Engine Tuning: Calibration Framework

Now that we have a good idea of what an engine needs to operate smoothly, it is time to actually begin to construct the framework of the calibration. Before the engine is even started, the calibrator must provide the PCM with some parameters that are close enough to the optimum setting to begin checking operation. “Getting into the zip code” of the actual ideal settings can be one of the toughest tasks in the whole process. Whenever possible, it is always easiest to start with a known good calibration for the engine combination being tuned. If a similar engine combination has been calibrated before, it saves a lot of time in the following steps. Making small changes to an existing tuning file also reduces the chances for wholesale miscalculations and no-start, no-run conditions.

 


This Tech Tip is From the Full Book, ENGINE MANAGEMENT: ADVANCED TUNING. For a comprehensive guide on this entire subject you can visit this link:
LEARN MORE ABOUT THIS BOOK HERE

 

SHARE THIS ARTICLE: Please feel free to share this article on Facebook, in Forums, or with any Clubs you participate in. You can copy and paste this link to share: http://www.musclecardiy.com/performance/engine-tuning-calibration-framework/


 

Some aftermarket tuning packages have built-in tools that can often generate maps based on engine component specs (port size, cam specs, compression ratio, etc.) that are close enough to the optimum setting to get started. Other programs allow for changes to be made to a starting OEM file to accommodate hardware changes (new injectors, different MAF sensor, addition of a supercharger) providing the tuner with a closer estimation of what the engine requires in the calibration. This is also the time to review all sensors to be used on the engine to ensure that their outputs are properly recognized by the PCM. A quick key-on (turning the ignition to power up the PCM) to compare ECT, IAT, and MAP values against ambient conditions prevents confusion after startup.

Fuel injector parameters should also be checked at this point. Make sure that fuel pressure has been set properly and injector size parameters in the PCM reflect the components used and current rail pressure. This is also the time to double-check the voltage compensation curves and dead-time characteristics of the injectors being used. Proper modeling of actual fuel delivery makes later modeling of airflow much easier.

 

Some software packages allow the calibrator a head start with the necessary changes by loading predefined tables. The SCT Advantage package shown here allows the user to select from a list of MAF sensors as a starting point for the airpath model.

Some software packages allow the calibrator a head start with the necessary changes by loading predefined tables. The SCT Advantage package shown here allows the user to select from a list of MAF sensors as a starting point for the airpath model.

 
The target air/fuel ratio is set to a constant value in all areas where the engine may operate during initial tuning. This helps reduce the confusion associated with trying to hit a moving target when correcting the airflow model.

The target air/fuel ratio is set to a constant value in all areas where the engine may operate during initial tuning. This helps reduce the confusion associated with trying to hit a moving target when correcting the airflow model.

 

One of the primary keys to properly calibrating any EFI system is the accurate modeling of engine airflow. This is the most important point to remember. The majority of all fuel and spark calculations is based upon airflow or engine load. If the PCM does not know how much air is passing through the engine at any time, it has little hope of accurately metering fuel or controlling ideal spark advance.

To begin calibrating airflow, it is usually easiest to start above idle, often 1,500 rpm or more. Idle is one of the most difficult conditions for the PCM to control, so it is easier to skip right past this to get things “in the zip code.” Running the engine at slightly elevated speed decreases the chances of stalling as a result of less than optimal air/fuel ratio or spark advance. When first starting the airflow modeling process, it may also be helpful to lock the timing to aid stability. The goal here is to build maps representing what the engine does under stable operation. We will return later to smooth out transitions. To accomplish this, the fuel tables are set to deliver a constant air/fuel ratio (usually λ= 1) under a wide range of speeds and loads. With known fuel outputs, it can safely be assumed that any deviation from the desired air/fuel ratio is a result of an error in measured airflow. If the target ratio is λ= 1, using a wideband air/fuel monitor makes this process easier.

