TRAP

BACKGROUND

[0001] The present invention relates to internal combustion engine design and methods of operation. More particularly, the present invention relates to systems and methods for operating an engine in a variable displacement mode of operation while treating the exhaust using a Lean NOx trap (also referred to as an 'Absorptive NOx Trap', or simply 'NOx Trap'). The present system and methods apply equally well to Otto cycle gasoline engines, as well as, all other known engine types.

[0002] Various types of variable displacement engine operation are known. These range from shutting down a bank of the engine's cylinders (sometimes referred to as "variable displacement engine operation" or "variable cylinder management"), to control over cylinder firing on a working cycle by working cycle basis (sometimes referred to as "skip fire" operation). In typical variable displacement engine operation, cylinders that are deactivated do not receive fuel and the intake and/or exhaust valves are deactivated, thereby maintaining the composition of the exhaust emissions largely unaltered.

[0003] In typical skip fire engine operation, the rate of firing (or skipping) a given cylinder may render shutting the exhaust and intake valves impractical. This is due to the mechanical nature of valve operation which, typically, has some latency between active and deactivated modes.

[0004] Regardless of method of variable displacement operation, these engine operational modes can enable engines to operate in a more efficient manner (i.e. with better fuel economy). This is due to the fact that, in Otto cycle engines, the active cylinders must 'work harder' to achieve the same power output as an engine where none of the cylinders are skip fired. In order to produce more power from the active cylinders the throttle is more widely opened (less restrictive). This unthrottling of the air intake thereby reduces pumping loss, which is a major contributor of spark ignition engine inefficiency. [0005] All known types of engine operation gives rise to problems with exhaust emission pollutants. Particularly, the presence of non-combusted hydrocarbon (HC) compounds, carbon monoxide (CO), nitric oxide (NO), and nitrogen dioxide (N0 ₂ ) are of particular concern to the environment and human health. As a result, the presence of these emission byproducts are highly regulated by governments. Concern about emissions developed in the 1960s, particularly in Los Angeles and large urban locations where atmospheric conditions led to formation of a photochemical smog. Strictest regulations for emissions exist in the United States, Japan and Europe.

[0006] The generation of particular pollutants typically is highly impacted by fuel to air ratios within the cylinder. Richer fuel mixtures (mixtures with greater than stoichiometric levels of fuel), such as at the quench layer and cylinder crevices, typically result in the generation of greater hydrocarbon (HC) emissions. Likewise, carbon monoxide (CO) is typically generated at the flame front outer layer of the quench layer. These areas are locally fuel rich and thereby have incomplete combustion. Lean burn engines, such as those under development by Mercedes ^® , reduce the presence of both CO and HC in the exhaust by introducing larger amounts of oxygen (lean burn) into the combustion chamber. The excess oxygen reacts with the CO and HC in the presence of a catalyst to generate water (H ₂ 0) and carbon dioxide gas (C0 ₂ ). In general, most engines which operate near or leaner than stoichiometric level are able to adequately eliminate excess HC and CO molecules with a standard catalytic converter.

[0007] The formation of the nitrogen compounds such as nitric oxide (NO) and nitrogen dioxide (N0 ₂ ) is more complex, as is the elimination of these hazardous compounds. These nitrogen compounds, cumulatively, are known in the industry as NO _x . NO _x formation is dependent upon a series of reactions, such as the Zeldovich mechanism. In these reactions oxygen gas is split into free radical oxygen atoms. These oxygen atoms react with nitrogen gas forming NO and a free nitrogen atom. The free nitrogen atom may likewise react with oxygen gas to produce NO and a free oxygen atom. The rate constants for the thermal mechanisms for NO formation are very slow compared to those for combustion, and the formation of NO only occurs when the temperature is significantly high (roughly 1800 K). Thus, NO formation is typically limited to regions of hot combustion gases in the cylinder. Additional NO generation may occur when there is fuel-bound nitrogen. [0008] After combustion NO may cool and may be oxidized to form N0 ₂ . As noted, NO _x refers to the combination of NO and N0 ₂ resulting in the engine output. In most cases roughly 90% of NO _x engine emissions are composed of NO. In the environment, however, this NO eventually oxidizes to N0 ₂ . The nitrogen dioxide may then react with unburnt hydrocarbons in the presence of ultraviolet light to result in photochemical smog.

[0009] Given the desire for cleaner emissions, a number of after-treatments have been developed for use on engine emissions. The most widely used mechanism is the 3 -way catalytic converter. Typical catalytic converters include a porous ceramic matrix (often in a honeycomb pattern). The ceramic matrix provides a structural support and increased surface area for the catalysts. Aluminum oxide (Alumina) is the most common matrix material. Of course alternate matrixes, including metal and even paper, are known. Precious metal catalysts are then deposited to the surface of the matrix. Catalysts reduce the activation energy required for a chemical reaction. Further, the catalysts remain intact throughout the reaction. In the catalytic converter, the catalysts reduce the heat needed to undergo the chemical reactions for treating the emissions. Typical catalysts include platinum, palladium, rhodium and iridium.

[0010] Palladium and platinum promote oxidation of the carbon monoxide and hydrocarbons. Rhodium promotes the reaction of NO _x in one or more of the following equations:

[0011] NO + CO→½ N ₂ + C0 ₂

equation 1.1

[0012] 2NO + 5CO + 3H ₂ 0→ 2NH ₃ + 5C0 ₂ equation 1.2

[0013] 2NO + CO→ N ₂ 0 + C0 ₂

equation 1.3

[0014] NO + H ₂ → ½ N2 + H ₂ 0

equation 1.4

[0015] 2NO + 5H ₂ → 3NH ₃ + 2H ₂ 0

equation 1.5 [0016] 2NO + H ₂ → N ₂ 0 + H ₂ 0

equation 1.6

[0017] Note that many of the above reactions rely upon the existence of carbon monoxide and hydrocarbon products in order to reduce the NO _x gas.

[0018] Use of a standard 3-way catalyst also limits the ability to operate the engine lean, because excess oxygen gas in the combustion chamber oxidizes the hydrocarbons and carbon monoxide, and thereby limits the efficiency of NO _x reduction in the catalytic converter. Thus, a typical engine's operation is very strictly controlled.

[0019] For a traditional engine operating with a three way catalytic converter, the engine control unit (ECU) fluctuates between running the engine in a rich fuel to air ratio and a lean fuel to air ratio (rich-to-lean cycle). These fluctuations are very rapid in order to permit proper catalytic converter functioning; often there are multiple rich- to-lean cycles per second of engine operation. The control over rich-to-lean cycling relies upon feedback from one or more oxygen sensor.

[0020] Due to the rapid rich-to-lean cycling required for catalytic converter function, there is little leniency in engine control in these traditional engines. Thus, even small deviations in engine control leads to maintenance issues, failure of emissions testing, and user hassle. Also, the great reliance these traditional engines place upon oxygen sensors may result in the frequent display of a 'check engine' light when small deviations in oxygen levels are experienced. The 'check engine' light leads to driver confusion, stress and annoyance. Further, drivers may ignore the check engine light given its propensity to turn on over small oxygen fluctuations. This is also problematic, in that more serious mechanical problems may remain undiagnosed since the check engine light does not differentiate between these more serious mechanical problems and oxygen sensor alerts.

[0021] Given the fickle nature of catalytic converter function for lean engines, alternative NO _x after-treatments have been developed in order to address this problem. For example, Exhaust Gas Recirculation (EGR) is one such method for NO _x reduction. In EGR the exhaust gas may be reintroduced into the combustion chambers, thereby limiting the new oxygen introduced into the system. This basically operates the engine at stoichiometric oxygen to fuel ratios. [0022] Alternatively, Specific Catalytic Reduction (SCR) has been utilized in lean burn diesel engines to reduce NO _x emissions. In SCR, a reducing agent is injected into the exhaust flow within the exhaust system. Common reducing agents include raw fuel, hydrogen gas and urea solution. In urea-based approaches, the urea decomposes into ammonia which reacts with the NO _x across a catalyst located downstream from the injection point. The problem with SCR is that there is a separate reagent tank required for reducing the NO _x emissions. SCR leads to an associated increase in cost, system complexity and user hassle.

[0023] Lastly, lean NO _x traps have been developed for the treatment of NO _x pollutants. Lean NO _x traps typically include a ceramic matrix substrate, such as aluminum oxide. A wash coat may be applied to the ceramic substrate to further increase surface area. Lastly an absorber and catalyst may be applied to the wash coat. As with a catalytic converter, the catalysts typically employed by a lean NO _x trap are platinum and rhodium based. Unlike the catalytic converter, however, the lean NO _x trap includes an absorption material which chemically reacts with the NO _x . This absorption material enables the lean NO _x trap to "soak up" large quantities of NO _x . Catalytic converters lack the ability to absorb NO _x gas through chemical reactions. Once saturated, a regeneration of the lean NO _x trap is performed by running the engine in a fuel rich series of combustion cycles.

[0024] The absorption materials typically utilized in a lean NO _x trap includes zeolite, alkali or alkaline earth material. Of course additional known, and to be discovered, material may be utilized to absorb the NO _x gases. Zeolite materials are microporous aluminosilicate minerals which have structures which accommodate a wide variety of cations. These positive ions are loosely held by the zeolite and may be readily exchanged in a contact solution. The zeolite may then bind with the NO _x gases until regeneration.

[0025] Alkali and alkaline earth materials (also known as 'soberer' catalyst components) likewise react with the NO _x gases until regeneration occurs. One commonly used alkaline earth material includes barium salts. For example BaC0 ₃ has been utilized as a soberer. When NO and 0 ₂ are present, the barium compound may be oxidized (referred to as sorption) resulting in bound BaN0 ₃ . Carbon dioxide is released during sorption. Example sorption equations for a barium based lean NO _x trap are provided below: [0026] 2NO + 0 ₂ → 2N0 ₂

equation 2.1

[0027] 4N0 ₂ + 2BaC0 ₃ + 0 ₂ → 2Ba(N0 ₃ ) ₂ + 2C0 ₂ equation

2.2

[0028] Regeneration occurs when sufficient reducers are present (rich engine operation), thereby returning the barium salt to its original chemical state. Below are example equations for the regeneration of the Barium based sorbent:

[0029] Ba(N0 ₃ ) ₂ + CO→ 2N0 ₂ + BaC0 ₃ equation

3.1

[0030] N0 ₂ → NO + 0 ₂

equation 3.2

[0031] 2CO + 2NO→2C0 ₂ + N ₂

equation 3.3

[0032] 4CO + 2N0 ₂ → 4C0 ₂ + N ₂

equation 3.4

[0033] Lean NO _x traps become saturated after the engine has been running lean for a prolonged period of time (often 30 seconds to 2 minutes). Regeneration must then be performed to prevent NO _x emissions. Regeneration typically takes less time (roughly 1-10 seconds). The lean NO _x trap enables an engine to run lean for a much longer period of time, thereby providing flexibility to engine operation.

[0034] One issue of lean NO _x trap design and implementation is the operational temperature of current lean NO _x traps is significantly cooler than other engine systems. Particularly, the lean NO _x trap usually operates at 250-450°C, whereas catalytic converters typically operate at 600-800°C. Further, given the high temperature of typical exhaust gas (often 900°C or more) there is a need for artificial cooling of the exhaust before the lean NO _x trap, in typical engines. This cooling has been accomplished using passive cooling loops between the catalytic converter and the lean NO _x trap, in some engines. Alternate systems include the use of refrigeration and radiator fluids to cool exhaust prior to the lean NO _x trap. Cooling requirements make the addition of lean NO _x traps to existing engines costly and problematic. [0035] Currently, skip fire engines are operated at or near stoichiometric fuel to air ratios. Sophisticated and advantageous skip fire operational modes may be impossible to implement in vehicles due to emission regulations and the limitations of standard 3 -way catalytic converter technology. These sophisticated skip fire engines may benefit from NO _x after-treatment to both comply with state and federal regulations for engine pollutants and yield simpler and more fuel efficient operation.

