LTFT is 22 WOT

[Screamn03]

So my LTFT is 22 under WOT. I'm looking around at the different tables under the fuel tab and starting to get confused. What table(s) do I want to start editing first edit.

[Torch]

I believe LTFT's are not stable or accurate at WOT. They apply mainly to idle and part throttle. The most common method for "street method" for measuring A/F ratio is to monitor the o2 sensors.
.800 - Lean
.880 - .890 - Ideal (Good Luck)
.900 - Rich

What are your LTFT's at idle? They need to be slightly negative.

[jimmyblue]

You're not confusing FTC (Fuel Trim Cell) 22, which is the
one that WOT uses, with LTFT B1 and B2, by any
chance? The FTC will always be 22. The two LTFTs ought
to be zero, but under abnormal conditions can show
positive values. They should not be that "cute" though,
nor that high, and the two sides should be different by
some random little bit.

With that kind of WOT LTFT, you'd either have a really
bent MAF or a fuel delivery problem in closed-loop
cells. I'd say that most of the time this ends up being
a MAF issue. Either an aftermarket MAF (GMAF), a
home port job, MAF ands - all of these will skew the
airflow measurement. The other "bad boy" is long term
use of a greasy K&N filter, which will slime up the sense
elements. All of these result in gross lean error that the
closed loop has to take out.

Anyway, it's not all bad news. If you have the log data
then get it into Excel. You want LTFT vs MAF frequency
(raw), MAF gm/sec and FTC (fuel trim cell). I don't know
what the actual proper PID labels are here but you get
the drift.

Now take the whole log strip for these values and sort
by the FTC value. Forget cell 22. Cells 0-15 are the main
closed loop ones. Now make a scatter plot of the LTFT
vs MAF Freq and of FTC vs MAF Freq.

These charts should reveal an LTFT vs MAF frequency
that has a fairly regular"profile" to it. Now you can take
the LTFT value here, and it's the correction you should
make to your MAF table at any given frequency, to get
right. Like if it says LTFT = +5 to +7 at 8000Hz, it means
the value at that point in the table needs to be scaled
by 1.07 because thePCM has found, over time, that it
has had to add the fuel corresponding to an extra 7%
of airflow. You want to use the most-positive value so
in the end you have a slight negative LTFT bias to keep
them from popping above zero on occasion.

Now get your MAF table as-is from the VCM_Editor into
the same Excel sheet, nearby. Parallel to the flow/freq
"strip", make a row and start entering the correction
factor. Go down it, cell by cell, entering the scaling the
LTFTs call for.

Then below it, you make a product row where each cell
ends up being the original * (1 + LTFT), the LTFT being
pointwise-right by frequency. This product row should
look almost like the original unless there is a really gross
LTFT showing up.

Now, you'll find that the LTFTs come to an end before
MAF frequency does. You either have to populate the
high frequency end of the table with the scaling value
you got off the highest displayed frequency or, if the
scatter chart indicates the error is not constant, put
an extrapolation drawing line on it and try to eyeball the
relationship as best you can, and populate the scale
factor row with that, pointwise w/ freq.

So with a shiny new product row that represents what
would make the O2 loop happy, take that and back-
stuff it to the VCM_Editor MAF table it originally came
from. You should see a happy set of LTFTs. Save the
'sheet for next time in case some need for further
"detailing" shows up.

But if you have a K&N filter, clean your MAF and the
ducting leading to it, first and reestablish clean LTFTs.

[SloppyRob]

JimmyBlue,
I autotapped my car the other day before recieving this product and found my 0-15 FTC ltft's averaged slightly positive (about +.8). Can I use this data to scale the Maf Table or am i gonna have to rescan the ltft's with all the correct parameters? I have a nice beefy 30min log that i don't wanna just do over.

[Screamn03]

Sorry guys I got that backwards. WOT it is in FUEL TRIM CELL 22 and at idle it is at 4. LTFT at idle is 10.2 B1 and 9.4 B2. WOT LTFT is 9.4 both B1 and B2.

Jimmy Blue-Thanks for all the info. I'm gonna get started on this later on this afternoon but like you said I'm gonna make sure I don't have an oily (G)MAF.

I'm starting to think about putting the stock MAF back on the car to help eliminate that variable so I can get the tune closer for the motor itself and then add the (G)MAF later that way I can just deal with that issue alone.