It has been my experience that it is easiest to perform initial fuel corrections in open loop. That way, instantaneous lambda output equals the necessary correction factor needed for the current operation point. The self-correcting action of the OEM lambda controller (closed loop operation) often becomes more of a nuisance than tuning aid. If the stock HEGO is not providing an accurate signal, the PCM’s correction may push against the calibrator’s initial changes. This can be more prominent in cases where a larger duration camshaft is being used or header lengths have required the HEGO to be moved farther away from the cylinder head.

Most PCMs can be forced to stay in open loop operation by adjusting the closed loop enable tables to values at the extreme of their range. For example, minimum throttle for open loop can be set to 0% or minimum coolant temperature for closed loop set to 300 degrees. After the majority of airflow modeling is complete and fueling errors are inside of a couple percent, changing back to closed loop operation should not significantly change engine operation.

It is possible to perform these corrections in closed loop, but it adds another multiplication step and requires more data recording. In this case, both short and long-term closed loop fuel trims must also be monitored. These trims must be applied to the actual lambda before calculating the necessary multiplier for airflow corrections. If the closed loop trims have enough authority to bring the engine to λ = 1, then these corrections themselves become the multiplier to be applied to airflow for correction. This process works best on vehicles where the exhaust manifolds, camshaft, and HEGO installations remain the same as stock. Since the PID (Proportional Integral Derivative) control loop running the HEGO feedback and lambda trim in closed loop is rarely stable at 0% correction, I often find it more useful to simply correct in open loop with a more stable delivered lambda.

The bulk of the work in calibrating the airflow model should be done at nominal temperatures. The engine should be warmed up and any startup enrichment routines should be expired before proceeding. Coolant, oil, cylinder head, and intake air temperatures should be stable to avoid interference from other multipliers while adjusting fuel delivery. Adjustments for cold start, varying air temperatures, and other conditions are handled later in the process. This is simply the time to model the base characteristics of the engine and its airflow measurement.

 

Mass Air Flow Modeling

For engines that use an MAF sensor, airflow modeling means making sure that the transfer function in the PCM matches the actual output of the sensor. The ideal method is to have data from the exact MAF sensor to be used, as flowed with the exact intake tract from filter to throttle body, and input this data directly into the PCM’s MAF table. Since this data is rarely available, the next best thing to do is create this data by actual testing on a dynamometer. Something as simple as a change in air filter design or MAF sensor clocking relative to bends in the plumbing can shift MAF output to the PCM by 20% or more. Verifying actual performance on the exact vehicle in question gives the best accuracy.

 

MuscleCarB

 

Actual MAF output calibration is best done by using the dynamometer to hold the engine exactly at one of the calibration points of the MAF sensor, waiting for the engine to stabilize, and finding the error. The calibration points in question may not necessarily line up with fixed RPM or engine loads. All that is needed is a static airflow rate. This means that simply driving the vehicle in a single gear on a chassis dynamometer can yield a wide range of calibration points simply by changing throttle position and consequently speed. Once the engine stabilizes, the MAF table is adjusted at this point and we move on to the next calibration point. This process is repeated for as much of the MAF sensor range as possible, always waiting for the engine to reach steady state operation before making an adjustment to the MAF table values. If it is not possible to adjust the MAF table in real time, make a spreadsheet showing MAF calibration points, target λ, actual λ, and percent error. (Figure 9-1) This makes for fast adjustments to the data tables and a quick return to the dynamometer to confirm or adjust again until error is acceptable.

 

Correcting the airflow model is best done in open loop with a target ratio of lambda = 1 and real time feedback from a wideband. This makes the math for finding the correction factor very simple.

Correcting the airflow model is best done in open loop with a target ratio of lambda = 1 and real time feedback from a wideband. This makes the math for finding the correction factor very simple.

 
Figure 9-1 Making a table of MAF breakpoints and the necessary corrections at steady state can ease the tuning process. Any trends can be extrapolated upward to save time later during WOT calibration.

