HOW EFI WORKS
Edited from the ACCEL EMIC Manual
By Ray T. Bohacz
Why Fuel Injection?
Even though in racing, fuel injection is usually associated with the ultimate in performance, the reason behind its proliferation onto mainstream automobiles is for quite the opposite purpose. With increasingly stricter emission standards and the EPA’s implementation of a required minimum fuel economy, the days of the carburetor were ended. Without electronic engine controls, it simply wasn’t possible to accurately command all the parameters of fueling and spark that were needed to remain compliant, without having finite control of the administration of the fuel into the engine. The advent of cold start emission tests made apparent the needed decrease in catalytic converter light-off time and made critical the cold start and intermediate warm-up fueling.
Even though fuel injection made its debut in America in the mid 1950’s, with General Motors using it on certain Chevrolet and Pontiac models, today’s injection bears no resemblance to its ordinary beginnings. The Europeans, working in conjunction with Bosch, were the pioneers of today’s injection systems. The high cost of fuel in Europe and the taxation of large power plants forced the development of fuel stingy small displacement engines. With no need to respond to the same conditions, American manufacturers were late to the injection game and first introduced electronic fuel injection as opposed to their earlier mechanical systems in the mid 1970’s with Cadillac and Lincoln leading the way. These were very crude non-feedback systems that utilized very slow analog microprocessors. The next foray by the industry was to feedback carburetors with a transition to throttle body injection and then on to today’s multipoint injection systems. What does the future hold for fuel management? The next logical step is the development of direct port injection where the fuel is injected under extremely high pressure directly into the combustion chamber allowing for mixtures in the 40:1 range. With computer technology making great strides in the 1980’s it was now possible to interface electronics with fuel management and today’s fuel injection systems were born. Not only did that prove to be a major factor in gaining EPA compliance but it also spawned a Muscle car renaissance with the arrival of the 5.0 Mustang, Buick GN, TPI Chevrolets and Pontiacs. It was only a matter of time before these late model injection systems found their way under the hood of an old hot rod.
IDENTIFYING THE COMPONENTS
Since an engine has no idea of how it is being fueled, most of the dynamics of the fuel requirements stay constant whether the engine is injected or carburetored and still need to be applied. The only area that does change is that heat is no longer needed in the intake manifold to help vaporize cold fuel and the function of load needs to be inputted independently to the computer as part of a decision making process in lieu of a dynamic action. The common Delco two-wire coolant sensor is mounted in the intake manifold water jacket. It is a thermistor, which is the opposite of a resistor. Where a resistor’s impedance increases with exposure to heat, a thermistors resistance decreases when heated. A 5 volt analog signal is sent out to the coolant sensor and returns back to the computer. Due to the design nature of the circuit, if it fails open it will be interrupted as a extremely cold coolant temperature. The coolant sensor value is extremely important and is used to invoke warm-up fueling, Idle Air Control starting position, fast idle speed, closed loop parameter and idle spark functions.
Air charge temperature - This sensor is identical to the coolant sensor except that it is mounted in the intake plenum in most cases, and is used to calculate incoming air temperature as a correction factor. Air temperature correction is used as a trim function to the base fuel map to compensate for the decrease in Volumetric Efficiency due to the heating of the charge air.
Manifold absolute pressure (MAP) - This Delco sourced piece is available in 1, 2 and 3 Bar configurations and is used as an input to determine load on the engine. This 3-wire analog output sensor reads pressure changes in the intake manifold. Since the input voltage of the signal stays constant at 5 volts, resolution is cut down proportionally when 2 or 3 bar sensors are used.
The computer originates the 5-volt reference signal. The sensor outputs a return signal (with an internal ground supplied by the computer). MAP sensor input is used by the computer to calculate load in both the fuel and timing matrix and is also used as an input to invoke load based fuel enrichment. If the computer is configured to read in Alpha-N then the MAP sensor is retained but is not connected to a vacuum source and is used for barometric compensation.
