Tuning MAF and VVE at the Same Time

[Cringer]

I tried the Smokeshow method and it taught me a lot, but ultimately I felt it had limitations. Could be my setup though. In the end, it lead me down a much simpler path and gave me more accurate results. Here is where the path has taken me. Thoughts?

https://youtu.be/xRZaTblSQ_0

Here is the math, copy and paste into your own math xml file. Also, be sure to use the proper filters no matter what method you use!

MAF CL
(([2320.71]+([2320.71]*([50116.156]+[50114.156])/100))-[50040.71])/[50040.71]*100


MAF PE* for Lambda
(([2320.71]+([2320.71]*([50119.238]-[50118.238])/[50118.238]))-[50040.71])/[50040.71]*100


MAF PE for AFR
(([2320.71]+([2320.71]*(([50120]-[50121])/[50121]))-[16.71])/[16.71]*100




VVE CL
(([2320.71]+([2320.71]*([50116.156]+[50114.156])/100))-[2311.71])/[2311.71]*100


VVE PE* for Lambda
(([2320.71]+([2320.71]*([50119.238]-[50118.238])/[50118.238]))-[2311.71])/[2311.71]*100


VVE PE for AFR
(([2320.71]+([2320.71]*(([50120]-[50121])/[50121]))-[2311.71])/[2311.71]*100

[Filters to use in Scanner to get rid of transients and CL/PE]

CL: [2517.161.avg(1500)]=0 and [2517.161.avg(-200)]=0 and (abs([50090.156.slope(1500)])+abs([50090.156.slope(-500)]))<2 and ([6310]=9 OR [6310]=14 OR [6310]=15)=0

PE: [2517.161.avg(1500)]=0 and [2517.161.avg(-200)]=0 and (abs([50090.156.slope(250)])+abs([50090.156.slope(-250)]))<2 and [6310.avg(250)]=9


*Notes
You will need to substitute the correct PIDs for your setup (wideband and anything else, your ECM may be different than mine).

[Alvin]

Great work.

Please note I don't mean to discourage you when I say this.

This is a problem of solving for 2 unknowns with only 1 function. Without a description of how the two unknowns are related... its just not possible..

Dynamic airflow = Somefunction(MAF, VVE)

Maybe the best example I can give is like the classic elementary math problem.

2 trains leave the station at 5PM on a 50 mi stretch of rail.

When do they collide?

Without knowing train A is 5 mph faster, slower, or even just the same speed we cannot solve.

_________________________

So without knowing dynamic is mostly coming from VVE or MAF, what the weighting is, etc.. We just don't have enough to solve. I still think isolating one variable (VVE) by eliminating the other (MAF) is the best strategy. After VVE is independently a good airflow model you can turn the MAF back on and finish the calibration.

[Cringer]

Great work.

Please note I don't mean to discourage you when I say this.

This is a problem of solving for 2 unknowns with only 1 function. Without a description of how the two unknowns are related... its just not possible..

Dynamic airflow = Somefunction(MAF, VVE)

Maybe the best example I can give is like the classic elementary math problem.

2 trains leave the station at 5PM on a 50 mi stretch of rail.

When do they collide?

Without knowing train A is 5 mph faster, slower, or even just the same speed we cannot solve.

_________________________

So without knowing dynamic is mostly coming from VVE or MAF, what the weighting is, etc.. We just don't have enough to solve. I still think isolating one variable (VVE) by eliminating the other (MAF) is the best strategy. After VVE is independently a good airflow model you can turn the MAF back on and finish the calibration.

I am not sure I follow you...it doesn't matter what the weighting is behind the scenes, or how the ECM arrived at whatever the dynamic airflow is. All we need to know is the DynAir and how far off it is (the fuel trims)...then that is the *correct* airflow regardless of the source. From there we just have to update the MAF and VVE to match the newly identified correct airflow.

[smokeshow]

Someone want to type out the function for me? I don't do youtube videos.

Alvin, I am in agreement about being unable to solve for two unknowns at one time. That's why the 'smokeshow method' (really, its just exploiting some basic arithmetic and physics...not really a method) can be useful...at steady state, its ALL MAF...which as I said in that thread, you should be at steady state. It is not magic. It just assumes the reader follows directions

[Cringer]

Someone want to type out the function for me? I don't do youtube videos.

