Adjusting MAF with Open Loop Fuelling
 

A tip from @Kodename47 in this thread:
http://www.ft86club.com/forums/showt...=114760&page=3

Splitting this topic out in its own separate thread.

My MAF is pretty well scaled and produces consistently low fuel trims. However, I had some curls in the critical 3-volt section that I couldn't get rid off, presumably because of AVCS.

My MAF looked like this:

https://s29.postimg.org/fygvy0ps7/Sc...t_18_17_10.png

After a bit of fiddling with the OL fuel table (moving a row and rescaling + adjustments), off I went for a drive:

https://s27.postimg.org/q5tv6gheb/Sc...t_18_17_20.png

I had brought my laptop. Pulled over, uploaded to datazap, did some adjustments in RomRaider, flashed, and did another pull:

https://s24.postimg.org/huozmid51/Sc...t_18_17_31.png

The OL fueling table looks like this:

https://s24.postimg.org/jtcvqxfxh/Sc...t_18_44_32.png

I'll sacrifice another row and move things a bit around, so I can put in another point to make the first part of AFR command more pointy to counter the dip.

It was a bit scary to flash the car while pulled over some random place. And moving rows around in OL fueling I preferred to do at home. So I am not done yet.

Before anyone says it's too lean, keep in mind this is without learning - just flash and go. The entire AFR curve moves down where it should be once the ECU learns the fuel trims. Usualy I end up with +1-2 after very little driving.

I just wanted to share my experience for anyone wanting to take this route too. I'm done messing with MAF. As long as it's a nice exponential curve and produces low LTFTs I think this is an easier way to bring AFR into line.

This is a really nice result! Can you post the table when you are done? Our AFR curve looks 100% the same on stock Wayno STG2.

Yes, sure. I'm just working on it now.

I'm changing OL fueling, so AFR command will look like the red line to get AFR to hopefully follow the target blue (green was actual and made it too lean at 3200 rpm):

https://s23.postimg.org/85bdzhiu3/Sc...t_20_22_49.png

OL fuel looks like this (note different rpm scaling!!!). Also this is what I just made and haven't been flashed yet:

https://s27.postimg.org/g5ldq4iwz/Sc...t_20_28_14.png

Don't know if you can copy paste this. Try (be sure to copy it all, it scrolls horisontally):

(changed 3900 to 3950)

Unfortunately copy pasting that doesn't work. Maybe you could send me your tune when you are finished with it (I'll PM you my details :) ), then I'll just copy paste it over.

I tried copy pasting from here too. It worked for me, are you sure you select the whole text before copying? You need Table 3D too.

Yes, now it works, I had to change my Windows settings (decimal and point for numbers) ! Or maybe I did it wrong the first time .. one way or another, it works :)

Edit: Which Wayno MAF scale do you use? The default one (+2%) or the lean one ?

Wayno used logs from my car (among others I presume) as a test subject when making the new MAF scales for the tunes with the new injector ratios. So I have one from those tests.

I tried the +1.5 and it gave a really nice result too (still stg 1):
http://datazap.me/u/tor/stg1-102-v12...zoom=9515-9631

As you can see I had +2 LTFT with the +1.5, and near perfect AFR. LTFT doesn't matter diddly squat with the new injector ratios because there is no split at 5000 rpm which usually messed up AFR if the LTFT were off on either side. The only thing that matters is that AFR is correct in OL, which can be adjusted easily by adjusting the MAF above 3.20 volt.

Even though LTFT of +2 doesn't matter, I decided to use the custom MAF that came from testing on my car for another reason. The benefit for me is that it practically doesn't have to learn anything (see, the last log screenshot - +0.28 LTFT and the AFR is where it should be). So I can pretty much "flash and drive" without thinking about learning. I don't drive that much and disconnect the battery a lot so it has value to me.

Anyway, I don't think that it will have an influence on the adjustment made in OL fueling here which scale you use. Since it's more of an adjustment for the AVCS (or whatever is causing the bumps).

P.s.
I use German Shell V-Power 100 octane. It contains a whopping 5% additives, which may be the reason that there was never a "ready made" MAF that really suited my ECU completely (only 95 percent fuel injected where assumed 100%).

that rich area is a fake rich . keep ur eyes on any activity knock

This turns out to be more complicated than I thought - as usual.