[0036] In view of the foregoing, systems and methods for skip fire operation of an engine with a lean NOx trap after-treatment are disclosed. The present invention provides a novel system for enabling enhanced control of cylinder fueling and combustion events to satisfy the need for greater fuel efficiency yet comply with engine emission regulations.

SUMMARY

[0037] The present invention discloses an engine operational system with exhaust after-treatment. More particularly, the present invention teaches systems and methods for skip fire engine operation with a lean NOx trap after-treatment. The skip fire and lean NOx trap after-treatment system may be utilized to modify current engines to operate at higher engine efficiency levels and yet produce clean or ultra-clean tailpipe emissions.

[0038] In some embodiments, the present system may include engine control circuitry capable of fueling and firing subsets of the engine's cylinders. Other cylinders are not provided fuel and are not fired (i.e. "skip fired"). This skip firing of selective cylinders increases the load on the working (fired) cylinders. In throttled engines, this causes the throttle of the working cylinders to be more open in order to produce the required power output. This unthrottling of the cylinders results in a reduction of 'pumping losses', thereby increasing engine efficiency.

[0039] In some embodiments, skip firing may include known variable

displacement systems where banks of cylinders are 'shut down'. Shutting down of cylinders includes deactivation of intake and exhaust valves for these modes of operation. This results in the deactivated cylinders behaving as air springs.

[0040] In other embodiments, the skip fire mode of operation may rely upon adaptive predictive control circuitry, such as a sigma delta engine control. This control circuitry may determine which cylinders are to be fueled and fired on a working cycle by working cycle basis. In some of these modes of engine operation, the intake and exhaust valves may remain active, thereby passing uncombusted air through the skip fired cylinders.

[0041] Some embodiments of the system include an exhaust manifold for channeling the exhaust from the engine. The exhaust may be channeled through a multistage catalytic converter. This may include a standard two-way or three-way catalytic converter. From the catalytic converter the exhaust may be channeled through a lean NOx trap. The lean NOx trap is able to chemically absorb the NOx emissions for regeneration according to a regeneration protocol. Lean NOx traps may include a substrate, an absorption material (also referred to as a 'sorbent') and a catalyst. The sorbent may include a zeolite material, an alkali and/or an alkaline earth material. For example, in some embodiments, a Barium (Ba) salt may be utilized as the absorber. The catalysts utilized in the lean NOx trap typically include precious metals, such as platinum and rhodium.

[0042] The regeneration protocol may, in some embodiments, include monitoring NOx emissions downstream from the lean NOx trap. These NOx emission values are then compared against either a threshold or a lean NOx trap saturation model. Once the NOx emissions reach the threshold, or deviate from the model, a regeneration cycle may be performed. In some alternate embodiments, the regeneration cycle may be performed after a set period of time, or number of engine rotations.

[0043] The regeneration cycle may include addition of a reducing agent to the exhaust system. Reducing agents could include hydrogen gas, urea or fuel, to name a few. Alternately, regeneration of the lean NOx trap may be performed by running the engine in a rich fuel to air ratio.

[0044] Note that the various features of the present invention described above may be practiced alone or in combination. These and other features of the present invention will be described in more detail below in the detailed description of the invention and in conjunction with the following figures. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The present invention is illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar elements and in which:

[0046] Figure 1 is an example structural block diagram of a system for skip fire engine operation with lean NOx trap after-treatment in accordance with an embodiment of the present invention;

[0047] Figure 2 is an example structural block diagram of the engine utilized in the system for skip fire engine operation with lean NOx trap after-treatment in accordance with an embodiment of the present invention;

[0048] Figure 3 is an example graph of common pollutants generated during engine operation dependent upon air to fuel ratios in accordance with an embodiment of the present invention;

[0049] Figure 4 is an example graph of lean NOx trap efficiency dependent upon operational temperature in accordance with an embodiment of the present invention;

[0050] Figure 5 A is an example block diagram a lean NOx trap undergoing sorption in accordance with an embodiment of the present invention;

[0051] Figure 5B is an example block diagram a lean NOx trap undergoing regeneration in accordance with an embodiment of the present invention;

[0052] Figure 6 is an example graph of operational engine fuel to air ratios over time for the regeneration of the lean NOx trap in accordance with an embodiment of the present invention;

[0053] Figures 7 A and 7B are example structural block diagrams of engine control systems in accordance with an embodiment of the present invention;

[0054] Figure 8 is an example structural block diagram an engine control coprocessor in accordance with an embodiment of the present invention;

[0055] Figures 9 A to 9C are alternate example circuit diagrams of drive pulse generators in accordance with an embodiment of the present invention; [0056] Figures 10A to IOC are alternate example structural diagrams for connectivity of the engine control co-processor in accordance with an embodiment of the present invention;

[0057] Figures 11 A and 1 IB are alternate example wiring diagrams for the engine control system in accordance with an embodiment of the present invention;

[0058] Figure 12 is an example graph illustrating fuel input and power output over time for an engine operating under traditional throttled mode and in skip fire mode in accordance with an embodiment of the present invention; and

[0059] Figures 13A to 13E are example flow chart diagrams for the process of skip fire engine operation with lean NOx trap after-treatment in accordance with an embodiment of the present invention.

[0060] In the drawings, like reference numerals are sometimes used to designate like structural elements. It should also be appreciated that the depictions in the figures are diagrammatic and not to scale.

DETAILED DESCRIPTION

[0061] The present invention will be described in detail with reference to selected preferred embodiments thereof as illustrated in the accompanying drawings. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. It will be apparent, however, to one skilled in the art, that the present invention may be practiced without some or all of these specific details. In other instances, well known process steps and/or structures have not been described in detail in order to not unnecessarily obscure the present invention. The features and advantages of the present invention may be better understood with reference to the drawings and discussions that follow.

[0062] The present invention relates generally to systems and methods for the operation of spark ignition internal combustion engines in a skip fire mode of operation with a lean NOx trap after-treatment of the engine exhaust. In particular, the lean NOx trap after-treatment includes channeling the engine exhaust through an absorptive lean NOx trap, and periodically regenerating the lean NOx trap when it reaches, or near, saturation. [0063] Also of note is that in the remainder of this application particular attention will be placed upon internal combustion engines for use in automobiles, trucks, locomotive, ship and other vehicular uses. It is important to realize that internal combustion engines are utilized in a wide range of other applications and that embodiments of the present invention are versatile enough to be utilized in a myriad of applications. This includes small appliance applications, such as portable generators, power washers, compressors and lawn mowers, for example. Additionally, the present invention may be useful in association with industrial applications, such as manufacturing equipment and construction machinery, for example. Likewise, while an Otto cycle style engine is described by way of explanation, the present invention is likewise usable in conjunction with other engine types including two stroke, and other spark ignition engine types. It is thus intended that the present invention is usable in conjunction with any viable engine regardless of eventual use or application.

[0064] The present application describes a number of engine designs and control arrangements for effectively controlling the operation of an engine in manners that permit some of the engine working chambers to operate in a skip fire mode. The various described embodiments include implementations that are well suited for use in: 1) retrofitting existing engines; 2) new engines based on current designs; and/or 3) new engine designs that incorporate other developments or are optimized to enhance the benefits of the described skip fire operational capacity.

[0065] In skip fire methods of operation one or more of the engine's cylinders operate at or near optimal efficiency. In Otto cycle engines, this is typically with maximum air intake (unthrottled) and at, or near, stoichiometric fuel to air ratios.

Operating in this manner reduces inefficiencies of engine operation resulting from pulling air through the engine (i.e. pumping loss). However, cylinders operating under these conditions tend to produce much more power than is required. By selectively not fueling or firing particular cylinders and operating others at optimal efficiency, overall engine efficiency may be improved while maintaining the desired power output. This method of engine operation is typically referred to as "skip fire".

[0066] Referring briefly to Figure 12, a graph illustrating an example of analog (1200A) versus digital (1200B) fuel injection amounts in reference to the

instantaneous torque output is presented. Axis 1210 is cylinder firings over a time period, and axis 1208 indicates the fuel injection amount. The dashed lines 1202 indicate the optimal fuel injection amount. Line 1206 indicates the average miles per gallon (MPG) and line 1204 indicates instantaneous torque output.

[0067] As may be seen, the injection patterns in the digital fuel injection plot 1200B all occur at substantially optimal injection levels. This results in a

substantially identical torque output as compared to the analog injection plot 1200A. However, the efficiency of the individual chambers firing at optimal injection amounts results in a higher overall efficiency, which is reflected in the higher MPG of the digital injection plot 1200B.

[0068] In some embodiments, the skip fire engine is operating near stoichiometric fuel to air ratios. However, in some skip fire modes of operation, skipped cylinders pass uncombusted air through to the exhaust system. Thus, while a standard three way catalytic converter may be capable of dealing with the NOx emissions from the combusted cylinders alone, when combined with the uncombusted air, the emissions include too much oxygen. The oxygen reacts with some of the carbon monoxide and hydrocarbons to generate water and carbon dioxide prior to reaching the three way catalytic converter. This leads to too few reducing chemicals (CO and HC) to properly react with the excess NOx. Thus, excess NOx may pass through the catalytic converter and then through a lean NOx trap.

[0069] Below is provided a number of example systems and methods of operation for a skip fire engine with lean NOx trap after-treatment. For the sake of clarity multiple subsections with corresponding headings are provided. These subsections are provided solely in the interest of clarity and are not intended to limit the present invention in any manner.

I. SKIP FIRE ENGINE WITH EXHAUST AFTER-TREATMENT

[0070] Figure 1 is an example structural block diagram of a system for skip fire engine operation with lean NOx trap after-treatment, shown generally at 100. Here a user is seen providing a Driver Input 110 to the Engine Control Unit 108. The user will also be referred to as 'driver' throughout this application given the present system's particular applicability to vehicular engines. Note, however, that the user may likewise include a machinery operator, remote pilot or even a computer system, such as a cruise control system. [0071] The Engine Control Unit 108 may receive Engine Sensory Data 112 along with the Driver Input 110. The Engine Control Unit 108 is then able to generate fueling, and in the case of spark ignition engines, firing instructions for the Engine 104. The process for generation of the fueling instructions will be described in greater detail below.

[0072] The Engine 104 may receive combustion air from the Air Intake Device 102. The Air Intake Device 102 may include an air intake manifold, turbo charger, or other air intake device. The incoming air may be throttled in the Engine 104.

Throttling instructions may likewise be generated by the Engine Control Unit 108, in some embodiments. The Engine 104 may likewise draw fuel from the Fuel Tank 106. Fuel, in most cases includes at least one of: gasoline, diesel fuel, hydrogen, natural gas, ethanol, biodiesel, and propane. Note however, that any fuel type which is able to power an engine in a skip fire mode of operation may be entirely suitable for use with some embodiments of the present invention.

[0073] The Engine 104 may combine the fuel and air within cylinders in known air to fuel ratios. The timing and respective amounts of air and fuel mixture may be controlled by the Engine Control Unit 108. The Engine Control Unit 108 may likewise control the ignition of each cylinder through spark plug firing. Combustion events and the subsequent expansion of the combustion gasses results in the power output of the engine.

[0074] In the present invention, the Engine 104 is configured to operate in a skip fire mode. This means that, at least during portions of the engine operation, one or more cylinders is not receiving fuel or undergoing a combustion event. Variable displacement is one known form of skip fire engine control; however, other more dynamic skip fire methods of operation are also considered by some embodiments of the present invention.

