Question: best practice for correcting upper MAF fuel calculation

[freerunner]

Hi there,


I need your help. What's the best way to correct the fuel calculation error in the upper maf range?


Do you accept the logarithmic curve flattening off past 4V or is it best to counter with adjusting fuel flow (GDI tables)?


Prequel: I was doing some experiments to the GDI multiplier tables. Y-axis re-scaled. The single, most bottom right 0.797 had to be decreased to 0.3xx before any significant influence was visible in my log files. Now I wonder if the GDI multiplier tables are the right approach to do that. Maybe I'm being picky but I don't think forcing a direction change of inclination in the maf scaling table is the right idea (though it works).


Example screenshots:

[Tor]

It only runs open loop/high rpm in that end of the scale. I don't see any problem with correct it with MAF as it won't have any influence anywhere else in the tune.

[Kodename47]

Or just accept that it does it and tune around it? Does it matter if the AFR commanded and actual AFR are different? Are you sure it's not due to the pressure in the exhaust system that it's not reading richer than it actually is?

[Tor]

Originally Posted by Kodename47

Or just accept that it does it and tune around it? Does it matter if the AFR commanded and actual AFR are different? Are you sure it's not due to the pressure in the exhaust system that it's not reading richer than it actually is?

I know Wayno suggested it could be differences in MAF sensors as well.

I would correct it anyway. My argument is: What if it indeed running too rich? How would you know without correcting it?

If it then starts to knock, it's easier to adjust POL to get it to stop than tinkering with the MAF scale. I.e. I'd rather have a slightly false reading that matches target for the ease of adjustability. Or maybe a false reading was even corrected: Win in any case.

[Kodename47]

Originally Posted by Tor

I would correct it anyway. My argument is: What if it indeed running too rich? How would you know without correcting it?

So how are you correcting it if you don't know it's off in the 1st place?

Originally Posted by Tor

If it then starts to knock, it's easier to adjust POL to get it to stop than tinkering with the MAF scale

Kind of my point, you could just change the POL table anyway. The AFRs will change without any of the knock on changes that come with changing the MAF. Do you need the commanded AFR and output to match at high RPM? I would say not. So long as the AFRs are hitting what you deem optimal then you don't need anything else and you can do that by changing the fueling table alone.

[freerunner]

Originally Posted by Kodename47

Or just accept that it does it and tune around it? Does it matter if the AFR commanded and actual AFR are different? Are you sure it's not due to the pressure in the exhaust system that it's not reading richer than it actually is?

In thery it would not. But the Monk in me forbids.

To be honest: I'm not sure *any* of the stock AFR readings are sane. And I have got evidence. Little excourse:
After dialling in port injector scaling and latency to the *theoretical* optimum, my car answered with a weak idle. After running in circles for a while, I made the decision that the front O2 reading is playing tricks on me. I altered the front O2 sensor scaling (open-source) by moving stoich AFR by ~0.1 to the lean side. How could have anybody known - the values in the log were perfect. The culprit was masked behind two different errors, cancelling themselves out.

Back on topic: at this point of engine speed I doubt the trend of running excessively rich is due to exhaust pressures. My thesis: the amount and the speed of the exhaust gases produced is probably high enough to rule out reversion. And - correct me if I am wrong - got the assumption that an O2 sensor technically could only read leaner, not richer.

Originally Posted by Tor

I know Wayno suggested it could be differences in MAF sensors as well.

I would correct it anyway. My argument is: What if it indeed running too rich? How would you know without correcting it?

If it then starts to knock, it's easier to adjust POL to get it to stop than tinkering with the MAF scale. I.e. I'd rather have a slightly false reading that matches target for the ease of adjustability. Or maybe a false reading was even corrected: Win in any case.

I do know it is running too rich, but I don't know the actual amount! Leaning it out resulted in a power gain of about 15 vdyno-hp, while still being slightly *below* AFR_CMD. Of course AFR following AFR_CMD would also ease up comparison if I ever pick up a proper wideband.

What really bothers me is why does it turn excessively rich in the first place. I don't believe it is an errorneous output of the MAF sensor because it is able to read up to about 5V in its stock form (of course never hitting that NA-only).

Is it because VE is already decreasing?

[Kodename47]

@freerunner I can only feedback what I've seen and know. Your theory about the O2 sensor scaling at stoich is an interesting one as the only thing that should be guaranteed is that at 0mA the sensor is reading stoich as this is inherently in the design. Perhaps your car likes to idle leaner than stoich or it's just resolved another issue.

As for the sensor, I have seen that on the dyno on a supercharged car that the AFR output from the OEM sensor becomes inaccurate about 5k compared to a WBO2 at the tail, IIRC it was rich. No amount of scale optimization could correct this or you end up with a horrendous O2 sensor scale that is very wrong. Whether this happens on your NA car, who knows. However this just means we leave the fueling to run with the settings we know work by adjusting the fuel map to get the fueling we want and pay no attention to the output or commanded AFR as this isn't a true representation to what we want. I know the rough translation between the sensor output in logs and what is really going on and that all the matters. It will never follow the commanded AFR and that doesn't matter. So long as it's not too rich to stop making power or too lean that it's causing knock, then is it an issue? What's to say that it's not the O2 sensor scaling that's partially causing this as well? You could probably lean up the cells below 11.6 in the scale a touch and then it will look near on perfect..... Does that then mean it was the O2 scale that was problematic?