Yes, something funky is going on. I tried out the last version that I pasted in the previous post and it doesn't work the way intended. It apparently wants to make a dip shape there no matter what. And now it makes a lean problem instead. No knock though. But another problem that becomes apparent is the influence on load:

https://s24.postimg.org/nnngpz6o5/Sc...t_19_09_40.png
http://datazap.me/u/tor/tor-204?log=...zoom=2636-2742

Compared to the normal OL fueling (that flat spot between 3600 and 3900 rpm).

https://s29.postimg.org/jtzrck26f/Sc...t_19_26_00.png
http://datazap.me/u/tor/tor-201-1-ia...zoom=5894-5985

It looks like there is some benefit to correct the lean spot above 4000 rpm. This was the more moderate first attempt:

https://s27.postimg.org/nueayb8pf/Sc...t_19_28_22.png
http://datazap.me/u/tor/tor-202?log=...7&zoom=627-725

So as usual, it's probably better to keep it smooth. I am going to revert this and forget about the (perhaps fake) rich dip below 4k and only concentrate on smoothening the lean spot above 4k, which seems to give better load values.

Maybe this dip shape is more to do with load limits?

http://datazap.me/u/tor/tor-204?log=...mark=2657-2661


smooth avcs ext to correct afr . otsv3

It's definitively load related. Curiosity got the better of me, so I changed went back to the normal OL fuelling and tried to change the load scaling instead. Changed 3700 to 1.25. Both stock and OTS has 1.3 already at 3200 rpm so I thought it would be okay.

Oh boy! It felt like the engine was running at half power. Never has the car felt so sluggish!:

https://s23.postimg.org/3rzxrfh5n/Sc...t_23_09_18.png

It did however, correct the dip. :lol: It used to be between the two yellow lines. Obviously it throws the whole MAF scale off as well and now it's running insanely rich.

http://datazap.me/u/tor/205?log=0&da...mark=3487-3510

To top it off, I heard for the first time real audible knock. Didn't sound like "ping" though, rather a sizzling sound. Ptjuzzzz combined with a noticeable bogging down. And this was under light load as can be seen in the log:

https://s23.postimg.org/658lfbesr/Sc...t_23_06_36.png

Next up: Reverting to standard loads and OL fuelling and forgetting about the dip. I am going to have to go with Wayno and aagun on this one, that dip is probably just somehow a wrong representation of what is actually going on in the combustion chamber.

But hey, learning by doing... It was an interesting experience.

Okay I shouldn't be posting here late at night, let alone mess with flashing and logging.

The load change log I posted above is totally messed up. The AVCS is not active at all. I don't understand why. I followed my usual procedure. After flash:
- Ignition on, wait 20 secs, then off again.
- Ignition on, wait 20 secs, start engine.
- Without touching anything let the car idle for 1 minute.

I even timed the one minute yesterday because it was late and I didn't want to let the car idle for too long unnecessarily (neighbors bedroom facing my driveway).

Anyway the log shows AVCS is not active at all:
http://datazap.me/u/tor/205?log=0&data=22-33-34-35-36

So I suppose 1 min wasn't enough for the calibration to complete?

If AVCS stayed at 0 during the drive, I suppose that would throw everything off including MAF?

P.s.
The load change is the only thing changed (OL fuelling back to normal).

normal . avcs ERROR . u cold start the car right. it will back Normal stat after worn up the oil temp

i found when i patch adle code that the ecu do alot of testing and checking when start in the engine . on AVCS.
keeping the engine out of troubles is one of major job by ecm . any disactives happens in this case

Yes, it was a cold start yesterday. I just started the car again. Was a bit nervous because I never had that problem before. Now it's working again: :happyanim:

http://datazap.me/u/tor/avcs-wake?lo...zoom=3552-4083

So the previous post with the logs can be disregarded. Going for a drive and will post how the increased 3700 load limit works out.

Note, this is an experiment to see how the load limit affects AFR.

Haha, the AVCS problem led me to the completely wrong conclusion. It's not load limit related at all, and the load doesn't go higher than 1.05ish even with the limit raised:

http://datazap.me/u/tor/tor-205-avcs...zoom=4743-4849

I didn't notice any adverse effect of the raised limit by the way. No hesitation or anything. Which makes sense, since it doesn't exceed the value anyway.

Dip is still there too. So aagun is right, it has to be the AVCS (now that I've driven with it inactive by accident it must be pretty safe to conclude).

Since I have been wrong several times, I am going to give the OL fueling adjustment one more try. I guess the easiest way to find out if the dip is real or not, is to eliminate it as best as possible and observe the effect.

Anyone that can point me to a log from someone with an OpenFlash Header? I would be curious to see how it compares to the Gruppe S.

Just an after thought:

Maybe the load is going down, because it is dipping rich? That would explain the flat spot after somewhat fixing the dip. With the dip fixed, it then hangs at the limit.

