Why bother dialing in the VE table?

[pauly24]

Ok stupid question, but why bother? LTFT dial it in for you?

When the car first injects the fuel and its mixture is off, the cars drive ability will be slightly off because of the wrong mixture.
But once the LTFT have been learnt, the second time round it injects fuel it will be the correct amount. So there is no stumbling.

So whats the difference if you have it dialed in with the VE or you leave the LTFTs on?

[mr.prick]

I guess if you don't mind cold start fueling being overly rich or lean then yeah, why bother. :cucumber:
High LTFTs cause problems too.

[swingtan]

Because LTFT's are simply an "average" of all the STFT's and then might be applied to the PE fuel. Also remember that there are no STFT's during PE mode, so there is no "learned data" for LTFT's there. The fueling in "guessed" from the trims learned in completely different load cells. Think of LTFT's like the "Octane Scalar" of fuel correction, it works, but not particularly well.

Simon.

[WeathermanShawn]

Even though I think the concept of Trims can be advantageous for a car driven on the street, remember the corrective fueling is initially applied after the fact (STFT). In other words you are driving, O2's say lean..then the fuel correction is applied.

Radically large swings in LTFT's can make for a very poor driving vehicle. It will buck, hesitate, then bog. Your PE and WOT will be hosed.

Personally I like to get Trims (LTFT's) as close to zero as possible. If your VE and/or MAF is dialed in..thats called a tuned vehicle.

Enjoy..

[swingtan]

The other option is to simply try it out. Do a full VE ( & MAF is running one ) tune up and don't touch any spark tables. Then once finished, save the tune and flash the stock tune back in and for good measure do a PCM reset and ensure all trim data is cleared out. Then go for a test drive noting how the car feels. Pull over, flash in the "fixed fueling" tune, reset and clear trims again and do the same drive. That should give a pretty clear reason why it's good to dial in the VE.

Simon.

[pauly24]

So what is the best indication of a properly tuned vehicle, when you turn the LTFT on and they stay close to zero? Or do you guys leave them off completely and just turn the STFT on?

Also is there any difference if you use the LTFT to zero in the VE table or the BEN AFR correction?

[swingtan]

One indication of correctly dialed in fueling is that the trims remain within +/- 5%, this goes for both STFT's and LTFT's. In the past I have turned off the LTFT's and only use the STFT's, lots of people do this as if the STFT's are very close, the LTFT's will not make much of a difference. In fact ATM I'm running full OL so don't use any trims at all. I do have a permanent WB gauge though so I always can see what is going on.

LTFT's should not be used in any correction work as they will not give a fine enough result to accurately adjust anything. When adjusting the VE or MAF, turn off all trims and us a WB sensor.

Simon

[pauly24]

I logged my WB signal and used that to create the BEN factor.
Once I dialed in my VE, I got all the BEN factors to around +-1-2%

But when I turned on my fuel trims, it got all the BENs to 0, doesnt this prove that the fuel trims are trimming each load point to the EXACT fueling thats required?
Hence they would make a good offset for the VE wouldn't they?
I guess I can experiment, ill dial in the VE using the fuel trims, then ill turn them off and log the WB BEN factor, if the BEN factor is the same results I get dialing in the VE using the BENs, then using the fuel trims is just as good.

[WeathermanShawn]

I have had more success apply Trims (STFT or LTFT) to MAF. Less success with applying them directly to the VE Table.

If you stick with closed-loop you can do either STFT or LTFT. Again if you are not running MAF, applying the full LTFT correction to the VE Table may take some more mastery..

[swingtan]

Pauly, if you turn on the trim and then attempt to do a BEN, the BEN will "always" be "0".

read up on exactly what the BEN is and you will understand. The WB-BEN is a "Fueling correction factor" that is averaged over many data points. It is based on the commanded AFR and the measured WB AFR, the result being a correction factor. If you have STFT's turned on, then they will always adjust the fueling to be stoichiometric and if you are commanding a stoichiometric AFR, then the WB-BEN MUST always be "0". This proves nothing in terms of how accurate the fuel mapping is, just that the trims can correct ( hence why CL fueling is used for a generic tune ). The only time you can really use a WB-BEN factor is when no other "fueling correction" is taking place.

