Dumb beginner questions...

[Rich Z]

GM.APP (Accelerator Pedal Position) seems to tie up with the idle parameters (cracker/follower) that I see.

BTW, that was exactly what I needed! Thanks! Now my pedal position actually shows zero percent when at idle.

[Rich Z]

Truth be known, I would rather capture ALL of the data streams in a log, and then simply filter which ones I want to look at at the dashboard and map level (instead of the PID data level) when I am analyzing the data after the log was taken. This just seems to be a better utilization of my time than making multiple passes for data logs, setting up the needed subset of PIDs for each subsequent run, and HOPING that I have all the data collected in the same run that I might want to compare the interactions of at a later date. Obviously some of the PIDs will have to be duplicated in each run, regardless, which is just a waste of bandwidth. Anyone who has done database programming probably understands my perspective of this.

Found myself thinking about this the other day and it brought a BASIC "how the heck" type of question to my mind.

HOW does the PCM know what sort of data to send out when we are doing a data logging capture? Does EFILive tell the PCM what data stream it wants via PIDs selected for logging? Kind of like "Hey PCM! This is EFILive. I'm going to be needing the data for ONLY the following PIDs. OK?" I have to assume that ALL of the data is not available at all times coming from the PCM, so I am presuming that there is some sort of hand shaking data taking place that sets up the PCM to generate the data stream that is wanted.

Yes/No?

[joecar]

Yes, scantool tells PCM which pids it wants.

[Rich Z]

I know I've probably looked this up or figured it out in the past at least a dozen times, but do positive trims (STFT and LTFT) mean that the fuel mixture is actually LEAN or RICH? I can see how it could be either way, depending on how you look at it. On one hand it could be saying that it is lean and the PCM is adding fuel (therefore "+"), or it can mean that it is rich (therefore "+") and the PCM has to subtract fuel. There seem to be instances of negative logic being applied throughout this tuning stuff, with little guidance of WHEN that is the case.

[Rich Z]

And speaking of negative logic, I need to make certain that I am looking at something correctly.

When I created my map from B0101, I'm using the data parameter from BEN factor Bank 1, Serial Wideband, LS1 style (factor).

After doing data logging, I run the suggested filter and then cut and paste with labels the values in my map. Then I paste with multiply those map values against the B0101 table.

From what I can figure, it looks like the map values over 1.00 are telling me that the cells are showing lean, which when multiplied to the respective cells in B0101 make those VE values LARGER, indicating that the PCM has to add more fuel for those affected cells. In the case of the map cell values being lower than 1.00, multiplying them against the relevant VE cells in B0101 makes those cell values SMALLER, indicating that the PCM needs to add LESS fuel because the map data was indicating richer mixtures.

In other words, the larger the VE cell values, the more fuel is needed to be able to reach stoich. If the VE value in a cell is too small, the data in the map will indicate that the cell is running too lean, so fuel needs to be added by increasing the efficiency value. When the efficiency value in a cell is too large, then the collected map data will indicate that the cell is running too rich and LESS fuel needs to be provided by decreasing the efficiency value.

Of course, it's entirely possible that I have this COMPLETELY backwards in my mind.....

[joecar]

This is a trick question that everyone gets wrong:

I know I've probably looked this up or figured it out in the past at least a dozen times, but do positive trims (STFT and LTFT) mean that the fuel mixture is actually LEAN or RICH? I can see how it could be either way, depending on how you look at it. On one hand it could be saying that it is lean and the PCM is adding fuel (therefore "+"), or it can mean that it is rich (therefore "+") and the PCM has to subtract fuel. There seem to be instances of negative logic being applied throughout this tuning stuff, with little guidance of WHEN that is the case.

Neither...

but rather the STFT and/or LTFT mean that airmass is either:
- under-computed: PCM adds a proportional fuelmass, but this is under sufficient, in which case the HO2Sx1 voltage spends most of its time low, this causes LTFT to go positive to maintain trim;
- over-computed: PCM adds a proportional fuelmass, but this is over sufficient, in which case the HO2Sx1 voltage spends most of its time high, this causes LTFT to go negative to maintain trim;

i.e. the trim is an indication of the cylinder airmass being under or over actual airmass;

maintaining trim means that the LTFT is compensating for cylinder airmass being incorrect (remembering that fuelmass is proportional to airmass).

[joecar]

And speaking of negative logic, I need to make certain that I am looking at something correctly.

When I created my map from B0101, I'm using the data parameter from BEN factor Bank 1, Serial Wideband, LS1 style (factor).

After doing data logging, I run the suggested filter and then cut and paste with labels the values in my map. Then I paste with multiply those map values against the B0101 table.

From what I can figure, it looks like the map values over 1.00 are telling me that the cells are showing lean, which when multiplied to the respective cells in B0101 make those VE values LARGER, indicating that the PCM has to add more fuel for those affected cells. In the case of the map cell values being lower than 1.00, multiplying them against the relevant VE cells in B0101 makes those cell values SMALLER, indicating that the PCM needs to add LESS fuel because the map data was indicating richer mixtures.

In other words, the larger the VE cell values, the more fuel is needed to be able to reach stoich. If the VE value in a cell is too small, the data in the map will indicate that the cell is running too lean, so fuel needs to be added by increasing the efficiency value. When the efficiency value in a cell is too large, then the collected map data will indicate that the cell is running too rich and LESS fuel needs to be provided by decreasing the efficiency value.

