TAQuickness

iTrader: (28)

I'm starting this thread as not to hijack someone else's. If anyone already has the answer - please post the link to it and I'll STFU

Here are some of the excerpts from the other thread...


So, to pick up where the other thread left off - I just finished reading RHS doc at www.allmod.net/hpt. Although it was great info (enjoyed the individual thinking bit) I missed any reference to how a bone stock MAF with a bone stock MAF table would incorrectly report air flow by the addition of a cam, heads, LT's, etc...

Seems to me that if you added only a cam to the car, then corrected your VE table for the change, the MAF could be plugged back in and not require re-calibration; however, the real world indicates a different scenario

txhorns281

iTrader: (12)

the VE table does not control as much as you think during MAF operation. It's mainly used as a reference against the MAF for error. You can manipulate low end and big end fueling quite directly by tweaking the MAF curve.

As to why it must be recalibrated, unless you were a little air particle passing through the MAF itself and could explain in better detail, the way the MAF meters air is through a frequency measurement, then a reference to a preset value in the MAF table. There are many factors to consider here though, velocity, temp, density, volume, etc. All these things can make a difference in true raw airflow metering , however the MAF most likely cannot sense all these variables through it's frequency sampling. In Red's write up, if you download the spreadsheet (it's been so long since I've checked I hope it's still there) you will be introduced as to the observation of how a motor's requirement/use of airflow changes with modding (even for boltons) and how the MAF does not adapt. For instance, at 5000Hz the MAF may be programmed to meter (more like suggest) that there is XX.XX airflow, when it could be the case that after modding, the true amount of airflow could be greater or less than that reported value. thank god for O2s and trimming then, as this undetected change in airflow will be detected in A/F production and compensated for through trimming.

If you ever get the chance, look at trims on a car before and after a lid change, or manifold swap, or headers. You will see mild adjustments occurring in the trimming, but if you really want to see the MAF do it's stupid thing, check trimming before and after descreening/porting one out (or using MAF ends or whatever). You will see that the MAF's calibration and metering ability will be totally blind to the 20-30-and in some places 40% increase in airflow through the measuring plane. The PCM has a nifty little fail-safe air calc of it's own that is derived from a greater more encompassing VE equation (dynamic airflow). This is how SD fueling is determined (MAF-fail mode) and this is also where we get clues on what sort of true airflow the motor is consuming and how we can align the MAF's calibration to better serve as a measurement device.

Bad30th

iTrader: (13)

Thanks so much for starting this thread/discussion and helping me get this figured out !

Cheers,
Rob (Bad30th)

TAQuickness

iTrader: (28)

TX - I agree with the majority of your post.

At the moment, I can not download Red's spread sheet (will be doing that when I get home).

Again, my sole purpose of this thread is to understand/figure out what in "real life", as opposed to "working on paper" would cause a bone stock OEM MAF, with an unmolested MAF table, to inaccuratly report air flow when the engine's breathing charactoristics are changed. The answer could be as simple as the stock MAF table is only accurate for what the PCM does with it, or the rate at which the MAF reports to the PCM is not fast enough/too fast, or a million other things.

To make sure I'm on the same page as the rest of you guys, tell me if you agree with my explanation of how the MAF works.

A stock 75mm MAF has 3 heating elements, a screen, and an air foil. The purpose of the screen and air foil is to create laminar air flow across the heating elements so the MAF can report an accurate airflow reading.

When the MAF is powered up, voltage is applied to the heating elements to acheive and maintain a constant temperature. The voltage required to achieve the constant temperature, at a known/constant ambeint (for lack of a better term) temperature, when no air flow is present, is a pre-programed constant within the logic circuits in the MAF (from here on I will refer to this voltage as CV)

As air begins to flow over the heating elements, the heating elements cool thus requiring a variation of the CV to maintain the constant temperature. The difference (or other mathmatical equation) between the CV and the applied voltage to maintain the constant temperature of the heating elements is then converted into a frequency.

This frequency is then read by the PCM, and the PCM looks up the respective air flow value from the MAF table.

Sound right?

TAQuickness

iTrader: (28)

no problem - I'd like to know why this is despite the fact I don't have a MAF

jimmyblue

iTrader: (11)

The MAF (on these cars) works by hot wire resistance.
The electronics try to maintain a fixed resistance (this
being a strong function of wire temperature) on the
sense element, against the cooling power of the airflow
(this being proportional to mass airflow and to air temp).
The power put to keep the resistor at the setpoint is
a function of airflow. Power is delivered in pulses, the
frequency of the pulses is the power delivered. The
frequency is a nice, difficult-to-corrupt signal to send
to the PCM.

If you look at the curves it's pretty obvious that it is not
a linear function.

But nonetheless given decent materials control and maybe
a bit of in-manufacturing trimming the MAF can be made to
be reasonably accurate. I have been told manufacturing
batches run to about 5%, by an outfit that did test Delphi
MAFs, a high quality calibration lab.