Alvin, I am in agreement about being unable to solve for two unknowns at one time. That's why the 'smokeshow method' (really, its just exploiting some basic arithmetic and physics...not really a method) can be useful...at steady state, its ALL MAF...which as I said in that thread, you should be at steady state. It is not magic. It just assumes the reader follows directions

Basically:
DynAir + Fuel Trims = Corrected Airflow
Then match MAF and/or VVE to that Corrected Airflow using this method




[My Version for MAF and VVE at the Same Time]


MAF CL
(([2320.71]+([2320.71]*([50116.156]+[50114.156])/100))-[50040.71])/[50040.71]*100

MAF PE* for Lambda
(([2320.71]+([2320.71]*([50119.238]-[50118.238])/[50118.238]))-[50040.71])/[50040.71]*100

MAF PE for AFR
(([2320.71]+([2320.71]*(([50120]-[50121])/[50121]))-[16.71])/[16.71]*100


VVE CL
(([2320.71]+([2320.71]*([50116.156]+[50114.156])/100))-[2311.71])/[2311.71]*100

VVE PE* for Lambda
(([2320.71]+([2320.71]*([50119.238]-[50118.238])/[50118.238]))-[2311.71])/[2311.71]*100

VVE PE for AFR
(([2320.71]+([2320.71]*(([50120]-[50121])/[50121]))-[2311.71])/[2311.71]*100



*[50119.238] is the PID for my WB. You will need to substitute the correct PID for your setup.

[Cringer]

Someone want to type out the function for me? I don't do youtube videos.

Alvin, I am in agreement about being unable to solve for two unknowns at one time. That's why the 'smokeshow method' (really, its just exploiting some basic arithmetic and physics...not really a method) can be useful...at steady state, its ALL MAF...which as I said in that thread, you should be at steady state. It is not magic. It just assumes the reader follows directions

Otherwise, you are just applying the fuel trim correction (from the DynAir) back to the MAF, which is not really correct. I agree that in steady state throttle the MAF = DynAir, but in my experience this still takes several logging sessions to get ironed out and in the end still not as accurate as the method above.

Bottom line:
The fuel trims are reporting the error against DynAir. Nothing else.

The MAF and VVE are just inputs, but the ECM is going to do what it wants to do.

[smokeshow]

Dynamic airflow should never be used as a basis for correction... It is the result of a predictive kalman filter.

[JBZ]

Cringer, surprised you didnt add this to this thread: https://forum.hptuners.com/showthrea...ation-for-gen4

a pretty log thread on tuning VVE as well as MAF simultaneously. which of course is Smokeshow's thread.

killer thread BTW. ive been reading it a and trying to wrap my simple mind around it. just getting back in the tuning game since im throwing a new Cam in my CTS-V and did it the old school way last time with MAF then disable half the damn program to run MAF-less and tune VVE.

[Pekka_Perkeles]

Dynamic airflow should never be used as a basis for correction... It is the result of a predictive kalman filter.

Too old to find out where's "thumbs up" button in this forum.

Thank you for clarifying this.

[Cringer]

Dynamic airflow should never be used as a basis for correction... It is the result of a predictive kalman filter.

Why not? This is the airflow that ultimately determines the injector pulse width, right?

Otherwise, what is the DynAir even used for in the ECM?

[smokeshow]

Why not? This is the airflow that ultimately determines the injector pulse width, right?

Otherwise, what is the DynAir even used for in the ECM?

It is used for transients to improve airmass estimation where MAF and VVE were determined to be insufficient to meet requirements.

[Cringer]

where MAF and VVE were determined to be insufficient to meet requirements.

Yes, that sounds like the perfect thing to use to me!

Edit to add: I would also argue the only thing the VVE table exists for is to handle transients (or as a backup to the MAF in case of a bird strike lol). So DynAir should equal VVE.

[Nanserbe]

Is there any reason you can't/shouldn't replace the LTFT+STFT portions with your wideband? Some people in the smokeshow thread seem to do that.

Or does that break the relationship you're trying to correct?

[Cringer]

Is there any reason you can't/shouldn't replace the LTFT+STFT portions with your wideband? Some people in the smokeshow thread seem to do that.

Or does that break the relationship you're trying to correct?

Narrowbands are more accurate. They should always be treated as better quality data than a wideband. But regardless of their accuracy, the narrowband input is what the ECM uses to control CL fueling. So you should be seeing and tuning with the same data as the ECM.

