This article describes the methods by which one can reduce their vehicle’s fuel consumption utilizing certain technologies. The main component required to do this is a Wide-Band Zirconia oxygen Sensor. When used in conjunction with the appropriate controller, the Innovate Motorsports LC-1 in this case, and a proper signal conditioner utilizing closed-loop feedback, one can vastly reduce their fuel consumption. These benefits will be realized primarily under light loads as seen while maintaining constant speed on a flat road, light acceleration and while at idle. The air flow value being fed to the motor’s control system will be varied depending on inputs from the sensor itself, the oxygen sensor, throttle position and engine coolant temperature. A signal matching that of the original oxygen sensor’s for a stoichiometric mixture condition will be supplied to the engine control unit during operation to avoid engine fault codes. This device is intended to be used in conjunction with stock engine control electronics. The control functions will be handled using the low-cost PIC16F819 microcontroller.

This article is presently in its development stages. The main consideration here is the reduction of the amount of fuel being consumed by means of manipulating electronic engine control input and feedback systems. It does not consider the potential environmental impacts due to lean-burns producing more NOx emissions presently. Initial research suggests that lean-burns under low loading conditions will not significantly increase NOx emissions, and can potentially reduce them depending on the degree of ignition advance in low-load situations. Once the concept has been thoroughly analyzed and is possible to implement, the emission considerations will be factored in. Initial research was taken from Chapter 15 of "An Introduction to Combustion: Concepts and Applications" by Stephen R. Turns.

The vast majority of modern vehicles utilize a Zirconium Dioxide (Zirconia) oxygen sensor. This is more commonly known as a Narrow-Band oxygen sensor. It is known as such due to its limited ability to sense the amount of oxygen present in exhaust gases. The only point at which it gives a truly accurate reading of this is at stoichiometric air/fuel ratios. It will indicate a rich or lean condition, but cannot supply an accurate measure of the magnitude of deviation from stoichiometric. A Wide-Band Zirconia oxygen sensor, or Wide-Band oxygen sensor uses electrochemical pumps and a complex control system to provide highly accurate, fast readings of the air/fuel ratio. See figure one for a graphical explanation of the sensors’ output characteristics.

We will now investigate how most electronic engine controls utilize the Narrow-Band oxygen sensor. Figure two shows a plot of the air/fuel ratio versus time for a 1991 BMW 318iS. The vehicle was equipped with the LC-1 Wide-Band kit, allowing direct acquisition of the air/fuel ratio while simulating a Narrow-Band sensor to supply the engine control’s Exhaust Gas Oxygen (EGO) input. The given plot was taken while maintaining 55mph in fifth gear. The green line is the air/fuel ratio, and is plotted on the right axis.

As you can see, the air fuel ratio varies constantly, and seems to have a center point around stoichiometric. Being designed to work with a Narrow-Band sensor, the engine controls will adjust for lean and rich conditions when they deviate beyond a certain threshold, only changing the richening/leaning a few times per second. This is acceptable on a stock vehicle. The car averages a safe air/fuel ratio of 14.1 over time, so it is not necessarily wasting fuel (in this case it is, but the concept’s intent is correct). The intent of this project is to allow the vehicle to run much leaner under low load conditions where detonation and overheating components is not an issue.

The LC-1 allows for its analog outputs to be calibrated in any manner that the user specifies. This means that you can reprogram the Narrow-Band output to tell the engine electronics that they are running at stoichiometric when in fact they are running richer or leaner. The BMW M42 has been safely run at an air/fuel ratio of 17:1 by some enthusiasts. Unfortunately, one cannot simply reprogram the LC-1 for an adjustment this dramatic. The engine electronics will perform as they were intended, and the air/fuel ration will swing richer and leaner than the desired ratio. When approaching a ratio of 18:1 or more, detonation and overheating of components becomes a threat. This swing in the air/fuel ratio must be eliminated. To keep the project practical, it is the intent is to utilize the stock engine management system without reprogramming its software.

The only other sensor signal playing a major role in the fuel injection rate, and able to be manipulated effectively, is from the air flow meter. In order to maintain a constant lean air/fuel ratio while under the proper conditions, the air flow signal must be manipulated according to the deviation from the desired air/fuel ratio. While this is occurring, a constant value of .45VDC will be fed to the engine control electronics in an attempt to minimize its own fueling adjustments in steady state operation. The modulation of the steady air flow signal will allow for more precise air/fuel ratio control, and thus allow a constant lean condition.

