Cobb BIG SF intake fuel trim issues

[SeeeeeYa]

Originally Posted by anjuna

AFR will always look good out of boost. It's in closed loop, so it's hitting the target.

You won't see MAF V correction - that's not a thing.

You'll want to monitor your A/F Correction and Learning related parameters. These will show how much the ECU is changing the amount of fuel injected from what the tune is telling it, to what the actual amount of fuel is required.

The tune does it's calculations based off the MAF. The MAF says "hey, there's this much air coming in" and the ECU says "okay, let me look up how much fuel I should inject. Here we go!" and then the oxygen sensor comes back and says "hey bro, you're like, 9% lean" so the ECU says "well then I should add 9% fuel" and that's where your A/F correction comes from.

A/F learning is like a trust issue between the MAF and the ECU. It's getting tired of hearing that it's wrong from the oxygen sensor, so it learns to just ALWAYS add X percent (which is an average value over time). So then, it's just taking the MAF value, adding or subtracting the learned value, and then making closed loop corrections after that from the oxygen sensor.

You would make a great teacher. ^^^

I'll add...

And you "hope," unless you have properly determined it, that that global AFR Learning correction is correct... unless you KNOW by having tuned your ENTIRE MAF Calibration curve... the that learned value is on your side. If that correction provides more fuel than necessary... it's not so bad, only your power will suffer. But if the correction is negative, and the global correction removes fuel... because your MAF Cal at some point in open loop is off in a negative direction, by even a small amount... that's when the engine will suffer. Repeated long enough and pistons and ringlands will let you know how much.

The point is, tuning... including the MAF Cal, is important to the long-term reliability of your engine. OTS tunes are NOT so calibrated. Cheap tuners with cheap tunes don't provide it either. Get it done right, or ... roll the dice.

If it were as simple as ABC... there would be no problem, no need for TUNING. All you would need is to list your mods and get a cookie cutter tune that was perfect. But everyone knows that has gone through a tuning process that that is not the way it is. Why then believe your intake is immune to that reality.

If you keep it simple, and stay inside the envelope with a modest tune like a stage1, you can usually get away with it. But when you go higher, not so much.

* Registered users of the site do not see these ads.

[simpleJ]

I made another scale tonight. I logged 45 minutes of mixed driving before my vacation. I have an hour commute tomorrow and I hope this new scale will be dialed in nicely after my longest logs and 4th revision.

Basically I am seeing what I believe is turbulence in the intake around 1.8v and 3.0v right before we make the open loop switch. Nice bumps in these regions on the cal...

I'm having fun with this.

[SeeeeeYa]

Originally Posted by simpleJ

I made another scale tonight. I logged 45 minutes of mixed driving before my vacation. I have an hour commute tomorrow and I hope this new scale will be dialed in nicely after my longest logs and 4th revision.

Basically I am seeing what I believe is turbulence in the intake around 1.8v and 3.0v right before we make the open loop switch. Nice bumps in these regions on the cal...

I'm having fun with this.

What you're doing is important, however you approach it. The MAF Cal IS THE MOST IMPORTANT REFERENCE in the ECU.

Typically, however, it is dialed in using wideband oxygen sensor data. Having not had experience with the FA, which has a WBO2, I am not certain how its data can be precisely correlated with the fuel table. I have always had a dedicated WBO2 with which that correlation is direct and obvious. Comparing what the precise cell-by-cell requested fuel is with what the actual WBO2 says is there is how one knows which direction and how much to alter the MAF Cal.

Any other approach is like using smoke to find a fire rather than looking for the flames. You can get there... sooner or later, but will often be left going in circles.
There are other inputs that contribute to refinement of the MAF Calibration... theoretical guidelines. Working from both mathematical ideals and actual results provides a referencing overview that guides the process and keeps it on track.

Keep at it. The more you know and the more you apply that knowledge the better your tuning.

[simpleJ]

I ended up finding a lovely tool to help scale things up using AFR errors in open loop, and compares the fuel look up tables to massive closed loop data logs to help you generate a baseline scale for closed loop fueling. From there it is basically inputting and then tightening the scale each time to come to a dialed in cal.

Now, where I am running into trouble is around that 1.8v mark, which is where our cars tend to run when cruising, fueling will wander up and down depending on the load. My scaling tool is adding too much fuel in that range. So I keep bringing it incrementally closer to where it needs to be.

Get a bit above that and A bit below am in the +-4% range.

Last night I moved to the v301 map which could explain why my scales are off a little more than expected today. I took like a half hour-ish log of traffic and freeway driving at varied speed.

