Tuning Your Ford EFI System - Inside The Black Box Part 1

Mass AirWith the SD systems, the instantaneous mass airflow was calculated by the ECU; the only risk in the calculation was the programmed value for VE. If it's not correct, the fueling will be off. Therefore, anything you change on the engine that affects the VE (like a better flowing intake manifold or camshaft for example) will throw off the fueling calculations, unless the VE can be reprogrammed accurately. For that matter, it's a tough task to program all the VE values for every combination of MAP and rpm.

With a Mass Air system, the airflow is actually measured, using (you guessed it) a Mass Air Flow (MAF) sensor. Injector pulse width is still calculated in the same manner as shown previously, however now the airflow is actually measured instead of calculated. The big advantage of a MAF system is that you can change things on the engine that affect airflow, the MAF sensor will realize the change in airflow, and the fueling will still be correct. It makes the MAF system the most forgiving for engine modifications. However, just as the SD system is only as good as the VE programming, the MAF system is only as good as the MAF sensor accuracy, and, unfortunately, many aftermarket MAF meters have poor accuracy.

Now you may be wondering about aftermarket MAF meters that are "calibrated" to work with larger injectors. For example, say you have a MAF and injector combination for 36-lb/hr injectors, and you install that into an original 19-lb/hr setup, with no other changes. It somehow magically works without having to reprogram the ECU, right? How it really works is by fooling the ECU. The recalibrated MAF sends roughly half the voltage signal to the ECU, causing the ECU to "think" only half the air is flowing into the engine. If there's only half the air, then only half the fuel flow is required, so the ECU will command roughly half the injector PW. This works out fine with the 36-lb/hr injectors because they flow about twice what the 19-lb/hr injectors did, and in the end, roughly the same amount of fuel still flows into the engine. The problem with this approach is the inaccurate load calculation, which can screw up the spark advance and desired A/F ratio.

Loading UpLoad is analogous to VE with the mass air systems. In the later EECs, load is exactly calculated as VE, but in some of the earlier EEC-IV applications, load was a "normalized" VE (often called "Percent Load"), where the raw load (VE) was divided by a maximum expected load for each rpm, at sea level. This Percent Load (if the peak sea-level load function was programmed correctly) would then calculate 100 percent load as full power output, at all values of rpm.

In some ways, this is a more accurate description of load, since at any rpm, whenat 100 percent load, the engine is giving its all. With raw load, you would see load values over 100 percent on supercharged applications, which might seem confusing at first. Another advantage of Percent Load is the ability to use all rows in any column of a programmed fuel or spark table (more about this in Part 2).

In the SD systems, load was analogous to MAP. At 100 percent load, you'd be at zero manifold vacuum (note: this system doesn't work well for boosted applications). In some other oddball applications, load is actually calculated from Throttle Position (i.e. at WOT, Load = 100 percent).

Load is used for many purposes in the ECU. The important ones for our discussion are to properly manage the A/F ratio and spark advance, i.e., at low loads the A/F ratio can be stoichiometric (or even leaner), and spark timing can be advanced much farther to achieve Maximum Brake Torque (MBT). At higher loads, the A/F ratio should get richer, and spark advance reduced for MBT (and to guard against combustion detonation).

Open Loop vs. Closed Loop Fuel ControlAll of the previous discussions on how the ECU calculates fueling are based on an Open Loop (OL) control strategy. In OL mode, the ECU reads all its sensors, checks with its programming to determine the desired A/F ratio, calculates the required injector PW (as detailed previously), then sends that PW to the injectors, hoping for the best. It never really knows if it achieved the desired A/F ratio or not.

