07 Hawkeye STi with EJ257 built engine - erratic AFR values

Hey Everyone,

So after a good couple of month trial and error I wanted to ask you before I gave up on this issue.

In short, my problem is that the AFR in closed loop is all over the place, but nothing like stoich. It actually fluctuates between ~13 and ~16.5 and it does that randomly (at least to me it seems like).

Now my build is quite far from stock, and the reason might lie somewhere there, so let's start with that.

07 Hawkeye EDM Sti - ECU Calid:A8DK100Z

EJ257 closed deck block, 14mm headstuds - nothing special here

Forged internals with 8.5:1 CR pistons

Cosworth heads (CNC ported) with Cosworth cams and +1mm Supertech valves

LM450 twinscroll turbo with Tomei EL header, 3" exhaust

Front O2 sensor moved to DP just after the turbo

DW850 injectors, Aeromotive FPR, AEM 340 fuel pump

APS 70mm CAI, rotated inlet manifold (TGVs deleted) with FMIC (custom made short pluming)

EGR removed, rear O2 sensor removed

That's probably it as far as the majority of spec goes.

I started with a stock map, disabled all CEL codes that had to be disabled, removed a little timing from stock, and simplified the primary fuel map to have 14.59AFR for light load and 11.47AFR for high loads to simplify injector and MAF scaling. Also removed all boost, so the setup makes only 1Bar MRP on wastegate spring pressure and per injector pulse witdth compensation A, B, C, D - zeroed out.

Scaled the injectors with stock airbox (without resonator), then moved on to MAF scaling (and I am still struggling with that).

Disabled the rear O2 sensor, disabled overrun fuel cut, played with the front O2 smoothing, and although I could get the idle trims to average to 0.x%, it fluctuates between +/-10%, just like light load which is not desired I think.

Now, the Cosworth cams obviously are more aggressive then stock, but they are not that aggressive so that they should mess up the idle this much, not to mention the light load cruising (at least this is what I read).

I obviously checked for vacuum/boost leaks, I changed MAF sensor, I tried a different throttlebody, I tried stock injectors, I purchased a set of IAG fuel rails with pulsation dampers (though only the right side is installed yet), but nothing made a difference.

The front O2 sensor is about a year old and although reads differently then my LSU4.9, it kinda tracks the shape of the AEM wideband and it is within an acceptable tolerance I think.

I also tried to force open loop operation to see if this is related to the front O2 sensor at all, but even in OL there is a +/-10% difference compared to both the stock and the AEM wideband. The values are a littlebit shifted compared the CL log, and that is probably caused by the difference between the stock and the LSU4.9 sensor reads, but again I don't think this much of variation is acceptable, especially considering that my flywheel is stock and I have a fluidamper cranckpulley.

I actually have an idle that I consider not too bad. Although idle speed is set to 1100RPM at operating temp, the car idles fine, and does not want to stall/die. Also it is kinda smooth, apart from an occasional aberration, that my enginebuilder says is some sort of ignition or fueling issue.

In the logs I can also see that the engine speed is varying around 1100+/-~50 and the ignition advance is also fluctuating around the base idle timing +/-~5degrees and this is where I got stuck finally as I am not sure which one was first, the chicken or the egg.

In general I know the ECU might adjust the ignition advance at idle to control torque/engine speed, but in my case I am not sure if this is the case or simply the ignition and the AFR is changing because of the engine load fluctuation.

A couple of weeks back I tried to check OL operation at cruise, but I quickly revert back to CL as the AFR leaned out to 18+ without any change in load.

So I am pretty sure I have some hardware changes/issues that is causing all this, but can't figure it out.

My last thing that I still need to check is the CAM/CRANK sensor signals as their failure could easily explain all the fluctuations (I think), but that is still to be checked when I get back home.

I was also thinking about exhaust leaks, but I can't see any signs of that throughout the piping of the exhaust.

Oh, and one last thing which I am not sure about, why are there +/-spikes in the Final Fueling Base. I thought those will go away after I disable all the rear O2 effects, but those did not go away.

So I would appreciate any insights of what could cause this issue and how I could resolve this and get to a relatively constant actual air fuel mixture.

I attach couple of logs and I also attach my current rom.