I am going to try to fix the dip and keep the raised load limit. In my virtual dyno curves I do have a slight torque dip in the same range still. Maybe this is the way to eliminate is?

https://s24.postimg.org/lhsc9jw11/Sc...t_05_17_50.png

Great effort for trying to find the cure for the dip. In case it doesn't work out, just a tip I got is to change the CL to OL transition, but having no knowledge on this stuff I have no idea where to start, but maybe you have.

This is one of my 102Ron Wayno tune, log after around 100km's of driving.
Mods:
- OpenFlash Header
- Milltek resonated primary exhaust
- Modified intake-tube with stock filter and box
- Crawford BPB
- Cusco 4.556 Final Drive

Datazap log

one minute . u wanna take off that dip. ........
that think Give you huge power at low mid range . happed becose of overlap the firing whats that mean . it means the intake and exhaust cam open in the same time or before that . means the intake gas go to exhaust Directly without burn the o2

what we read is rich
what the cp is lean . so the reason of knocky range at low mid area

u try to lean it more . it will lose power .

how to fix that dip . the only way is lower pushing avcs . it coz lose power little . same oem tune

what i recommend in this is rich ur tune more Maybe in ur will 12 From 1 load . test and check knock . then add timing . will make ur car fly on the sky

Yes, it's back to square one, i.e. if it worth spending time on or not.

I just drove the car to and from work with the version with the raised 3700 load limit and normal OL fueling. It feels great, dip or not.

Thanks elBarto for the log. It's a bit hard to see with the resolution of the OFT logs, but it seems to dip too at 3700 rpm, though generally it is leaner below 4300 so it doesn't dip that rich. The dip appears to be smaller too. The reason I wanted to see an OFH log is that it's my understanding that the AVCS is tuned on that header. Maybe the same values are just not optimal with the Gruppe S header? I don't want to change AVCS randomly, though, so I guess I am stuck using other options or leaving it alone altogether.

I do think there was a difference with the corrected lean spot just above 4000 rpm. But it could be that I imagine things. I will for sure try to pull that down again.

Infact, I already made a new attempt at correcting the OL fueling, but didn't flash it because of the inactive AVCS ordeal. I think I will flash that next and see how it goes, including the raised load limit. I have to work quite a lot the next days, so next time I have some spare time.

I think the mistake here, was pulling the bend too high up at 3950:
https://s23.postimg.org/85bdzhiu3/Sc...t_20_22_49.png

Going to try with something more like this:
https://s27.postimg.org/dzj7k4acz/Sc...t_19_28_22.png

In EcuTek, Maps by Function > Fueling > Open Loop. Threshold settings are there.

By default, there are delay timers preventing immediate operation of open loop. There are also RPM thresholds that need to be passed before open loop is allowed to operate. Default settings are no open loop until around 5k RPM in 1st, and 4k RPM in the rest of the gears. Check fuel system status when logging to verify what mode you are in.

@Tor - Verify your car is hitting open loop mode when you are testing these changes. If you are not in open loop, the primary fuel map table you are adjusting won't apply. Check your open loop thresholds and delay timers, and minimize these for immediate open loop transition. This will help the fueling issues you are seeing, but will not solve them.

You are adjusting fueling incorrectly. You do not want to adjust the fuel map to correct commanded vs actual AFR. If your actual AFR does not match your commanded AFR in open loop, either the top end / that problematic area of your MAF is not scaled properly, or your injectors are not scaled properly (CC/min and lag times). Ecutek CC/min setting is not linear from my testing, so 1000cc injectors != setting this table to 1000. My ID1000's worked best with this table set to around 600 to 650 after I verified my injector latency was correct. Still not sure why, as EcuTek claims this is linear. Other posts here back up my experience with this. Maybe someone else can chime in, but I was able to nail down my port injector fueling with this setting on ID1000's.

To fix your high end MAF curve, run the car on 100% DI and scale your MAF against fueling error (only for N/A cars, DI alone isn't enough fueling for FI so this trick wont work fully). Once your fueling errors are minimal, switch to 100% port injection, and use the cc/min and injector dead time tables to adjust fueling error. Do not adjust MAF at this point as you've already scaled it properly on DI. Once your MAF is set, you don't really want to mess with it too much as this changes the entire engine load calculation when adjustments are made.

Do new pulls at this point and your AFR error should be minimal. You don't want to trick the ECU to get the AFR you want by setting the fueling table to really high / really low settings to get the values you want. You want to avoid these kids of large spikes as much as possible (interpolate!).

