by Greg Banish
Fuel delivery in the WOT tuning process remains consistent with the model used at medium and low loads. The airflow model must now be adjusted in the upper range to match actual mass flow through the engine. Again, since the actual fuel delivery is known, the assumption is made that any difference between the commanded lambda and delivered lambda is a result of error in the airflow model. Just like at part load, this error is multiplied by the predicted airflow to bring it in line with actual flow. An accurate, temperature-compensated wideband oxygen sensor is absolutely necessary for accurate measurements during the WOT tuning process. Standard HEGOs lack accuracy in the target lambda range for these tests and only serve to mislead and confuse.
The basic process for WOT tuning is much like part throttle tuning. The primary difference is that the target A/F ratio is usually much richer than stoichiometric. (Nate Tovey)
A safe (lower than the expected maximum) amount of ignition timing is chosen and target lambda is set to a slightly rich value. Running slightly rich and with less-than-ideal ignition advance to start reduces the risk of detonation and helps keep exhaust gas temperatures safe. I usually use λ ≈ 0.85 (12.5:1 a/f) for naturally aspirated engines and λ ≈ 0.77 (11.3:1 a/f) for supercharged engines to start. It is helpful to change large areas of the load maps for desired lambda to the target WOT ratio at first. That way, even if calculated load at WOT is relatively low at low RPM, the target ratio remains the same. This simplifies the math when correcting MAF or VE values to achieve the desired ratio. Other fuel adders such as catalyst and piston protection strategies should be disabled for this part of the testing. They can be reactivated after the completion of WOT airflow mapping if needed.
For example, a naturally aspirated engine is desired to be run at
λ = 0.85 under WOT conditions. First, all target lambda values in the fuel map are set to 0.85 for loads over 50%. The engine is warmed to normal operating temperature and run in a WOT sweep from ~2,000 rpm to redline, provided there are no signs of detonation or lean mixture. During this run, a datalogger is used to record RPM, MAF, MAP, and actual lambda. If the engine achieves λ = 0.85 at 2,500 rpm, no change is necessary for those cells. If a lean condition of λ = 0.9 is seen at 4,000 rpm, an adjustment is calculated by dividing actual lambda by target lambda and using this as a multiplier for the corresponding airflow value. If a mass air system is used, this number is multiplied by the value in the MAF transfer function that was recorded at 4,000 rpm. If the value recorded by the datalogger is 3.2 v at 4,000 rpm, the curve is adjusted locally by:
Airflow correction = (0.9)/(0.85) = 1.0588
This yields a ~6% correction to add to the airflow number in the MAF transfer function at 3.2 v. Looking at the MAF transfer function, we might see a value of 30.0 lbs/min airflow at 3.2v. The airflow correction is applied to find the new airflow value to be entered into the MAF transfer function in the PCM.
New MAF value = (30.0lbs/min) x 1.058 ≈ 31.8 lbs/min
The target air/fuel ratio is set to a constant value in all areas where the engine may operate during WOT. This helps reduce the confusion associated with trying to hit a moving target when correcting the airflow model.
Speed density applications simply point to a cell in the volumetric efficiency table at 4,000 rpm and the corresponding MAP value (usually 95 to 100 kPa in this case) that can be adjusted by the same 1.058 multiplier. This process is repeated at all points where the delivered lambda varies from the target by more than one or two percent. Within a couple pulls at WOT, this process should bring the engine into almost exact desired fueling conditions.
The PCM likely rounds whatever new value you enter slightly to a step size allowed in its software, so do not be alarmed if it is not exact. The point is to get the airflow model as close to reality as possible. Most PCMs have small enough step sizes to get within 1% accuracy if you have the precision and patience.
Once airflow has been accurately mapped all the way up, the true search for power can begin. New target ratios can now be set in the PCM and verified on the dynamometer. The leaner the mixture, the closer to the knock limit the engine gets. For forced induction applications, the line between best power and detonation can be very thin. It is up to the calibrator to decide exactly how much leaner he is confident in letting a certain engine operate. I typically aim for λ ≈ 0.87 (12.7:1 a/f) for naturally aspirated engines, λ ≈ 0.78 (11.4:1 a/f) for positive displacement supercharged engines, and λ ≈ 0.80 (11.7:1 a/f) for centrifugally supercharged or turbocharged engines, but individual cases vary.
With the desired lambda reached, ignition timing can be added to find the best power without knock. Two-degree increments usually work well to quickly find out if more timing helps. During WOT testing, actual spark advance should be recorded as well as knock sensor signal, if available. The OEM knock sensor can be a very useful tool in determining the maximum allowable advance for an engine, and should be used whenever possible.
Most OEM knock routines are very accurate, just slow to recover. OEM knock strategies often aggressively retard timing as soon as activity is detected from the knock sensor. For performance applications, it can be beneficial to reduce the attack rate of knock retard as well as increase the recovery rate that increments timing back to normal. Even though spark advance tries to return to the desired value sooner, a repeat offense simply starts the routine over with another round of spark retardation. By not changing the actual threshold, this technique preserves engine protection while reducing the duration of knock strategy intrusion on the driving experience.
Low amounts of spark advance at high engine speeds often lead to high exhaust gas temperatures because part of the charge is often still burning when the exhaust valve opens. Increasing ignition lead forces combustion to happen earlier inside the cylinder and reduces exhaust temperatures. If exhaust gas temperatures are high enough to require advancing the timing, the knock threshold may limit total advance. The solution is to richen the target lambda to reduce burn temperature and allow the increased timing lead.