Now, 5% +/-, means your 12.7:1 (say) AFR could be 12.0:1
or could be 13.3:1 so I guess we would like better than
luck-of-the-draw. Odds are though that your unit is near
normal.

The hot wire MAF has embedded in its calibration a basic
assumption. That is, that the physical form and critically,
the distribution of airflow across the orifice is the same
now as it was during design and/or manufacturing trim.
The sense resistor occupies a tiny fraction of the orifice
and the sensible portion of the airflow is part of the cal.
Change that fraction by changing the throat effective area,
bend the cal. Change it by allowing the air stream to crowd
to one side, non-uniformly, and bend the cal. The "screen"
is meant to enforce a more even distribution of air and
fight this (I believe especially at high flows, the little
channels see a lot of turbulence and a square law or cubic
drag, which strongly favors redistribution to any less-crowded
cell. Unfortunately, this requires back-pressure).

The good thing about a MAF is, once you have it calibrated
and locate it in a good place, it doesn't much care about
what happens out back. It just punches the tickets as
they come through the door, enjoy the ride.

Speed density on the other hand needs the VE table updated
for every airflow mod downstream of the intake manifold.

Using a good speed density tune to correct the MAF, is only
done because you can't shut off speed density entirely.

You could make some assumptions such as, any scaling
needed to tune in the MAF from 4000RPM on up is equally
valid down low. But this is only an assumption and hard
to prove. So basically, get the speed density "good enough"
and then attribute any deviation to a MAF error, and fix it.

In a lot of cases this may all just be "trying too hard". Since
there are available MAFs with "known good" calibrations,
you could just pick one and go with it. I have had great luck
with my $30 eBay truck MAF, -screen, 6.0 truck table. But
some folks just have to go down the hard "free" road, or get
suckered into buying an uncalibrated piece with no such
support. To which I'd say, throw your own $30 after a
known MAF and sell off the pain-in-the-*** one, if your
conception of karma will let you.

Bink

Couple of things to consider:

I don't believe the voltage delta/difference is linear
(Is the frequency conversion factor a constant?).

The same MAF has multiple tables supplied by GM for varying applications.
The MAF is an inexpensive, mass produced sensor that provides a reasonable value for fueling when crutched with the airmass calcs in the PCM (below 4000 RPM).




jimmy posted his mucho better info as I was considering this.
His posts are rarely without flare!!

Another_User

You would be very surprised how much VE is factored in during idle and low speed operation. Enough to where your car can run like crap even with trim correction. Getting cammed cars to idle it 33% running airflow, 33% spark, 33% matching your VE to your MAF table, and 1% bulls^%ing.

TAQuickness

iTrader: (28)

Jimmy - thanks for the more detailed explanation.

So getting back to what in the "real world" would skew MAF readings. Well, as Jimmy indirectly pointed out - anything upstream of the MAF can skew readings. The factory lid, for instance, has a sound dampening baffle and several "ribs" to direct air flow. For that matter, even changing/removing your air filter could alter MAF readings. Although I think the differences in MAF readings would be negligible by the addition of a lid/filter, I'd still like to measure it. I just need to find someone with an unmolested MAF & MAF table.

This offers explaination to the various MAF tables GM uses for the same part number MAF. As well, offers some explaination as to why descreening/porting the MAF yeilds hit-or-miss gains.

The question still lingers. If you take a 100% bone stock car, change the cam, correct the VE table via SD tuning, then return to MAF'd operation, why would the MAF reading be off?

Bink

Airflow characteristics of the intake manifold, heads (including valvetrain), cam/overlap, exhaust. And their interaction with the MAF
Airflow is not an easy parameter to dissect. F1 teams spend 24 hours a day, 7 days a week. in the wind tunnel. Some teams run two tunnels!
FWIW.

gameover

also remember that the MAF senses airflow in both directions!! Airflow in the intake is not as one way as you may think....

also think about where the maf is positioned and what the PCM is trying to estimate from the MAF input. As you open and close the throttle the manifold "fills and empties" at a simplistic level (ie. the airflow the MAF reads may not necessarily equal the sum of all the air going into the cylinders).

i've asked a few GM folks how accurate the MAF is and most say either +/- 5% or +/- 10%. As jimmy said this is pretty much linear error on how much your fueling could be out due to the MAF. Generally this error will not be constant and depends very heavily on airflow characteristics within the MAF.

There is a lot of research going on in the area of "volume velocity meters" mostly driven out of the acoustic field. The trouble with acoustics is no one cares about DC flow (using an electrical analogy here). The engine problem is basically a modulated turbulent airflow (ie. it has a DC and an AC component and turbulence and other fun stuff thrown in as well).