Figure 9-1 Making a table of MAF breakpoints and the necessary corrections at steady state can ease the tuning process. Any trends can be extrapolated upward to save time later during WOT calibration.

 

A word of caution here is to monitor engine temperature and actual lambda. Stop testing if the engine is operating at either high temperature or an excessively rich or lean condition. Running the engine excessively rich (λ > 0.8) for too long during this process can cause bore wash and ring damage. Likewise, running too lean can generate high exhaust gas temperatures that may damage valves, manifolds, or catalysts. This part of the tuning process is not intended to cover airflow ranges approaching WOT. It should be possible to map the majority of airflow conditions without exceeding about 60% engine load. This should keep temperatures safe.

This is the very coarse tuning part of the calibration process where changes often resemble a lumberjack’s chainsaw as opposed to the whittler’s knife and sandpaper to be used later. Don’t be afraid to stop and make a global 30% or greater change to the MAF curve if it looks like the first guess was really that far off. Look for trends in MAF error to save time. If MAF output appears to be 10% low at all ranges tested so far, go ahead and add 10% all the way up to the maximum values. This saves time later when calibrating the WOT airflow errors.

The process of refining the accuracy of the calculated air mass to match the actual consumption of the engine is repeated until the errors become very small. Less than 5% error is desirable, with less than 3% being ideal. OEM applications usually have less than 1% error across the operating range. The more accurately the airflow modeling is done at this stage, the less work there is to be done later, and the better the vehicle drives. Good calibrators spend most of their time on this part of the process to avoid needing “patchwork” fixes to cover strange engine behaviors later.

 

MAF Scaling

Some PCMs are coded with a maximum calculated value for total airflow that cannot be changed in the calibrated values. When modifying the engine to make more power, such as adding a supercharger or larger displacement, this number can be exceeded. In order to retain full function of the PCM, one must find a way to stay within its calculated limits while still moving large amounts of air. The answer is scaling.

If a PCM is limited to 62 lbs/min of airflow (as seen in many GM LS1 applications) and the engine is capable of moving 70 lbs/min of air (not unusual for very high output supercharger systems), some creativity must be employed. The first and most important step is to use an MAF sensor with an accurate range to at least the maximum predicted flow rate. In our example, an MAF sensor with accuracy up to approximately 72 lbs/min is desirable to leave room for a little extra airflow under extreme conditions. To make this new transfer function work in the limited PCM, it is scaled down. In this case, a multiplier of 0.861 (13.9% reduction) is applied to the actual airflow values in the MAF transfer function. That way, even at the new predicted maximum of 72 lbs/min of airflow, the PCM is still within its allowable range.

To compensate for the reduction in calculated airflow, a similar reduction must be made to fuel delivery. Injector size should be reduced by the same amount, a 0.861 multiplier in this case. Do not change the voltage compensation or dead time at this point, since these should not be affected by the scaling.

 

The basic component of the speed density airpath model is the VE table. This Accel DFI GenVII table shows relative VE values versus speed and MAP.

The basic component of the speed density airpath model is the VE table. This Accel DFI GenVII table shows relative VE values versus speed and MAP.

 

Changing both the airflow and fuel delivery values keeps the delivered air/fuel ratio on target. However, this scaling also has an effect on load calculations. Even though the PCM is delivering the proper amount of fuel to go with the actual air entering the engine, the calculated load is not correct at this point. Load must be corrected since many of the base fuel and spark tables use calculated load as a primary input. To correct the load calculations, the calibrated engine displacement must also be reduced by the same amount.

With all of this done, the PCM now sees an exact scale model of everything. It takes what it thinks is 86.1% of actual airflow and delivers a theoretical 86.1% of matching fuel flow to an engine 86.1% of the actual size yielding exact desired lambda and load calculations. The primary benefit to this scaling technique is that any future fueling corrections can still be performed on a percentage basis. This means that even on a scaled MAF calibration, a 10.0% change to desired lambda still yields a 10.0% change to delivered lambda as long as adjustments are made as multipliers. Although a slight reduction in airflow measurement precision (13.9% in this example) results from this method, OEM transfer functions usually have more than enough break points to accurately model the sensor input. The added accurate measurements at high flow rates outweigh the minor drawback of precision loss across the spectrum.

