How should I proceed from a dirty learning view?

After not checking on my infamous stage 2+ (Cobb catted DP, AVO TMIC) tune for a winter's worth of driving, I see the following learning views. Looks like I have knock all over the place and I'm not sure if AF learning is acceptable. Dynamic advance multiplier (or IAM) is 1.

So, what should I do? Do I have some hardware issues? Bad winter gas? Should I clear the learning views and do some logging? I figure I should stay out of WOT to be safe, but am not sure what my learning views are telling me.

Finally, does anyone know if Cobb AccessTuner Race's learning views match up exactly with RomRaider? Why are the RPM and load ranges so ridiculous with Cobb AccessTuner Race?

Thanks in advance!

The a/f learning in the D range is definitely not good as it is leaning it out 8% across the board at ranges above 40g/s. My guess is you're already targeting close to 11.1 at wot so another 8% leaner could be a problem and the cause of some of that knock.

I've found that a/f learning D can be fine for 90% of driving, and then you hit one hill on a highway that you're holding closed loop at 50~g/s and you can watch a/f learning in that range go rapidly negative. I fixed this on mine with some careful logging and adjusting of the manifold relative pressure compensation table. Pretty much all of the intakes I've tried required setting cells in the 0-1.93 columns at 2600-3300 rpms to somewhere between -5% and -8%.

My guess is if you reset the ecu, everything will appear fine for a while until you go through that range that causes a negative dip and you may start seeing some knock again as that D range learns itself leaner and leaner. The rest of your af ranges are fine.

with the fuel trims the way they are, what do your boosted and/or WOT AFR's look like? If they are good, just leave it alone.

With the knock I saw from the learning view, is it smart to go WOT? Although I was thinking about going WOT in order to see if I can hit target boost.

Do a reset to get rid of the -8%. Then go WOT and monitor your AF ratio if you have a wb02 to verify what it is with no correction. Considering it's a tuned car I believe the ratio is going to be close to 11.1 or so. With -8% correction you could be closer to 12 which is too lean imo.

I wouldn't go WOT without resetting the ECU first.

Do a reset to get rid of the -8%. Then go WOT and monitor your AF ratio if you have a wb02 to verify what it is with no correction. Considering it's a tuned car I believe the ratio is going to be close to 11.1 or so. With -8% correction you could be closer to 12 which is too lean imo.

 

I wouldn't go WOT without resetting the ECU first.

Unfortunately, I don't have a wideband.

So, I think I should proceed by:

• resetting the ECU
  
• logging some part throttle driving and checking learning views, and also when knock occurs (if it does)
  
• depending on if the previous step came back clean, do a couple of WOT pulls and check the logs
  

In the meantime, I'll make sure all connections are secure between TMIC and compressor/throttle body hose. If I don't find anything, I may have to source the parts to do a pressure test (never did one before).

Sound like an ok plan?

Well, I did do an ECU reset and drove around a bit. I did a couple of WOT logs and pulled the learning views after about 50 miles of driving. Attached are my learning views.

It seems that again 40+ g/s AF ratio is learning to go lean again, although I'll have to keep an eye out after further learning.

I'm also worried about the learned knock appearing after so little driving. My logs show some knock in the low rpm range. Does this have to do with me jumping on the throttle instead of easing in?

I'm hitting target boost (~18 psi) so it doesn't appear that I have any major leaks. Should I still test for leaks in the intake?

Any ideas on how to proceed? I'm in the midst of communicating with Infamous1, but he is quite busy also so I want to see if some others experienced with tuning can chime in. Thanks.

datalog8.csv

datalog9.csv

The -.5 isn't too big a deal but I don't doubt that it will continue to go negative if part of the scaling is off either on the maf table or the mrp compensation table. If you're able to, log the stored D af learning as well as MAF g/s and Manifold relative pressure. Scroll through the log and when you see the stored D start to go negative, check and see what the MAF g/s is and what the MRP is. If it always goes negative in the same MAF/rpm range then you might need to adjust the maf scale a bit. If it's always doing it in the same MRP/rpm range then you would adjust the mrp compensation table.

The knock isn't good. I'm sure if you're working with Infamous he will address it in future revisions. I'm not sure that most tuners really get into scaling the MRP table though which as I mentioned was clearly the issue with mine slowly finding the D range dropping into the -5 to -8% area, which I think is more an issues with Canbus ecus.

The -.5 isn't too big a deal but I don't doubt that it will continue to go negative if part of the scaling is off either on the maf table or the mrp compensation table. If you're able to, log the stored D af learning as well as MAF g/s and Manifold relative pressure. Scroll through the log and when you see the stored D start to go negative, check and see what the MAF g/s is and what the MRP is. If it always goes negative in the same MAF/rpm range then you might need to adjust the maf scale a bit. If it's always doing it in the same MRP/rpm range then you would adjust the mrp compensation table.

 

The knock isn't good. I'm sure if you're working with Infamous he will address it in future revisions. I'm not sure that most tuners really get into scaling the MRP table though which as I mentioned was clearly the issue with mine slowly finding the D range dropping into the -5 to -8% area, which I think is more an issues with Canbus ecus.

