2015+ WRX MAF Scaling attempt accesstuner

Hey thanks for clicking on that link! I'm looking for some sage wisdom on CL MAF scaling with accesstuner on a stock 2018 WRX. What it seems like I'm doing is doing a best fit with the MAF sensor while holding injector specific parameters, variable valve timing, ignition timing, and stock wideband o2 sensor accuracy/latency as givens...written in stone as the way it is and must be. I assume the MAF sensors have been extensively tested and calibrated for the average production target accuracy. Possibly one of the more accurate and finessed sensors involved due to the simplicity of controlling the testing conditions. Stock, I see learned values as high as +/-8%. Is this the right approach or just an easy attack vector for the DIYer?

Tell me if i have this procedure correct;
If learned fuel trim is 7.00% (stable) at MAF voltage 2.00 (random value) which has a g/s value of 2 (random value) decrease g/s half the fuel trim learned 7.00*-0.5=-3.5 so 2*0.965=1.93 g/s as a sanity chech to make sure adjustment goes in the intended direction. So before, this MAF table dictated 2.00 g/s at 2.00 volts and was running lean because more air was entering than being dictated for, the ecu called for a fuel enrichment of 7.00%. Now, by adjusting the g/s value to a smaller value it expects less air at that voltage, allowing for the ecu to only be enriching by 3.5% more or less. Do we want the ecu to learn cut or add values? Prefer it to learn fuel cut values to err on the rich side of any blind side? It just feels like there are some other factors that weigh heavily on this like valve timing and egr.
Looking beyond that what about OL MAF scaling? Say I have finished my CL MAF scaling and begin logging for OL. How exactly do I merge the overlapping areas between Cl and OL? I would be using mafscaling-2.5.1 program or possibly accesstuner to ecu. Is OL going to require a dyno for any accuracy?

Feel free to move this thread to a more appropriate area.
 
Last edited:

HolyCrapItsFast

Drinks beer!
It is very easy and you don't need to think allot about it...

For a given MAF voltage, if the learned value is 7% then the g/s value at that voltage should be increase by 7%. It is that simple. It is a one to one relationship between trim error and g/s. Also keep in mind that the total trim error is A/F Correction Plus A/F Learned so to get the actual error you need to add the two. Typically when I scale the MAF, I do a rough correction first using only learned (long term) data and then I do it again and make a fine correction using both short and long term trims.

Another thing to consider is if the whole log run is say 7% then it may be prudent to apply a global correction. Most people would apply a 7% correction to the entire MAF scale to affect a global change, and though this will achieve the desired results, it is not ideal because you alter the calculated load value by doing this. The proper way to apply a global correction would be to apply a -7% change in injector size to add fuel over all.

Beyond that the real key to successfully scaling the MAF is to filter your log data correctly and that you have enough but not to much. It helps to graph everything out as well to get a better visual of what is really going on.
 

HolyCrapItsFast

Drinks beer!
Also you need to do Closed loop scaling separate from Open loop. The logging parameters are slightly different and you require the use of a wide band AFR gauge and be able to log from it. For open loop data you will need to filter out anything that is less than 100% throttle. Don't worry about open loop at lower throttle ranges. You should cover all of the voltages up to the open loop range during your closed loop tuning.
 
Last edited:
I had attempted to do a CL MAF calibration by logging, adjusting, loading new MAF table values, log some more, check/adjust. But I found it much more fun to do it in accesstuner live. Its hard to hit some of the cells though, long enough to get good learned values. When a learned value is steady and corrections are not fluctuation I make the adjustment and continue at the pace/load required to hold the cell, for confirmation that the correct adjustment has been made.

If I'm reading into your post correctly, if a consistent correction is called for across enough cells then the fuel injector size should be reduced? What exactly is being adjusted on the fuel injection tables? Or are we going from a g/s to a g/rev component?
 

HolyCrapItsFast

Drinks beer!
If I'm reading into your post correctly, if a consistent correction is called for across enough cells then the fuel injector size should be reduced? What exactly is being adjusted on the fuel injection tables? Or are we going from a g/s to a g/rev component?

Not necessarily. If a similar correction is needed throughout the entire maf scale then you change injector scale instead. If there are a range of cells that are similar but another range of cells are significantly different then you need to scale the maf.

The other thing you need to look at is latency because that is another requirement all together. For instance if your errors are low in the idle and light cruising area and then high in the open loop region, you need less latency and if they are high at idle and light cruising and then low during open loop then you need more latency. For stock injectors this requirement is not necessary so you need not concern yourself with latency. Just focus on MAF/Injector scaling.

Real time tuning the closed loop region of the MAF is tedious to me. I prefer plugging the data from a log into a spread sheet that spits out the new maf scale for me. I have such a file and I can send it to you.
 
Last edited:
Not necessarily. If a similar correction is needed throughout the entire maf scale then you change injector scale instead. If there are a range of cells that are similar but another range of cells are significantly different then you need to scale the maf.

The other thing you need to look at is latency because that is another requirement all together. For instance if your errors are low in the idle and light cruising area and then high in the open loop region, you need less latency and if they are high at idle and light cruising and then low during open loop then you need more latency. For stock injectors this requirement is not necessary so you need not concern yourself with latency. Just focus on MAF/Injector scaling.

Real time tuning the closed loop region of the MAF is tedious to me. I prefer plugging the data from a log into a spread sheet that spits out the new maf scale for me. I have such a file and I can send it to you.

I don't see tables that would directly adjust pulse width. It seems to be a given value not available. Or am I blind and its right in front of me.

Injector section.jpg
 

HolyCrapItsFast

Drinks beer!
Oh right... Direct injection. Dhu!

Ignore my previous posts. You could just focus on MAF scaling for now

If you do decide to adjust global injector scale then the two tables for "fuel injector trim (multiplier)" and "fuel injector trim (offset)" would be the tables to touch. Just simply apply a percentage change to all values in both tables. Keep in mind that in this scenario a positive percentage value will effect an increase in fuel.
 
Last edited:
Top