Trying to tune MAF
Originally Posted by RedHardSupra
you guys really need to look in the advanced section under 'complex math' thread, there's a lot of info there.
the main idea is that your resulting AFR is going to be a result of VE, MAP, TEMPS, IFR and IPW. VE and TEMPS and IFR are highly nonlinear functions, start mulitplying them together and you have a nonlinear hell. so replacing tuning all this stuff with just IFR or just VE is just silly, because it will never cover all the cases needed to run properly. that's why VE just doesn't converge for a lot of people.
i'm working on a model where we'll be able to tune all these things at the same time. it's not going to be perfect, but it's gonna be better than what we've seen using the AFR% corrections.
the main idea is that your resulting AFR is going to be a result of VE, MAP, TEMPS, IFR and IPW. VE and TEMPS and IFR are highly nonlinear functions, start mulitplying them together and you have a nonlinear hell. so replacing tuning all this stuff with just IFR or just VE is just silly, because it will never cover all the cases needed to run properly. that's why VE just doesn't converge for a lot of people.
i'm working on a model where we'll be able to tune all these things at the same time. it's not going to be perfect, but it's gonna be better than what we've seen using the AFR% corrections.
I agree here. I believe VE is just a start and a small piece of this complex puzzle which could explain why I see so many VE tables that are smoothed done with VE tuning alone and certain areas seem to suffer with higher/lower BEN's in cells but are accecptable. What it seems I'm finding is that with my larger injectors and H/C package that I needed to use your spreadsheet to get IFR's to where the engine liked and it started and ran fine. Then started AutoVE and as I smoothed the table I noticed certain KPA rows that needed tweaking but would really put a good wave in the table. My thought now is to adjust the IFR table in those KPA ranges in small increments to bring BEN's closer. Sorta tweaking the IFR table like VE needs to be tweaked. I believe this will bring BENs closer but there will still be the issue of temps and PW's to deal with. "One step at a time."
If I seem off the wall here at all, please inform me. I'm just a novoice compared to ALOT of you who's just on a learning curve.
Most of the inaccuracies I see in tuning my VE come from IAT error. It seems that when my IAT heatsoaks, things lean out because the PCM assumes the air entering the engine is warmer than it really is. This false assumption of less dense air means less fuel required and therefore a lean condition arises. This is nothing new. I just wish there were an easy cure for it.
After reading alot more of your Blog redhard I now have more of an understanding of what you are saying. I'm not one who can actually do all the number crunching but do have an understanding of it for the most part. I now see that I should leave that IFR table alone and keep the figures in from your updated spreadsheet then work on getting the required items for live fuel pressure logging, read up the tutorial on it then change those accordingly. I just wander where most of us like myself would be without the hard work you have put into all of this and making your finding available for us all. That I'd personally like to that you for.
Originally Posted by Violatorno1
COS5 if you can calibrate A0014 correctly. Something I need to do,just havn't read into it enough.
). My AFR was spot on. Once I started getting stuck in lights downtown, heatsoak set in and my car leaned out.I think we notice it more often in the summer months because things get warmer under the hood a little easier when it's not 55* outside. I honestly believe if we could relocate the IAT to a better, still accurate place not subject to heatsoak, it'd be the solution (like maybe mounting it somewhere in the front bumper so that it hangs in the air - maybe secure with some fishing line or sometihng).
Then you'd skew your numbers by moving it so far away from the intake. Have to calculate for the distance traveled temp change. That's why they want it close to the intake itself from what I read.
This again is something Marcins been trying to tell us. GMVE...... They calculate everything based on a hybrid and we're just tuning half of what needs to be done reading into his blog.
it's not just IAT, it's how IAT blends with ECT to estimate the temp of air in the manifold. newer cars blend table has been expanded from being based solely on airflow to airflow AND speed, so this is getting even more complicated. unless this is fully figured out, the PCM will never be able to estimate how much fuel is needed to match the _real_ airmass, thus AFR will always be scattered.