Wide-open throttle spark conditions are especially prone to misfire. During misfire, the spark event simply never sets off the chain reaction that creates power. When engine loads are highest, the mixture density inside the combustion chamber is also greatest. These dense mixtures, though packed with tremendous energy potential, make it more difficult for the spark to jump the gap between electrodes. The first solution is to reduce the gap size, making for an easier strike with the same electrical potential. It is desirable to operate the engine with the widest spark plug gap permitted by normal operation in order to start combustion from more spark surface area. Sometimes the inability of the spark to jump this gap forces the reduction of its size. A compromise is made to burn efficiency in order to support higher power levels. Most naturally aspirated engines operate with a normal spark plug gap of 0.050 inch to 0.080 inch. Forced induction engines may need gaps as small at 0.028 inch to operate properly. If an engine misfires under heavy load with gaps any smaller than this, another solution should be found to preserve good operation at light loads.
The second solution to misfire is to increase the electrical potential (voltage) of the ignition system. This can be done by increasing the size of the ignition coil or input voltage to the coil itself. There are many aftermarket ignition systems that can be easily retrofitted to most applications. One of the primary benefits of multiple-coil ignition systems can be seen in the form of better WOT coil saturation. This is due to the increased time each coil has to recharge before striking when only supplying a portion of the engine’s total ignition needs.
Additionally, insufficient spark advance can have the burn taking place too late as the piston is already rapidly moving downward, expanding cylinder volume. The resulting unstable burn yields poor power delivery and often feels exactly like a misfire to the driver. A heavy “bucking” under load that is not accompanied by obvious sounds of detonation may be improved by advancing spark lead. Remember that misfire conditions appear as lean to a wideband oxygen sensor due to the large amounts of unburned oxygen present in the exhaust.
The balance between lambda and spark lead can yield similar power levels out of the same engine with two different calibrations. It may be that λ = 0.90, 27 degrees of advance, λ = 0.87, 28 degrees of advance, and λ = 0.84, 32 degrees of advance may make the same power on an engine with obviously different fuel consumption rates. The difference is in how close each calibration rests to the knock limit and maximum capacity of the fuel system. Special care should be taken when considering possible loading and environmental changes the car may see. This is to make sure some degree of safety margin is left in the tune to prevent engine damage under unforeseen circumstances.
Changes in the delivered air/fuel ratio can be measured with a wideband oxygen sensor on the dyno and used to determine best output. Notice the minimal change in power with change in lambda between the runs.
There are a couple of serious caveats to consider during WOT, high speed airflow modeling. First, we have made the assumption that fuel flow is known and consistent. If the fuel injectors are forced to go into static flow and the maximum limit of fuel mass delivery is reached, the engine leans out as airflow increases. It is up to the calibrator to check injector duty cycle and make sure that this is not the cause of a lean condition under load. Insufficient fuel delivery can also result from a loss of fuel pressure at the rail. This is usually a result of insufficient fuel pump flow, but sometimes results from fuel delivery lines undersized for the necessary volume.
The second serious consideration for WOT airflow modeling is sensor range. As mentioned earlier, some MAF sensor ranges may be exceeded by engines that make large amounts of power. The same can happen when adding boost to a speed density system where the MAP sensor only has one bar of range. The ideal solution is to select a new sensor with a wide enough range to prevent “pegging” at high flow rates and input this new transfer function into the PCM. Always allowing the PCM to know exactly how much airflow the engine is developing allows it to maintain the correct lambda even if airflow changes slightly above what would have been the previous measurable limit.
For example, adding a supercharger to an engine increases power 50% above stock, and the MAF is replaced with a new unit of similar range. Running this engine at WOT on a cold day with denser air can yield another 5% increase in air mass entering the engine. If the MAF sensor is able to register this increase, all is well and the delivered air/fuel ratio never changes. If the PCM does not have any way to see this increased air mass delivery, it assumes that nothing changed and fuel delivery remains constant. This is a doubly dangerous situation, as the actual air/fuel ratio will lean out, bringing the engine closer to its detonation threshold, and more total power is being made, placing extra stress on internal components at the same time.
That same supercharged engine may also encounter belt slippage that reduces the air mass being force fed into the manifold. If the MAF and MAP sensors are both within range at this time, delivered lambda remains as commanded. If the engine is beyond the range of its sensors, this slippage results in a rich condition (often accompanied by too little spark advance for the reduced load), leading to further reduced power and possibly misfire.
If it is not possible to change the MAF or MAP sensor to accurately measure exact airflow at WOT, a few tricks can be applied to compensate. Admittedly, these are definitely not the ideal way to calibrate a car, but they work when the limits of the hardware prevent the correct solution from being an option.
Since most PCMs have either a separate table for wide open throttle fuel and spark, or at least a separate row/column in the base maps, WOT can be calibrated independently from the rest of engine operation. These are sometimes labeled as WOT maps, or “power enrichment,” and may be triggered by TPS input or load calculation. Under ideal calibration conditions they are merely set to the exact desired lambda. When the range of the MAP or MAF sensor is exceeded and the engine runs leaner as airflow increases, this effect can be counteracted by commanding a richer than actual mixture in these tables or cells. The MAF is mapped and WOT lambda is set normally up until the speed that coincides with maximum MAF value. Beyond this point, commanded lambda values are set intentionally richer to maintain a constant delivered air/fuel ratio. This speed versus enrichment method works well as long as the amount of error remains consistent. Just like any other time with a sensor out of range, slight changes to the actual value that are still out of this range are not registered by the PCM. This means that any increase in airflow (colder inlet air temperatures, higher boost pressure) will lean out the actual delivered ratio. The calibrator must plan ahead and decide how much safety margin to leave in the tune for just such occasions that inevitably happen. Running slightly richer seldom costs too much power, but it can save an engine from destruction the first time the driver floors the gas on a cool autumn night.