NefMoto

Since I pretty much do not tune anymore, here's some info on how to scale aftermarket injectors, when you have no datasheet for them.

Let's say you have injectors on the car for which you do not have any clue what the TVUB could be, maybe you're not 100% on the flow rate either.
Let's assume for the moment that everything else on the car is linearized - because if not, then good luck to you. It is still doable, but without a huge amount of experience and knowledge you will not find out what's scaled wrong, so it is outside the scope of this topic.
Also before starting, pressure test the car and make sure there are no air leaks whatsoever anywhere, otherwise you're gonna have to do this all over again after fixing the leaks.

1. Set FKKVS to 1 everywhere.
2. Set KRKTE into the ballpark that you think it could be. By far the easiest way is to divide the OEM injector size by the current injector size and multiply KRKTE by this number, instead of trying to do theoretical calculations.
3. Take TVUB from some random injectors, that are at least the same type - e.g. EV6, EV14.
4. Start the car, if it doesn't start, you probably messed up really badly, so try to correct your error (or remove and clean the sparkplugs if you totally flooded it).
5. Let it warm up fully, so it's completely hot and there is no load on it.
6. Find KRKTE value so that you have more or less 1-2% fuel trim on idle with the hot engine.
7. Put some load on the engine, e.g. turn on the aircon, rotate the power steering, slip the clutch a little, put it in D if it's an auto instead of P/N, whatever. Observe what happens.
a) If the fuel trim goes positive, increase KRKTE, decrease TVUB at the voltage where you are testing (or just multiply the whole table), so FT is still good on idle, but does not drift away on load.
b) If the fuel trim goes negative, decrease KRKTE, increase TVUB at the voltage where you are testing (or just multiply the whole table), so FT is still good on idle, but does not drift away on load.
Usually it is good to change TVUB in 2% increments. Repeat this process until it's more or less stable, then proceed to 8.
8. Set the car on idle, fully warmed up. Increase the RPM, to 3000, then 6000 rpm without any load on the engine.
a) If the fuel trim goes positive, increase TVUB, decrease KRKTE to compensate, so that you still have good fuel trim on idle.
b) If the fuel trim goes negative, decrease TVUB, increase KRKTE to compensate, so that you still have good fuel trim on idle.
Repeat this process until you can slowly increase the RPM on hot idle and the fuel trim doesn't drift away.

And you're pretty much done. The rest compensate using FKKVS. Beware of large changes (more than 5% on high RPM), this most likely means your fuel pressure is dropping -> verify with a fuel pressure gauge.
Make sure you do not make any sudden movements/load changes, as your tip-in/tip-out enrichment might not be correct - those can be adjusted via KFBAKL/KFVAKL by making sharp throttle movements and observing what actual lambda is doing, but only after you have set KRKTE and TVUB correctly.

Very interesting, thanks for taking the time to share  :)

Thanks prj, excellent summary. I should add it to the wiki.

Tiny suggestion to add remark for ethanol content in fuel - it will impact FC values by ethanol settling to tank bottom on car standstill/phase separation(water content).

Completely irrelevant if the car has been running until warmup, the fuel will be evenly mixed on any fuel system with a return.

Good stuff, thnx  :)

I'm like a week too late to the party, maybe this is useful to someone.. Thanks prj!
http://nefariousmotorsports.com/forum/index.php?topic=320.msg151363#msg151363 (http://nefariousmotorsports.com/forum/index.php?topic=320.msg151363#msg151363)

I think there are some inaccuracies in this algorithm above.


It is not specified which trim to watch? If you mean STFT, it will be 1-2% after several minutes of running the engine while warming up.
While the engine warming up Idle LTFT will be adjusted and STFT will become 1-2%.


Doesn't point 8 make point 7 meaningless because it is the opposite of what we did a while ago?
If we set up an Idle and now upset it, where will it end?

Total fuel trim. Disable LTFT if it's easier.

No, it is not the opposite at all.
Understanding how the hardware works helps to avoid asking such questions.

The last step is fine tuning. Read the whole post.