[0075] The exhaust gas from at least one of the cylinders is channeled through an exhaust system through a Standard Multistage Catalytic Converter 116. The Standard Multistage Catalytic Converter 116 may include any known catalytic converter type, including a two-way catalytic converter or a three-way catalytic converter.

[0076] After passage through the Standard Multistage Catalytic Converter 116, the exhaust passes through the lean NOx Trap 118. The lean NOx Trap 118 includes an absorption agent (sorbent) and a catalyst supported by a substrate. Often a washcoat may be applied on the substrate to increase surface area for the sorbent and catalyst. The substrate may, in some embodiments, include an Alumina silicate honeycomb ceramic structure. However, other suitable substrates may be utilized. The catalysts typically include precious metals such as platinum and rhodium. Other known catalysts may likewise be utilized.

[0077] The sorbent material in the lean NOx trap may include any number of appropriate materials. For example, in some embodiments, zeolite, alkali and alkaline earth materials may be utilized. In discussions related to this application examples of a barium salt sorbent will be used. Note that the use of this alkaline earth material is an example only and is not intended to limit the scope of the invention. It is considered that as technology of NOx absorption matures there will be an expansion of suitable sorbent materials. Some embodiments of the present invention are intended to work with known and future known sorbent materials.

[0078] After passing through the lean NOx trap the Exhaust Emissions 120 may be seen exiting the exhaust system into the environment. Due to the exhaust treatments the emissions will be substantially comprised of the following compounds: H ₂ 0, N ₂ , C0 ₂ and 0 ₂ . Levels of CO, HC and NOx in the Exhaust Emissions 120 will be reduced to below, or significantly below, federal and state emissions standards.

[0079] Although not illustrated in Figure 1 , a Diesel Particulate Filter (DPF) may be utilized in some embodiments of the invention. The DPF may exist prior to the Standard Multistage Catalytic Converter 116, after the lean NOx Trap 118 or between them. Further, it is considered, in some embodiments, that the Standard Multistage Catalytic Converter 116 exists after the lean NOx Trap 118 thereby switching positions of these parts in the exhaust stream.

[0080] Note that the lean NOx Trap 118 is proximally located next to the

Multistage Catalytic Converter 116, in some embodiments. In typical engines with a lean NOx Trap 118 there are cooling loops, or other external cooling systems, used to reduce the temperature of exhaust gas prior to lean NOx trap treatment. Conversely, in some embodiments of the present invention there is an absence of said cooling mechanisms. This is due to the lower exhaust temperature inherent to some skip fire engines. This will be discussed in more detail below in reference to Figure 4. [0081] Data may be compiled from the Engine 104, Standard Multistage Catalytic Converter 116 and lean NOx Trap 118 to generate the Engine Sensory Data 112. For example, the Engine Sensory Data 112 may include engine speed, engine temperature, air intake pressure, oxygen sensor data from the engine and the catalytic converter, and NOx levels downstream from the lean NOx Trap 118. This Engine Sensory Data 112 may be utilized to ensure proper fuel injection timing, smooth skip fire operation, and that the lean NOx Trap 118 is properly regenerated once it is sufficiently saturated with NOx.

[0082] Progressing now to Figure 2 where an example structural block diagram of the Engine 104 utilized in the system for skip fire engine operation with lean NOx trap after-treatment is provided. Here Engine Control Output 216 from the Engine Control Unit 108 is seen being provided to the Engine 104. The Engine Control Output 216 may be seen as being fed to a fuel line Pressure Valve 204 and each Injector Valve 212a-212d. Additionally, pressure sensors from the feed line and the rail may be provided back to the ECU. Likewise, fuel temperature may be provided to the ECU.

[0083] Fuel from the Fuel Tank 106 may be pressurized (typically about 2000 psi for rail injection) by one or more Pumps 206. The pressure may be regulated by the Pressure Valve 204. The pressurized fuel may be supplied to a common Rail 208. The individual injection Valves 212a-212d may open to allow fuel injection into each of the cylinders 214a-214d within the Engine Block 210 of the Engine 104.

[0084] As is well known in the art, the piston within the cylinders compresses the intake air, resulting in the air temperature rising. In a diesel engine, fuel may then be injected at or near top dead center (TDC) whereby the fuel ignites due to the elevated air temperature. In modern engines fuel dispersion may be enhanced through common rail systems, multiple port injectors, group hole nozzle injectors and pulsed injection. In pulse injection a series of fuel pulses may be injected into the chamber before (primer) and at top dead center. Multiple pulses of fuel may enable more homogenous burn patterns. This method of engine operation is also useful in

Homogenous Charge Combustion Ignition (HCCI) style engines. Some degree of recirculation of exhaust gas may also be desirable in some embodiments. [0085] In spark ignition engines the compression of the cylinders 214a-214d is not sufficient to cause the fuel and air mixture to spontaneously ignite. Thus, fuel may be injected into the cylinders 214a-214d prior to TDC and a spark is utilized to initiate the burn. Spark timing may likewise be controlled by the ECU. By injecting fuel into the cylinders 214a-214d earlier, in spark ignition engines, the fuel typically mixes more with the combustion air (as compared to non- spark engines). This greater mixing of fuel with combustion air causes the reduced particulate emissions from spark ignition engines.

[0086] Turning briefly to Figure 3, which illustrates selected emissions characteristics of a representative internal combustion engine at different air/fuel ratios. As can be seen therein, the amount of carbon monoxide (CO) in the emissions tends to increase as the mixture becomes richer and increases quite significantly in mixtures that are richer than stoichiometric. The amount of Nitric Oxide (NO) tends to be highest at a near stoichiometric mixture ratio and fall off relatively quickly as the air/fuel ratio becomes leaner or richer. The amount of hydrocarbons in the exhaust also generally tends to increase relatively quickly with increases in the mixture ratio beyond stoichiometric conditions. Many catalytic converters require the presence of a certain amount of carbon monoxide to run efficiently. If the engine runs lean, it is possible that the catalytic converter will become depleted and won't work efficiently due to the lack of carbon monoxide. The fuel to air ratio utilized in skip fire engine operation is often near stoichiometry. However, due to non-combusted air passing through skipped cylinders there may be insufficient CO and HC to reduce the excess NOx within the catalytic converter. This may lead to problems with state and federal emission control violations if the exhaust is not treated NOx reducing system.

[0087] As noted above, the most widely used mechanism to treat engine emissions is the 3 -way catalytic converter. Typical catalytic converters include a porous ceramic matrix (often in a honeycomb pattern). The ceramic matrix provides a structural support and increased surface area for the catalysts. Aluminum oxide (Alumina) is the most common matrix material. Of course alternate matrixes, including metal and even paper, are known. Precious metal catalysts are then deposited to the surface of the matrix to provide a reduction in reaction temperatures. Typical catalysts include platinum, palladium, rhodium and iridium. [0088] Palladium and platinum promote oxidation of the carbon monoxide and hydrocarbons. As previously noted, rhodium promotes the reaction of NO _x in one or more of the following equations:

[0089] NO + CO→ ½ N ₂ + C0 ₂

equation 1.1

[0090] 2NO + 5CO + 3H ₂ 0→ 2NH ₃ + 5C0 ₂ equation 1.2

[0091 ] 2NO + CO→ N ₂ 0 + C0 ₂

equation 1.3

[0092] NO + H ₂ → ½ N2 + H ₂ 0

equation 1.4

[0093] 2NO + 5H ₂ → 3NH ₃ + 2H ₂ 0

equation 1.5

[0094] 2NO + H ₂ → N ₂ 0 + H ₂ 0

equation 1.6

[0095] Note that many of the above reactions rely upon the existence of carbon monoxide. Without sufficient carbon monoxide, hydrocarbons, or other reducing agent there is insufficient reagents to eliminate the excess NOx.

[0096] Given the narrow range of chemical environments a standard catalytic converter is capable of operating in; typical engines must operate by cycling from rich to lean modes of operation very rapidly. Typically, an engine may cycle from rich to lean operational modes multiple times per second. Further, the cycling from rich to lean engine operation typically relies upon feedback from one or more oxygen sensors. The narrow range of engine operation, and the heavy reliance on these oxygen sensors, results in frequent engine maintenance alerts. These alerts include the 'check engine' light on the dashboard of he vehicle.

[0097] In lean burn engines various NOx after-treatment systems have been proposed, as previously discussed. These systems include Selective Catalytic

Reduction (SCR), exhaust recirculation and lean NOx traps. As skip fire engines typically operate near stoichiometric air to fuel ratios, NOx after-treatment has not utilized in skip fire engines previously. The present invention, however, combines lean NOx trap after-treatment with skip fire engines. Lean NOx traps will be discussed in more detail below.

II. LEAN NOX TRAP DESIGN AND OPERATION

[0098] Figure 4 is an example graph of lean NOx trap efficiency dependent upon operational temperature, shown generally at 400. Lean NOx trap efficiency is shown as very effective in a rather narrow band between 250° and 450°C. This temperature tends to be significantly below combustion gas temperatures which range as high as 800° C or higher. Thus, in existing lean NOx trap systems, the catalytic converter is typically before the lean NOx trap, as typical catalytic converters operate best at higher temperatures (functioning begins after 400° C or higher).

[0099] Additionally, in some known lean burn engines, the lean NOx trap systems requires cooling loops in order to further reduce exhaust temperature to the

operational range of 250° to 450°C. Other cooling systems may utilize refrigeration or fluid cooling (radiator cooling). This narrow, and relatively low, temperature range for lean NOx trap functionality leads to a number of potential issues related to engine design. Particular methods of skip fire, however, results in uncombusted air being passed through the skip fire cylinders. This uncombusted air is significantly cooler than combusted exhaust, thus, in these embodiments of skip fire systems, the exhaust temperature may be within the functional range of the lean NOx trap without any artificial cooling system.

[00100] Skip fire systems which pass air through the skipped cylinders are typically controlled via adaptive predictive logic circuitry, as is described in considerable detail below. This is because such control logic is able to determine whether a cylinder is to be fired or skipped on a working cycle by working cycle basis. Such rapid control over cylinder operation is easily implemented by the fuel injection systems of most vehicles, as the fuel injector systems are digitally controlled.

Likewise, control over the firing of the spark plug is rapidly controllable. However, deactivation of the intake and exhaust valves requires some level of latency.

[00101] Of note is that there are many known techniques for disabling a cylinder's intake and exhaust valves. For example, for pushrod designs, when cylinder deactivation is called for, hydraulic valve lifters are collapsed by using solenoids to alter the oil pressure delivered to the lifters. In their collapsed state, the lifters are unable to elevate their companion pushrods under the valve rocker arms, resulting in valves that cannot be actuated and remain closed. Likewise, for overhead cam designs, generally a pair of locked-together rocker arms is employed for each valve. One rocker follows the cam profile while the other actuates the valve. When a cylinder is deactivated, solenoid controlled oil pressure releases a locking pin between the two rocker arms. While one arm still follows the camshaft, the unlocked arm remains motionless and unable to activate the valve.

[00102] Since both of these systems rely upon hydraulic deactivation of the valves, the latency involved with this process may make cylinder valve deactivation impractical for rapid continuous variable displacement. Thus, uncombusted air passes through the skipped cylinders in these engine embodiments.

[00103] The narrow functioning temperature range of a lean NOx trap is related to the chemical reactions utilized to absorb the excess NOx gas. Figures 5A and 5B further illustrate the chemical equations.

[00104] Figure 5 A is an example block diagram a lean NOx trap undergoing sorption, shown generally at box 502. Here a segment of an example lean NOx trap is shown. The Substrate 510 may consist of alumina, silicates, and other suitable ceramic material. Of course, other suitable materials, such as metals and composites may be utilized for the Substrate 510. A washcoat may be applied over the substrate (not illustrated) in order to increase surface area and modulate surface properties, if desired.

