EPROM Tuning Theory

Kyle Tarry

The following procedure outlines my preliminary thoughts on the best possible way to tune a setup, when presented with the capabilities to modify features within the stock 1g DSM ECU code.

One of the key ideas that makes all of this possible is the ability to keep the ECU from going into closed loop.  By zeroing out both of the maps that chose which mode the ECU is in, you are able to lock it into a constant state of open loop operation.  This allows you to tune based on pure A/F ratios, without the added complexity of fuel trims.  Once the tuning is done, the ECU can be put back into closed loop under the proper conditions, to compensate for a changing environment.

All of these ideas also assume a stock OEM MAF, or a properly calibrated aftermarket unit.  As such, this is not as applicable for a system with a hacked MAF, or a VPC-like device.  Some of the theories can still be applied with great success, but the level of precision will not be as high.

First, the injector deadtime must be solved for and compensated.  While the deadtime and the injector compensation can be done together to some extent, changes to the deadtime will cause small changes to the injector compensation values, and as such the deadtime really needs to be locked in stone before global fuel settings are made.  Since deadtimes have an increasingly greater effect as the injector pulsewidth becomes smaller, the proper way to figure out the deadtime for a certain set of injectors is through a comparison of the A/F ratio to the desired A/F ratio and the pulsewidth at that point.  By driving around and recording the A/F ratio at several different load levels, and also recording the pulsewidth, you can figure out the deviation from the desired A/F ratio, and this deviation can be applied to the pulsewidth in order to find the margin of error.  By plotting several datapoints, you can attain an average margin of error, which will be the deadtime of your injectors.

With the deadtime of the injectors solved for, the deviation from the desired A/F ratio should be a fairly constant percentage at all points (load and engine speed).  This is when you finalize the global injector settings, by choosing a setting that will allow the A/F ratio to match the target A/F ratio as closely as possible at as many points as possible.  The more data points that you have to average (to from which to remove high and low extremes), the more accurate this global setting will be.

At this point, any single point deviations in the A/F ratio will be primarily due to inconsistencies of the MAF sensor.  These can be simply ignored if they are not large enough to worry about, or they can be compensated for using the MAF compensation table.

At this point, the ECU should be able to consistently and accurately calculate the correct pulsewidth in order to achieve the target A/F ratio at any given airflow.  Besides the obvious positive effect of the most efficient fuel delivery possible, I feel that this has an additional benefit.  If the ECU can reach a target A/F ratio consistently, the user as the ability to vastly reduce the extent to which he needs to “guess and check” his A/F ratio with a wideband when tuning.  This allows the tuner to set a target A/F ratio anywhere in open loop, and also gives the ability to perfectly tailor fuel curves.