New to EFILive, Help increasing AFR

[joecar]

Your LTFT's are mostly zero or slightly negative, and your NGK indicates 12.4 at WOT, and log indicates it's hauling.


I'm wondering why the NGK drops to 9.05 when throttle is released...


Let me have a closer look later today.

[joecar]

Do you have two successive .tun files...?

[sonicgroove]

Your LTFT's are mostly zero or slightly negative, and your NGK indicates 12.4 at WOT, and log indicates it's hauling.


I'm wondering why the NGK drops to 9.05 when throttle is released...


Let me have a closer look later today.

Thanks joe that would be great. Btw, why is it in the beginning of my log ltft are so high, then later begin to drop and vary within that ideal +-5 range...?

I know my knowledge on tuning is incredibly dull and broken but from what I currently understand if wo2lam readings are under 1.0 it means my wbo2 sensor is reading rich of stoich? So for example in my log using 0.9883 of the ngks stoich value 14.57 my afr would be 14.3995 yet in this same area longft1/2% are reading positve of 3.9 correlating with my ltftben being 1.039 hence my calc.vet wanting to richen up? Isn't this kind of backwards? If my wo2lam is under 1.0 my ltft should be negative along with ltftben so calc.vet will slightly subtract from the ve table? Or is it inaccurate to do this comparison with ltft and more appropriate to log stft to check if ltfts are calculating correctly for my ve table?

And just to clarify, wblam does not affect daily driving for calc.vet. It is only WOT or when in PE (cells usually above 85/90kpa) that wb readings are used to adjust calc.vet?

If what understand is correct. Wouldn't it be far easier or at least more accurate to have wbo2 perform calc vet rather than ltft for the first few tunes then use a series of ltfts calc.vets followed by adjust via stft to fine tune?

-Shaun

[joecar]

LTFT's seem to go high while idling.

Yes, Lambda < 1 indicates rich.

In CL, the WB should be indicating Lambda 1.00 (or very close to it) as log as the LTFT's are not pegging.

Yes, if you look at how CALC.SELBEN is defined, it uses LTFTBEN in CL, and WO2LAM1 otherwise (non-CL which includes PE/WOT).

You could disable CL/trimming (see B3801, B4205), and use WO2BEN only (and doing the same procedure as Calc.VET, i.e. correcting MAF and calculating VE).

[sonicgroove]

I would like to give that a try and see if I get more accurate afr/lamba readings from my wb and do a comparison between the calc.vet numbers. So essentially I'll be forcing PCM into OL. Do I need to adjust my calc.pids accordingly or just disabling cl/trimming I could use the same pid setup to perform calc.vet?

Btw, I have some past tune files but I was more dense and had less of an idea of what to do compared to now. So using them to determine how I tune probably won't be as accurate compared to if I were to create a new log and tune file for you to compare. I can easily do this for you if you wish and think it would help.

I also do think I need to fix my driving techniques a bit.

-Shaun

[joecar]

In the calc_pids.txt file, edit the CLC definition of CALC.CL to read "0" (i.e. never CL)... and use the same pids (CALC.SELBEN and CALC.VET) in the maps.

Also, LA fuel might contain 5%-10% alcohol, so try setting B3601 to 14.3 or 14.4.

[joecar]

Take note: when in CL, and LTFT's are not pegging, then wideband should read lamdba 1.00... if not the there is some offset in the wideband value.

[sonicgroove]

In the calc_pids.txt file, edit the CLC definition of CALC.CL to read "0" (i.e. never CL)... and use the same pids (CALC.SELBEN and CALC.VET) in the maps.

Also, LA fuel might contain 5%-10% alcohol, so try setting B3601 to 14.3 or 14.4.

Hey Joe,

I did the following so far to force openloop but am unsure how to "In the calc_pids.txt file, edit the CLC definition of CALC.CL to read "0" (i.e. never CL)..." so I can calc.vet based of wb02. Do I simply change "{GM.EQIVRATIO}" to = 0?