LS1 Tech Stories
The Best V8 Stories One Small Block at Time
Gas Monkey Built a 6-Wheel Ferrari Testarossa With a Corvette LT4 Engine
Verdad Gallardo
7 Most Reliable High-Performance Engines GM Has Ever Built
Verdad Gallardo
Amazing '71 Camaro Restomod Is Modern Muscle Car Under the Skin
Verdad Gallardo
6 Common C5 Corvette Failures and What's Involved In Repairing Them
Pouria Savadkouei
Retro Modern Bandit Pontiac Trans AM Comes With Burt Reynolds' Autograph
Verdad Gallardo
Top 10 Greatest Cadillac V Series Performance Models Ever, Ranked
Pouria Savadkouei
Top 10 Most Powerful Chevy Trucks Ever Made!
Hennessey's New Supercharged Silverado ZR2 Has 700 HP
Verdad Gallardo
Coachbuilt N2A Anteros Is an LS2-Powered C6 Corvette In Italian Clothes
Verdad Gallardo
Originally Posted by RedHardSupra
it's not just IAT, it's how IAT blends with ECT to estimate the temp of air in the manifold. newer cars blend table has been expanded from being based solely on airflow to airflow AND speed, so this is getting even more complicated. unless this is fully figured out, the PCM will never be able to estimate how much fuel is needed to match the _real_ airmass, thus AFR will always be scattered.
I know this is an IAT/ECT blending problem. Wouldn't though without your site. That being the case, how much more can changing the coolant temp affect what we're trying now. One would figure that GM numbers and calculations would be set for a higher expected temp and with the addition of a 160 t-stat and lowering fan turn on temps, expected or calculated temps would be different than actual. Am I right in thinking this?
If so then everything calculed by temps would be skewed such as injector tables and VE, ect....
If so then everything calculed by temps would be skewed such as injector tables and VE, ect....
Another thought.........In EFILive we can disable charge temperature blending leaving the PCM using IAT exclusively as the charge temperature. Could this make calculating the amount of fuel needed for proper AFR's? I figure with the use of A0014 in EFI's COS5 set properly that fueling can be done from IAT temps. orrrrr.....maybe this is dangerous to do? Just trying to through some ideas toward that number crunching mind of yours.
no, because it's still not reflecting reality. the idea is not to dumb down the engine so we can deal with it, but to get smarter to we can understand (thus predict) what the hell is it doing.
Ok.....So maybe, and this may sound stupid but, start looking into what is programmed into the 01 & up MAF sensors looking for programming made to work in conjunction with the PCM programming to calculate charge temps. Looking at a 99 tune compared to a 02 tune they eliminated backup VE and added charge temp filtering. Could this be where they determine MAF failure fueling?
nah, it's a perfect approach for learning it, start simple, once you get a feel for it, try to improve it.
gotta remember few things tho:
the fact you don't see it, doesn't mean it's not there. there's plenty of tables that get used but we dont get to know even exist, see, yet alone alter.
same goes for the inverse. the fact that something's there, doesnt mean it gets used. like cylinder volume in e40 pcm's, it's there, but it's useless.
labels are misleading. some tables are not used as the label might indicate, others are just plain wrong. OLFA==open loop fuel ADDER? adder my ***, it's a divider.
so with all that said, simple models are fun to play with, but they should really be used for building deeper understanding. you can just use TEMP=IAT for your model, and simulate what kind of fueling decisions it would make. how off would it be? is it better on average? but how bad are the really bad mistakes? after all it only takes few bad fueling decisions to blow **** up. which way does it err? rich might not be good at it, but lean is far more likely to blow it up, so you want to see 'tendencies' of your model.
then make another model, TEMP=IAT+(IAT+ECT)/2 and do the same for that model, compare results, errors, rinse repeat...
gotta remember few things tho:
the fact you don't see it, doesn't mean it's not there. there's plenty of tables that get used but we dont get to know even exist, see, yet alone alter.