You can however set up a "STFT-BEN" that uses the STFT averages to create a BEN factor. It can be used to massage the VE and/or MAF where the NB-O2's are in use but cannot be used for correction in PE mode or if OL fueling is used.

Simon

[pauly24]

Yes thats my point exactly, the fuel trims will take the BEN to 0 every time. And thats what you are trying to do by massaging the VE table with an AFR BEN.
Hence the fuel trims that have been recorded are the exact adjustments needed to fueling to bring the BEN correction to 1.

I just wanted some more info about the cons/pros about using fuel trims for dialing in the VE compared to BENs in CL mode of course.

Weathermanshawn, can you explain why you've not had as much success applying fuel trims BENs to the VE table? I understand people saying its not as good as using AFR BENS, but being told is one thing, but to believe its true I need to be given a reason why AFR BENs are better to be used then fuel trims BENs as I can't see the difference in CL mode.

[WeathermanShawn]

I could have probably stated it better.

Here is what I meant. I have had more experience applying LTFT in a MAF-Enabled car. They fall in line pretty easy. For instance if your car is averaging +10% LTFT's across the MAF Frequency..adding 10% to the MAF Table usually gets your LTFT Average to zero.

What I was trying to say is that if you tuned your VE Table 100% Open-Loop (with wideband) and then went to close-loop and attempt to use your narrowbands utilizing Trims..it is not always a 1:1 Ratio. It could be the location of the narrowbands, accuracy, or inevitable fuel/exhaust delay that makes it difficult.

In the U.S. with EPA requirements, emissions, etc., many people run 100% Closed-Loop. Some choose to retain MAF, others go MAF-less..but Closed-Loop. That is know as CLSD. If you were tuning 100% Closed-Loop (stoich portion of VE Table), then applying LTFTBENS to the VE Table has far greater success.

I do not know if I over-explained it..but here is the bottom line. If you tune Closed-Loop, techniques like STFT's and LTFT's can successfully be used. If you know how to tune Open-Loop with a wideband..you are set. Its trying to mix and match that creates problems. I.E., using Closed-Loop techniques for open-loop do not work as well.

Does that make sense?

[joecar]

The above are good reasons to make sure the VE table correctly models the engine.

Also, consider these:

The initial calculation for airmass is made (using the VE and/or MAF tables) regardless of trims... trims are applied to fuelmass later, just before calculating the injector pulsewidth.

What happens when the PCM goes to the VE table for airmass (i.e. during transients)...?

Transient air/fuel is what makes good or bad throttle response (which involves steady state VE and wall wetting/evaporation)... why should it be denied a good chance to work properly...?

What happens if for some reason the PCM faults into a OLSD mode and your VE table is not correct...?
(this has multiple side effects: with an auto trans, the running pressure will be low if the VE table is low, this allows slipping at high load).

[Redline Motorsports]

Don't loose sight to the fact that factory narrow band o2's have a accuracy of +/- 3% vs. a high quality WB. Its not unusual to get everything in line with a WB and then reable trim control and see a change.

Regardless of trims, not correctly modeling the engines airflow mapping is going to create issues everywhere. When you do a moderate throttle roll on, trims are not really working that quick vs. having the wrong reported airflow.

Your best bet is OL and use a good WB

[joecar]

In absence of a wideband, MAF seems to correct via trims ok, VE seems to be better calculated (from MAF, MAP, DAT, RPM).

[pauly24]

Thanks guys, I don't question because I'm trying to say your wrong or argue, I just question because I'm trying to learn.
Im not one to follow what someone says blindy, if someone says something to me, but they can't back it up with an explanation I usually disregard whats been said. (Which I presume what most people would do if they are trying to learn something anyways)

[WeathermanShawn]

In absence of a wideband, MAF seems to correct via trims ok, VE seems to be better calculated (from MAF, MAP, DAT, RPM).

Thats what I was trying to say..:grin:.

Joe, I always like your 'Power-Point' type explanations. You use fewer words but you always get your point across.