This method is so effective that it has even been used by Ford in the OEM calibration of the high output GT supercar using a standard EEC-V processor. Keep in mind that this is a vehicle making over 550 hp with 91- octane fuel while passing all CARB and Federal emissions standards.

 

Speed Density Airflow Modeling

For speed density systems, airflow modeling means building the PCM’s volumetric efficiency reference table to accurately represent actual mass flow at all speed and load points of the engine. Load is expressed in terms of manifold absolute pressure, received directly from the MAP sensor. Absolute pressure gives the PCM the density of the air mixture in the manifold, so mass flow can be calculated using the basic gas law once temperature is known. Some systems use Alpha-N mapping that replaces MAP load values with TPS position.

Much of the process is the same as used in building the MAF table, except that the dynamometer is used to hold the engine at various speed and load intersections rather than constant airflow points. To do this, speed is usually set at one of the calibration points in the base fuel map first. After the engine stabilizes at the lowest load point for that speed, changes are made to the corresponding load point. Speed is held constant, throttle is opened to increase load to the next point, and the process is repeated until engine or exhaust temperatures prohibit. All of this mapping should be done at a constant inlet air temperature. Changes in IAT significantly affect the density of the air being ingested by the engine, so it is best to attempt to keep IAT steady during VE mapping.

Much like mapping the MAF sensor transfer function, it is helpful to recognize trends in the volumetric efficiency table. This table should not really have any drastic changes between cells. When viewed as a three dimensional map, it should appear generally smooth with higher values near the engine’s torque peak, where efficiency is highest.

 

MuscleCarB

 

One key that is often overlooked by aftermarket calibrators is resolution of speed density maps. If the PCM has a limited number of cells in the volumetric efficiency table, it is important to choose where these points are best served in the calibrator. The VE table represents curves of engine efficiency versus speed and load. While some engines can be mapped with evenly spaced calibration points every 500 rpm or so, many cannot. To properly map these curves, it is necessary to plot enough points near the tightest bends to reduce errors. In areas where engine efficiency can change rapidly, such as just off idle, points should be set closer to each other in order to properly represent actual airflow changes. This gives the PCM better resolution to compensate for changing efficiency as the intake manifold and camshaft tuning becomes more influential as speed increases. I have often found it useful to have set points for speed at crank: 100 rpm below target idle speed, idle speed, 100 rpm above idle speed, and a few more added below 3,000 rpm to improve resolution. The changes become more linear at higher engine speeds, so it doesn’t hurt accuracy very much to map this region with points that are slightly farther apart.

If the camshaft being used only allows the engine to develop a limited amount of vacuum, the load scale should also be adjusted to reflect the actual range of engine inputs. There is no benefit to having a VE table with 10, 20, 30, and 40 kPa load points if the maximum vacuum seen is only 55 kPa due to cam design. In this case only one load point for extreme vacuum should be necessary, leaving room for more resolution between 50 kPa and 100 kPa. Better planning for set point distribution here can help alleviate drivability problems later when the PCM has trouble interpolating between two widely spaced points. In this case, a middle point to better define the curve would help. Most engines exhibit quasilinear efficiency changes with load, so spacing can be even on the load scale as long as the range is reasonable.

When first building a base VE table from scratch, the calibrator can safely assume that values increase with higher load and RPM. This means that if part load testing yields a VE of 60% at 70 kPa and 3,000 rpm, that same 60% value can be copied across to the remaining RPM cells at 70 kPa for an initial guess. This gets the calibrator a closer guess before attempting to fine-tune VE at 3,500, 4,000, 4,500, etc. The trend should continue upward as well. Taking our example of 60% VE at 70 kPa at 3,000 rpm, we also know that VE values should steadily increase from 60% as we work upward in load at 3,000 rpm. It may be possible to perform WOT (≈ 100 kPa on a naturally aspirated engine) calibrations at low engine speeds in this manner. But as engine speeds get higher, temperatures climb quickly at high load, preventing steady stat  mapping of those cells. Follow the trend to develop a best guess at what the values might be near 100 kPa at higher engine speeds to save time during WOT calibration later.

