NefMoto

KFMI_UM has to be updated. I know this because without changing KFMI_UM I got a lot of jerkiness and also torque monitor 2 intervention.

If you increase IOP, you will more likely have torque intervention if you also increased IRL

Explain to me then why cars that approach more the higher load values on the axis have higher TQ values? like in the RS4?

I'm sorry but if the last load # in IOP is @ or below your max load you want to have you want to make sure you're upping the values in IOP.   It's always worked well for me and the engine most definitely develops more TQ.  If course IOP changes means other things have to be changed to jive in the ECU.  But lowering IOP in almost all cases neuters the car from what I've found.

Or the easy way: just shift the axis values for IOP:

http://nefariousmotorsports.com/forum/index.php?topic=970.msg8536#msg8536

develops more TQ = the car feels much stronger.

That said scaling up IOP towards the end shouldn't cause any part-throttle issues.

You want to REDUCE IOP, not increase it.

^^ that makes no sense, my tunes have always been stronger when increasing IOP when you want more TQ especially in the higher end of the load scale.

I am starting to think we would be better off not even touching IOP.

And you'd get torque intervention if you bumped too much of IRL.

The effect is well documented.

The only real solution is to leave most of IRL alone, except for the top lines, and shift the axis of IOP around accordingly.

Purpose of the Optimal Torque Map

What it does, how it does it, why it does it, and how to make it work for you!



Purpose: Take the calculated torque value at a given RPM & load, then compare it to the value that is present on the map. The ECU then corrects for the difference by adjusting the 02 correction value, the requested boost value, and the timing value.

02 correction value: This value is the change in lambda that the ECU requires to reach it's current TARGET lambda, If an 02 correction value of 0 is recorded that means 1 of 2 things has occurred.

1. Your datalogging and reading values from the engine just turning on, the engine requires some time to begin correcting through the 02 correction value, lambda sensors must heat up etc,etc.

2. The correction the ECU needs is SO great it has gone off the 25% range the 02 correction has to offer. A significant change in the value in this area is going to be required.
This map is in place to check the MAF sensor's readings to the ECU. The Mutiplitive and Additive fuel trims are derived from the 02 correction values, The ECU then uses these fuel trims, along with timng and requested boost to effect the Actual Torque of the car so that the driver does not feel the effects of air tempurature, increased load driving(going up hill) and such on the power production of the engine.

You have to understand, bosch doesn't trust the millions of people who bought cars with their ME7 ecu system in it to understand the adverse effects real life variables have on the torque production of a turbocharged engine. What they DO understand is that if the average person presses the pedal down 50% of the way, they want the car to "FEEL" like it is producing 50% of it's power wether your driving up to the mountians or passing some one on the freeway in florida.

*in the average drivers mind 50% pedal means 50% power*

Most of this simulation is processed and converted by the ECU on the Acceleration pedal map and the target filling map.

In order for the ECU to know whether or not the it needs to make the engine produce more or less torque in order to comply with drivers demand of current Load(value measured by MAF or calculated if running mafless), it needs a value to check against. This is the value you are adjusting on the optimal torque map.

Now since the optimal torque map effects many variables off a single value, it is safe to assume that you can adjust for 02 correction and know that the effect of requested boost and timing from the optimal torque map will also be in line and if you have problems in these areas after getting fueling set you need to begin to adjust other maps to affect those values.


Types of Fuel Trims and how they work

The ECU has 2 Fuel Trims: Additive and Multiplicative both are applied globally to the fueling but they differ in HOW they are applied

Additive: A static value applied to the trim UNAFFECTED by changes in load.

Multiplicative: A value applied to the trim that is AFFECTED by changes in load.

The additive trim is used to make corrects that are happening at the same rate irregardless of load or at lower load levels like idle and cruising

the Multiplicative trim is used to make corrections that happen at an increasing rate as load is increased and in the higher load ranges.

What all this means is that when your tuning your 02 correction values you want them to be ether ALL positive or ALL negative.

If your 02 values swing between +/- your optimal torque maps need some work.

THe Ecu can only make a trim if a problem is reoccuring.

If your 02 correction value at the same load at
2000rpm is -5
2500rpm 3
3000rpm -4
3500rpm 2

while is might not seem to bad in reality this is not only improperly tuned, your actually causing the ECU to fight against your tune when it creates it's fuel trims!

