Great data Pete

This is also what I have done, but I have found that the MAF transfer function data is very dependant on the source of the information, and using MAF transfer data stripped from binary ecu files can introduce an unexpected error. For example the MAF transfer function in the 4DED v's 1DGA is 5.63 ± 0.01% leaner across the entire voltage range 0.5 - 5.0V (ie a [very] constant bias factor), yet they use the exact same MAF housing and sensor. Obviously this can not happen in reality, the same MAF will give the same voltage output v's airflow. So the programmer has deliberately introducted a bias, but which one do you use for comparison purposes? Was something else in the 4DED binary changed to partially (or fully) dilute the bias? Catch 22

Overlaying both these these curves over an A9L MAF transfer curve shows that using an A9L sensor and housing in a 94-95 EF with a 4DED ecu, could potentially cause a 9.5% leaner condition at WOT where there is no closed loop AFR correction. Not quite so bad on a 1DGA EB falcon, only 4.5% leaner. However in the less important closed loop portion of the curves the 1DGA peaks out at 14% richer, the 4DED only 8% richer. So the A9L curve is a different shape to 1DGA & 4DED.

Here are the curve formulas from my excel worksheet

1DGA=(4.751*POWER(G26,3))+(8.866*POWER(G26,2))+(18.65*G26)-0.499
4DED=(4.489*POWER(G27,3))+(8.451*POWER(G27,2))+(17.55*G27)-0.411
A9L=(4.23*POWER(G28,3))+(18.03*POWER(G28,2))-(8.021*G28)+13.63

In all cases, the correlation coefficient for the curve fitting the raw data points was 1 (no discrepancy between actual and calculated results).