Narrowbands cannot be used during PE, and during that time the ECM is flying blind. So you have no option but to use the wideband in that scenario. Only use the wideband when the ECM is NOT using the narrowbands.

In general the NB and WB should be agreeing with each other, and as a sanity check, you should look at both of them to make sure they are on the same page.

Edit: important to note that you can tweak how the ECM interprets the NB inputs on tables [ECM] 12530 and [ECM] 12531. You can decrease/increase the switching voltage to make the engine run leaner/richer and you will see a corresponding shift in the WB signal. Just something else to keep in mind when comparing NB and WB inputs against each other and trying to reconcile the differences.

[Nanserbe]

When using this method, should DFCO and Catalyst Temp Protections be disabled?

[Cringer]

When using this method, should DFCO and Catalyst Temp Protections be disabled?

Yes

[Alvin]

Edit to add: I would also argue the only thing the VVE table exists for is to handle transients (or as a backup to the MAF in case of a bird strike lol). So DynAir should equal VVE.

Transients.. Which is what the MAF is bad at.. but we need the VVE to be good so transients are good... but we are using the MAF to dial in the VVE ???

This assumption that the fueling is also weighted mostly towards MAF during steady state is just that... an assumption.


All of this said respectfully. I just think this and the other method are flawed and some careful consideration should be made before prescribing this as a "method" or strategy without vetting. By vetting I mean it should be proven on 10's of vehicles at least.. not just one.


VVE in pure speed density mode is so easy to dial in with just a few revisions. Maybe I should make a video on how I do it.


EDIT. Its easy to read what I wrote as an attack. Please don't take it that way, its not meant to be.

[KillboyPowerhead]

Transients.. Which is what the MAF is bad at.. but we need the VVE to be good so transients are good... but we are using the MAF to dial in the VVE ???

This assumption that the fueling is also weighted mostly towards MAF during steady state is just that... an assumption.

Just finished going through a log from last night; at steady state (flat road, cruise control) at times my lambda was showing 1 - 2% rich, at other times at the same airflow/frequency (but higher pressure ratio due to different gear) I saw 7 - 8% lean. Needless to say you can't tune a MAF with such data. The extra lean areas were always in one particular zone too (otherwise everywhere else was within 3%). So I gotta think VVE is definitely an influence at steady state. Decided to tune only the VVE and not touch the MAF in these areas, so I guess I'll see how much of an effect the VVE has next time I hit that zone.

[Cringer]

Transients.. Which is what the MAF is bad at.. but we need the VVE to be good so transients are good... but we are using the MAF to dial in the VVE ???

This assumption that the fueling is also weighted mostly towards MAF during steady state is just that... an assumption.


All of this said respectfully. I just think this and the other method are flawed and some careful consideration should be made before prescribing this as a "method" or strategy without vetting. By vetting I mean it should be proven on 10's of vehicles at least.. not just one.


VVE in pure speed density mode is so easy to dial in with just a few revisions. Maybe I should make a video on how I do it.


EDIT. Its easy to read what I wrote as an attack. Please don't take it that way, its not meant to be.

Alvin, I appreciate your feedback. I don't take any of it as an attack! I want you to share your thoughts and input so we all get better, including me!

The only way my method can be wrong is if there is some other means by which injector pulse is controlled. As far as I am aware, DynAir is the ONLY input that used to directly control injector pulse width. MAF and VVE indirectly control DynAir. They are merely inputs into the Kalman filter.

So unless the ECM ever ignores DynAir and uses some other source, then my method is not correct. But I am not aware of an instance where DynAir is tossed to the curb and raw MAF or VVE is used.


Edit: I also 100% agree that going SD to dial in the VVE is the best way.

[Cringer]

Just finished going through a log from last night; at steady state (flat road, cruise control) at times my lambda was showing 1 - 2% rich, at other times at the same airflow/frequency (but higher pressure ratio due to different gear) I saw 7 - 8% lean. Needless to say you can't tune a MAF with such data. The extra lean areas were always in one particular zone too (otherwise everywhere else was within 3%). So I gotta think VVE is definitely an influence at steady state. Decided to tune only the VVE and not touch the MAF in these areas, so I guess I'll see how much of an effect the VVE has next time I hit that zone.

Can you post the before and after tune as well as the log file?