This is what the ecu is thinking

2000rpm take away fuel
**** 2500 quick add fuel
wtf 3000rpm take away fuel
seriously... 3500 add fuel

this type of adjustment does not fall in line with the 2 available adjustments the ecu can make with the fuel trims.

Obviously we want to get our fuel trims as close to 0 as possible. But it must be done in the correct way or else your going to assume your tune is spot on when in reality it's just that the ecu can't make an adjustment because your optimal torque map is so off!

How to make this all work for your Tune
Adjusting the Optimal Torque map using 02 corrections as a guide

You MUST deactivate any corrections on your main fuel correction map, this is the pre correction map for the lambda targeting fuel algorithm and will skew results if any value but 1.0001 is present in the cell.

*Note* the values of this table compare to ACTUAL torque seen by the torque model, NOT the REQUESTED torque derived by the acceleration pedal map, I will explain the different in a different post.


*NOTE for setting up Y axis, The final column value should be the LOAD value where you car produces Maximum torque, this just happen to be the max load that my car reached... with my turbo (ported ko3s) this over shot my target filling value by 5% so while the highest value I request for load is only 200% my Y axis, last column on optimal torque is 205.5 since that is the max load my car will reach that I have seen, if it is completely unrestricted by the max VE table. This number has risen as my tune can getting more spot on, I can only assume this means I am producing more power, butt dyno agrees heavily



Do NOT fall into the trap off thinking you much adjust lower range values if you change the value of higher columns, you shouldn't have to unless you are raising the last column because of a pure 0 reading when approaching max torque, a very rare instance indeed, other wise you should not need to adjust values in lower load area's as produced torque should be the same at other load levels.

First you want to make sure there are NO recorded 0's in your datalog in the area we are going to be adjusting.

Any 0's must first be dealt with or your gonna be chasing your tail, A pure 0 is an indication of the adjustment running out of range... A BIG change will be required to bring this load and rpm point into line so that we may begin fine tuning. Use common sense with these adjustments, if your adjusting 151% load out of a max defined 200% and your opt torque value is 30%... you probably need to raise not lower to get it back scale.

*note values of 0.23 etc are perfect and infact mean your basically spot on at that point*

*NOTE* Other maps affect fuel trims so make sure you account for acceleration fueling map and it's effect on the fuel trim during sudden throttle requests. It is better to only tune portions of the opt torque map in which a steady load over a good portion of the rpm range is logged.

Now once you have no pure zero's the next step is to make sure your 02 corrections are either ALL positive or ALL negative. I shoot for negative because I rather be richer and have the ecu remove a small amount of fuel then rely on the ecu to add fuel, that's just me how I feel safer, also that's how I know I'm not "missing" out on power so to speak.

a Datalog that shows you have all the 02 correction values either positive or negative you have a choice:

Be done!, yup you could fine right here, if your getting recorded values of all negative or positive numbers your in line "enough" that the ecu can take care of the rest and you'll have a pretty decent tune.

Not be a slacker and really dial that **** in!

Okay I made a post about interpolation of points between mapped points and how the ecu comes up with these. An understand of this concept will greatly reduce your dial in time if you don't have access to an interia dyno where you can run the car at a set load through an rpm range constantly.

What your going to do is look at RPM points in the data log that you have defined on your X axis. Then check the calculated load at that point.

For MAFLESS people: *NOTE* I have not even turned the key on a mafless car let alone start to tune one this is a guess BUT I'm thinking this check would be how you would dial in your TB tables and such... You would want to make sure the calculated load matched what your spec load would be that way you know you are on the right spot on the optimal torque table. But that's tuning theory and I can't get into that atm(no time and outside of the scope of this already long and probably confusing post)

then you check the 02 correction, if it's negative your gonna to decrease the value of your optimal torque table in that cell.

Now the reason you are looking for defined RPM area's is because if you have all those in line the values in between those areas will fall in line too.

I gotta end the post here and get to class, but that's really all the basics, if you understand interpolation you can use the changing of 02 correction values to estimate what changes need to be done to cells of the map you might not have logged,

if you don't just stick to the defined values and only adjust based on calculated load, if calculated load falls between 2 defined values, average the defined values, if the calculated value is higher then average adjust the higher defined load cell, or vise versa.