The default RPM vs load scale on the fueling table is perfectly fine for N/A cars and does not need to be adjusted. If you're hacking up table scaling to fix issues, something else is off in your tune. After you have corrected your fueling issues, you can tweak ignition and cam timing from there to improve the torque curve. Once your fueling is corrected, your commanded vs actual AFR should be very close to each other throughout the entire fuel map.

Oh, and the load limiters won't really do much with adjustments on N/A. I believe N/A cars never really go much beyond 1.2 load, so raising this higher wont do anything. This is mostly a setting for FI where load will easily go in to the 2.0+ range.

FWIW, After multiple flashes sometimes it'll take take longer than usual. I just wait until the DI injectors kick in than drive normally until they pop on

@BRZoomTX thanks for taking your time to write in details. I didn't change the timers, but I am sure Wayno did. As can be seen the ECU runs in OL as soon as I put my foot down.
http://datazap.me/u/tor/tor-205-2-mo...mark=9384-9403

I am not blanket adjusting fueling with OL. The MAF is already scaled to provide solid low LTFT and generally, AFR follows AFR command.

The point is to make an adjustment to one specific problematic area. If fixing this with MAF I would have to make the scaling extremly lumpy, if it would even work.

The idea came from Kodename47, and the point of the thread is to test out the idea:

Better repesentation of the problem area:
http://datazap.me/u/tor/tor-205-2-mo...9403-9393-9374

Rich dip at 3,7k probably vanishes if you move your load limit closer to 1.0 in that area.
Lower = leaner. In terms of "ecu picks a cell further on the left"

If the engine doesn't respond to it, I think the maf scale is probably off.


Regarding that lean spike around 4,1k, I'm not sure. :iono:
It looks like the afr curve follows the shape of the exhaust cam. Coincidence?

4th attempt...

I think this time it looks pretty successful. Whether it's a good idea or not, I don't know. But it is indeed possible to steer AFR with OL fueling.

https://s29.postimg.org/cpaywjnuf/Sc...t_00_57_26.png
http://datazap.me/u/tor/tor-206-3-lo...zoom=9038-9141

But I got a feeling that this is just patching an underlying problem: That the standard OFT AVCS' are not correct for the Gruppe S header.

I am getting more and more tempted to get it dyno tuned if only there was someone here that would do it with BRZEdit and leave it unlocked so I could keep messing with it myself afterward. Alas, that's not going to happen. :(


I accidentally made a 4th gear pull. How do you do that by accident? Well, I was rolling on at 50-60 km/h as usual, and I somehow missed that it was in 4th. I did feel it going a bit faster than my usual 3rd gear pulls at the top, but I didn't look at the speedometer. So it wasn't until I got home and looked in the log that I was actually sure. That's how badass the car pulls now! :D :D

Topping out at 170 km/h it was good that this was of course done at a closed off test circuit. :thumbsup:

But it revealed that there is a lot of work to do. Temperatures have risen, so my IAT is now 12-15 deg C instead of 5 degs, and the engine doesn't like that as far as timing goes:

https://s30.postimg.org/vdbbndb9d/Sc...t_01_10_28.png

http://datazap.me/u/tor/tor-206-3-lo...zoom=7558-7762

Somehow my otherwise predictable slight positive LTFT are now negative so that probably also counter productive. Could it be due to the load change at 3700 rpm? In any case, I am going to lower it back to where it was as there is no point in having it at 1.25.

Next test could be to revert it all completely and instead to lower exhaust AVCS a deg or two. Is there any effective way to adjust AVCS without a dyno?

...and of course, I had to throw that 4th gear pull into virtual dyno. :)

Correcting the dip/top didn't have any effect on the curve (but I finally have my 185 whp :D ).

https://s28.postimg.org/n3fbvxjwd/4thgear.jpg

50 to 170 km/h in 18.6 seconds without changing gears isn't all too bad me think. :)

Load limits will not have any impact on the load that is logged, you can set it to 0.5 everywhere and you'll get the true loads that are calculated from the MAF calculation in your logged parameter. What the load limits do are set the maximum look up values in the tables and in the fuel (load) calculation. This is worth noting especially in the fuel and timing tables.

If the OL fuel table values are the same for all the load columns, then lowering the load limit will make it run leaner. A rich spike can be countered by doing this. The opposite can be done by raising the limits, so long as the loads aren't lower than the set limit.

Don't forget that changing the MAF scale will change the engine load, this can have a knock on impact with the load limits etc.

You must also consider the delay between cylinder and sensor, when changing the correct load and RPM cells.

Thanks a lot, that really helped my understanding of load limits.

Since you are the one starting me on all this, what is you opinion about the corrections I did to OL?

