• Before microcontrollers
• Why electronic control?
• Into the `90s. Power and memory
• Up-to-date data collection

Electronic engine control first came to light in 1978 in what was called a "closed loop" carburetor. It was a response to the oil crisis and promised marginally better fuel economy. At the same time, national concern over airborne pollutants propelled investigation into the origins of automotive exhaust components, how pollutants were formed, and how they could be abated.

Why electronic control?
Gasoline-powered Otto cycle engines are simple. Combine measured amounts of air
and fuel in a confined space and ignite them. The pressure of expanding gasses forces a piston to move. This motion is translated into rotational movement. Like all simple things, there's more than meets the eye.

In the 70's, engines relied on mechanically generated signals to ignite the fuel/air mixture. Electrical energy from a battery was stepped up from 12 volts to many thousand volts by a coil. A mechanical "distributor" selected the appropriate spark plug and sent a signal along a wire. That selection "window" was wide, the ignition signal (spark) could be initiated any time within many degrees of rotation as a "rotor" contacted mechanical switch points inside the distributor.

Science knew that the outcome of combustion - both power and pollutants - was greatly affected by how precisely the fuel/air mixture approached theoretical perfection and when the ignition event took place. To get clean air, fuel had to be precisely mixed in a "stoichiometric" 14.7:1 ratio. The mixture had to be ignited at a precise instant that varied with load, speed, and other factors. Mechanical devices could not achieve the required precision and automakers soon approached Intel, a manufacturer of microcomputers and microcontrollers. (A microcomputer chip with a single preprogrammed task is referred to as a microcontroller.)

Early discussions centered on sensor data~ what information an engine microcontroller would require. Critical needs included the rotational position of both crankshaft and camshaft, and air flow. Throttle position and rate of throttle position change (the transmission wants to know when you need to accelerate quickly) were needed, too.

Rotation sensors i.e, crankshaft, camshaft, and ABS sensors at the wheels utilize a wide variety of technologies. Optical sensors may use infrared Light Emitting Diodes to peer through a slotted wheel. Other sensor designs interpret the rise and fall of magnetic energy as a metallic part approaches and departs. Interestingly, many very precise sensors receive only 4 signals per 360~ of rotation - exact position is calculated mathematically, predicting not only position, but whether the engine is accelerating or decelerating. This accuracy allows the fuel/air mixture to be ignited at a precisely selected moment appropriate for engine power and emissions control.

Other data signals are critical to powertrain control. The microcontroller has to know temperatures in the engine's water cooling system along with oil temperature and transmission temperature. Fuel injection requires knowing atmospheric density and how quickly air is being drawn into the intake manifold. Air temperature affects air density. Hydraulic pressure information is sent to the microcontroller by automatic transmissions, as are battery voltage, road speed, and oil pressure.

Every signal adds calculation complexity as it increases the precision of control.

Within the last five years engine microcontrollers have also been required to determine shift points as the industry installed electronically controlled (vs. hydraulic/mechanical) automatic transmissions. The precise management provided by digital control means every automatic transmission will soon be under numeric control.

As engine control advances, so does data complexity. Oxygen sensors enabled controllers to accurately mix air and fuel based on combustion results. Now, a second oxygen sensor placed down stream from the catalytic converter infers the state of the catalytic converter (e.g. converters work best when hot; they suffer contaminant damage, even aging.)

At one time it was thought that some kind of sensor would be added to each cylinder to monitor every combustion event. This would have added enormous cost and complexity to engines. Instead, by increasing the power of software - placing added burdens on the engine computer/controller - events can be inferred or predicted. This increase in computational power places great demands on microcontroller performance.

Computer chip families or "architectures" may be understood by an analogy to a subdivision. Every house may look different, yet it is built from common components. And the core structure - placement of furnace, water and sewer - may be identical in every building. Computer chips are built the same way with a core framework embellished by appropriate structural add-ons. Consideration is given to growth (both in raw processing power and memory) and additional input (more sensor data links.) So a microcontroller can add more memory or calculation power up to its architecture's limit, just as a growing family can
add rooms or central air conditioning.