Instead of

*CLC-00-032
factor 0 4 .0 "{GM.EQIVRATIO}=1"

to

*CLC-00-032
factor 0 4 .0 "{GM.EQIVRATIO}=0"

B3601 Stoich Changed from 14.63 to 14.3 for California 91 Octane Fuel
B3801 Changed from Enabled to Disabled
B4206 Changed from Disabled to Enabled

After this I could just log and perform a calc.vet how I did previously, correct? (Using all the same pids if I were doing narrowband ltft calc.vet)

-Shaun

[sonicgroove]

In anycase that is what I did, haha. I kind of don't think I did it properly because I was still running in closed loop per the log. Still seeing CL of 1 but I pushed on anyhow and made the calc.vet changes. When I checked out the log, I have to say, the raw data looks pretty good. The changes appear to be something I would expect. However, once I applied the filters... all hell breaks loose. The data gets skewed to a rich tune. I am thinking this may be due to the fact that the filter excludes any data below 1% throttle so it throws out my deceleration data. That said I changed the filter from 1% to 0% and reapplied. Not much changed from the raw data. I want to try using this mildly filtered data because the changes do look to be in the direction I THINK the tune should be going. What makes me feel more confident is the fact that my selbens has soooo many 1.00. More than I have ever seen. It leads me to believe the raw data mildly filtered is accurate. I am going copy and paste the calc.vet map with minimal changes to my VE table then change PCM settings back to normal and do a log to check out ltft and my wb02 readings. If it all checks out then I'll be doing it once more but this time hitting all cells as much as I can.

On a side note, since I changed 3601 to 14.3 I decided to change commanded fuel to 1.13 to maintain an afr of 12.65 during PE/WOT.

Here are the files you requested.
Initial Tune
Attachment 14690

Corrected/Calculated Tune
Attachment 14691

Data Log
Attachment 14692

Omg... I think I just had an epiphany... stoich of california gas is likely 14.1-14.4 (14.13 to be more accurate) because in california since 2010 most gas stations supply e10 or fuel with 5-10% ethanol (Found this out after research). I think this is now mandated since 2012. Btw Joe, the research was prompted after you told me to change 3601 to 14.3/14.4 and I thank you for this.) Since there is a discrepancy between lambda stoich of gas vs lambda stoich of the ngk afx I will almost never see a lambda value of 1 in my data logs if tuned correctly. 14.3/14.57 = 0.9815V. If this is true, it makes sense why my wblam values are consistently in the .97-99 range. This also clarifies why my calc.vet tunes have been telling me to add more air in my VE tables. Because my LTFT are trying to get stoich of pump gas which is 14.3. Seeing that 14.3 was not 14.57 or the 3601 14.63 on my ngk afr display box I thought something was wrong but apparently not. What was wrong was my understanding of how tuning for stoich works. This really made clear the disadvantages of seeing AFR values versus lambda/eqr. Please correct me if I am wrong I am really trying to understand this stuff, haha. If this new found knowledge is correct, my brain is begging clarification of 3601 and LTFT. When PCM is in normal operating closed loop, do the ltft's adjust towards 3601 which is suppose to represent stoich of current fuel? So if ve tables are tuned for lamba 1 (or 0.98=14.3/14.57 in my case) of current fuel (approximately e10) then ltft should be close to 0 correct? Now... if I did not have a WB and I performed calc.vet with a factory 3601 value of 14.63, the narrowband ltft based calc.vet would actually take air away because the pcm thinks the car is running rich because stoich of gas is 14.3 so fuel trims likely be negative to reach factory 3601 14.63. This in reality would be lean. God I hope this is right...

This probably does not explain why I am going rich on decel. I think this may be related to the map filter I mentioned above. Or so I hope...

Sorry for jumping all over the place but the order of what I wrote chronological describes my thought process at the time, haha.

-Shaun

[joecar]

Hey Joe,

I did the following so far to force openloop but am unsure how to "In the calc_pids.txt file, edit the CLC definition of CALC.CL to read "0" (i.e. never CL)..." so I can calc.vet based of wb02. Do I simply change "{GM.EQIVRATIO}" to = 0?

Instead of

*CLC-00-032
factor 0 4 .0 "{GM.EQIVRATIO}=1"

to

*CLC-00-032
factor 0 4 .0 "{GM.EQIVRATIO}=0"

B3601 Stoich Changed from 14.63 to 14.3 for California 91 Octane Fuel
B3801 Changed from Enabled to Disabled
B4206 Changed from Disabled to Enabled

After this I could just log and perform a calc.vet how I did previously, correct? (Using all the same pids if I were doing narrowband ltft calc.vet)

-Shaun

Hi Shaun,

I actually meant this:

*CLC-00-032
factor      0       4      .0           "0"

this will cause the SELBEN pid to always use WO2BEN (which is the correction pid from your wideband).

You will also have to disable trimming/LTFT/STFT/SOL, like this:
- set B3801 to disable,
- set B4205 to max temp,
- set B4206 to disable (if not already).

Yes, around So.Cal, 91 contains 5-10% alcohol, so setting B3601 to 14.3-14.4 is a reasonable compromise.

Yes, correct, just perform the Calc.VET procedure as usual.