Thanks a lot in advance and sorry for being so lengthy.

Béla

So I opened CL-CRUISE-ecuEdit_SSM_2018-09-22_18-21-58.csv and I see the lean spike at time signature 14,500 . One area that has my attention is the MAF voltage. It's definitely oscillating. That can be the result of pulsation in the MAF tube or some kind of electrical shielding issue. Overlap doesn't help either. I don't see AVCS angles in the logs I checked out. Can't open your .BIN file at this time, sorry. Are you using AVCS? Is it deleted or something? What happens if you zero out AVCS? What does the MAF housing and intake system look like? Have you made any alterations to the wiring harness?

Raymond, first of all thanks for taking the effort to read my not too short post and check into the log.

So, yes the MAF voltage is oscillating, but I thought that is due to the RPM changes of the engine and that goes back again to the chicken and egg dilemma. So is the MAF voltage oscillation caused by the RPM oscillation which is caused by the Ignition timing oscillation which is caused by not sure what or the other way around?

the AVCS angles I checked temporarily, it is 0 degree for both cams (if I remeber well, but once I get home from Malaysia, I certainly will check again), and no, AVCS is not deleted. They are almost stock values as I only took off a couple of degrees to be on the safe side from a valve to piston clearance perspctive.

The MAF housing is an APS 70mm diameter cold air intake, sucking air from behind the fender and yes, I made some wiring changes. I stripped off all the unnecessary wires from the engine harness and checked all joints and replaced some of the connectors with new ones.

So I guess if I will check the MAF voltage directly at the sensor with an oscilloscope, that should confirm or deny the wiring issue of the MAF, although it still could be the pulsation. On the other hand people on this side of the world say the APS CAI units were one of the best available on the market back then, so personally I would not suspect that. Now I have all my previous logs saved, so I will look for one with the stock airbox on, to check what was the AFR and MAF voltage oscillation with that. As far as I can remember this was my issue from the very beginning with stock airbox and stock injectors too. But I will try to check and come back on this.

Thanks again

So you have stock values in your AVCS table but you have more aggressive cams. What happens if you zero out your AVCS table? From the perspective of combustion stability and pulsation, I would expect a totally 0 AVCS table to have less problems, or just no improvement.

I'm a little concerned about the "simplified" wiring. You could have disturbed the shielding and grounding by removing "unnecessary" wires.

I certainly will check and try the AVCS vales/tables.

Now as for the wiring, I removed wires for EGR (secondary air pump), PCV valve, and TGVs which would have not been connected anyway as the solenoids, the motors/sensors and valves were removed no matter what and I payed attention to shielding when I had to extend the section of the DBW connector.

I will attached couple of pics of the wiring, but I need to download them first from OneDrive.

Thanks again for the inputs/help.

Here come the pictures I promised, and don't get me wrong, I am not saying wiring is not causing the issue, but I just wanted to show what I did, how I did and see what you think.

So on the original loom picture you can see the connectors marked with yellow-red tape are the ones I removed. TGV motors and position sensors, secondary air solenoid, PCV valve/solenoid. I also had to remove the resistor box of the old peak and hold injectors and had to replace the old Bosch connectors with Denso connectors for the DW850s. On the last picture you can see that while I had to extend the DBW throttle connection, I used the same kind of shielded wires to do that. and made sure the old and extended wire shields have a solid connection.

Just checked an old log where the OE Filter housing and MAF tubing was fitted to the car and I certainly can tell, that with the APS CAI the MAF voltage is more jumpy/erratic, while it is kinda smooth with the OE housing.

The question in my mind, does this really matter? I mean does 0.1V erraticness (if that is a word) really causes that much of a problem? And in the meanwhile I think I will answer my own question, yes. Because 0.1V in this range is roughly 1.5g/s airflow mass and that is a ~20% variation in the measured air mass.

The other thing though that I just spotted is that there are a lot of spikes +/- in the MassAirFlow measurement for what there is no equivalent (or anything close to that) MAF voltage. For example the negative spikes are around 2.5 g/s which would equal to 1.08V, but the MAF voltage never drops under 1.3V in the latest log. Any idea what is causing that? My only idea is that I have too many PIDs selected for logging and even though I have the fast polling turned on, the recorded values are inconsistent. This is something I will need to try too when I get home.