Remember small adjustments are best, even with recorded zero's I would only make 5% adjustments at most

One last thing if you have Positive numbers where you have it marked as 100% opt torque(last column is the only place this is gonna happen) you need to RAISE the Y axis value of the last column! The car is telling you if produced max torque at a higher load level then the defined axis you either

need to bring the axis up and readjust the other values in the column accordingly, (you want to tune for more power)

just allow for the ecu to make that correction in that case (a positive value less then 2 in this case is probably just from the car not completely cooling down between runs and heat effecting power output)

Limit the max VE to avoid hitting that load level again ( you want to limit power produce or boost for reliability reasons etc.

damn that's longer then I expected, I'll check it over for spelling and clarity after I get back from class, post any areas of concern or questions I can help with and I will try and add the info in

Interesting. I didn't think ME7 made fueling decisions based on target torque.

- IRL/IOP relationship : milsol_w only

- IRL torque intervention : rlsol_w = rlmin_w only

Given the IOP/IRL torque intervention thread, what causes torque intervention (other than ARMD)?

Yes, understood, but under what conditions, specifically. Are you saying the information in the IOP thread is wrong? And the logs are wrong?

Do you have any logs showing torque intervention and a clear explanation about what conditions lead to it?

Where does IRL come into play though?

miist goes to CAN only.

misol and mifa are definite contributors and IOP influenced.

I'm not necessarily trying to prove anything one way or the other, but I would like to clearly define how it occurs so that we can properly tune and not just find something that works.

I'm pretty sure it doesn't.

If any details were included it would be much easier to try to figure what the author is talking about.

because you need to make it easier on the ecu to control fuel at WOT (imagine the ecu attempting to go from lean to rich to lean to rich +/- 5% all the way through to redline

Bull. There are no short term trims when in openloop, by definition.

Which is why I welcome constructive criticism backed up with factual data/information.

I don't understand how you can say a VE table has no influence on fueling.

KFMIOP - Titled in maestro editor as the optimal engine torque map. Which is the main Volumetric Efficiency map.

KFMIRL - target filling map which is the driver requested torque based off of throttle input and requested kpa map value.
and  KFMIOP - Titled in maestro editor as the optimal engine torque map. Which is the main Volumetric Efficiency map. RPM vrs. cylinder filling that is a function of engine load. For those that don't know VE is an indicator of how much pumping is being done by the engine at any point since the engine cannot always fill cylinders completely. Any modification to the intake design, intake/exhaust temps, exhaust, charge piping, or cams that would change air density, velosity or flow this map will be out of calibration(how much will depend on application). Which is all common knowledge in EFI tuning. Airmass is a function of engine volume, number of cylinders, map value in KPA, and intake air temp. Total airmass and required fuel should be derived from that calculation.

KFMIRL - target load from torque/rpm. The torque input is far from driver's request.

KFMIOP - The output is torque and just like optimal ignition angle it is almost exclusively used in intervention.

Once we can see the actual torque put out by the engine in the data this will be easier to see. if it doesn't the car will begin to make fuel and timing changes to try and get that in line

I'm not sure I agree with the maf being the only input into airmass calculation(all the time). My theory: Bosch uses both speed-density and MAF readings to achive requested AFR's dependant on the situation where they are most useful (cruise,WOT and part throttle).

I'm starting to agree with your ideas behind torque monitoring it has a direct effect on fueling, timing and boost under certain conditions such as when actual exceeds spec load.

I have analyzed and logged most of the torque model and it seems most interventions come from the CAN variables being less than mifa_w. Basically, the goal in MDKOG is to get mifa_w to the end. I'm going to shift my focus to MDFAW.

The ecu is constantly calculating an internal torque value, on a scale of  0 to 100%
This torque value is then corrected for actual lambda (vs. 14.7:1), actual timing (vs. Optimal timing, and this is often a theoretical value that cannot be achieved due to the knock limit)

This internally calculated torque value is then compared to the value from the optimum torque table, and the ecu may then take steps to alter engine torque if the values do not match.

Interesting... Yes I'm sorry I should start adopting the correct abbreviations into my responses to steer away from confusion. I'll work on that, considering that I've only used maestro working with me7. So let me ask you this... What are you looking to accomplish? Are you simply looking for the source of torque intervention? From my understanding as long as your KFMIOP is properly calibrated as well as the inverse KFMIRL you shouldn't see any torque intervention via fuel, timing or throttle.

Why does KFMIRL have to be the inverse? It doesn't it is just logical. KFMIOP is load to torque and KFMIRL is torque to load.

If KFMIOP is NOT adjusted to be the proper inverse of KFMIRL (btw stock isn't entirely the perfect inverse), is undesired/unexpected torque intervention the result?

This calculation uses the *optimum torque* table as a starting point.

