UPDATE -- Computers in Racing
Chapter in Race Car Engineering and Mechanics

(See index at http://www.pvanvalkenburgh.com/Race Car Tec/engineer.html

Computerized Engine Control

Computer-control of both engine spark timing and induction will eventually become almost universal -- even in racing, probably even in NASCAR, as they find it more meaningful to the manufacturers and as easy to regulate as carburetors. However, it will only be covered briefly here. As mentioned in Chapter 8, racing engines are most often bought as a package from specialist builders with a fortune in dyno equipment. So engine electronics alone deserves its own book. Telling you how to reprogram your engine's computer is almost like telling you how to design your own tires -- at least it used to be.

Today, if you want more performance out of a production engine, there are aftermarket chips, and plug-in devices to rewrite the spark and fuel maps. To satisfy government emission and mileage requirements, and to control all the other electronic devices in a production vehicle, the microprocessor has to handle hundreds of functions, from cold start to the air conditioning compressor. And these computers have to be produced inexpensively by the millions for mass production automobiles. And they have to be bulletproof. Therefore, the development cost of the hardware is minimized in racing applications, if it can simply be adapted from existing components.

Even when aftermarket performance devices aren't available, it's still possible to trick a system, since the computer requires signals from a number of engine sensors which are accessible. For example, the mass airflow sensor may produce a DC signal that is proportional to the airflow rate. Say the signal is 5.0 volts at maximum air flow. If a leaner mixture were needed, another potentiometer might be added, and adjusted so that at maximum flow the reading was only 4.8 volts. In other words, the computer would think there was less air, and therefore provide less fuel at that point.

Another modification possibility is constant ignition advance by moving the distributor or crank sensor, or changing the mixture only at wide open throttle by moving the throttle potentiometer. Although tricking sensor signals can be used to make broad across-the-board mixture or timing changes, this is a crude last-ditch measure. Dyno tests would more likely show a need for complex changes in the advance or mixture curve, which would justify using a user-programmable aftermarket engine control unit.

For pure racing engines, the rapid advancement of inexpensive computer power, and the availability of more computer-literate racing engineers, is revolutionizing electronic engine controls. Competition components and software become cheaper and easier to work with as more and more companies enter the market. Often both the engine control system and data acquisition are available from the same companies because of the overlap in sensors and processing. Because of these constant advancements, this chapter would soon become obsolete if specific products were described. So it will instead focus on general capabilites and selection criteria. For information on making the general manufacturer, software, and hardware decisions, refer to the previous DAS section.

There are two primary computer control functions available: spark timing as a function of rpm and throttle angle, and injector timing as a function of rpm and throttle -- or airflow. Functionally, these aren't much different from the old methods of mechanical distributor and carburetor controls, which used centrifugal weights, vacuum actuators, or pressure differentials. However, the computer can be faster and much more precise, by using 3-D "maps" which are plots based on dyno runs and stored in memory for rapid lookup. In addition, these maps may be modified in real time, depending on changing conditions such as temperature or pressures or other variables, actually making them maps of 4-D or 5-D or more. Also, other functions besides spark and fuel flow can be controlled by the computer, such as a turbo wastegate, or the length of variable inlet bells (or possibly exhaust pipes), or perhaps in the future, a variable cam-advance mechanism.

In selecting a system, the first considerations might be hardware, or the number and types of input sensors that can be used to calculate the output controls. Pressure sensors are obvious, especially barometric pressure, fuel pressure, and possibly low oil pressure as a safety factor. Temperature sensors might be used for the coolant, the ambient air, and the fuel, which can become surprisingly hot in return-flow systems. Also, the exhaust gas temperatures of each cylinder can be valuable, as commonly used in dyno development. "Lambda" sensors, or post-combustion oxygen sensors, might be installed in one or more tailpipes for "self-mapping closed-loop" control systems. A knock sensor could be used, or even one per cylinder, just as they have become more common in passenger cars. And where mileage is important in longer races, a fuel-flow meter could be important, if fuel is not measured by recording the injector flow. Finally, a driveline torque sensor may be used for a more direct measure of performance. So check the system specs for number and types of input channels available, and check the catalog for the transducers you might need.

Besides the obvious control map functions above, there may be other software "bells and whistles" to influence your decision. Field adjustment of the engine control parameters should be easy, using a laptop to edit the maps, or at least a "trim pot box" to make adjustments. There should be an option for one or more alternate reference maps in memory, possibly driver-selectable, for either engine development or race strategy changes in mileage, or running under the yellow flag. Sensor input signals should have easy calibration adjustment, probably including non-linear calibration curves. Programming the maps can be easier if there is interpolation between points, or adjustable-width "break points" on both the throttle and rpm axes. And as engine development becomes more precise and sophisticated, it will be critically important to have complete and independent control of spark and fuel for each individual cylinder.

Other possible features: Given the onboard processor and memory, it might as well include system diagnostics, as is common on production cars, with such functions as sensor tests and verification of harness continuity. With a constant record of fuel usage, an obvious function is preprogrammed fueling or pitstop strategies. Some systems are said to have "self-learning," with the ability to extrapolate beyond programmed functions, or to identify trends and feedforward in time. There are other control functions that can be very useful -- when not restricted by the rules. A shift interrupt might be used to soften the blow on driveline components in clutchless shifting. A rev-limiter could be adapted for use as an automatic pit lane speed controller. And, of course, the well-known and generally banned traction and stability controls, which reduce engine power or applies the brakes, based on various sensor indications of wheelspin. Not too far out is the use of driveshaft torque sensors to optimize performance based on immediate racetrack feedback.

In development: Fiber optics to connect sensitive sensors on the engine, without picking up the tremendous EMI noise from high energy spark production. Going beyond individual cylinder control to adaptive control of the spark and fuel on each power stroke, based on instantaneous cylinder pressure, or implied pressure based on spark energy feedback. Real tracktime recording of the engine power curve, given the driveline torque sensor, and previously input data on vehicle and rotating masses, plus rolling and aerodynamic drag. The time may come when the race car becomes a real time engine development tool as accurate as any engine dyno.