barely changed ve & went from rich condition to super lean?

[Imdominant]

Ive never posted in the 'tuning' section before so hello to all , i need some help with what i think could be a tuning problem.

I am in the process of doing a little naturally aspirated tuning to get my <100kpa values correct before adding in boost. I was previously having a "rpm wall" problem but got that completely sorted out. The car has been revving to where i need it to (and does a good job of doing so), so i go make a log... Its a street car that will see drag racing duty so i am basically making "passes" meaning the ve travels in pretty much the same path every time, repeatable, with little variable. Anyway so i make a baseline run and afr reads a bit rich (10.0 ish) so i did a run analysis in megalogviewer and it brought back those ve cells a little.. when i say a little i mean from a ve of around 80 to 76. so i save that tune and go try again hoping to work it down slowly....

Heres what happened.. the car would barely rev as if something were wrong.. i had made a datalog of this and afr is now very lean like off of the charts lean. (22+)

What do you guys make of this? please keep in mind i previously could make a full hard pass (if it didnt SAY 10.0 afr you would never have thought it) , but a small ve change leaned it out that far? i dont think so - and i cant have these sort of troubles when i ultimately put boost on (make a small ve change and it runs 22+ lean afr)

please advise!

[subwoofer]

Rich misfire... Too much fuel for it to burn properly.

IMHO, 10.0AFR isn't "slightly rich", more like pig rich.

[Imdominant]

Well, yea i consider it pig rich too - and trying to work it from a 10 afr to a 12.6-12.9 seemed easy enough but it went super lean with a small ve change. do you think it should have shifted that far to be that super lean? what sort of a ve cell change would you think it SHOULD take to go from 10 to 11 or, 10 to 12 , or even 10 to 13?

Is the run analysis feature typically this inaccurate? I started to work them down by hand but i figured that first jump could be handled in one swoop just by using run analyzer.. am i wrong in thinking this?

[subwoofer]

If it misfires, it reads dead lean no matter the reason. Take out 15-20% fuel across the board and work towards your target from there.

[mschlang]

You were running about 20% rich, but only took out 5% fuel. If your original ve was 70, your new should be around 58. Never added this, but here goes:

[dontz125]

If it misfires, it reads dead lean no matter the reason.

Just to expand - the O2 sensor reads (oddly enough) oxygen. (Who knew?! )

In the event of a misfire, (pretty much) none of the fuel is burned, (darned near) all the O2 remains, and your sensor reports that you are suddenly extremely lean.

[billr]

Also, O2 sensor systems can be wrong, for a variety of reasons. Pay attention to the plug coloring and how the engine runs.

[Imdominant]

Ok sorry guys , ive been having pretty good luck but have been mostly manual tuning.. Since having fixed my previous "rpm wall" , ive figured out my entire tune is acting different. SMALL changes to VE now make a pretty big impact on AFR readings (observed during logs)..

Before, i could move a ve value a whole number (example 83.0 to 84.0) and this would correspond to a very slight change in afr... that size change may would change it from a 12.7 to a 12.5) , but NOW a change like that would go from 12.7 to 11.7... so without knowing exactly why it is just way more sensitive. this makes for very overcompensated adjustments when using the "run analyzer" feature which i personally have used between autotune sessions to hone it in and bring it to where it needs to be quicker. when it didnt do this and took me from pig rich to 22.0 lean , i thought to just pass it by you guys...

All it boils down to is if i make smaller adjustments and observe the changes i will be fine. It does make me wonder if i should back up and put injector data into mstuner so all of the algorithms will have a chance to function properly...

[Imdominant]

newscheme4.msq

2nd.msl

[Imdominant]

Also, O2 sensor systems can be wrong, for a variety of reasons. Pay attention to the plug coloring and how the engine runs.

Bill how do you suggest that i figure this out reliably enough to trust tuning a TURBO engine ? I would like for the afr that is reported to be correct even if i have to skew it somehow to do so

[subwoofer]

Again:

Measure your injectors, make sure size is appropriate for the job at hand. Run sequential if you can and start over. With all data correct and the default tune in place you should be in the ballpark, fuel wise, within an hour or so on the street if you use VEAL.
Ignition really needs a dyno.

[billr]

Alas, I know of no magic. My belief is that a dyno is the only way to tune and engine for optimum power, longevity, and economy; the best compromise of those for your particular requirements. My point was that the WB system is fallible, so keep that in mind if you don't want to damage an engine while squeezing it for power. My understanding is that the color and surface appearance of the spark-plug insulator is the simplest and most reliable way to tell how things are going on in the combustion chamber. Maybe even a dyno doesn't eliminate the need to "read the plugs".

[Imdominant]

FYI, i attached a MSQ and a datalog of how it was doing as of a couple days ago... Like i said the ve map is just WAY more sensitive now... smaller changes yield bigger results. I cant use ego and i havent been using veal either , and i certainly cant use the "run analysis" feature in mlv.

I am contemplating an injector flow test , but if i did it it would have to be done on car and my hoses are a bit tight for all of that. im still trying to work out a way to do it.

[elaw]

It's a bit hard to tell much from that datalog because it's so short. Two things that would be more useful would be a log showing the problem as you drive, and/or a log where the engine is at a steady-state condition (like idle) and while logging, you vary the VE number up and down slowly so we can see its effect on PW and AFR.

Have you looked at your injector deadtime vs. voltage? The value of 0.9 in the deadtime box looks suspiciously like the default, and the deadtime-vs-voltage curve in your tune is just plain weird. For the most part it's straight where usually it's more like an exponential curve.

An incorrect deadtime setting could explain what you're seeing, or at least part of it.

Also another thought: is your wideband O2 sensor calibrated properly? If it has a display, does the displayed value match what you see in Tunerstudio and the log?

[Imdominant]

I intend on building a jig and testing some injectors so i can input the correct amounts into the deadtime field. yes i have a figure of .9 in there and the manufacturers spec is .7 i didnt figure this out until recently.... do you think i should enter the manufacturers spec into the field until i can get a rig built for injector testing???????

[elaw]

Manufacturer's spec is definitely better than a random number!

But be aware of two things: 1) the deadtime varies by voltage so not only must the single deadtime number be correct, but the voltage curve must be correct too. And 2) make sure the spec you're looking at is definitely deadtime and not opening time. Those two tend to get confused a lot - deadtime is the difference between opening time and closing time and if you enter just the opening time you won't get correct results.

The test-jig approach is definitely the best though... injector opening and closing times depend on the drive circuit, so the only way to get truly correct numbers is to test with the actual driver circuit you're using.