Of course, it's entirely possible that I have this COMPLETELY backwards in my mind.....

In a manner of speaking.

BEN > 1: VE cell is lower than actual, so BEN multiplier is increasing VE cell;
BEN < 1: VE cell is higher than actual, so BEN multiplier is decreasing VE cell;

(i.e. just like you said, but in terms of airmass)


then note that fuelmass is proportional to airmass.


Also note that when not at stoich fueling, the cylinder airmass still has to be correct to achieve the desired fueling (e.g. from PE table).

[joecar]

i.e.
- first, VE table models the cylinder airmass;
- then, fueling is determined by looking it up (stoich or OL fueling table or PE fueling table);
- then fuelmass is calculated from those two;

keeping airmass and fuelmass separate makes it easier to think about and to tune.

[Rich Z]

OK. I think.

The way I've been looking at this might not be the way that tuning actually works.

But perhaps the logic path I have taken to get there is wrong.

I'm looking at airmass as being a factor that is DETECTED (via MAF) or CALCULATED (via VE table) and fuelmass as being a factor that is CONTROLLED (via fuel injector pulse width commands from the PCM). With that in mind, I just think of this in a relationship whereby the PCM is able to CONTROL the air/fuel ratio in the ONLY way it can: By controlling the pulse of fuel from the fuel injectors. The PCM cannot CONTROL airmass at all, and is only REPORTED that value either directly by the MAF or interpolated via the VE table. When the PCM then detects the ACTUAL afr as reported from the O2 sensors, it can either reduce the pulse width if the afr is richer than it should be, or increase the pulse width if the afr is leaner than it should be.

So under the above assumptions, I tend to look at the fuel trims as being exactly what they infer: FUEL trims. Meaning, FUELMASS is being adjusted to accommodate an inaccuracy of a previously determined airmass for which the PCM applied what it thought was the correct amount of fuel.

The act of "tuning" is actually calibrating the airmass (via a correlation between the frequency signal of the MAF and the airmass that references in a conversion table, or values in the VE table that are predicting the actual airmass based on engine rpm and engine vacuum) in order to try to make the PCM's calculations of what fuelmass to use be more accurate.

Hmm, let me back up a bit, thinking about how the fuel trims actually work. Well, I've been thinking of the STFTs as being pretty much an instantaneous determination of the afr (with inherent delays that are going to take place from the time the PCM determines what amount of fuel is needed at a given instant for a particular cylinder to the time that the exhaust that results from that cylinder firing actually gets to the relevant O2 sensor), and the LTFTs as being sort of an averaging value of a set of STFTs. But there is a complication. Aren't both STFTs and LTFTs basically a SINGLE value that applies no matter what the MAF is detecting and which value is being selected from the VE table based on RPM and kPa? So unless the engine is running in a perfectly stable condition with airflow or engine rpm and intake vacuum CONSTANT, the fuel trims will never be accurate across every possible value that the MAF can detect for airflow, and not for every cell in the VE table. Which, of course, is why the MAF table is not a single constant, nor is the VE table a single value. The STFTs and LTFTs are correcting for errors across the entire spectrum of possible conditions for which the PCM is required to supply a calculation for fuelmass. Which, of course, is the purpose for calibrating the MAF table and the VE table. But heck, even if you got those calibrations PERFECT, can the output of the O2 sensors EVER be flatlined directly dead center between showing too rich and too lean?

So why do the O2 sensor outputs always seem to be showing either too rich or too lean? It would seem to me that TOO much fuel is being added when the O2 sensors detect that the exhaust is showing lean, and then TOO much fuel is being subtracted when the exhaust shows a rich condition. Which makes the O2 sensors bounce up and down like yo-yos trying to compensate and are actually OVER compensating. I guess it seems OK to think that the average afr is actually stoich, but in reality, the engine is spending half of it's time running too rich and then the other half running too lean for every cycle of the O2 sensors.

So, hmm again. Why doesn't the signal of a wideband oscillate in the same manner that the narrow band O2 sensor does? For that matter, why don't the STFTs exactly match the O2 output signals? Are they doing some sort of averaging or in some way determining HOW much compensation to apply and any given instant?

Sorry, just doing some thinking out loud here as I try to put the pieces together in my mind.

[joecar]

Everything until you get to STFT/LTFT yes, basically.

STFT/LTFT: yes, LTFT is the average trend of the STFT... yes, both are applied, but remember that the STFT is zeroed everytime it bumps the LTFT.

O2 voltage swinging: this is as-designed since the O2 sensors are narrow band (have a very sharp slope)... the only way the PCM can determine the correct trim is by constantly cycling the O2 voltage around stoich (so the PCM fuels a little richer than stoich, watches the transition, and then trims a little leaner, watches the transition... as long as it sees the transition in the correct direction then it keeps tracking and trimming in CL.

When you use STFT only (e.g. as in SOL) the wideband does tightly oscillate around stoich since the STFT are instant and not learnt, see the attached image.

When you use LTFT the wideband does not appear to oscillates because the LTFT that was previously learnt is being applied, so you don't see the oscillation.