Must be natural. My explanations never are that concise. And that was a very good way of explaining it..:cheers:

[joecar]

Thanks guys, I don't question because I'm trying to say your wrong or argue, I just question because I'm trying to learn.
Im not one to follow what someone says blindy, if someone says something to me, but they can't back it up with an explanation I usually disregard whats been said. (Which I presume what most people would do if they are trying to learn something anyways)That's fine, we can see you're asking good questions and the discussion that follows is excellent... we all learn from questioning and discussing... questions make people think. Blindly following doesn't promote learning; understanding why/how promotes learning. :)


Thats what I was trying to say..:grin:.

Joe, I always like your 'Power-Point' type explanations. You use fewer words but you always get your point across.

Must be natural. My explanations never are that concise. And that was a very good way of explaining it..:cheers:That comes from either being an engineer or not being able to type... :D

[joecar]

...

Regardless of trims, not correctly modeling the engines airflow mapping is going to create issues everywhere.

...This is a very important point that summarizes the whole point of it all.

:cheers:

[redhardsupra]

Heh, this question is simple, but only on the surface. The real question here is 'why is GM's ECU operating in a hybrid of open- and closed-loop fueling corrections?' But let's back up a bit first to make sure you got the basics covered.

Trims are corrective mechanisms. A fully closed-loop system doesn't have a starting condition, it is purely reactionary; it takes a measurement, compares to the desired effect, and restarts the process. In contrast, open-loop is predictive. You predict a certain outcome, and you stick to your predictions, regardless of the result.

It is easy to see the problems of both systems. Open Loop will never get better, and will never adjust to any changes in the system. Closed Loop doesn't have a good 'starting' condition, and depending on the system, the corrections might never converge on the stable solution. It is also 'naive' in the sense that a wrong batch of data will lead it astray, if it doesn't have a mechanism for sanity checks.

This is why GM came up with a hybrid of the two: you start with a prediction (airmass based on either MAF or VE/MAP/TEMP), you let the mixture burn, but then in a reactive fashion it observes what happened, and it remembers that correction for the next round. With both short and long trims, the 'remembering' process actually gets more complex, as the ECU tries to classify the observed change as either a temporary fluke, or a permanent condition. If that wasn't enough, there's sanity checks in the form of MAP and MAF and CYLAIR filters and limits.

So as per usual, a singular, simple mechanism isn't enough to catch up with the complexities of the natural phenomenons we are up against. We are forced to get creative and mix all the tricks we got, or we're not going to cover every possible condition, situation, fluke, act of nature, or sheer stupidity the ECU and its sensors will go against out in the wild. Don't fear complexity, master it instead.

[pauly24]

See this is the bit I don't understand, or believe a proper explanation has been given for.

Lets just say your in CL mode, car injects X amount of fuel, through the feedback system it realizes it needs X+2%, this 2% becomes the LTFT value.

Next time around, it will inject X+2%=Y, so now it injects Y amount of fuel straight off the bat and this is the correct amount of fuel. No stumbling of getting the incorrect mixture, now every time it goes to that same load point it injects Y amount of fuel being the correct amount.

So once all the LTFT have been set after a long drive, its not incorrectly injecting X amount of fuel resulting in incorrect mixture THEN correcting, its already corrected first time round so the amount of fuel injected is correct first time, just like it would have been if you massaged the VE table.

I dont see how injecting Y amount of fuel from X value + LTFT is any different to injecting Y amount of fuel that is calculated straight from the table because it would give the same result wouldn't it?

[redhardsupra]

What you're missing in your model is noise. And lots of it. Stuff you cannot account for. Wear, tear, bad batches of petrol, petrol temperature, wall wetting, and million other variables. You can set your correction to +2%, and guess what, 2 seconds later it might need -3%. It's about having a fully automated way of chasing that moving target, doing it precisely, and doing it a robustly, no matter what gets thrown at the sensors.

[WeathermanShawn]

See, this is what I can not understand about the Open-Loop vs Closed-Loop 'debate'. If a tuner is utilizing stoich throughout the drivable portion of the VE Table, how is a wideband advantageous at stoich. If we assume that at stoich a narrowband can be just as accurate as a wideband, why not go closed-loop. The PCM can accurately correct at stoich.