[00105] Coupled to the substrate are an Absorber Material (sorbent) 530 and Catalyst(s) 520. Typical Catalysts 520 include platinum, rhodium, palladium, iridium, and other precious metal compounds. The Absorber Material 530 may include a zeolite, alkali, or alkaline earth material. As noted earlier, barium salts are known absorbers, and will be utilized throughout this example and associated chemical equations.

[00106] During sorption nitric oxide (NO) may combine with oxygen gas (0 ₂ ) in the presence of the catalyst to produce nitric dioxide (N0 ₂ ). The equation for this reaction is given by:

[00107] 2NO + 0 ₂ →2N0 ₂

equation 2.1 [00108] The generated nitric dioxide (N0 ₂ ), along with the already present N0 ₂ may subsequently be absorbed by the absorber (Barium salt in the present example). The equation for this reaction is given by:

[00109] 4N0 ₂ + 2BaC0 ₃ + 0 ₂ → 2Ba(N0 ₃ ) ₂ + 2C0 ₂ equation 2.2

[00110] As can be seen, carbon dioxide is released during this sorption process. The carbon dioxide is released to the environment as part of the emissions.

[00111] Figure 5B is an example block diagram a lean NOx trap undergoing regeneration, shown generally in box 504. Again, here a segment of an example lean NOx trap is shown. The Substrate 510, Catalysts 520, and Absorber Material 530 are may still be seen interacting with exhaust chemicals.

[00112] Regeneration occurs when sufficient reducers are present (i.e. rich engine operation), thereby returning the barium salt to its original chemical state. Carbon monoxide may react with the barium dinitrate to release the nitric dioxide and leaving barium carbonate given the following example equation:

[00113] Ba(N0 ₃ ) ₂ + CO→ 2N0 ₂ + BaC0 ₃ equation 3.1

[00114] Some portion of the nitric dioxide may form oxygen gas and nitrogen oxide given the following equation:

[00115] N0 ₂ →NO + 0 ₂

equation 3.2

[00116] Lastly, the nitric dioxide and nitrogen oxide may be reduced by carbon monoxide or hydrocarbons (or other suitable reducer) given the following equations:

[00117] 2CO + 2NO→ 2C0 ₂ + N ₂

equation 3.3

[00118] 4CO + 2N0 ₂ →4C0 ₂ + N ₂

equation 3.4

[00119] 8HC + 10NO ₂ → 8C0 ₂ + 5N ₂ + 4H ₂ 0 equation 3.5 [00120] Thus, after regeneration the resulting emissions include water, nitrogen gas and carbon dioxide.

[00121] Figure 6 is an example graph of operational engine fuel to air ratios over time for the regeneration of the lean NOx trap, shown generally at 600. Note that the graph of Figure 6 is an example graph only, and is not necessarily to scale. Lean NO _x traps become saturated after the engine has been running lean for a prolonged period of time (often 30 seconds to 2 minutes). Regeneration must then be performed to prevent NO _x emissions. Regeneration typically takes less time (often 1-10 seconds).

[00122] In the exemplary graph, two lines are provided. Line 610 may indicate a skip fire engine operating at near stoichiometric levels during the bulk of the working cycles. Regeneration peaks may be seen occurring every hundred seconds. During regeneration of skip fire engines the fuel to air ratio becomes dramatically richer (more fuel than stoichiometry).

[00123] In contrast, line 620 may indicate fuel to air ratios typically found in a lean burn engine, such as the Mercedes ^® lean burn engine. In these engines the cylinders typically operate at very lean air to fuel ratios. During regeneration, the fuel to air ratio shifts to a rich fuel to air ratio.

[00124] What is important to note is that the use of a lean NOx trap in a lean burn engine is known, however, the use of a lean NOx trap in a stoichiometric burning engine is generally thought of as unnecessary, costly, and requiring significant redesign due to temperature requirements for lean NOx trap functioning. Thus, the addition of a lean NOx trap into a stoichiometric burning engine has not been previously attempted. The present invention is unique in this capacity.

III. ENGINE CONTROL SYSTEMS

[00125] In traditional engines, the engine control unit (ECU) provides strict fueling and firing instructions to the engines to ensure proper catalytic converter functioning, and to provide the necessary engine output. The fueling instructions include rapidly cycling the engine between rich and lean burn cycles. This rapid cycling occurs multiple times every second in order for the catalytic converter to properly function. Any prolonged periods (more than a second or so) of rich or lean burns effectively overwhelm the catalytic converter. The resulting emissions will include greater NOx (for lean burns) or greater HC and CO (for rich burns) than is acceptable by regulatory standards. In order to control the cycling between rich and lean burns, traditional ECU's rely heavily on oxygen sensors located throughout the engine and exhaust systems. In particular, oxygen sensors located downstream from the catalytic converter may indicate the efficiency of catalytic converter function.

[00126] One drawback of such a tightly controlled system is there is very little room for error. Thus, even slight deviations from the necessary oxygen levels results in an engine malfunction. To the driver, these malfunctions are basically

imperceptible; however, a malfunctioning engine may fail during smog testing.

Moreover, the engine malfunctions may trigger the 'check engine' light to turn on. This can be very frustrating for the driver. Moreover, if the driver assumes the check engine light is on for an emissions problem, more serious mechanical problems may be ignored. This could be a costly error, and may endanger the driver as well.

[00127] In variable displacement style engines (i.e. skip fire), the ECU may likewise provide fueling and firing instructions to the engine. However, the skip fire ECU may indicate that only certain cylinders in a given working cycle are to be fueled and fired. These may be referred to as 'activated cylinders'. Other cylinders may not receive fuel or a spark. These cylinders are termed 'deactivated cylinders'. In some cases the intake and exhaust valves of the deactivated cylinders remain functional. Thus, uncombusted air is supplied to the cylinder and vents through the exhaust system. This may substantially reduce exhaust temperatures, as well as alter total exhaust chemistry. In particular, the uncombusted air includes an abundance of oxygen gas.

[00128] Alternatively, it may be possible that the deactivated cylinders have the intake and exhaust valves decoupled from the cam via hydraulic mechanisms. These cylinders do not pass air through to the exhaust system.

[00129] The ECU's (and engine control co-processors), in some embodiments of the present invention, may include more leniency on degree and duration of rich and lean burn cycles. This is because, unlike the traditional engines, the additional NOx trap enables greater latitude of operational conditions. Likewise, the inputs from one or more of the oxygen sensors may be ignored, or overridden, as the data supplied by these sensors is less relevant. A benefit of this is that the instances of the 'check engine' light being activated are substantially reduced. [00130] Figure 7 A provides a structural block diagram for an example of an Improved Engine Control System 700A. Portions of the Improved Engine Control System 700A may be preexisting components found within current vehicle engines. For example, most vehicles include an Engine Control Unit 108 and a Fuel Injector Driver 730. Additionally, most engines include means for generating Engine Sensory Data 112 and Driver Input 110. Thus, for many current engines, an aftermarket system including the Engine Control Co-Processor 710 and the Engine Control Multiplexer 740 may be installed which complements the existing Engine Control Unit 108 of the vehicle. This design is particularly well adapted for retrofitting existing engines to incorporate the described high efficiency operating modes.

[00131] Note, that the Engine Control Multiplexer 740 may also interface with other optional systems, including Diagnostic Tools 780 and turbocharger systems (not illustrated), to name a few.

[00132] The described control system may be implemented in a wide variety of different manners. It may be accomplished using digital logic, analog logic, algorithmically or in any other appropriate manner. In some embodiments the high efficiency control logic will be built into the Engine Control Unit 108 (ECU, sometimes also referred to as an ECM, engine control module). In other

embodiments, high efficiency mode control logic may be built into the Engine Control Co-Processor 710 that is arranged to work in conjunction with an existing Engine Control Unit 108.

[00133] It is anticipated that as the technology develops, the high efficiency mode control logic will be integrated into the engine control units that are provided with new vehicles or engines, as illustrated at Figure 7B. This is particularly beneficial because it allows the ECU to readily take advantage of all of the features of the engine that are available to improve engine control using the high efficiency mode.

[00134] New ECUs that incorporate the high efficiency mode and other engine control modes may also be developed for vehicles that are on the road today (and for other existing engines and/or engine designs). When such ECUs are developed the existing engines may readily be retrofitted by simply replacing the existing ECU with an improved ECU that incorporates the enhanced engine control. [00135] Alternatively, as will be appreciated by those familiar with current automotive engine control design, the engine control units in most late model automobiles are arranged such that third party devices may interface with the engine control unit. These interfaces are often provided, at least in part, to facilitate engine diagnostics; however, a variety of third parties products such as turbochargers, superchargers, etc. include control co-processors that have been designed to utilize such interfaces to work with the engines without voiding the manufacturer's warranty. These interfaces may be used advantageously to allow a low cost fuel co-processor that incorporates the enhanced control logic to be installed as a retrofit to enable higher fuel efficiency operation for diesel vehicles on the road today.

[00136] When a new vehicle is designed, the entire Improved Engine Control System 700B may be installed. Here an Enhanced Engine Control Unit 708 may be utilized, as described in Figure 7B, which incorporates the functionalities of the Engine Control Unit 108 and Engine control Co-Processor 710.

[00137] The Engine Sensory Data 112 and Driver Input 110 are provided to the Enhanced Engine Control Unit 708 for determining the eventual desired Engine Control Output 216. Engine Sensory Data 112 may include limited information as engine speed, or may include a much wider set of information such as current operational mode, MAF, MAP, Exhaust Oxygen levels, vehicle speed, weight, slope that the vehicle is on, towing load, road friction resistance, ambient humidity, ambient air pressure, ambient temperature, operational information and other relevant vehicle information.

[00138] The Driver Input 110 may come from any suitable source that may be considered a reasonable proxy for a desired engine output. For example, the input signal may simply be a signal indicative of accelerator pedal position taken directly or indirectly from an accelerator pedal position sensor. In vehicles that have a cruise control feature, the Driver Input 110 may come from a cruise controller. In still other embodiments, the Driver Input 110 may be a function of several variables in addition to accelerator position. In other engines, that have fixed operational states, the Driver Input 110 may be set based on a particular operational setting. In general, the desired output signal found in the Driver Input 110 may come from any suitable source that is available in the vehicle or engine being controlled. [00139] The Engine Control Unit 108 often includes look up tables for the fuel injection driver. As will be appreciated by those familiar with the art, the designs of the existing ECUs and their respective interfaces vary significantly and accordingly, the Engine Control Co-Processor 710 may be adapted and designed to work with the particular ECU provided for the engine.

[00140] When operating in a high efficiency skip fire mode, the Engine Control Co-Processor 710 effectively overrides the fuel injection level instructions calculated by the Engine Control Unit 108, and instead orders the fueling and valve control determined to be appropriate by the Engine Control Co-Processor 710. The Engine Control Co-Processor 710 also may correct for other inputs (such as the oxygen sensor input when applicable) as appropriate to insure that the rest of the engine's systems run correctly.

[00141] Additionally, the Engine Control Co-Processor 710 may be enables, in some embodiments, to monitor NOx levels downstream from the lean NOx trap in order to determine when to regenerate the lean NOx trap. This regeneration may include the addition of a reducing agent (such as fuel, urea or hydrogen) or may include the operation of the engine in a rich fuel burn. In addition to monitoring NOx levels, the Engine Control Co-Processor 710 may rely upon lean NOx trap saturation models to assist in the determination of timing and length of regeneration.