Into the `90s. Power and memory.
If controlling what happens in an engine and automatic transmission were not sufficient challenges, a greater one has emerged from legislation. Specifically, On Board Diagnostics II (second generation) laws. These require monitoring automotive systems that affect emissions. Not only does your cars' engine control unit have to "watch" what goes on and record troubles for service technicians as they happen, OBD II rules require the prediction of deterioration of the following: catalytic converter, fuel delivery and evaporative emissions systems, crankshaft and camshaft position sensors, oxygen sensors, manifold air temperature sensor, ignition system and others.

Let's look at the most difficult problem, ignition misfire.

A four-cylinder engine cruising on the freeway at 65 mph revolves at approximately 3,200 revolutions per minute or 6400 spark events per minute. Because of moisture, voltage drop, or a variety of other factors, some misfire is inevitable. The engine microcontroller monitors combustion and should, for instance, 6 of 20 sparks in a row fail in any one cylinder, the engine controller must notify the driver "SERVICE ENGINE SOON." But what if you splashed through a deep puddle and a damp wire caused a temporary misfire?

Sophisticated algorithms examine error data and query whether the event is recurring. If it was an isolated incident, as in our damp spark plug wire example above, the check engine light is extinguished.

This benefits the consumer. When a car arrives for service, modern computer-powered diagnostic equipment can directly interrogate the on-board microcontroller and elicit a specific response. This could enable the mechanic to go directly to, for instance, spark plug number 4 and begin repair.

This sophisticated diagnostic power requires roughly the same computational capability as the engine controller itself.

Because of increasing complexity, responsibilities, and the sheer number of calculations, engine microcontrollers need increased power to avoid being overwhelmed. Just as desktop computers evolved from early 8088 PC Jr. machines into today's Pentium® processor powered models, so have engine microcontrollers changed.

The latest generation of microcontrollers, like Intel's 83C196EA, exemplify this vastly improved processing power and communications ability.

Up-to-date data Collection
Microcontrollers utilize digital signals. Sensor data (voltage, temperature, linear and rotational velocities, etc.) must be converted from continuous or analog information into the digital form required by microcontrollers. This Analog-to-Digital conversion is typically performed within the microcontroller.

Another critical need is the ability to capture and compare events as they happen. And high-speed input and output channels transport data to the microcontroller for action. The microprocessor core itself needs raw power and speed as it executes the complex instruction codes called algorithms.

Once sensors, algorithms and the microprocessor core have done their work, control signals are sent out to tell the engine what to do. High-speed channels specify when and how long to fire spark plugs, when and how long to send fuel through the injectors, or when to shift to a different gear. And there is a need to query sensors for updated data in return.

This information is typically stored in flash memory at "key off" time, along with other updated parameter values. In addition to this, flash is the nonvolatile memory of choice for Engine Control Units for the operating code because it can be updated in the field if needed. As well as having the ability to download a final program into the unit at the end of the module assembly line after using optimized diagnostic code for unit test.

Intel's new 83C196EA microcontroller harmonizes with modern engine control needs. It has 16 Analog-to-Digital converter channels built in. Its new higher speed core zips along twice as fast - 32 MHz vs. 16 MHz - as its predecessor. This kind of speed is optimum for power train control, particularly its high-speed I/O for controlling electronic transmissions. The family features 17 event capture-and-compare channels and eight capture-only channels for a total of 25 high-speed Input/Output channels.

Intel's history in engine control electronics is lengthy, from the first 1983 Ford EEC-IV based on the 8061 microcontroller to today's modern 8065 chip set, also known as the EEC-V. Today, Ford is still using the 8065 in many of their new vehicles produced through the start of the next decade. The 16-bit architectures, such as the 8065 and MCS(R) 96 controllers, have been widely accepted in Europe and the U.S in ABS, engine control, and networking applications. From the introduction of the 8061, Intel has continued to provide more innovative and highly integrated microcontrollers in order to help their customers with evolving powertrain applications. Specifically, the 83C196EA doubles the performance of our existing microcontrollers while adding more functionality. The 83C196EA will provide designers with lower development costs and time-to-market.