Pauly, I understood the first part of your question, but not the last. Closed-Loop can be very accurate if tuned correctly. I think the advantage of Open-Loop is when you have a genuine need to command an AFR other than stoich for your drivable portion of the VE Table.

Closed-Loop is a misnomer. In reality you are intelligently utilizing Trims to correct for Fuel type, Fuel temperature (watch your Trims/AFR after a gas fill-up), engine wear, etc.

Closed-Loop tuning is still Open-Loop tuning in PE Mode and WOT.

Pauly, I think you are mixing and matching tuning types in the same question. Trims can be your friend if you run stoich in your VE Table..

[joecar]

The trim is only applied to the calculated fuelmass before the injector pulse width is calculated.

[pauly24]

thats exactly what I'm getting at, whats the point of running open loop mode if you want to command stoich for the main drive ability part of your maps.

Isn't having the feedback system in place, with a correctly tuned VE table always going to be better then having a correctly tuned VE table and having the trims turned off?

I think maybe its just my misunderstanding of the general consensus on this board that everyone thinks fuel trims are bad, but I don't see it that way at all.
If you want to run stoich then fuel trims turned on are always going to be help full correct?

And the only reason why they should ever turn them off is because you want to command something else than stoich.'

Joecar: thats my point aswell, if your VE tables are slightly off, once all the long term fuel trims have been learnt, the computer will be injecting the correct amount of fuel first time around, just like it would be doing if the VE table was correct in the first place. So I don't see how drive-ability is affected either way?

[WeathermanShawn]

Well, those are exactly my conclusions (the first portion of your point). I think Closed-Loop gets a bad reputation because many in the past have just used it as a 'crutch' to avoid properly tuning the VE Table and MAF.

In my case, my wideband and narrowbands are in sync. If I read stoich (14.63 AFR) on the wideband, my narrowband are right at zero Trims. I like to think the 'new' breed of Closed-Loop tuners get their MAF and VE Tables as accurate as possible, but intelligently use the narrowbands superiority in Trimming. No human being can Trim as fast as the PCM. In the 'old days' we would have probably paid hundreds if not thousands of dollars to be able to have a computer successfully Trim to stoich.

My point to others who have a bad feeling about Closed-Loop is..I aim to get as accurate an Airflow Model with the MAF and VE Table..I use the narrowbands for the stoich portion, and Open-Loop for PE Mode and WOT.

Pauly, I guess we agree..:)

[redhardsupra]

See, this is what I can not understand about the Open-Loop vs Closed-Loop 'debate'. If a tuner is utilizing stoich throughout the drivable portion of the VE Table, how is a wideband advantageous at stoich. If we assume that at stoich a narrowband can be just as accurate as a wideband, why not go closed-loop. The PCM can accurately correct at stoich.

narrowbands tell us that we are lean or rich. widebands tell us HOW MUCH lean or lean we are. at stoich they're even. how does the computer convert the O2 voltage readings into trim values is magic.


Pauly, I understood the first part of your question, but not the last. Closed-Loop can be very accurate if tuned correctly. I think the advantage of Open-Loop is when you have a genuine need to command an AFR other than stoich for your drivable portion of the VE Table.

...which is almost never, with exception of cases where a rich idle must be induced to battle fake lean. And if the idle is that bad, you should go alpha-n probably anyway.



Closed-Loop is a misnomer. In reality you are intelligently utilizing Trims to correct for Fuel type, Fuel temperature (watch your Trims/AFR after a gas fill-up), engine wear, etc.

Closed-Loop tuning is still Open-Loop tuning in PE Mode and WOT.

no, it is not a misnomer, you're simply not using it properly. When CL gets turned off, it's OL. When OL gains any sort of trims, it's CL. THE END.



Pauly, I think you are mixing and matching tuning types in the same question. Trims can be your friend if you run stoich in your VE Table..

that I agree with, i think he cant tell the difference between the basic airmass estimation, and the corrective measures. two orthogonal issues.