You don't need to understand how the hardware works to see that point 8 is the оpposite of 7 :)

Except it is not "the opposite".
If you knew the hardware, you would understand that point 8 is the final step, needs to be done only once and has a very small effect.
The reason it is there, is to account for non-linearity of injectors, and due to the fact that the engine behaves differently in different airflow situations. To make sure you do not bake that non-linearity into TVUB, where the non-linearity should be handled by FKKVS.

But hey, you can do it however you like.

Just keep in mind, that there is more than a decade of actual experience between every word and letter written here.
Something you seem to lack.

It needs to be done exactly in that order, period. If you do it in reverse order or you do one step, but not the other, you will never have a good result.
The input to the first part is increasing load at a fixed engine speed in a low airflow situation, the input to the second part is with much higher airflow but lower load. The engine behaves completely different with more airflow.
Two completely different inputs. The fact that the adjustment is worded one or the other way makes no difference. You are tuning the same two parameters. Your assumption that somehow 7 and 8 cancel each other out is BS, which shows your complete lack of understanding.

7 gets the rough KRKTE and TVUB parameters, 8 gets the correct/precise KRKTE and TVUB parameters from that. You can try to do only 8, but good luck to you. You better know all the math in your head eyes closed, and the engine better not have the tiniest air leak.

Good shit.

A little thing you can do if your dumbass decided to change the injectors and MAF at the same time is you can force your lambda target to see if your fueling inaccuracy lies in the MAF or injector scalings, for example:
-If you target lambda 0.99 (ie force open loop) and get lambda 1 on your wideband, you're MAF+injector scaling combined is accurate at that given load point (this is easy to accomplish)
-If you target lambda 0.8, and get lambda 0.9, then your injector isn't running long enough - your injector has a lower flow rate than you thought. Fix your KRKTE and adjust MAF scaling to compensate.
-If you target lambda 0.8, and get lambda 0.7, then your injector is running too long - your injector has a higher flow rate than you thought. Fix your KRKTE and adjust MAF scaling to compensate.

...or better yet, learn from that dumbass who did it and change one thing at a time :)

No, it does not mean anything. This is not how you check for correct MAF scaling. Both can be wrong.
No.

Yes precisely, perhaps I should have clarified that better when I said
What I mean is that together they are resulting in your target AFR being met, and that an error in both can combine together and cancel each other out as long as your lambda request/fuel trims stay the same as when you calibrated.
Was there a fact I got wrong here? Obviously you shouldn't calibrate this way, but is my reasoning for actual AFRs not matching requested wrong?

Yes it's wrong.

MAF affects load. Having incorrect MAF calibration means that everything in the engine from throttle and knock control to ignition is wrong.
Your approach of garbage in, garbage out makes tuning and getting anything right a nightmare.

This is a perfect example of how not to tune.

Yes, thank you for clarifying what I have said twice now.

In my particular case this is what worked best for me, as I did not have data for my MAF housing or my injectors beyond a rough estimate of what the injectors flowed. Yes, I should have done one at a time, but I still stand by this method of forcing AFRs to confirm where inaccuracies are hiding, particularly since we can't rely entirely on seeing if ps_w==pvdkds_w to verify MAF accuracy at light load, especially when we may also be fighting an unknown TVUB in these light load areas.

Shit, don't take this the wrong way but I could even argue your method in the OP is, the perfect example of how not to tune since you should just have correct injector data from the manufacturer instead of making assumptions, but that would be retarded and completely ignoring the fact that this is tuning and that there will always be areas that need to be tweaked here and there to get things right.

The only thing that manufacturer data will help you do is set TVUB and KRKTE, it will not help you with fuel system and injector linearity.
Furthermore, using the method in the OP will have the ecu calculate the correct load and correct fueling.

Your method is pure garbage in garbage out. As soon as load is wrong in the ECU, everything else is wrong too.
TVUB, KRKTE, FKKVS - this affects only fueling, nothing else. If you try to randomly adjust KFKHFM for changes in fueling system, load will be wrong everywhere.