IMO it is compared to misol, which *isn't* from the optimum torque table. Again, i think the "actual" torque is calculated using the IOP as a starting point.

I could be wrong on this, it is from memory and I dont have the FR handy to check.

misol != mifa means that mifa got adjusted on it's path through MDKOG or mifa > miszul (max permissable torque from MDZUL). This can trigger ignition angle intervention.

A complete understanding of the torque model. No more "it is my understanding" statements, witchcraft theories, assumptions, or being stuck on one line of text in the FR. I want factual, substantiated information.

You say properly calibrated KFMIOP, how does one properly calibrate this map (kinda rhetorical, we already heard that fairy tale)? Why is that the proper method?

Does any part of the torque model need to be calibrated? Why? What is gained from adjusting KFMIOP? Higher torque input to KFMIRL? Why not just make up for it in KFMIRL?

Why does KFMIRL have to be the inverse? It doesn't it is just logical. KFMIOP is load to torque and KFMIRL is torque to load.  mimax from KFMIOP can basically be the input to KFMIRL (mimax --> mivbeb --> mifa --> mifafu --> milsolv --> milsol).

No more myths. Just facts.

From memory, by "actual" i mean "mifa/miist"

I thought mimax and mifa/miist are used in places other than IRL.

What is the point to converting "actual" (mifa/miist) load to torque (through IOP) then back to "req" load (misol) again (through IRL)?

What side effects occur if IOP is not the inverse to IRL?

I wasn't attacking or attempting to insult you... I was asking you if adjusting KFMIOP can solve the issue with torque intervention why go any further? What is gained from KFMIOP is smooth/refined power delivery. Higher input to KFMIRL will result in more requested torque rlsol_w. You can change KFMIRL independantly from KFMIOP (I have done it with ill effects, havn't changed the axis though) but they do appear to be setup inverse to one another. I wasn't saying they MUST be inverse.

Are you using ME7L yet?

I though that if IOP isn't calibrated right, misol < mifa, which leads to torque intervention.

Not just yet but I am currently in the process although everyone in gods creation around western NY is APR tuned which makes reading the ECU file a problem. Have any pointers to get around this? I know I have to have a backup file incase the "progamming not finished" error pops up(does this only happen when trying to flash or reading as well?) or shall I just take the chance at reading the ECU and if the problem occurs go from there.

I now understand the questions that you are asking and I'll have to take some more time to read through the bosch document so that we can start speaking the same language here. Also get my car back on the road or grab a friends so I can start digging a little deeper and back my claims up with logged info. I appreciate you guys being patient with me.

What ECU?

If your original file is not already available on this site, there is a section for requests that does get good results.

Didn't see this one earlier...

misol < mifa is a sign that torque intervention has occured. mifa is desired torque at the beginning of MDKOG and misol is the result at the end after all the other torque requests have had the chance to adjust desired torque. If the end result is less than the original request then torque intervention is present and ignition angle intervention will be triggered.

All the other torque requests will be 99.99 unless they are requesting intervention, so mifa will make it to the end and if mifa < miszul then misol = mifa and there will be no intervention.

Obviously I'm summarizing and generalizing, but that is a good overview of what I've seen.

Thanks for the summary, the torque model is pretty intensive to go through. I have done a few fast skims through a couple of sections and was until now plotting mifa_w and mifab_w(didnt have mimax_w in my logger definition), and was aiming for mifab_w = mifa_w.

EDIT: I have made 2 files now with changes to KFMIOP to see how/if it affects my mixture control.

midab_w == mifa_w only shows that mifa_w <= mimax_w. Which should typically be the case. Unless maybe one of the sub-fuctions in MDFAW is actually raising desired torque after  the min(mivbeg_w, mimax_w).

However, mifab_w > misolv_w is also compared and enables some intervention conditions that I can't remember off the top of my head.

Yes, thats how I figured also.

Here is a plot with mifa_w, mifab_w and misol_w:
(http://img41.imageshack.us/img41/5940/bische120120912152253.png)

misol_w = mifa_w even though mifa_w is ~1.5% over mimax_w, I have to go over the functions more but this leaves me to believe there is some tolerance before intervention is induced.

I'm sure you haven't forgotten, but in all this discussion you should distinguish between Level 1 and Level 2 :)

Rick

I'm sure you haven't forgotten, but in all this discussion you should distinguish between Level 1 and Level 2 :)

Rick

Or have no way of removing a limit in Level 1  :)