Learning view is starting to go leaner. I've attached a log showing af learning d going more negative (it really drops quickly at 388 sec). Thing is, I'm using the stock airbox, and always have been. Maybe I need to check how the air filter is, and maybe clean the MAF sensor? Do stock intakes need MAF calibration or MRP adjustments?

My car was tuned by infamous1 last year, but I never pulled any learning views at that time (he only requests logs of WOT pulls from 2k-redline). So, for all I know, I could have had the problem with af learning then too.

Any help is appreciated. I'm gonna start to get the needed supplies to do an intake pressure test.

datalog1.csv

Looks very similar to what was going on with my car. I had this issue on the stock intake as well as every other intake I've tried. With the stock intake it didn't require as much correction though. In this case, either your maf scaling around 50 needs to drop a few percent, or your mrp around 2600rpm at 14.7 (0 boost) needs to have a couple percent of negative value to account for it.

rtbrjason, thanks for your help so far. I'm trying to see what infamous1 has to say. If I have more questions regarding either MAF or MRP scaling, I'll let you know.

^^^ wrong

For some reason your car is running rich at 40+ g/s. Then the ECU is pulling fuel, and of course learning it over time.

It is sometimes a bit difficult to tune the 40+ g/s range, especially if tuning remotely.

INfamous is more then capable of tuning this. it could be an air leak somewhere, but they tend to manifest themselves more in the lower learning ranges.

Assuming your air-filter isn't real dirty, and neither is your MAF sensor, just follow his instructions to scale the MAF better. I like to add back in the CL/OL delays as it makes it easier to scale the 40-80 g/s range.

FWIW I have seen this twice lately with the COBB airbox on 02-03 WRXs, suppose to use stock scalar, but trim D goes to about -8% or so. I have even seen -10% on trim D w/ COBBs OTS AEM intake tune.

From the looks of your post-ECU reset LV there is still some tweaking to do on the tune, as w/o the 8% off WOT fueling you are still getting some knocking.

FWIW ATR LV and normal LV are the same, just different ways of reading the data. Subaru on some cars uses some crazy values for rpm and load in the FLKC table, I personally change them.

I was looking at the first pic on my phone... the -8 looked like a 8

not a dead fuel pump

Infamous1 says he does not adjust MAF calibration on a stock intake. He says the AF learning is due to either climate change or winter gas. I can't argue with either, but that essentially means that I need a winter tune.

rtbrjason,

Can you tell me how the manifold relative pressure table works? I understand how to determine what cells to alter, but what does the z-axis (engine load compensation) mean?

The reason why I think I should adjust the manifold relative pressure instead of the MAF scaling, is because in my previous log, I was in the ~50 g/s range in multiple gears and rpm ranges, but only when in 5th gear at ~0 psi boost, did it start to learn lean.

Of course, let me know if I am off base in my thinking.

Infamous1 says he does not adjust MAF calibration on a stock intake. He says the AF learning is due to either climate change or winter gas. I can't argue with either, but that essentially means that I need a winter tune.

Well it is off. 8% is too much, period. I know all about winter gas, and E0 to E10 only changes things by about 4%.

I honestly believe that (assuming nothing is leaking anywhere) there is some point in the MAF that is scaled poorly (from the factory) and/or one of the MAF comp tables needs to be changed.

Regardless, you don't need two tunes, you need one good one. I do think that your tuner should be paid properly for his service, but if it were me I would tune the intake perfectly at 70F weather (where the comps are all 0), and then watch for changes in different weather, and change the comp tables from there.

Infamous1 says he does not adjust MAF calibration on a stock intake. He says the AF learning is due to either climate change or winter gas. I can't argue with either, but that essentially means that I need a winter tune.

 

rtbrjason,

 

Can you tell me how the manifold relative pressure table works? I understand how to determine what cells to alter, but what does the z-axis (engine load compensation) mean?

 

The reason why I think I should adjust the manifold relative pressure instead of the MAF scaling, is because in my previous log, I was in the ~50 g/s range in multiple gears and rpm ranges, but only when in 5th gear at ~0 psi boost, did it start to learn lean.

 

Of course, let me know if I am off base in my thinking.

It's not the manifold relative pressure table itself. That is your boost table.

If you look in your maf section you should see manifold relative pressure compensation. It is by RPM and Manifold relative pressure from -10 up to about 1-2 psi of boost. This is the table I used after rescaling my maf to solve all my maf issues. Like you, even with a well tuned maf I was seeing my D range go slowly more and more negative up to -5% to -8% (depending on intake used). I think this is more common to canbus as folks with 05-06 lgts don't seem to see this issue to the same degree.

If you look at my table above, you can see that around 0 boost and 2600-3000 rpms I have some negative compensation. That is the problem area as you can be in closed loop with steady throttle in the 50-60gs range here and watch the D range tank despite the 50-60g/s being otherwise dialed in.

This is the sheet I use to dial my MRP table in. I had to rescale my table a bit to make sure it was going into the 3600rpm range as that is about as high of rpm as you can maintain closed loop for.

This is another alternative filtering tool from Airboy.

I don't think it is that table. I have never had a problem tuning these cars w/o touching that table (that table is, IMO, more for weird stumble type problems, over-simplifying).

I tune our local cars in +100 and -40 weather, and every single guy I tune knows how to check LV, and I do not have that problem. But I do see it lots on some car with stock scaling.