 illustrates how the crank angle location of peak ion current changes with air/fuei ratio. In general, the peak current location moves closer to top center as the mixture becomes richer. Figure 6 is an example of the ionization curve integral from different air/fuel ratios at fixed torque. The ion current integral provides similar information to the peak current because of the characteristic shape of the curves and the fact that the majority of the measured current is due to the chemi- ionization (flame front) effect. Peak ion current, location of peak ion current, and ion current integral are three simple examples that correlate in some way with overall air/fuel ratio.

Ionization Current (µA) Crank Angle (degrees after top center) Ionization Current (µA) Crank Angle (degrees after top center) Ionization Current (µA) Crank Angle (degrees after top center) Ionization Current (µA) Crank Angle (degrees after top center) Fig 3a - 3d. ion current waveforms (40 cycle averages) as a function of air/fuel ratio at discrete load points Maximum Ionization Current (µA) Normalized Airffuel Ratio Figure 4. Peak ion current as a function of air/fuel ratio Crank Angle (degrees after top center) Norma@@zed Air/Fuel Ratio Figure 5. Crank angle location of peak ion current as a function of air/fuel ratio Ionization Integral Crank Angle (degrees after top center Figure 6. Integral of the ion current for different air/fuel ratio test conditions at a fixed load (5.4 ft-lbs) APPLICATION TO ENGINE CONTROL SMALL UTILITY ENGINES - Spark gap ionization sensing can be applied to achieve the lower hydrocarbon emission requirements of small utility engines. These engines as a class emit high levels of unbumed hydrocarbon for several reasons. For example, very rich air/fuel mixtures are typically used for piston cooling which may be critical at high power Conditions but may not be required at lower power conditions The relationship between air/fuel ratio and spark gap ion current can be used to control small engines at leaner than typical air/fuel ratios (which results in lower hydrocarbon emissions) while at the same time running rich enough to provide proper piston cooiing. When rich mixtures are not required for cooling or full power, the ignition system with ionization sensing can control an engine to run near the lean burn limit to minimize hydrocarbon emissions and maximize fuel economy. At full power conditions or when additional cooling is required (based on a lookup map or other sensor input), the engine can be controlled to run at richer mixtures again using ionization sensing for feedback control.

In the experiments presented here using a production utility engine, peak ion current appeared to exhibit a relative maxima between approximately A = 0.90 and 2 = 0.95. This allows for an easy control strategy to operate within this air/fuel ratio range by maximizing the ion current peak. When richer mixtures are desired, the fuel system would control to the rich side of this region (perhaps using a predetermined offset). Alternatively, the engine could be controlled to the lean side of the peak ion current using ionization sensing in a partial fire / misfire detection mode as the key feedback parameter to operate near the lean limit.

STATIONARY ENGINES - Large stationary spark ignited engines (frequently natural gas fueled) can also use ionization sensing for air/fuel ratio control. These engines, which already operate lean of stoichiometric for good efficiency and overall low emissions, are being pushed even leaner to meet stringent NOx emission requirements. Open loop air/fuel ratio control is typically used with pre-programmed safety margins to keep the engines out of lean misfire and detonation conditions.

Ionization sensing, on the other hand, can be used for feedback control of air/fuel ratio, on an individual cylinder basis, with correspondingly smaller safety margins to operate closer to the lean limit resulting in lower NOx emission levels.

AUTOMOTIVE ENGINES - Ionization sensing could also be applied to automotive engines for air/fuel ratio control - specifically cylinder-tocylinder mixture balancing.

Today's passenger cars require precise air/fuei ratio control for proper catalytic converter function. Exhaust gas oxygen sensors (one or more depending on engine application) are used as part of the engine management system that keeps the overall airlfuel ratio at stoichiometric conditions While this control strategy works well for catalyst operation, it does not ensure that each cylinder is operating at the proper air/fuel ratio Ionization sensing and its relationship to in-cylinder air/fuel ratio could be incorporated in a feedback loop on a cylinder-to-cylinder basis to maintain proper mixture balance across the engine. A global oxygen sensor will likely be required to maintain the air/fuel ratio precision required for the catalyst.