same goes for the inverse. the fact that something's there, doesnt mean it gets used. like cylinder volume in e40 pcm's, it's there, but it's useless.
labels are misleading. some tables are not used as the label might indicate, others are just plain wrong. OLFA==open loop fuel ADDER? adder my ***, it's a divider.
so with all that said, simple models are fun to play with, but they should really be used for building deeper understanding. you can just use TEMP=IAT for your model, and simulate what kind of fueling decisions it would make. how off would it be? is it better on average? but how bad are the really bad mistakes? after all it only takes few bad fueling decisions to blow **** up. which way does it err? rich might not be good at it, but lean is far more likely to blow it up, so you want to see 'tendencies' of your model.
then make another model, TEMP=IAT+(IAT+ECT)/2 and do the same for that model, compare results, errors, rinse repeat...
Originally Posted by RedHardSupra
....others are just plain wrong. OLFA==open loop fuel ADDER? adder my ***, it's a divider....
explain please
-edit-
unless you mean (not trying to be obvious but...) it's in EQ format where 14.7ish / 1.30 = 11.3:1 ..... but the 1.3 could be considered additive I suppose since that extra .3 over 1 is against stoich right?
Last edited by Frost; Jun 11, 2007 at 12:18 AM.
HPT defaults to EQ, EFI defaults to lambda, but in neither case it's an adder.
well i guess if internally it's stored as .3 while we see 1.3 then it it is an adder. i guess that's another good point: what you see is not necessarily how it really works.
well i guess if internally it's stored as .3 while we see 1.3 then it it is an adder. i guess that's another good point: what you see is not necessarily how it really works.
TECH Enthusiast
Joined: May 2007
Posts: 538
Likes: 0
Originally Posted by SSpdDmon
If I'm interpreting this correctly, this is why we tune the VE first and then the MAF. From what I understand, there's a VE calculated airflow value and a MAF calculated airflow value. Below 4,000rpm (or whatever value you set B0120 to), these are blended together with the factor you mention. EFI Live describes it as follows:
Originally Posted by SSpdDmon
By killing the MAF at first, we eliminate the MAF's influence and re-map the VE table's values (listed in grams*kelvin/kpa....not a VE%). Once those numbers are established, we re-enable the MAF with the assumption that the VE is now correct. Personally, I drop B0120 to a low value so that the VE blending factor doesn't interfere. Then, I can dial in the MAF.
Again your AIM is for AF ratios that are not real. IF you try to rationalize the amount of air actually in the cylinder and the amount of fuel in the cylinder then compare that to the AF your observing you will find a large error between the 2.
Originally Posted by SSpdDmon
Like I mentioned before, my experience shows fueling most definitely varies with RPM (rich down low and lean up top). Why wouldn't two of the main infuential factors like VE and MAF need to change accordingly? The airflow characteristics are nowhere near the same when you throw on headers, heads and a cam. It just doesn't add up for me.
I have no idea what your observing but its seems very inverted. How are you concluding with rich down low ? Higher MAP values at low rpm should always be richer then lower map values. the engine is more heavily loaded.
I have been side-lining this conversation, but I just have to ask/add the following. Please explain, and thanks for your time.
Again your AIM is for AF ratios that are not real. IF you try to rationalize the amount of air actually in the cylinder and the amount of fuel in the cylinder then compare that to the AF your observing you will find a large error between the 2.
Well then, explain why this approach yields such excellent results as far as real-world driving is concerned if it's so wrong... I've seen too many cars come out running like a champ to beleive that there is some large flaw here
I have no idea what your observing but its seems very inverted. How are you concluding with rich down low ? Higher MAP values at low rpm should always be richer then lower map values. the engine is more heavily loaded. while I tend to agree, the big issue is that the lack of low-end vacuum puts the PCM into an area around idle where the perceived load is much greater than the actual load, which results in overfueling
Originally Posted by LS1curious
Again your AIM is for AF ratios that are not real. IF you try to rationalize the amount of air actually in the cylinder and the amount of fuel in the cylinder then compare that to the AF your observing you will find a large error between the 2.