Looks like I got my work cut out for me. Thanks for the help guys. Mike

[jimmyblue]

JimmyBlue,
I autotapped my car the other day before recieving this product and found my 0-15 FTC ltft's averaged slightly positive (about +.8). Can I use this data to scale the Maf Table or am i gonna have to rescan the ltft's with all the correct parameters? I have a nice beefy 30min log that i don't wanna just do over.

I would get away from looking at averages and instead
try to see if there are particular FTC "bins" and/or MAF
frequency ranges that are especially troublesome. The
average is going to be skewed by how much time you
spend in each cell or at airflow; normal driving biases
the average to what's going on in the idle and low
cruise cells, where my (stock) LTFTs are more negative
(natural rich). If yours is similar then your slight positive
average may be obscuring a worse lean condition in the
upper closed loop cells.

I just prefer working with visual / graphical views, they
make more immediate sense to me. Try the scatter plot
of LTFT vs MAF Freq and see if something doesn't jump
out at you. If you don't have Excel, I believe you can get
an equivalent GPL-licensed spreadsheet tool from the
OpenOffice/StarOffice folks?

[Screamn03]

LTFT vs MAF frequency
(raw), MAF gm/sec and FTC (fuel trim cell).

The only LTFT I see is bountries.

[SloppyRob]

Ok thanks for the reply. I'll rescan with vcmscanner using the LTFT, Maf freq, Maf gms/sec and FTC. Parse the FTC's for 0-15 and plot the relationships you suggested and scale accordingly.

I sketched out on a piece of paper a sample graph for the ltft vs Maf freq and i see how this will allow me to pinpoint problem areas.

But what is the FTC vs Maf freq gonna tell me?

[SloppyRob]

I think the ltft's come from the scan tool along with the Maf freq. You take these and place them in excel in the order of the dump (no timestamp avail). Once you plot the graphs and find the problem areas you can bring up the Maf table and scale accordingly. What units are the x and y axis of the Maf table...I don't have the tool with me cause im at work?

Am I understanding correctly?

Oh and try the Airflow tab for the Maf Table!

[jimmyblue]

Yeah, you need the scan tool to be collecting FTC, LTFT B1 (and/or B2), MAF Freq and might as well grab O2s
(B1S1, B2S1) while you're at it.

I was just checking all of this out on an old EFILive log
I had saved; something weird going on there, for the
same FTC I would get a large variation in the LTFT -
like it wasn't settled in. Especially bad on cell 4, which
is low RPM, low MAP, the engine seems to spend most
of its time in 0 and 4 (idling), the cells that are the most
relevant to performance are 3, 7, 11 and 15 (RPM from
2000 up, across the MAP range) and 12, 13, 14, 15
(high MAP, across the RPM band).

This was a street driving log and it turned out to be
full of worthless crud. I think for performance tuning
you would want to log more like a "poor man's dyno" -
do a bunch of "pulls" at like 1/4, 1/2, 3/4 fixed throttle,
and as little as possible idle, decel, etc. Better quality
data, and you can cut-n-paste specific "pulls" to weed
out the garbage and make the plots less of a "cloud".

Like I had about 5300 time-ticks of which over 4300
were idle, and Cell 4 LTFT (which looks like your off-idle
light tip-in) was anywhere from 0 to -14 LTFT - trimming
its little heart out but totally irrelevant to performance.

Anyway, if you can snip out the high-relevance bits it
will clarify the MAF adjustment needed, and if you use
"street crud" like the log I'm talking about it may be a
kinda futile analysis. Also you will be messed up if you
do not have stable LTFTs in the cell you're looking at.

The MAF table is mass airflow (gm/sec) against the MAF
frequency index. This is how the PCM gets from the very
nonlinear MAF output to a usable air quantity to use in
fuel and other calcs. By scaling individual cells this lookup
can be made more true to the MAF you are using (or
made worse).

If you assume the fuel tables represent your car
properly (probably true enough for a stocker with a
clean fuel filter, no charging system issues and a sound
fuel pump) then any air/fuel deviation is due to error in
the incoming air data. This might be less of a valid
assumption at the highest fuel deliveries but here you
will have no valid trim data to use anyway, and the MAF
function will have to be assumed continuous, not a bad
bet, and extrapolated.

So where you have an LTFT @ FTC and a MAF frequency,
it is telling you how much the MAF-based fuel calc was
off, by how much fuel it had to add. LTFT = +4% means
that roughly there was 4% excess air that the MAF did
not declare, but the O2s saw. So, in theory anyway,
the MAF table entry at that frequency ought to be
increased 1.04X.