[00142] In this embodiment, the Engine Control Co-Processor 710 and the Engine Control Unit 108 include and/or are coupled to a Fuel Injector Driver 730 for each of the fuel injectors so that the Engine Control Co-Processor 710 itself may drive the fuel injectors. Thus, the Engine Control Unit 108 and the Engine Control Co- Processor 710 may operate in parallel, with each receiving inputs (i.e. Engine Sensory Data 112 and Driver Input 110) and both determining the appropriate engine control, which are fed to an Engine Control Multiplexer 740. When the engine is operating in high efficiency skip fire mode, the Multiplexer 740 is directed to only deliver the signals received from the Engine Control Co-Processor 710 to the fuel injectors (and any other components controlled by the fuel co-processor). Any time the engine is taken out of these enhanced skip fire control modes, the Multiplexer 740 is directed to only deliver the signals received from the Engine Control Unit 108 to the fuel injectors (and other components). Any components that are controlled by the Engine Control Unit 108 in both the normal and enhanced control modes may always be controlled directly by the Engine Control Unit 108.

[00143] The resulting signal from the Multiplexer 740 may include the Engine Control Output 216. This Engine Control Output 216 may include valve control information, fuel injection control and other information such as oxygen sensor input corrections when applicable.

[00144] In Figure 7B, a single ECU with Enhanced Engine Control 708 is illustrated. This Enhanced ECU receives the Engine Sensory Data 112 and Driver Input 110 to generate a wide range of output including fuel injection control and cylinder valve controls. The Enhanced ECU 708 may couple to the Fuel injection Driver 730 and produce final Engine Control Output 216.

[00145] Many methods may exist for wiring the instant invention with an engine control unit. As examples, Figures 11 A and 1 IB provide logical wiring and connectivity diagrams for the instant invention.

[00146] As will be appreciated by those familiar with the art, the designs of the existing ECUs and their respective interfaces vary significantly and, accordingly, the fuel co-processor must be adapted and designed to work with the particular ECU provided for the engine. Conceptually, the ECU typically includes an input cable having a plurality of input lines that deliver the signals and sensor inputs required by the ECU and an output cable that includes a plurality of output lines that deliver the control and other outputs provided by the ECU to other devices. In practice, the input and output cables may be integrated into a single cable bundle or multiple bundles that mix input and output lines, and/or may include some duplexed I/O lines.

[00147] As noted above, most late model automotive engine control units (ECUs) have external interfaces that permit third party devices to interact with the ECU. Often, this interface takes the form of a diagnostic interface. The Engine Control Unit 108 in the embodiment illustrated in Figure 11 A includes an external diagnostics interface 1116 and the Engine Control Co-Processor 710 communicates with the ECU through the diagnostic interface. Specifically, an ECU bus cable connects the Engine Control Co-Processor 710 to the diagnostic interface 1116. The input cable is connected to a splitter 1114 that delivers the input signals to both the Engine Control Unit 108 and the Engine Control Co-Processor 710. Therefore, the co-processor has all of the information available to it that is available to the ECU. When operating in the high efficiency skip fire mode, the fuel co-processor communicates with the ECU over the ECU bus cable and overrides the fuel injection instructions calculated by the ECU, and instead orders the fueling timing and amounts determined to be appropriate by the fuel co-processor. The co-processor also overrides other inputs (such as the oxygen sensor input when applicable) as appropriate to insure that the rest of the engine's systems run correctly.

[00148] Another fuel co-processor wiring embodiment is illustrated in Figure 1 IB. In this embodiment, the Engine Control Co-Processor 710, in addition to the Engine Control Unit 108, includes drivers for each of the fuel injectors so that the fuel coprocessor itself can drive the fuel injectors. In this arrangement, as noted above, the Engine Control Unit 108 and the Engine Control Co-Processor 710 operate in parallel, with each receiving inputs from the input cable and both determining the appropriate engine control, which are fed to multiplexor 1126. When the engine is operating in the high efficiency mode, the multiplexor 1126 is directed to only deliver the signals received from the Engine Control Co-Processor 710 to the fuel injectors (and any other components controlled by the fuel co-processor). Any time the engine is taken out of the high efficiency skip fire mode, the multiplexor 1126 is directed to only deliver the signals received from the ECU to the fuel injectors (and other components). Any components that are controlled by the ECU in both the normal and high efficiency operating modes are always controlled directly by the ECU.

[00149] In the embodiment of Figure 1 IB, the Engine Control Co-Processor 710 communicates with the ECU over ECU bus cable through the diagnostic interface 1116 and is arranged to override any input signals that need to be corrected for when the engine operates in the high efficiency mode.

[00150] Although specific wirings of the fuel co-processor are illustrated in

Figures 11 A and 1 IB, it should be appreciated that a wide variety of other wirings and/or couplings may be utilized. For example, in some cases only a portion of input signals may be delivered to the Engine Control Co-Processor 710, since some of the input signals may not be relevant to the operation of the Engine Control Co-Processor 710. Additionally or alternatively, input signals that are intended to be modified by the Engine Control Co-Processor 710 may be wired to first be input to the Engine Control Co-Processor 710 and then a (potentially) modified signal may be fed from the Engine Control Co-Processor 710 to the Engine Control Unit 108. That is, the Engine Control Co-Processor 710 may intercept some or all of the input signals and modify some of those signals as appropriate prior to their delivery to the Engine Control Unit 108.

[00151 ] In still other embodiments, some or all of the output lines may be connected to the Engine Control Co-Processor 710 rather than the Engine Control Unit 108. This is particularly appropriate in implementations in which the Engine Control Co-Processor 710 is designed to determine the fuel injection and/or valve control in all operations of the engine.

A. EXAMPLE ENGINE CONTROL CO-PROCESSOR

[00152] Now particular attention will be given to a number of embodiments of the Engine Control Co-Processor 710 and the control logic circuitry utilized by this example Engine Control Co-Processor 710. Note that this section provides examples engine control systems which are enabled to provide rapid adaptive predictive control of cylinder operation on a working cycle by working cycle basis. Of course other skip fire systems and control circuits may likewise be utilized by some embodiments of the present invention. Thus, for example, known variable displacement systems, which shut down a bank of the engine for prolonged periods of time, may be just as useful in some embodiments of the invention. Thus, the disclosed example skip fire control circuitry is not intended to limit the scope of the invention, but rather is intended for clarification purposes.

[00153] Figure 8 is a structural block diagram for an example of the Fuel Processor 710. In the illustrated embodiment, the Fuel Processor 710 includes a Preprocessor 812, a Drive Pulse Generator 814 and a Sequence Generator 816. The Engine Sensory Data 112 and the Driver Input 110 may be provided to the Preprocessor 812.

[00154] In some of the embodiments described, a signal from the accelerator pedal position is treated as the indication of the desired engine output that is used as the input to the control system (e.g., drive pulse generator 814, Engine Control Unit 108, etc). In such embodiments, the desired engine output signal (Driver Input 110) may be taken directly from a pedal position sensor on the vehicle, or it may be amplified in an appropriate manner. In other embodiments, the pedal position sensor signal may be combined with other inputs (such as the dither signal described below) before it is provided to the drive pulse generator 814. In yet other embodiments, the accelerator pedal position sensor signal may be provided to the Output Calculator Preprocessor 812, which either generates its own signal or does some level of processing on the pedal sensor signal. The output of the Output Calculator Preprocessor 812 may then be used as the input to the drive pulse generator, with or without an additional dither signal as may be appropriate for a particular design.

[00155] The Output Calculator Preprocessor 812 may be arranged to provide any desired type of preprocessing of the accelerator pedal position sensor signal. For example, it may be desirable for an automobile to provide a fuel savings mode where the accelerator pedal position signal is preprocessed in a way that helps operate the engine in the most fuel efficient manner. In another example, it is generally known that some drivers tend to relatively rapidly fluctuate the pedal position. For such drivers it may be desirable for an automobile to provide a smooth driving mode in which a preprocessor averages or smoothes certain pedal position fluctuations (e.g., the preprocessor may take the form of, or include a low pass filter). In still other implementations, the vehicle may include a cruise controller. In such vehicles, the cruise controller may be incorporated in the preprocessor or may serve as the source of the drive pulse generator's input signal when the vehicle is in the cruise control mode. In still other embodiments, anti-aliasing filtering of the pedal position may be provided in the Preprocessor 812. Of course, the preprocessor may be arranged to perform any other type of preprocessing that is deemed appropriate for the engine and/or vehicle being controlled. The preprocessor may, in some embodiments, determine the operational mode that the engine is to work in. Likewise, in some embodiments, the Output Calculator Preprocessor 812 may be configured to monitor NOx emission levels and determine when a regeneration cycle is needed. This process will be described in more detail below.

[00156] The Drive Pulse Generator 814 may be arranged to use adaptive predictive control to dynamically calculate a drive pulse signal that generally indicates when firings at specific fueling levels are required to obtain the desired output. As will be discussed in more detail below, the controller may be synchronized with the engine speed (part of the Engine Sensory Data 112) so that the generated drive pulse pattern is appropriate to deliver the power desired at the current engine speed, which may be constantly changing.

[00157] The drive pulse signal may then be provided to a Sequence Generator 816 that orders the pulses to provide the final Cylinder Specific Operation Data 802. Generally, the Sequence Generator 816 may be arranged to order the combustion pattern in a manner that helps prevent excessive or inappropriate vibration within the engine. As is well known in the engine design field, the order in which cylinders are fired may have a significant effect on vibrations within many engines. Therefore, as will be described in more detail below, the Sequence Generator 816 is designed to help insure that vibrations generated by the operation of the engine are within design tolerances. If a particular engine is enabled to be run using an arbitrary firing pattern (i.e., the cylinders may be fired in any pattern without generating undue vibrations), then the sequencer may potentially be eliminated and the Drive Pulse Generator 814 could be used to dictate the firing pattern.

[00158] Note that while "firing patterns" and "firing" of the cylinders is disclosed, in diesel and HCCI style engines an ignition event is caused by the introduction of fuel into the heated compressed cylinder. Thus, fueling may be considered synonymous with firing in compression ignition engines. In gasoline Otto cycle engines, firing refers to first fueling the cylinders and then providing a charge to the spark plugs.

[00159] The Drive Pulse Generator 814 is generally arranged to determine the number and general timing of cylinder combustion events that are required to generate the desired output given the current operating state and operating conditions of the engine. The Drive Pulse Generator 814 uses feedback control, such as adaptive predictive control to determine when cylinders must be fueled to deliver the desired engine output. Thus, the drive pulse signal outputted by the Drive Pulse Generator 814 effectively indicates the instantaneous displacement required by the engine to deliver the desired engine output.

[00160] The power output required by the engine will vary with operating conditions and may be based on both what has happened in the past and what is predicted for the immediate future. In various embodiments, the Drive Pulse

Generator 814 is generally not constrained to limit fluctuations in the number of cylinder firings that are required per revolution of the crankshaft to deliver the desired output. Thus, the effective displacement of the engine may be continuously varied by selecting which cylinders to fire and which cylinders not to fire, and fueling levels to initiate these firing events on a firing opportunity by firing opportunity basis. This ability to continuously vary the effective displacement of the engine is sometimes referred to herein as a 'continuously variable displacement' mode of operation.

[00161] A variety of different control schemes may be implemented within the Drive Pulse Generator 814. Generally, the control schemes may be implemented digitally, algorithmically, using analog components or using hybrid approaches. The drive pulse generator may be implemented on a processor, on programmable logic such as an FPGA, in circuitry such as an ASIC, on a digital signal processor (DSP), using analog components, etc.

[00162] One class of controllers that is particularly well suited for use in the drive pulse generator is adaptive predictive controllers. As will be appreciated by those familiar with control theory, adaptive predictive controllers are adaptive in that they utilize feedback to adapt or change the nature of their output signal based on the variance of the output signal from a desired output signal and predictive in that they are integrative so that past behavior of the input signal affects future output signals.