With Closed Loop (CL) control, the ECU has feedback to tell it whether it hit the desired A/F ratio target or not (well, sort of, as we'll find out). At part throttle, when the engine runs at the stoichiometric (chemically correct) 14.6 A/F ratio (required for minimum emissions, and maximum catalytic converter efficiency), the combustion products should ideally be only water vapor and carbon dioxide. If the A/F ratio was a little lean (too much air and/or not enough fuel), the excess (unburned) oxygen will show up in the exhaust gases. Using an O2 sensor, the ECU can now detect when the engine is running lean. Problem is, the factory O2 sensors have a narrow band for sensing oxygen. In other words, they can really only tell the ECU if there is some oxygen in the exhaust, but not precisely how much. So the ECU doesn't really know how lean the engine is running, only that it's running lean.

So what happens when the engine is running rich and there is no oxygen in the exhaust? The ECU will assume that if there's no oxygen present in the exhaust, then the A/F ratio must be rich. At that point, the ECU makes a conscious decision to lean the A/F ratio, that is, until it reads oxygen in the exhaust again. At that point, the ECU knows it went too far (it's now lean), so it will richen things up again until the oxygen disappears, but now it's rich again, so the ECU will again lean things out, and so on. In CL mode, the ECU will continually cycle the A/F ratio lean, then rich, over and over, hovering the A/F ratio closely around the stoichiometric point.

In CL mode, the ECU will still go through its normal PW calculation, then adjust the PW if necessary, based on what the feedback loop (O2 sensors) is saying. Any adjustment to the calculated PW is handled by the ECU as a Short Term Fuel Trim (STFT). With modern (second generation) On Board Diagnostics (OBD II), many ECU parameters can now easily be logged in real time, including the STFT. Depending on the ECU and data-logging system being used, STFT can be reported differently, sometimes as a plus or negative percent (minus STFT meaning the engine is running lean, so the ECU is reducing its calculated A/F ratio in order to get the actual desired A/F ratio), or as a number around 1.00 (STFT numbers less than one, meaning the ECU is correcting a lean condition).

Adaptive StrategyIf the ECU constantly needs to shorten calculated PW to achieve the desired 14.6 A/F ratio in CL mode, it knows its programming is calculating a PW too long (possibly from erroneous sensor inputs), and it will remember that correction for next time it makes a PW calculation for the same operating conditions of load and rpm. In other words, the ECU actually learns. The remembered corrections are known as Long Term Fuel Trims (LTFT). The LTFT work oppositely to the STFT, i.e., a plus LTFT (or LTFT greater than one) indicates the ECU is adding some to the calculated PW, in order to get the A/F it wants, based on what it's learned in the past.

Even though the ECU only knows it's hitting the desired A/F ratio when the A/F ratio is stoichiometric (in CL mode), the ECU can also apply the learned corrections (LTFT) any time the ECU is operating in OL mode, and commanding an A/F ratio other than stoichiometric. This Adaptive Strategy allows sensors to age and drift in their readings, but the ECU can now correct for the errors, and still hit the desired A/F ratio in the end.

In the real world, adaptive strategy can be a problem. For some ECUs, corrections learned at one set of operating conditions (e.g., idle) are also applied under other operating conditions (like high-rpm WOT). In that case, if you have a MAF sensor that reads too rich at idle (typical of many aftermarket MAF sensors), a leaned idle correction is learned, which if also applied at WOT, can spell disaster. For these applications, it's necessary to either rework the adaptive learning table in the programming to prevent learned corrections from being applied under other operating conditions, or make certain the MAF sensor calibration is corrected. The least desirable, last-ditch technique is to disable adaptive learning in the tune. Whichever way, a custom tune is required.

Closed Loop mode (and Adaptive Strategy learning) is usually only active during part throttle modes, when the engine and O2 sensors are up to stabilized temperatures, and the goal is an A/F ratio right at stoichiometric. During the first few minutes after the engine is first started, and during high-load periods (like WOT) when a richer A/F ratio is intentionally desired, OL fueling strategy is typically used. Thus for maximum power, we need to ensure the ECU calculates the correct OL injector PW, so we get the WOT A/F ratio right. This is where a proper tune comes in.

In Part 2, we'll cover some of the specific Ford EEC tuning parameters, then in Part 3 we'll actually put the science to the test, with a step-by-step custom tune of an '04 Cobra. Stay tuned.