[joecar]

...
the difference between the basic airmass estimation, and the corrective measures. two orthogonal issues.This is a key point.



PCM roughly follows this sequence:
- compute airmass from VE and/or MAF;
- lookup AFR to command;
- compute fuelmass required to meet that AFR;
- apply CL trim to fuelmass;
- look up injector flow rate;
- compute injector pulsewdith required to deliver that fuelmass;
- look at O2 response to determine new CL trim;
- repeat.

[pauly24]

Basic airmass estimation is through the VE table, it computes the Volumetric Efficiency, ie. what percent of the cylinder is full of air. Then through that can work out how much fuel to inject for correct ratio, rudimentary stuff.
Once all that is done, if the measured afr through the O2 sensor is incorrect, it trims the amount it injects next time round to match the desired AFR.

What im saying is, a correction is when it injects the wrong amount of fuel, and second time round it injects the correct amount of fuel. The second time round, fueling is correct from the get go, no incorrect mixtures are taking place. Hence once all the fuel trims are learnt, the correct mixture is injected first time round, even if it is calculating it from an incorrect VE table and a fuel trim.

[redhardsupra]

pauly, you're assuming you can nail the correction on the first try. you might. once. in your life. but not twice. Imagine using a sniper rifle, shooting a target far away, when gravity, air pressures, temperatures, the weather are changing 100 times a second. oh yea, and every time you shoot you must use a different weight of a bullet. do you still think one correction is all that's gonna take? ;)

[5.7ute]

Lets not forget that closed loop corrections are applied in Zones, not to singular load/rpm points. This in itself makes it a 100% reactive system. IMO, X+trim=Y is incorrect. X+trim=X. Hence the loop.

[WeathermanShawn]

Pauly, are you saying that a properly tuned VE Table and/or MAF with an absolutely correct airmass calculation..runs the same as a car that is constantly adding +20% Trims.

That I can not agree with. I still believe that the location of a narrowband (or wideband) makes it where the exhaust sampling and the PCM's response makes for a poor running car.

While I totally understand what you are arguing, a vehicle that is running +20 LTFT's..even though it technically will attempt to correct to stoich, will have O2 mv readings bordering on a Lean DTC. In addition PE will add +20% to WOT AFR.

There must be a delay, because a car running with +20 LTFT's will have wideband readings that go from 15.5-14.63, then 15.5-14.63. They oscillate.

I agree LTFT's can be a tool, but something sounds wrong in your logic..:confused:

[pauly24]

Weatherman im not saying that, ive been told otherwise and I'm not doubting it, but im just trying to figure out why. If what you said is correct about the AFR swingings then that would be the explanation why, but..

If you had +20 LTFT why would it ever swing from 15.5-14.63? wouldnt it stay constant at 14.63 because they fuel trim is constantly being applied every time before the injector fires.
Only the very first time it injects it would read 15.5 then it corrects it self, then from then on it will always be 14.63.

[5.7ute]

Pauly24, look at the operating zones for LTFT. 3 boundaries for map & 3 for RPM. Compare that to the 20x19 cells for the ve table & you will see the need for an accurate VE table.(or at least a consistant area in the cells that meet these zones). And lets face it, if it is consistantly wrong you may as well make it consistantly correct.

[pauly24]

ohhhh, see this I didn't know about the boundaries, I thought the LTFT were applied to each and every cell.

Where are these boundaries in the E38 ecm?

Whenever I have logged LTFT, my MAP columns were only 2 kpa apart and no two cells were ever the same, hence it made me believed that the LTFT were unique to each cell.

[WeathermanShawn]

Weatherman im not saying that, ive been told otherwise and I'm not doubting it, but im just trying to figure out why. If what you said is correct about the AFR swingings then that would be the explanation why, but..

If you had +20 LTFT why would it ever swing from 15.5-14.63? wouldnt it stay constant at 14.63 because they fuel trim is constantly being applied every time before the injector fires.
Only the very first time it injects it would read 15.5 then it corrects it self, then from then on it will always be 14.63.

Pauly, the problem is that my training is in Meteorology, so I lack the engineering background to properly explain.

