ME 102 Fall 2008
Electronic Fuel Injection System




I. Introduction

A. Motivation

Cal’s Super Mileage Vehicle (SMV) team is an extra-curricular project run entirely by undergraduates. Each year the team designs and builds a lightweight car with the goal of ultimate fuel efficiency, with past cars achieving up to 1068 mpg. The car is hand-made of a carbon fiber and epoxy laminate shell with a carbon fiber and Nomex honeycomb chassis.

The vehicle is powered by a single cylinder 25cc four-stroke gasoline engine manufactured by Honda. The engine previously used a carburetor to feed air and fuel into the engine allowing only limited control over the fuel to air ratio which can lead to inefficient operating conditions. This project aims to develop an electronic fuel injection (EFI) system that can provide a greater efficiency to the vehicle through more precise control of fuel usage.

Eventually, the fuel injection system should be able to allow the team to specify more exact fuel to air ratios, time the delivery of fuel to the engine, and adapt easily to changing needs using software that utilizes lookup tables. In this project, we focus on designing and assembling a basic fuel system that can later be modified to include a variety of capabilities that will improve the overall efficiency of the vehicle.

B. Overview

The system has been broken down into two categories: hardware and software. The hardware is comprised of the throttle body and the fuel line assemblies, while the software calculates the ideal fuel amount given RPM and throttle position.

II. Hardware

A. Throttle Body Assembly

The throttle body assembly consists of the throttle body, throttle, servo and air filter. The servo attaches to the throttle and is used to control throttle position, either via software or a hard-wired manual analog controller. Currently, we are only using manual control of the throttle. The body also features a barbed fitting that can be attached to an air pressure sensor with vacuum tubing; however, we are currently not using a manifold air pressure (MAP) sensor due to technical problems. The MAP would be used to monitor air intake, which can also be estimated directly from throttle position, which is known precisely because of the servo.

B. Fuel Line Assembly

The fuel line assembly consists of the fuel reservoir, fuel pump, two filters, a pressure regulator and terminates at the injector. The pressure regulator also has a return line to the reservoir. The fuel pump and injector are both powered by 12-volt batteries, such as those used in automobiles. In competition fuel pumps are deemed illegal and the team uses a pre-pressurized system, as the run time is short and uses very little fuel. This also means that there will be no fuel reservoir in the final system, only the fuel tank. But for simplicity we will be using the fuel pump and pressure regulator in bench top use.

C. Engine Assembly

The system utilizes an electric starter motor that replaces the pull-start ignition coil that originally came with the engine. The starter motor is attached to the engine via a chain and a one-way bearing. This means that once the engine is running it does not move the chain or the electric motor. The other side of the engine is attached to a bike wheel to simulate load (albeit light) and demonstrate the intended use of the system.

III. Software

Originally, our software was interrupt driven. In other words, we configured the DSP board to automatically respond to a low to high transition on the hall effect sensor (a sensor that returns a high voltage when a magnet is nearby). We placed a magnet in the camshaft, so that these “hall ticks” would let us know every time the engine went through a full cycle. However, we were unable to get even the most basic interrupt functionality to work properly on our board, so we abandoned this idea in favor of a more simplistic but slightly less robust polling method.

After setting up the proper registers and signal inputs and output, our code enters an infinite loop of air/fuel ratio calculation and injection event management. It polls the hall effect sensor until the magnet on the camshaft rotates into position. Our DSP waits for a specified delay and then sends a signal to the injector to begin injecting. The DSP would then wait for the length of injection pulse (typically 2.5-4.5 milliseconds) and then stop injecting. After this, the injection parameters would be re-calculated for the next injection event.

To calculate the injection parameters, we created a function that would map throttle position (essentially the amount of air entering the engine) and RPM to an injection pulse time. A greater injection pulse time causes more fuel to enter the engine, which creates a richer fuel/air mixture. We created an AFR (air/fuel ratio) table from tuning measurements and interpolated intermediate values describing the length of the injection pulse.

IV. Functionality

When the engine is running in standard operation, the inputs are RPM via hall tick and throttle position; however, we also can switch to a tuning mode where we utilize the oxygen sensor mounted at the exhaust to determine if the engine is running rich or lean. By keeping the throttle position constant and varying the amount of fuel injected we can look at the RPM and oxygen sensor to obtain efficient conditions to populate the fuel map. We chose not to use the oxygen sensor as a closed loop control while running because the update time for the sensor is much too long—by the time it produces a value the engine may be running with a completely different set of conditions.

V. Conclusions

In the future we hope to expand and perfect the fuel map to further increase the economy of the engine. With access to a dynamometer, we would be able to hold the engine RPM constant, allowing us to simulate a span of real world operating conditions and expand the fuel map. We also would like to run comparison tests between the original carburetor and our EFI system to confirm the effectiveness of our design. We can use these tests to optimize the software to obtain maximum fuel efficiency.

Group Members:
Steven Delong
Matthew Pies
Elizabeth Steele
Michael Watson