Mass Airflow Sensor - MAF always posed three major problems. The first being the room needed to package the actual sensor and the second being the flow limitations of the sensor. A stock GM TPI MAF is only capable of flowing approximately 550-600 CFM @ 28 inches of H2O.
A fully ported MAF will still only approach 750 CFM. The third obstacle with MAF is the inability to interface it with unique manifold designs, for example an individual runner intake. Intially, but not a major concern today, the Bosch supplied MAF that GM used on its early TPI systems had a 9 out of 10 failure rate. By employing a quality Delco produced analog MAP sensor, all of these issues were addressed at a substantially lower cost than with a MAF.
Throttle position sensor (TPS) - Hooked mechanically to the throttle body this 3 wire Delco sourced sensor inputs throttle angle position data to the computer. The TPS receives an analog 5 volt signal while a circuit returns the sensor’s output to the computer. Throttle position input is used to promote Idle Air Control tracking and is also used to trigger acceleration enrichment, which takes the place of an accelerator pump. Not only is the position of the throttle monitored, but also the rate of change to determine the amount of enrichment needed. Its input is also used as a threshold for transition out of closed loop and to engage and disengage the torque converter clutch and to invoke the idle spark function. Additionally, at 100% TPS, the coolant fan circuit that is controlled by the computer shuts off if engaged. If engaged during Alpha-N configuration, TPS input replaces MAP load as a function of fueling. Slotted TPS’s are used so that base values can be adjusted to coincide with closed throttle.
Oxygen sensor (O2) - Only used for closed loop stoicohmetric fuel control. This sensor is a 3 wire heated Bosch unit. By nature, a oxygen sensor will not operate accurately until it reaches a temperature of 600°F. The heating element brings the sensor on-line sooner and keeps its output voltage more stable. The heater also allows for more freedom in the placement of the sensor, which is critical in a performance application that utilizes headers. The computer receives the sensors output voltage and then correlates that to an air fuel ratio. The sensor has an accurate range from 0-1 volt with approximately .5 being 14.7:1. Output voltage degrades below .5 with air fuel ratios leaner than stoich and increase to over .5 with mixtures richer than stoich. The 12 volts that are supplied for the heater originate from a Relay terminal shared with the fuel pump.
Knock Sensor - The sensor its self is a piezoelectric accelerometer specific to either engine, and even though physically looks the same, is tuned to detect a different knock frequency in a Small Block vs. a Big Block. This sensor during detonation produces a low voltage sine wave output that is then converted by the Electronic Spark Control module to a digital signal. The output signal of the ESC module that correlates to detonation is then sent to theh computer and is processed internally to retard the ignition timing.
The following is a list of circuits that are controlled by the computer.
Injector firings - With a batch fire computer all of the injectors are fired simultaneously with ½ their programmed pulse width for each crank rotation. Therefore, it is given the name of Simultaneous Double Fire (SDF). Incorporated into the computer are four drivers with each one having the duty of pulsing 2 injectors. Later model sequential fire computers pulse the respective individual injector when called for in the ignition cycle rather than activating all on a bank simultaneously.
Fan control - The computer has the ability to control the fan circuit by the grounding of an independent relay. The fan enable temperature operates with a 10 degree F. hystersis. For example, if the fan is programmed to turn on at 180°F it will not turn back off until the coolant reaches 170°F. Also, at 99% TPS the coolant fan circuit is automatically shut off.
Fuel pump - The computer powers a micro relay to power-up the fuel pump and supply 12 volts to the heated O2 sensor. 12 volts is supplied to close the fuel pump relay and to turn the pump on. As soon as the ignition is turned on, the computer commands a 2 second prime signal to charge the fuel rail. If an RPM signal is not seen the relay is shut off. Once an RPM signal is active again, the fuel pump is turned back on.
Timing control - The computer controls the timing to the ignition module by varying the frequency of the square wave output.
Idle air control - The computer has the ability to have closed loop idle speed control through the use of a four- wire Delco stepper motor air bypass valve. The computer alternates the pulsing of an internal ground circuit to command the idle air control position. This function works in counts and is not actually aware of where the IAC pintle is. It just knows how many commands it has issued from a zero starting point.