But, what I think is occurring is a delay in the O2 reading (mv) and when and how the STFT's and resultant LTFT's update. My memory is good, as my car started out with Trims of +15 %. Vehicle is driven, Wideband reads 15.5-16.0, O2 sensor reads lean, Trims are added. LTFT's are updated..+15% will be added to next Injector Pulse. However, that Correction will not be added until the O2 mv reading hits the next O2 switchpoint..it is delayed by time (see attachments). When you are at say a +15-20% LTFT, it always seems that it does not add the fuel until it hits the next lean-rich switchpoint.

Basically, there is always a delay. No matter that it is adding, it adds it at the end of the O2 cycle. Pauly, I have probably butchered the pure engineering response. Perhaps one of our outstanding engineers will brief us. But I am 100% sure that you will constantly fighting very lean when adding +20 LTFT's, and ironically be very rich in PE Mode. When I first tuned with +15 LTFT's my wideband went 16.0-14.63-16.0. Now with +/- 0 Trims it goes 14.9-14.63- 14.2-14.63..etc..The average O2 mv went from 400 mv to an average of 540..

Here are a few attachments to ponder. Like I said I never had a mechanical engineering course in my life, so this is the best common sense answer I can give.

Can anybody clear it up for us?

[5.7ute]

You may log differences between cells when mapping out averages etc, but this isnt the case. LTFT will hold the last correction between those boundaries until filtered STFT's tell it otherwise. Not sure on the E38, I will get back to you.

[5.7ute]

After having a look there appears to be no defined boundary settings in the E38 cal. This doesnt necessarily mean that they dont have them though. Quite possibly they use some other method due to the way they calculate the airmass through coefficients. They could also quite easily tie them in to the operating zones. But that is just a guess & not good science.(Dont shoot me Marcin)
Any chance you can post up some logs Pauly24 with MAP,RPM,STFT & LTFT?

[swingtan]

The only real fueling "zone" tables in the E38 are B8020 and B8021. They may be dual purpose and work for both the VVE and for LTFT zones, they may not. I haven't run any LTFT's for so long now that I don't even think I have any logs with them on....

Simon.

[pauly24]

Sure, this is a quick log of the car with its stock tune.8451

[pauly24]

Theres probably an undiscovered table somewhere, this is the first GM V8 car Ive played with, and when I opened up an LS1 tune I was amazed and how many more tables there was to play with compared to the E38.

What does a log of LTFT look like on an LS1? can you distinctively see that there are boundaries where zones have all the same LTFT?

[joecar]

Sure, a car with +20% trims might run the same as a car with +1% trims during any steady state condition...

where those cars deviate is during transient conditions and conditions outside of closed loop.

[joecar]

Example LS1 log.

[redhardsupra]

After having a look there appears to be no defined boundary settings in the E38 cal. This doesnt necessarily mean that they dont have them though. Quite possibly they use some other method due to the way they calculate the airmass through coefficients. They could also quite easily tie them in to the operating zones. But that is just a guess & not good science.(Dont shoot me Marcin)
Any chance you can post up some logs Pauly24 with MAP,RPM,STFT & LTFT?
hehe, all good science starts with observation and conjecture, it's what happens next that's important.

On a more technical note, if you don't know what the modifiers are doing, graph them separately from the VE. If you've done any of the old school VE tuning with narrowbands, you probably noticed that the trims are different in every RPM/MAP cell corresponding to entries in the VE table. So I don't think any of the really old 'block trims' really apply anymore, even on a LS1 computer, yet alone on a modern E38/E67. But again, that's conjecture, and it will remain so until our resident lord and saviours Ross and Paul will tell us how the code works. This of course brings up other questions, like 'then what are all the LTFT boundaries for?' or 'what is the resolution of the trim tables?' and 'how do LS1 differ from E40 and E38/E67 in respect to fuel trim setting and application?'

[5.7ute]

Mapping the trims will still show different values in different cells regardless of the boundaries because they are reactionary, not because they arent "block learned". The LTFT from one loadpoint carries over into the next cell, & then is further modified to suit the different conditions at the time. You really need to follow the trace & not a map in these scenarios to see how they work.