[00163] A variety of different adaptive predictive controllers may be used to calculate the chamber firings required to provide the desired output. One class of adaptive predictive controllers that work particularly well in this application is sigma delta controllers. The sigma delta controller may utilize sample data sigma delta, continuous time sigma delta, algorithm based sigma delta, differential sigma delta, hybrid analog/digital sigma delta arrangements, or any other suitable sigma delta implementation. In some embodiments, the sigma delta controller's clock signal is arranged to vary proportionally with the engine speed. In other implementations, a variety of other adaptive predictive controllers including pulse width modulation (PWM), least means square (LMS) and recursive least square (RLS) controllers may be used to dynamically calculate the required chamber firings.

[00164] Looking now at Figures 10A, 10B and IOC, various feedback architectures may exist for the Engine Control Co-Processor 710. Figure 10A provides a functional block diagram that diagrammatically illustrates the drive pulse generator 814 and a sequencer 816. Driver Input 110 that is indicative of a desired engine output is provided to the drive pulse generator 814. The drive pulse generator 814 may be arranged to use adaptive predictive control to dynamically calculate a drive pulse signal 1010 that generally indicates when cylinder firings are required to obtain the desired output. As will be discussed in more detail below, the controller may be synchronized with the engine speed (Engine Sensory Data 112) so that the generated drive pulse pattern is appropriate to deliver the power desired at the current engine speed, which may be constantly changing. The drive pulse signal 1010 may then be provided to a sequencer that orders the pulses to provide the final cylinder firing pattern 1020.

[00165] In a first implementation, each cylinder that is fired is operated at or near its optimal thermodynamic efficiency. That is, air and fuel are introduced into the cylinder in amounts that allow the most work to be obtained from the cylinders per unit of fuel burnt while still meeting other constraints on the engine (such as emissions requirements, the effects of the combustion on engine life, etc.). Many vehicles include engine control units (ECUs) that determine (among many other things) the desired air/fuel ratios and the amount of fuel to be injected for each cylinder firing. Often the ECUs have lookup tables that identify the desired air fuel ratios and/or fuel injection amounts for a number of different operating conditions (e.g. engine speeds, manifold air flow, etc.) based on various current ambient conditions (including air pressure, temperature, humidity etc.). In such vehicles, the amount of fuel that the firing control unit causes to be injected into each cylinder in the continuously variable displacement mode may be the value stored in the fuel injection lookup table for operating the cylinder at full throttle under the current conditions.

[00166] The drive pulse generator 814 is generally arranged to determine the number and general timing of cylinder firings that are required to generate the desired output given the current operating state and operating conditions of the engine. The drive pulse generator uses feedback control, such as adaptive predictive control to determine when cylinders must be fired to deliver the desired engine output.

Components of the feedback may include feedback of the drive pulse signal 1010 and/or feedback of the actual cylinder firing pattern 1020 as generally illustrated in Figure 10B. Since the drive pulse signal 1010 indicates when working chamber firings are appropriate, it may generally be thought of as a signal indicative of requested firings. The sequencer then determines the actual timing of the requested firings. When desired, the information fed back from the actual firing pattern 1020 may include information indicative of the firing pattern itself, the timing of the firings, the scale of the firings and/or any other information about the cylinder firings that is desired by or useful to the drive pulse generator 814. Generally, it is also desirable to provide the drive pulse generator 814 with an indication of the engine speed (included in the Driver Input 110) so that the drive pulse signal 1010 may generally be synchronized with the engine.

[00167] Various feedbacks may also be provided to the sequencer 816 as desired. For example, as illustrated diagrammatically in Figure IOC, feedback or memory indicative of actual firing timing and/or pattern 1020 may be useful to the sequencer to allow it to sequence the actual cylinder firings in a manner that helps reduce engine vibrations.

B. DRIVE PULSE GENERATORS

[00168] As previously noted, the Drive Pulse Generator 814 may be any adaptive predictive controller capable of generating a firing pattern for the engine. Sigma Delta circuits have been found to be particularly suited for this, and below are provided a number of possible embodiments suitable for the Drive Pulse Generator 814. Again, the example circuits and control logic provided are not limiting to the scope of the invention, but rather are provide as a means for clarification.

[00169] Figure 9A is a structural block diagram for a first example embodiment of a sigma-delta control based Drive Pulse Generator 814A for the Engine Control Co- Processor 710. The Drive Pulse Generator 814 includes a sigma-delta controller 910 and a synchronizer. The sigma-delta controller 910 utilizes principles of sigma-delta conversion, which is a type of oversampled conversion. (Sigma-delta conversion is also referred to as delta-sigma conversion.) The basic theory of sigma-delta conversion has been described in what is commonly referred to as a seminal reference on the subject: H. Inose, Y. Yasuda, and J. Murakami, "A Telemetering System by Code Modulation: Δ-Σ Modulation," IRE Transactions on Space Electronics

Telemetry, Vol. SET-8, September 1962, pp. 204-209. Reprinted in N. S. Jayant, Waveform Quantization and Coding, IEEE Press and John Wiley, 1976, ISBN 0-471- 01970-4.

[00170] The illustrated sigma-delta control circuit 910 is an analog third order sigma-delta circuit generally based on an architecture known as the Richie architecture. Sigma-delta control circuit 910 receives an analog input signal that is indicative of a desired output (which might be thought of as desired work output or desired torque). Since sigma-delta controllers of the type illustrated are generally known and understood, the following description sets forth the general architecture of a suitable controller. However, it should be appreciated that there are a wide variety of different sigma-delta controllers that may be configured to work very well for a particular implementation.

[00171] In the illustrated embodiment, the desired output is indicative of accelerator pedal position as included in Driver Input 110 (although as described above, other suitable input signals indicative of, or proxies for, desired output may be used as well). The input signal is provided as a positive input to the sigma-delta control circuit 910, and particularly to a first integrator 914. The negative input of the integrator 914 is configured to receive a feedback signal that is a function of the output such that the operation of the sigma delta control circuit 910 is adaptive. As will be described later, the feedback signal may actually be a composite signal that is based on more than one output stage. The integrator 914 may also receive other inputs such as dither signal provided from a Pseudo Random Dither Generator (PRD) 904 which also will be described in more detail below. In various implementations some of the inputs to integrator 914 may be combined prior to their delivery to the integrator 914 or multiple inputs may be made directly to the integrator. In the illustrated embodiment, the dither signal 904 is combined with the input signal by an adder 902 and the combined signal is used as the positive input. The feedback signal is a combination of feedback from the output of the sigma delta control circuit and the controlled system.

[00172] The sigma delta control circuit 910 includes two additional integrators, integrator 916 and integrator 918. The "order" of the sigma delta control circuit 910 is three, which corresponds to the number of its integrators (i.e., integrators 914, 916 and 918). The output of the first integrator 914 is fed to the second integrator 916 and is also fed forward to the third integrator 918.

[00173] The output of the last integrator 918 is provided to a comparator 920 that acts as a one-bit quantizer. The comparator 920 provides a one-bit output signal that is synchronous with a clock signal. Alternatively, the output may include a multi-bit output which includes pulse amplitude variations. The amplitude may correspond to the eventual level of fuel provided to the cylinder at the fueling event. Alternatively, a downstream logic circuit may be configured to interpret signal pulses and generate fueling levels.

[00174] Generally, in order to insure very high quality control, it is desirable that the clock signal (and thus the output stream of the comparator) have a frequency that is many times the maximum expected firing opportunity rate. For analog sigma delta control circuits, it is typically desirable for the output of the comparator to oversample the desired drive pulse rate by a factor of at least about 10 and oversampling factors on the order of at least about 100 works particularly well. That is, the output of the comparator 920 is preferably at a rate of at least 10 times and often at least 100 times the rate at which engine firing opportunities occur. The clock signal provided to the comparator 920 may come from any suitable source. For example, the clock signal is provided by a crystal oscillator clock 906.

[00175] It should be appreciated that these clock rates are actually relatively slow for modern digital electronic systems and are therefore readily obtainable and usable. For example, if the controlled engine is a eight-cylinder engine that operates using a four stroke working cycle, then the maximum firing opportunity rate expected might be something on the order of 8,000 RPM x 8 cylinders x 1/2. The factor of 1/2 is provided because, in a normally-operating four-cycle engine, each cylinder has a combustion opportunity only once every two revolutions of the engine crankshaft. Thus, the maximum expected frequency of firing opportunities may be approximately 32,000 per minute, or about 533 per second. In this case, a clock operating at about 50 kHz would have nearly 100 times the maximum expected rate of firing

opportunities. Therefore, a fixed clock having a clock frequency of 50 kHz or greater would work very well in that application. [00176] In other embodiments, the clock used to drive the comparator may be a variable clock that varies proportionally with engine speed. It is believed that the use of a variable speed clock in a sigma delta controller is different than conventional sigma delta controller design. The use of a variable speed clock has the advantage of insuring that the output of the comparator is better synchronized with the engine speed and thus the firing opportunities. The clock may readily be synchronized with the engine speed by utilizing a phase lock loop 908 that is driven by an indication of engine speed (e.g., a tachometer signal).

[00177] The one -bit output signal outputted from the comparator 920 is generated by comparing the output of the integrator 918 with a reference voltage. The output is effectively a string of ones and zeros that is outputted at the frequency of the clock. The output of the comparator 920 (which is the output of the sigma delta control circuit 910 is provided to a synchronizer that is arranged to generate the drive pulse signal. In the illustrated embodiment, the sigma delta control circuit 910 and the synchronizer together constitute a drive pulse generator 814.

[00178] The synchronizer is generally arranged to determine when drive pulses should be outputted. The drive pulses are preferably arranged to match the frequency of the fueling opportunities so that each drive pulse generally indicates whether or not a particular working cycle of a working chamber should be exercised. In order to synchronize the drive pulse signal with the engine speed, the synchronizer operates using a variable clock signal that is based on engine speed from the Engine Sensory Data 112. A phase-locked loop 908 may be provided to synchronize the clock with the engine speed. Preferably, the clock signal has a frequency equal to the desired frequency of the outputted drive pulse signal. That is, it is preferably synchronized to match the rate of combustion opportunities.

[00179] The output signal of the sigma-delta control circuit is generally a digital representation of the analog input signal that is received by the sigma-delta control circuit 910. The digital output signal from the sigma delta control circuit 910

contains a certain number of "high" symbols it is appropriate to generate a positive drive pulse (i.e., to order the fueling of a working chamber). Thus, conceptually, a purpose of the synchronizer may be thought of as being to count the number of high symbols in the output signal and when enough symbols are counted, sending a drive pulse that is synchronized with the engine speed. Additionally, the amplitude of drive pulses may be used to dictate fueling levels. In practice, true counting is not actually required (although it may be done in some implementations). Additionally, the number of chambers utilized for a desired application may be calculated from the number of "high" symbols generated over a set period of time.

[00180] Another characteristic of the output of the described sigma-delta control circuit with a high oversampling rate when used in this type of engine control application is that the controller tends to emit long blocks of high signals followed by blocks of low signals. This characteristic of the output signal may be used to simplify the design of the synchronizer. In one implementation, the synchronizer merely measures the length (i.e., time or period) of the blocks of high signals emitted in output signal. If the length of the block of high signals exceeds a designated threshold, a drive pulse is generated. If the length of a block of high signals doesn't exceed the threshold - no drive pulses are generated based on that block of high signals. The actual thresholds that are used may be widely varied to meet the needs of a particular design. For example, in some designs the threshold may be the period of the clock signal which (since the clock is synchronized with the engine speed) corresponds to the duty cycle of the drive pulse pattern and the average delay between working chamber firing opportunities. With this arrangement, if the length of a block of high signals is less than one duty cycle, no drive pulses are generated

corresponding to the block; if the length of the block exceeds one duty cycle and is less than two duty cycles, then one drive pulse is generated; if it exceeds two duty cycles but is less than three duty cycles, then two sequential drive pulses are generated; and so on. In some embodiments, every output signal may have an amplitude modulation indicating the designated fuel amount.