I am going to undo the changes below 4000 rpm and instead try out this:

https://s28.postimg.org/glhqgr0v1/Sc...t_04_14_58.png

To not get things more complicated than it needs to (for a novice like me at least):
It could be of benefit to shift the exhaust and intake cam symmetrically, e.g. if you reduce your exhaust retard value, add same value to intake advance. Then you don't change the duration of the overlap event.
Think of an old 2-valve engine with a single cam.
You're just trying to close the exhaust valves soon enough to prevent the loss of the charge where the lean spikes are.

@Tor if it's working, then it's working. I would try and avoid massive swings in the OL table and try and work out what's happening and why. What seems to be overlooked here is injection volumes/IPW. This gives a better idea of what is going into the cylinder, rather than relying on just the output ;) The logged load gives a half decent idea of the torque curve and the VE. Therefore you should be expecting the fuel delivery to be similar shape.

Use VGI's tool in the WOT pulls section of the log view tab as it gives easy comparison. I just opened AFR, DI IPW, PI IPW and load on the same chart one pull (pull 7 of the log2_31 above). It gives you a good indication of the delays I mentioned and easy to see why the AFR output is doing what it's doing ;)

Thanks a lot again. I need a hint though as to why it's doing what it's doing, please.

I also opened pull 4 of log0006_9 where the OL fuelling was not corrected yet. It looks like this (delays not so prominen):
https://s27.postimg.org/p0t09ja4j/Sc...t_11_05_35.png
It's this log:
http://datazap.me/u/tor/tor-201-1-4-...?log=0&data=22

I also added AFR_Command to the selection of the pull you tried out:

https://s27.postimg.org/9w6n4gvoz/Sc...t_11_11_55.png

So now I'm confused as to what I am looking for? If it's the delay, then it seems I caused it by changing OL?

@freerunner thanks. My thought was actually just to smooth out the phasing in of the overlap, so it doesn't onset so abrupt.

Maybe I just plain and simple just straight over copy the OTS standard UEL AVCS instead. What I am running now is pretty much the OFT/OFH AVCS.

If I do change anything, I will be sure to study the ratio they are changed between the OFH and UEL tables. Thanks for the hint.

what is the secret on avcs? this is high technology knoledge on engine operation . on ur map there are 44° of overlap duration at uncontrollable area 4K rpm. that's mean in FA20 that the intake and exhaust cams opened 2mm at same time. in that error will push the air from intake to exhaust Directly without burn it . that air is full charge of fuel . (thats why it coz jumping afr to rich at that dip)
this is first reason .
2nd season there is low exhaust back pressure on aftermarket headers .
the 3rd is timing advance . it makes intake vacuum
4rd is where is the top of overlap

How can I calculate the overlap from the two tables? And wouldn't it be on purpose that the intake and exhaust valves are open at the same time to help scavanging?

Are these done in the same gear? Obviously the delays change if the ramp is different speeds? The delay isn't the cause, my point was more of the case that you may need to change lower cells than the point of issue, so if the log is showing an issue at 4k you may be better adjusting those of 3.6-3.8k etc. Same with load if it's sweeping through cells quickly.

As for some help, have in mind this: If the fuel quantity remains the same but the AFR output goes rich, then the cylinder is getting less oxygen. If it goes lean, it's getting more oxygen but not all of this may be remaining in the cylinder depending on the overlap so may get a "false lean". There may be instances where EGR can mess with AFR readings, but this won't be the case on WOT so can be discarded. If you add fuel but it stays "lean" then there is a good chance that the intake charge is passing through the cylinder. What I've been trying to suggest is that you want the in cylinder AFR curve to be smooth and not swinging rich to lean. You want to repeat tests in the same gear etc but even then the AVCS can vary so you get differences pull to pull. So you want to look at changes of fuel quantity, the AVCS activity and the output AFR to get a good idea of what's going on. It's also why I treat the OL fuel table as just a fuel additive, because I'd much rather have my fuel delivery how I want it than needing my O2 sensor reading matching the AFR tables.

This could explain why it goes lean after the rich dip, not why it reads rich. The sensor won't see the fuel in the exhaust. It's also highly unlikely at low RPM that there is enough vacuum to pull the charge straight through the cylinder, this is where the load logs will be of some use. Those logs show a dip in VE where these fluctuations are happening so it's not a sudden vacuum. Also, most of the fuel is provided by the DI. There is a much lower possibility (almost impossible?) of this being blown through the exhaust valves.

My AVCS tool

There's a formula in there you can add to your logs if you want to get the overlap into the log tools.

my tool for overlap calculation
CAM LUMPY TOOL