Torque converter clutch - An optional TCC clutch engagement signal is available for TH 700R4, 2004R, and 4L60E transmissions from some computers for cars with automatic transmissions.
Prior to concerning ourselves with programming decisions, lets follow the procedure the computer will take to gather data and then lets start to think like the computer.
As in any electronic fuel injection system all the computer does is gather values from its sensors and uses them to identify different points in what is called look up tables. A good analogy of a look up table is the chart that is used in a road atlas to determine mileage between two points. For example, if we needed to find the distance between NYC and LA using a mileage chart we can find the two points and see where they intersect and derive the mileage. Now according to the publisher of these maps, those values are the foremost direct route from city hall to city hall. So, to carry this further if I was traveling from Nassau County Long Island to Malibu, California the chart would only be a reference point and not truly accurate do to the different points that I am beginning and ending my trip. Well, that is exactly what is happening inside the computer. Even though we have base values inputted, we also have numerous trim factors that are used to adjust to conditions to zero in on the proper fuel mixture.
Since the computer is an electronic device it can only understand voltages and has no way to interrupt mechanical conditions. For that reason, there needs to be a sensor that has the ability to convert a mechanical conditional to an electrical signal for interruption by the computer. For example, the computer does not understand Wide Open Throttle but it does know what 5 volts output on the Throttle Position Sensor signifies. Since the computer is inanimate it has no way of determining if a sensor’s input is accurate and takes that value as the truth. As with any computer, garbage-in is garbage-out and proper decisions can only be made with accurate data inputs.
In essence, what the computer does is gather data from all of its sensors and then use that data to find a place in the fuel and spark look-up tables to determine an injector pulse width or spark advance command. The MAP sensor input along with the RPM signal are the only data that is needed to find a place in the fuel and timing look up tables. Other sensor inputs are used as trim tables to the core fuel program. For example, if the look up table identified a load and RPM cell that placed the base injector pulse width at 10 MS, and the coolant look up table identified at that given coolant value we should add 10% enrichment we would end up with a a gross injector pulse width of 11MS (Base pulse plus or minus the trim value = gross pulse width). Once we understand this, we can very easily break the computer program down into two simple categories - core and trim tables.
CORE TABLES TRIM TABLES
Since the computer uses the trim tables in conjunction with the core tables, it is essential to have the values in the core tables as close to being correct as possible, since the trim tables have limited control.
FUEL FLOW DYNAMICS
All late model injection systems utilize an electric fuel pump to maintain the constant supply of fuel that is needed. A common misunderstanding when dealing with electronic fuel injection is the concept of pressure vs. flow. To maintain a given pressure it is essential to maintain a certain volume of fuel flow. If fuel flow ramps down under load condition, fuel pressure will drop. To establish this fact, think of a simple garden hose. In a residential water system the amount of water volume to the hose spigot remains constant (the fuel pump). When hooking up a garden hose with a nozzle you are creating a orifice that causes a directing of the water (the injector). If a second hose is connected to the same source what happens to the flow from the original nozzle? As we all know, the amount of pressure and flow at the nozzle will decrease. In essence the same pressure flow relationship happens with a fuel injector. As load is applied to the engine, an increase of pulse width is required to supply the proper amount of fuel and if fuel flow is not sufficient in the rail, pressure at the injector will drop. For this reason, fuel pressure should always be checked at idle and then under load.
The function of the fuel pressure regulator is to maintain a preset pressure in the fuel rail. The way all electronic fuel injection systems function is that they pump more fuel than they consume, constantly returning fuel back to the supply tank.
The fuel pressure regulator is connected in series with the fuel supply and controls the amount of fuel returned to the tank. The less fuel returned, the higher the rail pressure. Referenced to an engine vacuum source for enleanment under light load and coast down, (lower pressure), the regulator consists of a diaphragm and a calibrate spring. An adjustable regulator has the ability to while maintaining Original Equipment levels of reliability. Change the diaphragm position and thus control the amount of return fuel to achieve a given pressure. If fuel flow volume is insufficient, then regardless of the diaphragm’s position in the regulator, fuel pressure cannot be maintained. Remember, to maintain the pressure set point, there needs to more fuel volume supplied than fuel consumed.