Back at it tonight. I'll post comparison spread sheets of the cals when I finish

[Perscitus]

Originally Posted by simpleJ

I ended up finding a lovely tool to help scale things up using AFR errors in open loop, and compares the fuel look up tables to massive closed loop data logs to help you generate a baseline scale for closed loop fueling. From there it is basically inputting and then tightening the scale each time to come to a dialed in cal.

Thanks for the update, can you share the tool or a link to it, threads about it?

I'm using the MAF Scaling (and much more) tool found below. Its based on many spreadsheets, and previous tools built by AirBoy and others over the years.

https://github.com/vimsh/mafscaling/releases
https://github.com/vimsh/mafscaling/wiki

Its great for CL, OL ranges, re-scaling (adding resolution where you need it, cropping the curve to a voltage range that makes sense for your application) and fine tuning Load Comp MP, IAT Comp, WOT VVT tweaking, etc.

[anjuna]

Originally Posted by simpleJ

I ended up finding a lovely tool to help scale things up using AFR errors in open loop, and compares the fuel look up tables to massive closed loop data logs to help you generate a baseline scale for closed loop fueling. From there it is basically inputting and then tightening the scale each time to come to a dialed in cal.

Now, where I am running into trouble is around that 1.8v mark, which is where our cars tend to run when cruising, fueling will wander up and down depending on the load. My scaling tool is adding too much fuel in that range. So I keep bringing it incrementally closer to where it needs to be.

Get a bit above that and A bit below am in the +-4% range.

Last night I moved to the v301 map which could explain why my scales are off a little more than expected today. I took like a half hour-ish log of traffic and freeway driving at varied speed.

Back at it tonight. I'll post comparison spread sheets of the cals when I finish

Be careful with your method.

If you're doing some sort of "target vs actual" type method for determining error, you're gonna have a bad time when you lift throttle (target remains 14.7 and actual goes to 22, resulting in an huge positive fueling error)

If you're working with your AF correction in closed loop, be careful as this goes to 0 in all your open loop stuff (WOT).

I've had good success in sorting by AF ratio and cutting all the stuff that is off throttle. Then you don't end up cutting idle data too.

[simpleJ]

Yes, all open loop is clipped for the closed loop scaling (when you do a quick lift off of throttle for example the car switches to open loop)

Open loop will use comm
Fuel, AFR, and AFR error to
Determine a scale

[simpleJ]

Some clean up still to do in the 1.7-1.9 volt range as well as in the 2.8v range (see data log-big lean spike coming into boost- what is with that???)

But we are getting damn close here boys...
http://www.datazap.me/u/simplej/log-...&data=3-4-7-15

[Nuked]

Some kind of throttle plate delta enrichment/compensation?

[anjuna]

Originally Posted by simpleJ

Some clean up still to do in the 1.7-1.9 volt range as well as in the 2.8v range (see data log-big lean spike coming into boost- what is with that???)

But we are getting damn close here boys...
http://www.datazap.me/u/simplej/log-...&data=3-4-7-15

Remember, when tuning AVCS, you can increase spool by increasing "blow through" in which some intake charge is passed through. This looks "lean" as it is charge air and/or unburnt fuel (yes, unburnt fuel shows as "lean" on a wideband) so this may or may not be bad.

Originally Posted by Nuked

Some kind of throttle plate delta enrichment/compensation?

The word you're looking for is "tip in enrichment" which shouldn't matter for a WOT run.

[simpleJ]

Alright, so I am pretty sure I have finished, I will make an excel comparison of the scales shortly and post them up. Anything left is just clean up. Occasionally I will see a very brief funky jump to +-10 while but it generally stays between +-4. probably still settling in from this afternoon's flash.

Here is open loop:

http://www.datazap.me/u/simplej/log-....93&tmax=15.72

If i can get a chance to take back to back to back logs one day I will to clean up open loop just a tad more we should be good.

This whole process was very very easy and I would recommend anyone having trouble with their maps gives it a shot.
There are some things on the scale- mainly a big old bump- that are perplexing. It is like there are 2 separate scales matched together, the tipping point around 2.8v's

Car also underboosts ALOT- may have to try the HWG map. Or maybe I should install and tune my own EBCS....

[Nuked]

Originally Posted by simpleJ

Car also underboosts ALOT- may have to try the HWG map. Or maybe I should install and tune my own EBCS....

Just tune your boost with the stock setup.

[Catoboli]

Hey Simple, This was a great read through.
I Have the same Cobb Intake installed and waiting till it warms up a bit more to throw on some other mods. Ive been datalogging pretty religiously since I got the car and I havent had as sever of an issue with my AFRs in OL, however I do have the crazy AF learned values, mine are typically around -10 (im assuming thats from Cobb OTS maps running rich)

My big question for you is how are you editing your maps & scaling as I would love to clean mine up a bit.