Modulation of the air flow signal will require certain conditions to be met prior to making adjustments. The throttle position will be monitored at all times. Modulation allowing an air/fuel ratio of 16:1 will be operating on throttle openings of 35% to 40%. Ratios of 17:1 will be seen at openings of 2% to 35%. Below 2%, the engine will be idling which corresponds to the lowest possible load while still supplying fuel. The ratio here will be targeted at 19:1. The theoretical decreases in fuel consumption will be 8.1%, 13.5% and 22.6% respectively with the car running at 14.7:1 stock. The conditioner will not operate under the following conditions:

• Throttle position is greater than 40%

• Engine coolant temperature is below 50 Celsius

• The sensed air flow rate is less than that seen at idle

The throttle dependent condition is a safeguard against detonation resulting from extreme ignition firing advances and high-load conditions with a lean air/fuel condition. Disabling the adjustments below 50 Celsius ensures that the engine controls are allowed to run their warm-up enrichment protocols. Although this uses additional fuel, it helps get the motor up to its prescribed operating temperature faster, and thus minimizes the additional mechanical wear associated with running “cold.” Finally, there will be a minimal air flow threshold. The Bosch Motronic 1.7 control unit used on the BMW M42 ceases all fuel injection when a negative load condition is sensed. This is indicative of a coasting condition, where the engine speed is above that of idling and the throttle is closed. Applying idle-corrections here could lead to potential fuel waste.

The following is a flow chart outlining the initial routine that will be tested.

It is possible that sending a constant stoichiometric signal to the engine control unit (ECU) will trigger a fault, and being the case there are two options. The EGO input to the ECU could be eliminated entirely. This would result in pre-programmed fueling maps being used that are dependent solely on the air flow and throttle position inputs. The modulation device could be used to monitor the air/fuel ratio from the sensor and adjust the air flow rate to maintain the desired ratios. This would result in a constantly-lit “Check Engine” light, which in turn would make it impossible to know if another, more crucial fault was occurring. Emission tests would be failed as well with an EGO sensor fault condition. The second option would be to have the modulator output a triangular analog signal with a frequency of roughly 0.8Hz, mimicking typical variation in the EGO sensor output seen by the ECU.

Presently, the method under investigation is that of sending a constant stoichiometric signal to the ECU and observing its behavior. Ideally, doing so would lower the amplitude of the fluctuations in the air/fuel ratio. The ECU would likely not hold a steady fueling rate since it is not programmed to deal with a constant stoichiometric condition, but the degree of modulation might be reduced without causing a fault. Since the apparent response time of the ECU and EGO sensor is roughly 600ms, these fluctuations should not pose a problem. The air flow rate, which has been shown to be checked far more often in other experiments, will be updated on a 5ms period. This should allow for sufficient damping of ECU-driven fluctuations in fueling rate due to EGO inputs.

The next focus will be on the algorithm outlined in the flow chart. When an action block appears saying “Get EGO Value,” or “Get Air Flow Value,” it is referring to using the microcontroller’s onboard Analog to Digital Converter (ADC) to capture the analog voltage level the sensor is outputting. Since the EGO and Air Flow inputs to the microcontroller both range from 0V to 5V, some simplifications can be made in how the air flow value is adjusted. Rather than calculating a signed percentage deviation of the actual EGO level from the specified one and changing the air flow value by this percentage, another method will be used.

The captured 8-bit value of the EGO level will be compared to the 8-bit value corresponding to the desired EGO level. The difference will be divided by four (by simply shifting the difference file register right twice). This value will be directly added/subtracted from the captured air flow value and sent to DAC-2. The value should be divided by four to effectively damp the response of the system. Sending the full difference to the Air Flow value could lead to output oscillations in some cases. While dividing the difference by four decreases the overall sensitivity somewhat, it was determined to cause a negligible swing in air/fuel ratio. This deviation was calculated to be approximately 0.8% worst-case. This was based operation in the portion of the air flow sensor’s transfer function corresponding to minimal load conditions where the curve can be approximated as being linear.

The switching points corresponding to required coolant temperatures and throttle positions will be dealt with as follows. The fully-closed throttle potentiometer output voltage and fully-open voltage will be recorded. From there, a formula can be easily derived for the throttle position based upon potentiometer voltage. The transfer function for the Bosch Negative Temperature Coefficient (NTC) coolant temperature sensor was derived experimentally and curves were fit to it. A Motronic 1.7 control unit was opened and the coolant temperature thermistor was found to be part of a voltage divider circuit. The fixed value resistor in the ECU was found to be R1000 Ohm. At a temperature of 50 Celsius, the thermistor would have a resistance of R900 Ohm. The voltage divider junction voltage would be 2.39V in this condition. At junction voltages less than this, the air flow modulator algorithm will be inactive as this would correspond to temperatures below 50 Celsius.

Currently, the initial algorithm is being written in PIC assembly code. It is very simple to code for the desired function in prescribed manner. Regardless of the method used to modulate the air flow value, there will always be some oscillation. This is due to three things. First, it takes time for the ECU to adjust the fueling rate and the intake charges with the new fuel ratio to combust and pass the EGO sensor. This will introduce an inherent delay between the modulator’s EGO feedback input and the corrections it applies. The other cause of the oscillations will be from the ECU making its own changes to fueling rates based upon its perceived EGO input and the modulator having to correct for them. Finally, the division by four in the difference of EGO values will introduce a minimum deviation required to change the air flow rate. Even while running at leaner air/fuel ratios, a deviation of 0.8% should be more than acceptable for safe operation.