Another area that comes into play in regard to pressure vs. flow is the voltage supply to the fuel pump. The pump itself is a rotary vane design that needs a constant supply of voltage to maintain the proper fuel volume. Applying Ohms’ law, if current demand increases, then voltage will drop if the voltage supply is constant. The following chart shows a direct correlation of fuel pump output in regard to supply voltage.
Not only does fuel pump voltage affect pressure, but the size of the fuel supply lines and the area of the fuel filter become paramount. Fuel injected engines always have large fuel filters in comparison to carbureted engines due to the volume of fuel pumped. Remember there is always more fuel pumped than used, and to allow sufficient area as not to disturb flow characteristics. Keep in mind that a fuel injection system has no real reservoir to store fuel like a carburetor does. The only storage it has is whatever is in the fuel rail.
The sequence that the injectors are fired also has a effect on the fuel rail dynamics. Simultaneous double fire is the least complicated system electronically and in theory has fewer components to possibly fail but is the hardest to keep the fuel rail pressure constant with and the rail charged. When eight injectors all open simultaneously, there is a great depletion of fuel from the rail and this causes the pressure regulator to close off the return quickly to maintain pressure. This surging of the rail may actually cause a phenomenon called fuel rail knock, specifically on an engine with large injectors. This knock is produced by the rail being shocked by the rapid discharging of the fuel supply and the rush of replacement fuel. Sequential injector firings are the most fuel rail friendly due to the minimal discharging of the fuel rail by the firing of only one injector at a time.
A common feature in these systems is the incorporation of a fuel pressure check valve that will maintain fuel pressure back to the regulator. This is used to aid in starting the engine by keeping the fuel system charged and to help combat percolation of the fuel in the rail and lines during heat soak conditions.
The Fuel Injector
Other than the computer, the main component of an injection system is its namesake, the fuel injector. It is customary to identify an injector by a number of different characteristics. Its flow capacity, fuel feed point, attachment to the fuel rail, electrical resistance and tip design. In GM OE applications, another criteria is the placement of the injector. General Motors uses completely different designed injectors in their port fuel injection systems vs. throttle body vs. central port injection. Before extrapolating on injector differences we need to understand the basics of an electronic fuel injector. At ACCEL we currently distribute injectors produced by Bosch and Siemens. They are generally classified by their flow and resistance values. All of our injectors are top feed with o-ring attachment to the fuel rails. All electronic injectors consist of the same core components and are simply a solenoid that is attached to a fixture that opens and closes a fuel flow orifice. When the computer completes the ground circuit, the solenoid is energized, lifting the fuel flow closing device and exposing an orifice to pass fuel. When the ground circuit is removed, the solenoid closes and fuel flow stops. When I spoke earlier about the core fuel table and the programmable amount of injector pulse width, we were describing the amount of time the circuit was grounded and the signal to open the injector was applied. This seems very straightforward, but there are other crucial factors that need to also be applied. The length of time the injector ground circuit is applied is measured in milliseconds or thousands of a second. That is referred to as injector pulse width. The length of time it takes for the solenoid to lift and completely uncover the fuel flow orifice is called the rise time. Even though to the human eye the opening of the injector is quicker than the eye and mind can capture, once broken down to the finite measurement of 1/1000 of a second the slowness of the injector opening becomes apparent. By applying Ohm’s law, we can see that with the voltage remaining constant, if we increase current flow we are able to charge the solenoid quicker and make it respond faster. By changing the resistance of the injector windings we are able to pass additional current and decrease injector rise time. Most OE applications use high impedance (12-16 ohm) injectors due to their lower cost and the ability to use saturated drivers in the COMPUTER. Lower impedance injectors (2-4 ohms) respond quicker (shorter rise time) but necessitate the use of peak and hold drivers which are not only more complicated but more costly to manufacture. By design, a saturated driver will keep current draw constant during its whole duty cycle. Conversely a peak and hold driver will initially surge the current up and then step it down to a lower value and maintain that value throughout the event. If a peak and hold injector or driver is rated at 4/1 amps that translates as 4 amps to open the injector and 1 amp to keep it opened. Historically, tests have proven that a low impedance injector will have a rise time of just below 1.5 MS, while a high impedance unit can approach 2 MS.
