So say I've got some maf sensor "Z" with a completely unknown transfer function, and I'm looking to calibrate it for the following situation. Focusing solely on the software/calibration aspect (and leaving hardware feasibility out of the discussion), tell me if the following looks good, and if there's anything you'd do to improve the process:

SITUATION 1:


Goal:  To create a drop-in MLFHM map for this new sensor intended for people who are currently running a tune based off of a stock Hitachi MLHFM map that has been multiplied by a simple scalar factor.  Ideally the new MLHFM would be plug-n-play, and be able to replace the previous 'scaled-Hitachi' MLHFM map with minimal tweaking (minimal tweaking consisting only of adjusting the new MLHFM by a slightly different scaling factor).


Assumptions:  Calibration car in step 1 has completely stock hardware (other than the new blow-though MAF spliced into the intake system), and the calibration car is on stage 1 software that has left all intake and fueling maps untouched. A Wideband O2 monitoring system installed with the ability to also log the new MAF sensor's voltage independent of the stock ECU.  The calibration car in step 2 is a stage 3(+) car with a Hitachi MAF sensor and an MLHFM map that is only modified by a scalar.   


Procedure:   

Step 1.  Calibrating up to the limit of the stock MAF and stock fueling:
     
     With the car running on the stock Hitachi MAF sensor, simultaneously log variable mlhfma_w (the variable that the MLFHM map spits out from the stock MAF voltage) and also log the new MAF 'Z's voltage.  Correlate these new MAF Z's voltage with mlhfma_w to create the new MLHFM.


Step 2.  Calibrating from the stock MAF and fueling hardware limit upwards:

     Since the stock MAF/fueling will have limits that are well below what the new Z sensor will have, the stock car can only be used to calibrate the bottom portion of the MLHFM map.  I'm not sure what the most reliable/accurate way to calibrate from here on up is, but I'm thinking the following sounds good:

Take a stage 3 car with an stock Hitachi element and 85mm (or bigger) MAF housing that is using a tune that has an MLHFM map that is a scalar form of the stock Hitachi MLHFM map (let's say they multiplied the stock map by scalar X).  Do some extensive logging at various loads and RPM's, simultaneously logging load, RPM, requested AFR and actual AFR (via wideband).  Make a 3D map that has load as the X-axis, RPM as the Y-axis, and the cells are filled with (actual AFR - requested AFR) [averages, as the logging should have more than 1 data point for each cell].  So we've got a 3D map that correlates load, RPM, actual AFR and requested AFR (we'll call this 'High Load Calibration Map').  Now, we plug the new sensor "Z" into the ECU and adjust MLHFM with the points we got from the stock setup in step 1 (multiplied by the scalar that was used originally in this stage 3 tune), and put guesses in for the higher MLHFM values that are still unknown for this MAF.  Now running the new MAF 'Z', we log the same variables as before and come up with the 'High Load Calibration Map' for MAF 'Z'.  You compare MAF Z's calibration map to the original calibration map that you made, and make adjustments to MLHFM (on the higher end of the map) based on the differences between the maps. Iterate, log, compare, tweak until you get a MLHFM that give you the same values in your 'High Load Calibration Map' as the original.

Once you've got that High Load Calibration Map matched up on the new sensor and new MLHFM, you divide the MLFHM by the original scalar used to get back down to the unscaled, stock MAF housing sized MLHFM map.  You can now append these higher values to help fill out the top end of the MLHFM map that you got from step one with the stock hardware.  If I'm thinking about this correctly, doing the above by comparing and linearizing just the AFR differences in the stage 3 setup makes it independent of any fueling, KFKHFM, FKKVS, KFLF, intake, intercooler etc, changes that they might have made. This way, it doesn't use their possibly (likely) inaccurate g/s reading to calibrate it.  Am I thinking about this correctly, and this process properly isolates MLFHM so I can linearize the new sensor to the Hitachi, even with the original stage 3 Hitachi setup under/over scaled?


Thought, comments, concerns?