The use of ionization sensing for cylinder-to-cylinder mixture balancing supplements other potential uses of the ion current signal. The ignition and ionization feedback system used in this small engine experiment continues to be developed for OEM passenger car applications. Detection of specific cylinder events such as misfire and knock has been demonstrated using the ion current signal. The ability to detect these abnormal conditions using engine-wide measures (e.g., block vibration, crankshaft speed fluctuation, etc.) is often complicated by external factors whereas ionization sensing provides a more robust detection technique due to its fundamental tie to the combustion event. In addition, MBT spark timing may be controlled using the ion current data because of its relationship to peak cylinder pressure, location of peak cylinder pressure, and indicated mean effective pressure. Historical and/or statistical ionization data can be used to slowly adjust for changes during the life of the engine or to diagnose problems. For example, an injector problem that would cause one cylinder to either run extremely rich or extremely lean could be detected with ionization sensing with the proper historical knowledge assimilated by an advanced engine management system. These are only a few examples of how the information rich ion current data can be implemented in automotive engine control strategies.

SUMMARY AND CONCLUSIONS A Dual Energy Ignition system which includes circuitry for ionization sensing, developed by Adrenaline Research, Inc., was successfully used in an experimental program at Wayne State University to study the relationship between engine operating conditions and the in-cylinder ion current flowing across the spark plug electrodes.

Empirical data show that there is a strong relationship between features of the cycle averaged ion current waveform and the overall (time averaged) air/fuel ratio.

Specifically, peak ion current, crank angle location of peak ion current, and the ion current integral are example characteristics that correlate with air/fuel ratio.

Ionization sensing is already used for engine control in a few limited applications. Techniques to measure the ion current with greater sensitivity, higher signal to noise ratio, and over the entire combustion event continue to be developed. The authors continue to study and refine the correlation of ion current with key engine parameters.

It is expected that ionization sensing will be applied more broadly and in a wider range of engine control strategies in the near future.

ACKNOWLEDGMENTS We wish to acknowledge the support of Wayne State's financial sponsors (Robert Bosch Corporation funded Kenny Ratton's thesis work). We also acknowledge Adrenaline's licensee, Standard Motor Products, Inc., who continues to fund the development of Dual Energy Ignition with Ionization Feedback for automotive applications. The support of MTS-PowerTek in donating a 15 hp DC motor is also appreciated.

CONTACT Dr. Eric N. Balles Adrenaline Research, Inc.

Three Brent Drive Hudson, MA 01749 978/568-8770 phone 978/568-8786 fax eric balles@alum.mit.edu or Prof. Ming-Chia Lai Mechanical Engineering Department College of Engineering Wayne State University Detroit, MI 48202 313/577-3893 phone 313/577-8789 fax lai@eng.wayne.edu REFERENCES 1. Saitzkoff, A., R. Reinmann, F. Mauss, M. Glavmo, "ln- Cylinder Pressure Measurements Using the Spark Plug as an Ionization Sensor'. SAE 970857. Society of Automotive Engineers, Inc., Warrendale, Pennsylvania, 1997.

2. Reinmann, R., et al., zLocal Air-Fuel Ratio Measurement Using the Spark Plug as an Ionization Sensor". SAE 970856, Society of Automotive Engineers, Inc..

Warrendale, Pennsyivania, 1997.

3. Saitzkoff, A., R. Reinmann, T. Berglind, M. Glavmo, An Ionization Equilibrium Analysis of the Spark Plug as an Ionization Sensor", SAE 960337, Society of Automotive Engineers, Inc., Warrendale, Pennsylvania, 1996.

4. Eriksson, L., L. Nielsen, "Closed Loop Ignition Control by Ionization Current Interpretation", SAE 970854. Society of Automotive Engineers, Inc., Warrendale, Pennsylvania.

1997.

5. Porreca, P., E. VanDyne, "Duel Energy Ignition System~.

U.S. Patent No. 5,197,448, March 1993.

6. VanDyne, E.. P. Porreca, SIgnition System with Ionization Detection", U.S. Patent Pending. 1997.

7. VanDyne, E., A. Wahl. D. Gardiner, B. Warshow, "The Science Behind Today's High Energy Capacitive Discharge Ignition Systems", presented at the SuperFlow Advanced Engine Technology Conference, Jan 1997.

8. VanDyne E., Porreca P., ZPerformance Improvement from Dual Energy Ignition on a Methanol injected Cosworth Engine", SAE Paper 940150, Society of Automotive Engineers. Inc., Warrendale, Pennsylvania, 1994.

9. Rattan, K. L., "In-Cylinder Method for Measuring Instantaneous Local Air-Fuel Ratio Using the Spark Plug as a Sensor', M.S. Thesis, Mechanical Engineering Department, Wayne State University, 1997.