Fuel delivery at or below the minimal injector rise time skews the atomization and leads to high emissions and poor idle quality. A standard 015013 computer incorporates 4 4/1 drivers and has the ability to fire 8 low impedance injectors for short periods of time. The second most important identifying aspect of an injector is its rated flow capacity. Think of injector flow capacity as you would a carburetor jet orifice dimension. ACCEL uses one of the three industry standards of lbs/hr to rate their injectors. The other two standards are gm/sec and cc/second. The rating of fuel in lbs/hr is always at a test pressure of 43.5 PSI, which equates to the metric equivalent of 3 bar. Given this, an ACCEL injector that is rated at 24 lbs/hr flows 24 lbs of fuel at 43.5 PSI pressure. If the pressure is increased above the test value the injector flow capability increases. Conversely if the test pressure degrades, flow suffers accordingly. Rule of thumb is that for our Bosch supplied injectors, anything over 36 lbs/hr are low impedance.
The previous mentioned flow values are not only at the given 3 bar test pressure but also at 100% duty cycle. That brings us to another area that is related to injectors. This subject is not a function of injector design but is affected by it. Because an injector has to supply fuel for a multitude of engine operating conditions it has to be very adaptive. It must respond fast enough (rise time) to supply proper atomization at short pulse widths during idle or light load, which is hard for it to accomplish, while also providing enough fuel flow at high RPM with very brief valve opening times. The easiest way to understand injector duty cycle is to think of it as the length of ignition event time that the COMPUTER is grounding the injector for. Since the COMPUTER is going to fire the injectors in time with the ignition events, it only has a given amount of time to get the fuel into the cylinder before the event for the next cylinder comes up. The length of time between ignition events is much shorter at 6000 RPM than it is at 3000 RPM. If the injector is staying open longer than the ignition event time, then it is considered to be going static and not able to control fuel flow. That is why injectors are offered in different resistance values and flow rates. If you cannot pass enough fuel during one event to satisfy the engine, you need to either raise the fuel pressure or the flow capacity of the injector to accomplish the task at hand. Decreased injector resistance allows for quicker response and in turn the ability to increase the amount of usable time between ignition events. Original equipment applications do not usually like to raise fuel pressure above 3 bar due to the possibility of fuel leakage past the closed pintle on high mileage injectors causing excessive coast down emissions. A good rule of thumb concerning injector duty cycle on an engine that will see short bursts of high RPM, is that you can run near static with usually no problem as long as the air fuel ratio is maintained. When sizing an injector for an endurance engine always allot for a maximum of of 80% duty cycle to allow for the injector and drivers to cool.
To properly choose the needed injector size for an engine, not only do you need to know all of the above, but also the brake specific fuel consumption data of the engine (BSFC) and accurate horsepower numbers. This will tell you how much fuel in pounds it takes to produce one horsepower. A textbook rule states that a normally aspirated engine will use ½ pound of fuel (.5 BSFC) for each horsepower produced, while a forced induction engine will consume slightly more fuel at .55 pounds. Then again, the equations at the end of this chapter will establish this fact further and a more indepth discussion of factors that affect BSFC will be covered in the tuning section of this manual. Using the .5 value, a 30 lb/hr ACCEL injector at 3 bar fuel pressure and 100% duty cycle can support 60 HP per cylinder, and a total of 480 HP on a 8 cylinder engine.