Of course there are problems; a single FTC will cover a
fairly large "band" of RPM, and so on, making it kind of
a judgement call. The MAF is going to be less consistent
(I believe) at low airflows, idle cells. I would try to use
the 3,7,11,15 cell set or 12, 13, 14, 15 cell set to cut the
number of variables from the picture.

[SloppyRob]

Ok, im trying to digest this last post.......

So, the ltft vs Maf freq table determines @ what Freq the ltft is off

Then you reference this Freq with a FTC from the FTC vs. Maf gm/sec table to see if the FTC is important.

You then open the Maf table and find the Maf freq and Maf gm/sec and alter the table value by a correction factor either greater than or less than 100% depending on the ltrm values.

Let me know if this is correct

[jimmyblue]

Ok, im trying to digest this last post.......

So, the ltft vs Maf freq table determines @ what Freq the ltft is off

Then you reference this Freq with a FTC from the FTC vs. Maf gm/sec table to see if the FTC is important.

You then open the Maf table and find the Maf freq and Maf gm/sec and alter the table value by a correction factor either greater than or less than 100% depending on the ltrm values.

Let me know if this is correct

I don't know as I'd say my advice is "correct" but
the FTC is just "in the mix" because you can get the
same MAF frequency and be in different fuel map cells.
Like wide open, 2000RPM is the same absolute airflow
as half throttle, 4000RPM (constant MAP*RPM "load
line&quot. Where the FTC is especially useful, is as a sorting
"hook" for Excel Data > Sort to use, you can move all of
the "garbage" to the end and chop it off, then resort
by the original time-tick, etc. and reconstruct the "pulls"
and get an LTFT that corresponds to the MAF freq, and
the MAF freq is the "index" for the changes you make
while the "most sensible" LTFT is how much to move the
table value. Which FTC's LTFT to believe when they
can't agree? Hopefully you can get clean enough data
over short enough time that it won't all be a bickering
mess like my log file....

Now the LTFTs are also crude "bins" and cover broad
MAF freq ranges so you have to be able to "smooth"
what they are telling you, pick the truth out of all the
chatter.

If you look across RPM at wide-open MAP pressures, the
fuel trim cells are:

12: 400, 800 RPM
13: 1200 RPM
14: 1600, 2000RPM
15: 2400 - 5600+ RPM

Dunno how the indices get used, whether these RPM
are upper, lower boundaries or center etc. but you get
the picture - cell 15 is huge and the most important /
relevant as it's the closest thing to WOT that has any
closed loop feedback to teach us. But all of the cells are
potential stepping-off points to WOT and so must be
pushed & kept "underwater". But if I were going to do
a straight MAF table scaling I would use Cell 15's LTFT.
And this might be these best way, use that guy first
and then hand-detail anything that's left sticking out
when the LTFTs have all re-settled.

Anyway, I hope this is helping. I think tonight I'll go out
and take my own advice and see if I can get coherent
data, and work through the exercise and come up with
some illustrations, sample Excel files, etc. That, and
make sure I haven't been telling you to try "stuff" that
don't work... but from working with it, it seems that it's
really about getting the good data. That's the key, good
data makes it easy to interpret and bad data will just
drive you nuts trying.

[SloppyRob]

Alright, thanks for the clarification. I too will try this stuff out, prob over the weekend as it's been rainy in NY lately. I first have to get the car to idle well which im in the process of doing because I just installed a cam.

[EC_Tune]

Something that may simplify the Trim Cell issue is to create one big cell. I used this method all the time on earlier PCM's and it works great.

When you are finished you can restore the Cell structure again.

[jimmyblue]

What do you get from the single-cell approach?
It seems this would be de facto averaging the
system fueling error (weighted by time-in-region
and error-in-region) and would tend to obscure
any specific problems in (say) low RPM/low MAP
vs high RPM / high MAP?

Just wondering because part of my plan is to go
the other way, pull the RPM thresholds in my FTC
boundaries list to activate the "missing" 8 cells
and move more of them down into non-PE regions.

[EC_Tune]

My idea behind this is that you don't get cell switching if you move slightly above or below the MAP/RPM settings.

Also by watching the short term value you can see if the cell learning is complete when under a constant MAP & RPM.

When the PCM switches cells it uses the previously stored value from the last time it was in that cell which starts the short term/long term learning all over again.