Mustang 5.0 Maf Transfer sanity check

Hey guys would appreciate 2nd or 3rd opinions on zeroing out the fuel trims via the maf transfer,

2017 5.0 mustang with pp2 and custom maf periods to increase resolution in areas the engine actually hits.

this is with one adjustment to the original maf curve after re scaling the periods. the fuel trims are driving me nuts.. in several spots on the same drive, the same period will register lean or rich. this really throws off the adjustment to apply!

im not sure using the entire log average is ideal and have started doing scaled views only to analyze stable, constant throttle/period/airflow sections.
how would you finish off this curve? switch to ltft only? use the average of the whole log?

example,
time stamp 7:24 zoomed in at 2s, period 220 is ltft+stft +1.63%
time stamp 4:18 zoomed in at 2 s, period 220 is ltft+stft -1.54%

Attachment 97587
Attachment 97588
Attachment 97589

A 3% difference really isn't that far off and both are +/- 2% correction. Your widebands constantly making corrections and torque control effects these as well.
If you want to rescale a few cells, just paste to excel and delete the least used row, then add in the maf period you think is useful and interpolate the data. Flash that and make corrections using your datalogs.

Basically use the data you have available. I only rescale the maf transfer on boosted cars where i need more resolution between 100us and 82us.
+/-1.5% correction, I'd call that damn near perfect. I'd call 5% good. You do want wot as close to 0 as you can.

I haven't checked any of your files but the fuel trim swings you mention are very minor.

Open MAF Temp Compensation and zero the rows for 5* to 104*.
Disable MAF Adaption.
You can also disable LTFT and adjusting using only the STFT.

In all my tunes I change the maf frequency points to be more linear in separation. Doing so reduces the chances of spikes and non regularities between points and also allows me to see inconsistencies visually since that's what I'm used to.

It might not help you may give you some idea.

This is what I have found works very well with the most resolution.

Code:

---

        1485
-647        838
-200        638
-100        538
-70        468
-50        418
-30        388
-20        368
-20        348
-20        328
-20        308
-20        288
-20        268
-20        248
-10        238
-10        228
-10        218
-10        208
-10        198
-10        188
-10        178
-5        173
-5        168
-5        163
-5        158
-5        153
-5        148
-5        143
-5        138
-5        133
-5        128
-5        123
-5        118
-5        113
-5        108
-5        103
-5        98
-5        93
-5        88
-5        83

---

Appreciate the time from both of you! Ive adjusted the settings for the maf adaption/temp comp and turned off LTFT. definitely seems to be getting more consistent and repeatable. Ill usually do two logs on the same drive to make sure the numbers im seeing are consistent and trending in the same direction.

certain areas still perplex me but it may just be the new highs and lows on the curve impacting the areas around them. The middle region between 268 and 340, graph on the right is log from last night, copied and multiplied at 50% ( but left the highlighted area alone, within 0.5% seemed acceptable!). Then logging this morning the rest of the curve took nicely to the changes as expected but the center area is swinging a bit with no changes made.

appreciate the example of the change in the curve point over point. I noticed the stock curves had some very bizarre spacing. Im going to check what mine is at and smooth it.

Do we know what math the ecu uses to calculate between two periods? is it linear/poly/exponential?

Attachment 97635

linear