The pictures of the wiring harness you posted look good, but I have no way to judge if the harness is ok or not underneath that. Still, we can set the wiring issue aside for a minute. Can you put the stock airbox and tube back in just to test and use the factory MAF scaling? Does it fit in there at all?

Try logging with just a few PIDs selected.

That's exactly what I am planning to do. I just need to get back the stock airbox as I only borrowed that temporarily for the initial injector scaling.

Hi Bela,

I read your original post and have a few questions/concerns for getting this car to run right on a MAF tune. I'd be happy to get you pointed in the right direction and I can send over a basemap once you answer some questions/change some setup items.

1. Why did you move the primary O2? This will certainly have a huge effect on CL and would require re-scaling the O2, but even then you might not get it to ever behave right. I would highly recommend keeping the primary O2 in OEM location with stock scaling to keep that as a known and trusted measurement.

2. Did you physically delete or just tune out the rear O2? On the 32bit ECUs this has an effect on the CL AFR target so again, I'd keep this stock for now.

3. What size are the cams? Unless they are massive there shouldn't be a need for that much idle RPM. Typically you can add some timing and play with AFR (within .5 or so of stoich) for a better idle without tons of RPM.

4. Are you positive you have DW850s? Are they the blue ones and did you buy them new? Injectors are something you definitely want known 100% or you can have a lot of headaches getting the car smooth. Just taking a glance, your latency is way off spec for a DW850.

Lastly, I noticed you mentioned trying both the stock airbox and stock injectors, so I assume you have these available to swap in? I would address 1 and 2 on my list first and if it's readily available go ahead and do stock airbox and injectors to make things even easier for us!

Looking forward to hearing back from you and getting the car straightened out!

Hey Chris,

Thanks a lot for your comments and willingness to help. Here are the answers to your questions.

1. The O2 sensor was already in the DP when I bought the car, so it was not me who did that. But aside with who did it, it was moved because there is a Litchfield (TurboDynamics) LM450 twinscroll turbo fitted to the car. So there was no other option I guess, and I would have moved it too if I were to choose this turbo.

2. The rear O2 is still in the exhaust, but I have no idea if it functions properly or not. So I just disabled it in the relevant tables.

3. For the cams I attached the spec sheet.

4. These are dark blue DW injectors that I bought new with the flow specs and yes, I know latency is off compared to DW's specs, but these values were the ones where I could settle with both idle and cruising having low trims with stock airbox.

5. I had a stock airbox and a set of stock injectors borrowed from a friend for the initial scalings, but I certainly will borrow those again.

One more thing I forgot to mention, but just got in mind regarding maf vs sd question is that I also have a TurboXS computer in the car, which can be easily configured and mapped for both operations. I just wanted to get the OE computer tuned properly first and then move on to the TurboXS.

Whoa there, slow your roll on the TurboXS computer. I presume that's a piggyback of some sort that's just going to make your life difficult. If you want speed density, use the Cobb Speed Density code.

The cam specifications you posted are mild. I would still zero out your AVCS temporarily just to eliminate that as a root cause.

On the TurboXS I am not in a hurry. I am messing with this thing for more then 8 month now (with little breaks here and there), so that was just an idea as it is in the car already and you mentioned SD.

Cobb is out for me (atleast to my best knowledge) as this is an EDM car and Cobb supports only USDM and some ADM models.

I finally got home though so did some checks.

MAF V directly measured at the sensor is the same what is logged and it is oscillating unfortunately. If I measure the V between the sensor ground and battery ground though, there is a ~300mV potential difference between the two, and I am not sure if this is good. I know in some cases/cars there is ~100mV on some grounds, but my measurement is 3x of that.

The other thing is that cylinder#2 has quite some misfire and #4 has some too. not sure what the count threshold is for the CEL to kick in, but the maximum count on #2 was 6 within 1 or 2 minutes of logging/checking. You can see that in the smaller log I attached.

I also attached an other log that I wanted to use to check if the air mass follows the MAFV, but even with minimal pids logged, there are spikes in the air mass that don't have a corresponding MAF V.

Lastly, I checked the VVT, and both sides are 0 degree.

I still need to go and grab the OE airbox, so I can retest these with factory MAF scaling.