[Catoboli]

Also thanks so much for turning me on to DataZap, tiz the greatest thing ever.... well almost.

[Catoboli]

Ive attached one of my more recent datalogs, It looks like the OTS is working pretty well for me. Just waiting for a few more bolt ons before I get a protune.

http://www.datazap.me/u/catoboli/log...ta=3-5-7-22-24

[simpleJ]

Jesus! Your boost plots look just like mine but you're VERY rich. What's your AF learns at when cruising?

[simpleJ]

Okay So I am finished, or very close to it.

Here is an open loop log in 3rd: http://www.datazap.me/u/simplej/log-...17&zoom=61-402

So to scale it, I used the Vimsh tool to get everything dialed in. this was done over the course of 15 different rev's. Though it would likely take me less using what I had learned. Now the use of the tool is detailed on the rom raider forums so instead I will talk strategy.

My first step was to use some of my multiple previous OL data logs to build a safe scale- with excess fueling up top- which was done automatically by the tool- I built closed loop off of this. My reasoning for this was to give me overhead if I needed WOT in traffic without blowing the motor.

From there I just logged my daily driving and produced CL scales using the tool- honing in each time. Towards the end of the process I began to smooth and manually adjust the scale to bring it in to alignment as it should be, in a curve. thus eliminating some bumpy scales resulting from some strange AFR errors or temps. Once you get CL pretty dialed- back to OL fueling. Now, as anjuna pointed out there are some strange phenomenon in the log that will throw your scales off- likely do to blow throw (scavenging strategies) to help you boost from low rpm. Ignore it- you'll end up with a lumpy scale this is where you need to bring things into alignment. Overall a very easy process.

Here are the scales for those of you to see the differences: though I would like update/smooth things out a bit more- this will be close to what I am sticking with.

[simpleJ]

Also those of you keen on an running an intake without a tune can see why this is a bad idea above, the X axis is the MAF reading while the Y represents the look up value for G/S of airflow based on the MAF reading- the difference is huge

So basically if you ran a cobb intake without a tune it could get 30% more air flow than is accounted for by the ecu.

[Changtownriot]

Originally Posted by simpleJ

Also those of you keen on an running an intake without a tune can see why this is a bad idea above, the X axis is the MAF reading while the Y represents the look up value for G/S of airflow based on the MAF reading- the difference is huge

So basically if you ran a cobb intake without a tune it could get 30% more air flow than is accounted for by the ecu.

Are the cobb ots tunes ok or would you recommend a protune?

[simpleJ]

It depends- some peoples cars run great with the OTS maps, some do not. A protune would eliminate those variations

[Catoboli]

Originally Posted by simpleJ

Jesus! Your boost plots look just like mine but you're VERY rich. What's your AF learns at when cruising?

My AF Learns are very negative, the issue that got me here. I would say the typical while cruising is around -7.5% , idle is usually -13%. Ive had trips where the AF learn has hit -19%. on warmer days I typically range -7%--1% for a trip. I almost never see positive AF learns.

Ive definitely noticed more negative Learns on colder days. (Seems odd to me).

[Catoboli]

Im definitely looking to get a Pro tune im just trying to hold off till I get the rest of my bolt on mods. (TGVD, EGRD, TBE, FMIC, maybe Exhaust Manifold).

Id rather do the tuning myself tho.

[Catoboli]

So interesting turn of events, I threw on a nameless muffler delete and now my AF learns are close to 0 when driving, still negative a bit at idle. I wouldnt have expected it to change the AF learns quite that much if at all.

Its so interesting to see how differnetly this car reacts to things. The last car I modded was a 1995 Mitsu Eclipse GSX.

[simpleJ]

That's kind of crazy- I do have a cbe on my car, I wonder if that's why it needs to much fuel- that is kind of crazy

[Catoboli]

Originally Posted by Catoboli

So interesting turn of events, I threw on a nameless muffler delete and now my AF learns are close to 0 when driving, still negative a bit at idle. I wouldnt have expected it to change the AF learns quite that much if at all.

Its so interesting to see how differnetly this car reacts to things. The last car I modded was a 1995 Mitsu Eclipse GSX.

Update to this. So this was an error on my part. I noticed when I took the car for the first ride I had a weird knock event, happened only once definitely an anomaly. however due to it it looks like it reset all my learned tables hence the 0 learned A/Fs for awhile, its not back to normal if not a little worse actually.