MAF based fueling should take VE into account, that's the benefit of a MAF based system. As the VE changes the airflow over the sensor will change and fueling is adjusted accordingly. You can see that the load output is a good indication of the torque output, which is the VE.

I'll close on this, what you are trying to do is potentially bandaid an issue that doesn't actually exist. Yes, in an ideal world the commanded AFR and sensor output AFR would be tight but the hardware on this car isn't necessarily up to it and lets face it, all tuning is really best effort. I started on this car years ago and realised that you can spend months chasing your tail to get that last 1%, only to find it doesn't actually work or make any real improvements. I've learnt that if something works, but isn't visually perfect in logs, sometimes it's just best to leave it. I remember someone years ago said that tuning is best done with the least interference possible, ie the less you change the less chance you have of messing up the result.

[tomm.brz]

So the answer to all that is an aftermarket wideband ans use it with ecutek racerom 10 an use CL fueling with that wideband :P

[Kodename47]

Originally Posted by tomm.brz

So the answer to all that is an aftermarket wideband ans use it with ecutek racerom 10 an use CL fueling with that wideband :P

I've had one sat in my shed for ages. Yet to bother to get it integrated as I'm really not sure the cost is worth the effort.

[tomm.brz]

Why not? A wideband is usefull to go FI also :P and doesnt cost too much

If one would integrate a wideband to keep it always there, would be best to put it where front stock lambda sits or on the secondary o2 hole?

[freerunner]

Originally Posted by Kodename47

@freerunner I can only feedback what I've seen and know. Your theory about the O2 sensor scaling at stoich is an interesting one as the only thing that should be guaranteed is that at 0mA the sensor is reading stoich as this is inherently in the design. Perhaps your car likes to idle leaner than stoich or it's just resolved another issue.

Except the other way around. Pushing 0mA to the lean side results in an enriched mixture.

Originally Posted by Kodename47

What's to say that it's not the O2 sensor scaling that's partially causing this as well? You could probably lean up the cells below 11.6 in the scale a touch and then it will look near on perfect..... Does that then mean it was the O2 scale that was problematic?

Hahaha, I'm not a jerk.
The comparison is not fair. 0 mA is 14.7 in the stock tables and in the community tables as well.

Originally Posted by Kodename47

I remember someone years ago said that tuning is best done with the least interference possible, ie the less you change the less chance you have of messing up the result.

I agree. If it ain't broken, don't fix it.

Originally Posted by tomm.brz

If one would integrate a wideband to keep it always there, would be best to put it where front stock lambda sits or on the secondary o2 hole?

Or drill a third hole.

[Kodename47]

Originally Posted by freerunner

Except the other way around. Pushing 0mA to the lean side results in an enriched mixture.

I assumed you meant that you were moving lambda 1 into the positive current range. So what you've done is the opposite? That would make sense as a slightly richer idle can help. I found that idle target throttle angle and target MAF changes really help the idle, but that is just correcting the mix as well without the need to fiddle the latencies and scaling so much.

Originally Posted by freerunner

Hahaha, I'm not a jerk.
The comparison is not fair. 0 mA is 14.7 in the stock tables and in the community tables as well.

I wasn't implying your scale that you did was off. Just pointing out that you could "fix" your issue by simply changing the O2 sensor scale. From my findings the scales posted on here or in any tunes are not perfect. I have definitely found a variance and are also different to EcuTek's verified scale as well. The OEM sensor can only really be used as a generic indication to what's going on. A little bit of variance is to be expected.

I spent ages fine tuning my MAF to the most commonly used O2 scale from on here. Went to the dyno and I was running so lean (like 0.5-1 AFR leaner) on the tailpipe O2 and my AFR curve was nothing like that in my logs.

[Tor]

Originally Posted by Kodename47

So how are you correcting it if you don't know it's off in the 1st place?

How do I know that it's not-not off? I mean, there is 50/50 percent chance it's off.

So if I correct it, I have 50% chance that I corrected a problem. If it's not off, I now have the added benefit of POL matching target/actual, for ease of adjustment.

I've tuned now 8 car with virtually the same setup. Those that need the dip above 3.9v all had an aftermarket intake hose by the way. Could be related.

They all run more or less the same POL. So I think it's fair to assume that they indeed were running rich.

Kind of my point, you could just change the POL table anyway.

Yes, well I suppose that's becoming a matter of philosophy. I would rather correct matching readings. Less mental gymnastics required. Also, my autism would prevent me from being satisfied with curves not following eachother.

"Three minutes to Wapner"

[Spec C Wannabe]

Trying to follow what you guys are talking about.

Anyway with does POL stand for?