After the base volumetric efficiency map has been constructed under stable conditions, the vehicle can be checked at different inlet temperatures to build the IAT compensation curve. If tuning was performed with steady 80 degree F inlet temperatures on a dynamometer and the vehicle exhibits a lean condition when driven outside in 50 degree F ambient conditions, the calibrator needs only to adjust the IAT fuel multiplier to compensate. Changing the base VE table at this point only causes a rich condition when inlet temperatures rise again. To further aid the process, the vehicle should later be allowed to completely cool (preferably overnight) to check the bottom end of this compensation curve for accuracy. If it is not possible to actually test with very cold inlet temperatures, a straight-line extrapolation usually works.

 

Spark Advance

After the majority of the airflow modeling has been completed, the calibrator can now look into spark mapping. For part load conditions, it is usually best to find MBT timing for most cases. This allows the engine to operate as efficiently as possible, with optimal fuel economy and good throttle response. The spark map should look like an inverse of the volumetric efficiency map, smooth and predictable. Using the dynamometer to find best torque output at a few points along the load and RPM curves, as well as filling in the remaining cells of the base spark table, is rather simple.

If spark advance is left too low in part throttle regions, the result can be a bucking or “trailer-hitching” sensation on the road as the engine misfires. Remember that larger camshafts increase the natural EGR of the engine and require more spark advance to compensate for the cooler combustion temperatures. High-octane fuels also require a global increase in ignition advance to compensate for their slower burn rate.

 

Most PCMs use speed and load for inputs to the base ignition map. The actual delivered timing may vary from this if other adders for temperature, lambda, or torque control are active.

Most PCMs use speed and load for inputs to the base ignition map. The actual delivered timing may vary from this if other adders for temperature, lambda, or torque control are active.

 

The PCM calculates total timing primarily from speed and load. Additional adjustments are made to correct for temperatures, fueling, and other conditions before aligning the spark event with the crank sensor output.

The PCM calculates total timing primarily from speed and load. Additional adjustments are made to correct for temperatures, fueling, and other conditions before aligning the spark event with the crank sensor output.

 

High compression and superchargers are seldom friends. This kit from Vortech engineering for the 11.0:1 Corvette Z06 works very well as a result of low charge temperatures and careful engine tuning.

High compression and superchargers are seldom friends. This kit from Vortech engineering for the 11.0:1 Corvette Z06 works very well as a result of low charge temperatures and careful engine tuning.

 

It is not usually necessary to plot every single load and speed point to find MBT unless working on an actual OEM release calibration. For areas outside the easily measurable range, extrapolate the trends seen in the known areas of the map. Timing should trend downward with increasing load and upward with engine speed.

During this process, monitor the IAT and ECT to ensure that the engine is operating near the nominal values. Actual temperature compensation curves can be finished after the base tables are complete. Don’t be afraid to use what looks like large values for ignition advance at part load—thisis often what the engine wants. Running as much spark advance as possible (up to MBT) results in higher overall engine efficiency, more torque, and crisper throttle response.

Knock is still a concern at part load. Although its presence is less likely at lower loads, it should still be avoided. Most engines exhibit a slight loss in torque at part load before actually hitting the knock limit. If steady state testing is being performed on a dynamometer, this can be seen while adjusting the timing in the search for MBT. Listen carefully for any signs of knock and reduce timing accordingly in the affected area.

 

Written by Greg Banish and Posted with Permission of CarTechBooks

GET A DEAL ON THIS BOOK!

If you liked this article you will LOVE the full book. Click the button below and we will send you an exclusive deal on this book. Deal-Button