[00181] It should be appreciated that with this arrangement, the "length" or time duration of a burst of high outputs from the sigma-delta control circuit will have to be longer in order to trigger a drive pulse when the engine speed is low than the length of a burst would need to be in order to trigger a drive pulse when the engine speed is high. That is because the duty cycle of the drive pulse signal is longer at lower engine speeds.

[00182] In other implementations, the threshold may be set differently. For example, the thresholds might be set such that any block of high outputs having a length that exceeds some designated percentage (e.g., 80 or 90 percent) of the duty cycle of the drive pulse signal causes a drive pulse to be generated, while shorter pulse lengths are effectively truncated.

[00183] At first review it may seem that ignoring portions of pulses in the manner suggested above could degrade the performance of the control system to an unacceptable level. However, for many engines, the high frequency of the firing opportunities and the responsiveness of the control system in general make it perfectly acceptable to use such simple synchronizers. Of course, it should be appreciated that a wide variety of other synchronization schemes may be used as well.

[00184] It should be appreciated that although the comparator output, the drive pulse signal and the actual fueling pattern are all related, their timing will vary and the general magnitude of the comparator output may differ from the others. The most accurate feedback in terms of reflecting actual engine behavior is the fueling pattern; however, there may be significant time delays (from the standpoint of the sigma-delta control circuit 910) between the output of the comparator and the actual combustion event of a working chamber. The next best feedback in terms of reflecting actual engine behavior is the drive pulse signal. Thus, in many implementations it will be desirable to heavily weight the feedback towards the drive pulse signal and/or the fueling pattern. However, in practice, the performance of the sigma delta controller may often be enhanced by feeding back some portion of the comparator output signal.

[00185] In some embodiments, it may be desirable to anti-aliasing filter the input signal and the feedback signal. The anti-aliasing functionality may be provided as part of the sigma-delta control circuit 910 or it may be provided as an anti-aliasing filter that precedes the sigma delta control circuit or it may be provided in any other suitable form. In the third order analog continuous time sigma-delta control circuit 910 illustrated in Figure 9A, the first integrator 914 may provide the anti-aliasing functionality. That is, it effectively acts as a low pass filter.

[00186] Another known characteristic of sigma delta controllers is that they sometimes generate "tones" which are cyclic variations in the output signal relative to the input signal. Such tones are particularly noticeable when the analog input signal varies slowly, which is often the case when driving and in many other engine control applications. The presence of such tones within the comparator output signal may be reflected in the engine firing pattern. In some situations, there is a risk that such cyclic variations in the drive pattern may generate undesirable resonances within the engine which may generate undesirable vibration patterns. In extreme cases, the tones could even be manifested as a noticeable fluctuation in drive energy.

Accordingly, various arrangements may be provided in an effort to help prevent and/or break up such tones. One option that may help prevent and/or break up tones in the sigma-delta controller output is to combine the input signal with a noise signal ("dither") that averages to zero over time, but whose local variations tend to break up the tones in the output signal of the sigma delta controller. A pseudo-random dither generator (PRD) 904 may be employed to generate the dither signal, but it should be appreciated that dither may be introduced using a wide variety of other approaches as well.

[00187] The output of the synchronizer is the drive pulse signal discussed above. The drive pulse signal effectively identifies the cylinder combustion events brought about by cylinder fueling (or instantaneous effective engine displacement) that is needed to provide the desired engine output. That is, the drive pulse signal provides a pattern of pulses that generally indicates when cylinder fueling is appropriate to provide the desired or requested engine output. In theory, the cylinders could be fueled directly using the timing of the drive pulse signal outputted by the

synchronizer. However, in many cases it will not be prudent to fuel the cylinder using exactly the same timing as pulse pattern because this could generate undesirable vibrations within the engine. Accordingly, the drive pulse signal may be provided to the Sequence Generator 816 which determines an appropriate fueling pattern. The Sequence Generator 816 is arranged to distribute the cylinder combustion events called for in a manner that permits the engine to run smoothly without generating excessive vibrations.

[00188] In still other embodiments, differential sigma delta controllers may be used. In such embodiments the synchronizer may be arranged to generate drive pulse patterns based on the differential signals outputted by the sigma delta controller. A wide variety of different differential sigma delta controllers may be used and generally they may include the variable clock and/or multi-bit comparator output features discussed above when desired. One advantage of differential sigma delta controllers is that they may often be configured to provide even smoother

performance than a corresponding non-differential sigma delta controller.

[00189] Figure 9B is a structural block diagram for a second example embodiment of a Drive Pulse Generator 814B for the Engine Control Co-Processor 710. This alternative embodiment of the drive pulse generator incorporates a variable clock sigma delta controller 910. The drive pulse generator 814B has a structure very similar to the drive pulse generator 814A described above with reference to Figure 9A. However, in this embodiment, the clock signal provided to the comparator 920 is a variable clock signal that is based on engine speed. The clock signal is generally synchronized with the engine speed by utilizing a phase lock loop 908 that is driven by an indication of engine speed (e.g., a tachometer signal).

[00190] As described above, it is desirable for the sigma delta controller to have a sampling rate, and therefore an output signal frequency that is substantially higher than the desired frequency of the drive pulse pattern outputted by the synchronizer 930. Again, the amount of oversampling can be widely varied. As indicated above, oversampling rates on the order of 100 times the desired drive pulse frequency work well and accordingly, in the illustrated embodiment a divider 922 is arranged to divide the clock signal provided to the synchronizer logic by a factor of 100 (for this example), and the output of the divider 922 is used as the clock for comparator 920. This arrangement causes the output of the comparator 920 to have a frequency of 100 times the frequency of the drive pulse pattern that is output by the synchronizer 930. Of course, in other embodiments, the divider can be arranged to divide the signal by any integer number that provides sufficient oversampling. In other respects, the other components of the sigma delta controller 910 may be the same as described above with respect to Figure 9A. Any of the designs of the synchronizer 930 and/or the sequencer 816 discussed above or a variety of other synchronizer and sequencer designs may also be used with variable clock sigma delta controller 910. One advantage of synchronizing the output of the sigma delta controller 910 with the engine speed is that it permits simpler synchronizer designs.

[00191] This embodiment of a sigma delta controller output may be configured to output a multi-bit signal. The multi-bit output of the sigma delta controller is used by the synchronizer 930 to generate partial drive pulses. In this embodiment, the sigma delta controller 910 may have a design similar to any of the previously described embodiments, however the comparator 920 is arranged to output a multiple bit signal. In some described embodiments, the comparator 930 is two bit comparator and accordingly the output signal is a two bit signal. However, in other embodiments higher bit comparators may be provided which would result in higher bit output signals. The actual number of bits used can be varied to meet the needs of any particular application.

[00192] The various states of the multi-bit output signal can each be set to have an associated meaning. For example, in a two bit comparator, a 0,0 output signal might reflect a zero output; a 1,1 output signal might be a full signal output - e.g. one; a 0,1 might be arranged to represent a ¼ signal; and a 1 ,0 might be arranged to represent a ½ signal. Thus, firing and amount of fueling may be coded as well as engine operation modality. Of course a two-bit comparator may readily be designed to have the various states represent different levels than the 0, ¼, ½, and 1 levels suggested above. In higher order comparators, many more states would be available. For example, in a three-bit comparator, 8 states would be available; in a four-bit comparator 16 states would be available, etc. Thus, with a comparator with sufficient bits, say a five-bit, a fine granularity of fueling amount instructions, operational modes and fuel injection timing may be coded by the output.

[00193] As will be appreciated by those familiar with multi-bit comparator sigma delta design, the comparator 920 may be configured to output some (generally controllable) percentage of the non-zero samples as intermediate level signals. These intermediate signals can be treated by the synchronizer 930 and the sequencer 816 as corresponding to requests for partial energy drive pulses and reduced energy firings (i.e. varied fuel amounts). For example, if the sigma delta controller 910 outputs a string of one half (½) level output signals that is sufficiently long to cause the synchronizer 930 to generate a drive pulse, the outputted drive pulse can be a half energy drive pulse. The half energy drive pulse, in turn, is used by the sequencer 816 to direct a half energy fueling. The same type of logic can be used for other (e.g. quarter) level output signals. When the comparator output is a multi-bit output, the synchronizer and the sequencer may readily be arranged to handle and output corresponding multi-bit signals. [00194] It should be appreciated that multi-bit comparator sigma delta controllers are typically arranged to generate extended strings of symbols having the same state. Therefore, any of the general sequencer logics described above can be used to output drive pulses having the same state as the signal fed to the synchronizer 930. That is, the drive pulse outputted by the synchronizer may be arranged to match the level of the signal input into the synchronizer that caused generation of the drive pulse.

[00195] Again, it should be appreciated that the logic of the synchronizer 930 can be optimized and widely varied to meet the needs of any particular application. In some applications it may be desirable to provide more sophisticated synchronizer logic to handle specific situations in a desired manner. For example, different logic may be provided to handle situations where the signal inputted to the synchronizer transitions from a higher level to a lower non-zero level that is held through the end of a drive pulse period. In some implementations it may be desirable to have the drive pulse output at the lower level in such situations. Similarly, when the signal transitions from a lower non-zero level to a higher non-zero level, it may be desirable to provide specific logic to dictate what happens in such circumstances.

[00196] It should be appreciated that the thermodynamic (and fuel) efficiency of the engine will be best when the working chambers are operated at their optimal efficiency. The optimal efficiency for typical compression ignition internal combustion (diesel) engines is roughly between 10% and 20% of total power output of the engine. For gasoline engines, optimal efficiency typically occurs at maximum power (unthrottled) since pumping losses are minimized. However, engine design and ambient conditions may vary the range of optimal efficiency. Therefore, it is generally desirable to have as many cylinders firings at near optimal fueling level as possible.

[00197] The comparator and synchronizer logic may also be arranged to take engine speed and/or the operational state of the engine (e.g. cold start, etc.) into account when determining the percentage of intermediate comparator output signals and/or partial drive pulses to output. For example, when the engine is idling or cold starting, it may be desirable to only output intermediate signals from the comparator so that only partial drive pulses are generated in those situations. Likewise, during regeneration cycles, the engine may be operated at fuel rich levels despite the negative impact this has upon fuel efficiency. It should be appreciated that the comparator and/or synchronizer logic may be arranged to accommodate a wide variety of different desired operational rules.

[00198] Figure 9C is a structural block diagram for a third example embodiment of a Drive Pulse Generator 814C for the Engine Control Co-Processor 710. The Drive Pulse Generator 814C illustrated includes a digital third order sigma delta control circuit 950. In this embodiment, accelerator pedal position indicator signal (as part of Driver Input 110) is inputted to a first digital integrator 956. The output of the first digital integrator 956 is fed to a second digital integrator 962 and the output of the second digital integrator 962 is feed to a third digital integrator 968. The output of the third digital integrator 968 is fed to a comparator 920 that may be arranged to operate in the same manner as either the single bit or multi-bit comparators described above with respect to the analog sigma delta circuits. In the embodiment illustrated in this example the first digital integrator 956 effectively functions as an anti-aliasing filter.

[00199] Negative feedback is provided to each of the three digital integrator stages 956, 962 and 968. The feedback may come from any one or any combination of the output of the comparator 920, the output of the synchronizer logic or the output of the Sequence Generator 816. Each stage feedback has a multiplication factor (954, 960 and 966) of L, M, and N respectively.