This statement shows again that you do not understand load in ME7.
ps_w in a modeled car is based on KFPBRK/KFPBRKNW/KFURL/KFPRG, not just on the MAF. If the MAF you use is not a random sensor in a random tube, but rather a larger OEM unit, then yes, you can enter the characteristic and then use the other maps to get ps_w more or less right at WOT. Furthermore on a car with a MAF ps_w matters very little, it's only used for a few things.

If fitting a random MAF, random injectors then you do not know anything about anything anymore.
People who can make that work right don't need advice from the forum. You are certainly not going to get it right. As I said before. Garbage in garbage out.
The method described in the OP is nothing like it, don't even go there.

I appreciate your technical posts prj, and I don't disrespect your insight because you don't say what I want to hear. I suppose we're starting to really go off tangent to the OP, but would you argue that relying on ps_w for MAF tuning is a poor practice? I will admit that my understanding of the load model isn't perfect, but as I understand it, (rlroh_w - rl_w) -> intake manifold integrator magic <-> ps_w -> VE model magic -> rl_w, where, generally speaking, rl_w will be pretty close to rlroh_w in steady state (this is why you say ps_w is partly irrelevant).

In my case, I did not tune relying on STFT+LTFT, rather forcing lambda to a specified value (typically 0.99 or 0.90) and logging lamsbg_w, actual lambda (thru the downstream ADC), ps_w, uhfm_w, nmot, and with enough data from street driving I was able to see where problem areas are hiding, either in the MAF or injector data - the key here being that errors that show up when lamsbg_w gets changed implies a problem in the injector data, while problems that persist implies a problem in the MAF data. The end result is that my ps_w is very close to actual manifold pressure, and whatever lambda I request is the lambda I get. My STFT+LTFT stays within 5% overall in closed loop, which I consider good enough considering how pump gas can vary - maybe this might not be the case on a very high powered car where going 5% lean could result in instant engine deconstruction.

I am a mechanic by trade and do not have access to a dyno or any testing equipment, and this was done some time ago as my first hands on tuning experience, using injectors I got for free and a MAF housing I paid something like $5 for off a Ford F250. Would I do it this way again? No, unless I wanted to double check a tune as it's just about complete. Do I think this could go wrong and give you garbage data? Yes, probably. When I do my turbo ford I do plan on actually testing the injectors I use to get a good baseline for what injector breakpoint, high/low slopes, offset, and all associated pressure compensations before installing them since figuring out injector data from scratch on the car sucks no matter how you do it.

Exactly, the only thing that the manifold model does here is some filtering, but rlroh_w and rl_w are almost equal. The whole rlroh_w thing is a hindsight thing, the ECU was designed speed density first, and then the MAF was tacked on. I would say that ps_w is largerly irrelevant on a MAF based car, because it is used only for a few things like brake booster. Everything else uses load. If ps_w is in the ballpark, then it's good enough. Totally different story on a speed density ECU of course, where the whole calculation matters a great deal, since load is calculated from it. On HFM the calculation is applied to get ps_w from rlroh_w and then the same calculation is inverted to go back to rl_w - you can probably see here that it does not matter a great deal what is in the VE model, as it does not really affect rl_w.

This is wrong. Injectors and fuel system are not 100% linear, so no, this does not imply anything whatsoever.
You have no way of knowing if the problem is due to MAF linearization or due to fuel system non-linearity in this instance, and that is the exact moment it becomes garbage in, garbage out.

Thanks a lot prj.
It helps a lot,good luck with your next life. :)

With this method, is the change between TVUB and KRKTE equal? If say you lower TVUB by 2%, do you increase KRKTE by 2% as well?

Definitely not.

Nice post prj.

Does this assume that the load is directly reflected in the changes in battery voltage? I must be stupid but I don't see any point in this, very bad way to adjust the injectors. Could you please enlighten me.

What I would do is modify the voltage regulator of the charger and add a potentiometer to it and adjust the tvub with it.

I think you need to do some more reading before coming to this topic, this went completely over your head.
I am describing how to get from point A to point B, you don't even know at this point what A and B are...