So far through this section of the manual I have been very careful to be vague in regard to making mention to any tip designs of injectors, often referring to it as the fuel flow metering device. The reason being, not all injectors incorporate the same fuel metering orifice that is opened up while the windings are energized and allowing fuel to pass. ACCEL injectors made by Bosch utilize a pintle to stop the fuel flow out of the injector. Rochester produced injectors utilize a ball valve with a director plate and Lucas uses a disk shape. Each injector design has its own weaknesses and strengths but the pintle design that ACCEL uses has proven to be the most accurate and repeatable out of the three, albeit the most costly to manufacture.
The next area of concern is injector stroke to stroke repeatability and its effect on engine performance. Most people take the approach of outof-sight out-of-mind and feel that as long as the injector is pulsing that all is well. When dealing with a port fuel injected engine, think of it as multiple carburation. Anyone who has worked with mutiple carbs, specifically of the Webber type, on IR runner manifolds has a full understanding of fuel distribution and the problems it can cause. Since each injector is responsible for all of the fuel requirements for that particular cylinder, any flow variation from injector-to-injector will cause differences in mixture for each bore. When injectors become dirty, they lose their ability to properly atomize the fuel, causing high emissions and idle instability. With today’s fuels, injectors tend to stay cleaner than they did 10 years ago, but tip fouling is still a concern. Even if the tip of the injector does remain clean, the constant charging and discharging of the injector solenoid will eventually lead to a lazy injector electrically with an increased rise time. The only true way to quantify injector performance is to remove the injectors and fire them on a test bench like our Asnu injector tester and cleaner. This will give you the ability to view atomization patterns and to measure flow accurately.
It is not uncommon for out-of-the-box injectors to have flow variations of up to 15%, and this is why ACCEL sells only matched sets of injectors.
To produce power in an internal combustion engine, we need to burn fuel. The byproduct of that combustion event is generally referred to as emissions. Whenever modifying a street legal vehicle, most states require it to still remain emission compliant. To the best of my knowledge, all states that do emission checks do so using infrared optical bench analyzers with the only difference being the actual test standard and the test cycle itself. With the growing implementation of the IM240 based test cycle, the relevance of understanding emission output becomes essential. Since normal emission test standards will measure either PPM (parts per million) or percentages in lieu of grams/mile, all references will be made to the more common state level standards. The five main emissions that we are concerned with are the following:
Carbon monoxide (CO) - During a tail pipe emission test, CO is read in a percentage. Simply put, CO is the byproduct of a rich air fuel ratio. When too much fuel is present in the combustion chamber, it cannot be completely burned and the fuel that is only partially burned is registered as CO. Any time a vehicle fails an emission test for CO, the cause is a rich mixture. Dirty fuel injectors, clogged air filter, high fuel pressure and a degraded O2 sensor are the biggest culprits.
Hydrocarbons (HC) - If CO is partially burned fuel, then HC, which is read in PPM, is raw unburned fuel. Anything that will affect the combustion processes will have a effect on HC production. Ignition problems, incorrect spark advance curves, mechanical engine problems, cam overlap, lean mixtures and large crevice volumes will all raise HC.
Carbon dioxide (CO2) - Read in percent, CO2 represents the bonding of the carbon and oxygen molecules that takes place during efficient combustion. It is the emission that you want the most of. The higher the CO2 of your engine, the better the burn.
Oxygen (O2) - Measured in a percentage, this reading works inconjunction with CO2. This value establishes the amount of O2 that does not have a carbon molecule to bond with. It is inversely proportioned to the CO2.
Oxides of nitrogen (NOX) - Formed from NO (nitrogen monoxide) when in the presence of oxygen. It is a byproduct of very high combustion temperatures.
The Catalytic Converter - The sole purpose of the catalytic converter is to clean up what does not get burned in the combustion chamber. For a catalyst to function properly, it must accomplish what is referred to as "lightoff " to start the chemical conversion process. To light off, most converters must see temperatures of 1000°F or more to operate efficiently. It is also important to note that a converter will only operate within a range of 2% of stoich; rich or lean. A good, efficient converter will cover a multitude of tuning errors if light off can be accomplished. The need to maintain catalytic efficiency is the reasoning behind the set point of 14.7:1 as the target air fuel ratio calibration of the oxygen sensor.Back to table of contents