[00200] Like the analog sigma delta control circuits described above, the primary input to the digital sigma delta control circuit may be an indication of the accelerator position or any other suitable proxy for desired output (from Driver Input 110). As previously described, the desired output signal is combined with pseudo random dither signal 904 in the illustrated embodiment in order to reduce the possibility of generating undesirable tones.

[00201] The primary difference between analog and digital operation is that the integrators in analog sigma delta are continuously active, whereas the digital integrators are only active at the beginning of each clock cycle. In some

implementations, it may be desirable to run the clock at a very high speed. However, that is not a requirement. Since the output that is ultimately desired has a frequency that is equal to the fueling opportunities that are being controlled, the clock may be synchronized with the fueling opportunities which may eliminate the need for (or simplify the function of) the synchronizer and/or the Sequence Generator 816. Thus, when a digital controller is used, the controller design may be simplified by running the clock at the frequency of the fueling opportunities being controlled.

[00202] Although analog and digital controllers have been described, it should be appreciated that in other implementations, it may be desirable to provide hybrid analog/digital sigma delta controllers. In a hybrid analog/digital controller, some of the stages of the sigma delta controller may be formed from analog components, while others may be formed from digital components. One example of a hybrid analog/digital sigma delta controller utilizes an analog integrator 914 as the first stage of the controller, in place of the first digital integrator 956. The second and third integrators are then formed from digital components. Of course, in other

embodiments, different numbers of stages may be used and the relative number of analog vs. digital integrators may be varied. In still other embodiments, digital or hybrid differential sigma delta controllers may be used.

[00203] In still other embodiments differential sigma delta controllers may be used. In such embodiments the synchronizer can be arranged to generate drive pulse patterns based on the differential signals outputted by the sigma delta controller. A wide variety of different differential sigma delta controllers may be used and generally they may include the variable clock and/or multi-bit comparator output features discussed above when desired. One advantage of differential sigma delta controllers is that they can often be configured to provide even smoother performance than a corresponding non-differential sigma delta controller.

[00204] In some circumstances it may be advantageous to operate in a mode that may be referred to as an implied differential sigma delta. In such a mode, either the synchronizer or the sequencer (or both) are constrained to limit the drive pulses and/or chamber firings to one at a time. That is, in this mode, each fired working chamber is constrained to be followed by a skipped firing opportunity (and/or each drive pulse is constrained to be followed by null pulse). This implied differential sigma delta is particularly useful when the engine is operating at a level where significantly less than 50% of the firing opportunities are required to deliver the desired engine output since it can help further smooth the engine output by insuring that two firings do not immediately follow one another when the required output is relatively low.

[00205] In some implementations it may be desirable to operate the engine in an implied sigma delta mode during some operational conditions, in a different type of continuously variable displacement mode during other operational conditions, in a conventional operating mode during still other operational conditions and a rich fuel (regenerative) mode during yet other conditions, as was described above. Of course, the number and nature of the various operational modes may be widely varied.

Therefore, it should be appreciated that the engine controller may generally be arranged to operate in a variety of different operational modes during different operational conditions.

[00206] The constraints provided by implied sigma delta may also be widely varied. For example, when very low engine outputs are required, there may be instances when it is desirable to constrain the firings pattern to skip at least two firing

opportunities after each firing. In other instances it may be desirable to allow two firings to follow one another, but not three. In still other instances, it may be desirable to require a firing any time a designated number of skips follow one another. Generally, it should be appreciated that the firing pattern for a particular engine may be constrained by the sequencer or the synchronizers in a wide variety of manners that are determined to appropriate to provide the desired engine output and the constraints may be arranged to vary with the load placed on the engine, the overall ratio of firings to skips, or any other factor that is appropriate for the control of a particular engine.

IV. METHODS FOR SKIP FIRE ENGINE OPERATION WITH EXHAUST AFTER- TREATMENT

[00207] Figures 13A is an example flow chart diagram for the process of skip fire engine operation with lean NOx trap after-treatment in accordance with some embodiments of the present invention. This example process begins at step 1310 where a subset of cylinders are fueled and fired. This results in some other cylinder being skip fired. Control over which cylinders are fired and which are skip fired may be performed by any of the above adaptive predictive control logic. Likewise, other known skip fire control systems, such as bank deactivation variable displacement systems may be utilized in this step. [00208] Exhaust generated by each of the cylinders may be collected and channeled through an exhaust system. In some alternate embodiments, only some of the cylinder's exhaust gasses are channeled through the exhaust system which includes the lean NOx trap. Thus, in such engines there may be multiple exhaust manifolds and pathways. The exhaust system typically includes a standard multistage catalytic converter, such as a two-way or three-way catalytic converter.

[00209] In some embodiments, the skip fire engine is operating near stoichiometric fuel to air ratios. However, as is noted above, the inclusion of the lean NOx trap enables greater leniency in engine operation, therefore enabling richer or leaner engine operation for longer durations than would be acceptable in a traditional engine. Furthermore, skipped cylinders may pass uncombusted air through to the exhaust system, in some engines. Thus, while a standard three way catalytic converter may be capable of dealing with the NOx emissions from the combusted cylinders alone, when combined with the uncombusted air, the emissions include too much oxygen. The oxygen reacts with some of the carbon monoxide and hydrocarbons to generate water and carbon dioxide prior to reaching the catalytic converter. This leads to too few reducing chemicals (CO and HC) to properly react with the excess NOx. Thus, excess NOx may pass through the catalytic converter. Therefore, the exhaust may be passed through a lean NOx trap, as is indicated at step 1320.

[00210] Due to the lower exhaust temperatures inherent to some embodiments of the present invention, the lean NOx trap may be proximally located to the catalytic converter. No external or artificial cooling apparatuses (cooling loops, radiators and refrigerants) are necessary in these embodiments. The absence of these additional cooling mechanisms enables greater vehicle design flexibility, and easier retrofits of existing vehicles.

[00211] The lean NOx trap, as previously discussed includes an absorber capable of chemically binding to the NOx gas. The absorber may include a zeolite, alkali or alkaline earth material. In some embodiments, a barium salt may be utilized as an absorber.

[00212] The absorption of the NOx continues until the NOx trap becomes saturated, which typically occurs between 30 seconds and 2 minutes of lean engine operation. As skip fired engines operate at or near stoichiometric fuel to air ratios, the levels of NOx may vary significantly from a lean burn engine. Thus, the amount of time it takes for the lean NOx trap to saturate may vary accordingly.

[00213] After lean NOx trap saturation the lean NOx trap may be regenerated according to a regeneration protocol, as indicated at step 1330. Regeneration may include running the engine under rich conditions or through the addition of a reducing agent to the lean NOx trap.

[00214] An inquiry is made, at step 1340, whether the engine is turned off. If so the process ends. Otherwise, the process returns to step 1310, where a subset of cylinders are fueled and fired.

[00215] Looking now at Figure 13B, where the process for fueling and firing a subset of cylinders is described in greater detail, as indicated at 1310. Here the driver input (desired engine output) is received at step 1311. Driver input includes an accelerator petal position, cruise control input, traction control input or any other indicator of the engine power required.

[00216] Additionally, engine sensory data may be received at step 1312. Engine sensory data may include any of engine speed, oxygen levels, NOx levels, intake pressure, humidity, load, or any additional engine sensory data that is pertinent to proper engine control. At a minimum, engine speed may be necessary to synchronize the drive pulses to firing opportunities.

[00217] Then, at step 1313 a firing pattern may be generated which meets the desired engine output. As previously noted, a sigma delta controller may be well suited to generate the firing pattern based upon the desired engine output. The firing pattern may then be synchronized to the engine speed at step 1314.

[00218] The firing pattern may be output as fueling and firing instructions to the fuel injector driver at step 1316. The fuel injector driver may then utilize these instructions to control the fueling and firing of each cylinder in the engine. This sub- process then ends by progressing to step 1320.

[00219] Now the sub-process for regeneration of the lean NOx trap (step 1330) will be described in alternate embodiments in relation to Figures 13C to 13E. These alternate embodiments of regeneration protocols are labeled 1330a, 1330b and 1330c, respectively. [00220] Embodiment 1330a may be seen at Figure 13C. This regeneration protocol relies upon the monitoring of NOx emissions downstream from the lean NOx trap, as indicated at step 1331. These monitored NOx emissions are then compared against a threshold level, at step 1332. Thresholds may be dependent upon saturation models for the lean NOx trap or may be based upon applicable pollution regulations.

[00221] An inquiry is made if the monitored NOx levels are above the threshold levels, at step 1333. If not the process returns to step 1331, where NOx levels are continually monitored. If the monitored NOx levels are above the threshold, however, then the process may progress to step 1334 where the lean NOx trap is regenerated. After regeneration the process may return to step 1340 of Figure 13 A.

[00222] Regeneration of the lean NOx trap may include operation of the engine in a rich fueling mode, or through the addition of a reducing agent to the lean NOx trap. A reducing agent may include ammonia (bound in a urea solution), hydrogen gas, or unburnt fuel to name a few. If rich engine operation is utilized, the system may run the skip fire cylinders at an extremely rich fuel to air ratio. Alternately, the engine may drop out of skip fire operation, and rather operate in a standard mode of operation. Then all cylinders may be operated as a slightly rich fueling mode. This method of operation (non-skip fire during regeneration) may be preferable from an engine efficiency standpoint.

[00223] At the regeneration embodiment presented at Figure 13B the regeneration protocol may rely upon set periods of operation rather than monitored NOx levels. In this process, the time since the last regeneration is tracked, at step 1335. Instead of timing, the number of cycles the engine has operated since last regeneration may be tracked, in some embodiments.

[00224] After a set time, or number of engine cycles, the lean NOx trap may be automatically regenerated, at step 1336. The timing between regeneration cycles may be based upon a NOx saturation model. Typically, such a method of lean NOx trap regeneration may result in more frequent regenerations since it is important that NOx is not emitted into the environment. As there are no actual readings of NOx levels exiting the trap, regeneration models should err on the side of caution, thereby resulting in more frequent regeneration of the lean NOx trap. After trap regeneration the process may return to step 1340 of Figure 13 A. [00225] The final regeneration protocol embodiment described is presented at Figure 13E, shown generally at 1330c. Here NOx emissions are again monitored downstream from the lean NOx trap, at step 1337. These monitored NOx levels are then compared to a saturation model, at step 1338. NOx trap saturation models may be stored as a lookup table within the ECU or Engine Control Co-Processor.

Regeneration may then be performed when the model indicates (after a set time/cycles) or when the monitored NOx levels deviate from the expected levels of the saturation model, as shown at step 1339. This method of regeneration is a hybrid of the previous regeneration protocols. After trap regeneration the process may return to step 1340 of Figure 13 A.

[00226] In sum, systems and methods for skip fire operation of a spark ignition engine with lean NOx trap after-treatment are provided. While a number of specific examples have been provided to aid in the explanation of the present invention, it is intended that the given examples expand, rather than limit the scope of the invention. Although sub-section titles have been provided to aid in the description of the invention, these titles are merely illustrative and are not intended to limit the scope of the present invention.

[00227] While the system and methods has been described in functional terms, embodiments of the present invention may include entirely hardware, entirely software or some combination of the two. Additionally, manual performance of any portion of the methods disclosed is considered by the present invention.

[00228] While this invention has been described in terms of several preferred embodiments, there are alterations, permutations, modifications and various substitute equivalents, which fall within the scope of this invention. It should also be noted that there are many alternative ways of implementing the methods and systems of the present invention. It is therefore intended that the following appended claims be interpreted as including all such alterations, permutations, modifications, and various substitute equivalents as fall within the true spirit and scope of the present invention.