Well then, explain why this approach yields such excellent results as far as real-world driving is concerned if it's so wrong... I've seen too many cars come out running like a champ to beleive that there is some large flaw here
I have no idea what your observing but its seems very inverted. How are you concluding with rich down low ? Higher MAP values at low rpm should always be richer then lower map values. the engine is more heavily loaded. while I tend to agree, the big issue is that the lack of low-end vacuum puts the PCM into an area around idle where the perceived load is much greater than the actual load, which results in overfueling
Originally Posted by Frost
I have been side-lining this conversation, but I just have to ask/add the following. Please explain, and thanks for your time.
I'm still trying to understand how unburnt O2 is making it's way into the exhaust because of my cam overlap.?.?.? What kind of pressure scenario would allow for a stream of unused intake air to make it into the exhaust stream? I mean...the piston is either pushing exhaust gas out or pulling fresh air in, right? And, both valves are closed at TDC, right? So, where does it get by? Not only that, but why would it proportionally vary based on RPM if the amount of overlap doesn't change (re: "Most cams usually 'cleanup' above 3000rpms)? Just trying to visualize it and I can't seem to see it in my head...
TECH Enthusiast
Joined: May 2007
Posts: 538
Likes: 0
Ok you are fimiliar with overlap. You seem to believe the reversion happens only in the intake manifold. It is the only place reversion happens I have yet to see it make its way out of the TB . What doesn't get discussed is low speed over scavenging that is from the overlap.
think of it like this.
If there is an exhuast pulse going down the exhuast system when a cylinder is in an overlap state that creates vacum in the adjacent header tube. That vacum and pull " much larger swing at low engine speeds" will actually suck fresh charge out of that cylinder.
Hence your oxygen wash.
It happens it exists and its real. Watch a header tube during the intake stroke and you can see the tube temp drop 40-50 degree from waste 0xygen hitting the pipe.
you can also quantify this by doing simple calculations based on Engine speed and airflow into the engine. You can see clearly it is overscavenging at low speeds.
think of it like this.
If there is an exhuast pulse going down the exhuast system when a cylinder is in an overlap state that creates vacum in the adjacent header tube. That vacum and pull " much larger swing at low engine speeds" will actually suck fresh charge out of that cylinder.
Hence your oxygen wash.
It happens it exists and its real. Watch a header tube during the intake stroke and you can see the tube temp drop 40-50 degree from waste 0xygen hitting the pipe.
you can also quantify this by doing simple calculations based on Engine speed and airflow into the engine. You can see clearly it is overscavenging at low speeds.
Originally Posted by SSpdDmon
I'll pass the torch off to you on this one Frost. I've all but given up on trying to explain why what we do works - right, wrong, or indifferent. Over in the Live forums, we've been watching this thread and nobody (from either site) has yet to fill me in on something new...
I'm still trying to understand how unburnt O2 is making it's way into the exhaust because of my cam overlap.?.?.? What kind of pressure scenario would allow for a stream of unused intake air to make it into the exhaust stream? I mean...the piston is either pushing exhaust gas out or pulling fresh air in, right? And, both valves are closed at TDC, right? So, where does it get by? Not only that, but why would it proportionally vary based on RPM if the amount of overlap doesn't change (re: "Most cams usually 'cleanup' above 3000rpms)? Just trying to visualize it and I can't seem to see it in my head...
I'm still trying to understand how unburnt O2 is making it's way into the exhaust because of my cam overlap.?.?.? What kind of pressure scenario would allow for a stream of unused intake air to make it into the exhaust stream? I mean...the piston is either pushing exhaust gas out or pulling fresh air in, right? And, both valves are closed at TDC, right? So, where does it get by? Not only that, but why would it proportionally vary based on RPM if the amount of overlap doesn't change (re: "Most cams usually 'cleanup' above 3000rpms)? Just trying to visualize it and I can't seem to see it in my head...