Bottom line is if you put a MAF on a flow bench (steady lowspeed DC airflow) it is very accurate. Put it on an engine and the conditions of the engine make it inaccurate.

I tossed around a couple of idea's relating to multi sensing MAFs that try to measure both the AC and DC flow components and vector sum them etc. The reality of all this techno talk is its far easier and cheaper to put an accurate and fast feedback loop on the exhaust side (ie. a WBO2) with a decent speed CPU in the PCM and trim the air/fuel/spark equation under all conditions on the fly. Now add a cheap, reliable and accurate torque sensor into the mix and you just made a hell of a lot of progress...

Chris...

SmokingWS6

iTrader: (1)

Is this a project in the works, say with a stock pcm? I'm sure the processing power isn't there but wouldn't that be nice

gameover

I wish... the trouble with the auto game is the stuff you see in the field is 5 years behind the times. We have enough other things to do than give ideas to multimillion dollar development teams... Also, everyone likes to think without regard to cost unfortunately if every development team that works on a car got an extra $100 item to make their perfect car, the average car would increase in cost by $1000's. Engineering is always fun until the accountants get involved...

TAQuickness

iTrader: (28)

Agreed. The day I removed my MAF I was quite surprised when I found motor oil in it. Reversion, deflection, reflection, etc... is very much present between the valves and air filter...

Sometimes all you need is a set of fresh eyes on a problem to figure it out.

Given the high probablility of error with a MAF, makes me wonder why manufactures even use them. They make sense as a sanity check (i.e. dry shot or other sudden increase in air flow), but for the average person (general public) commuter, what benifet does the MAF offer? Are manufaturers really able to reduce emmisions and/or improve performacne by installing a mass flow meter in less than ideal conditions?

IIRC - to properly install a mass flow meter, you need a section of straight plumbing equal to 5 times the diameter of the oriface.

gameover

The benefit is reduced calibration time and less dependancy on volumetric efficency modifiers over time.

Again the reality of closed loop operation (where 99% of all engines spend 99% of their time under normal conditions) the accuracy of the MAF is irrelevant when it comes to emissions and fuel economy. Its a case of near enough really is good enough as long as your feedback system is fast enough to keep the catalytic converter reaction happy. Some car makers today still use Alpha-N airflow estimation (TPS vs RPM) and manage to meet emissions requirements.

TAQuickness

iTrader: (28)

JimmyBlue, TXHorns, Red Hard Supra, Game Over - Thanks for your feed back. Believe it or not, the "real world" MAF makes a lot more sense now (even on paper )

Bad30th

iTrader: (13)

Yeah, thanks a lot guys for helping me make more sense of this.

Cheers,
Rob (Bad30th)

RedHardSupra

iTrader: (1)

great thread, first let me share some newer discoveries about the MAF behaviors, that I haven't included (yet) on my MAF tuning page.

1. tuning MAF for < 5000Hz is hard. because the air isn't under much pressure, you end up with a much greater spread of values. I've tried filtering it based on higher MAP values, and while it helped, there was still a huge difference in Trims between SD and MAF modes. so for now I actually left the MAF table alone <5kHz, and tune just the rest, which has a very clear and distinct pattern (as it's under pressure so there's little variation).

2. After doing #1, idle trims got better, but what I've been noticing lately that the trims go significanly richer (as much as 10 points!) as the car warms up, and I don't mean just ECT's going >180F, I mean like 15 mins of driving warm. Any clue why so much and why just the idle trims are temp sensitive?

3. I use the comparison of trims as my 'goodness' index when I'm trying to decide if MAF is dialed in, it's a good ballpark measure. Do the same driving routine with MAF enabled and disabled, and if they're close, you're dialed in.

4. we need data filtering in HPT badly! Lately i've been gathering larger samples than usual and trimming them, all kinds of sudden transitions, or deceleration, and unusually large/small values get kicked out. and after i'm down trimming all this junk out, i end up with a much cleaner dataset, even though there's less data. look at it on raw y(x) graphs and you'll see how patterns become simpler and data isn't spread anywhere near as much.
This is a broader thing, but it applies very visibly to MAF/DynAir data.

5. VE table is absolutely fundamental. It must be perfect, and not just <4krpm, but the full thing, even if you're going back to MAF. Why? Because Dynamic Air is a function of IAT, MAP, RPMs, displacement and VE. The first three are kinda just there and you measure them, displacement is hopefully a constant, but VE is what we alter and now we must make the PCM aware of the changes we've made. so the list of dependencies goes:
VE->dynamic air->MAF->effective AFR
If you dial in your VE only for the first 4krpm, then the dynamic air values for higher rpms are going to be off, so when you recalibrate your MAF, you'll have part old and part new values. So once you do a big change (cam swap), and you redo your _whole_ VE table in SD mode, you better recalibrate your MAF from scratch (no historical values if you're using them in my spreadsheet) or you're going to have some suprising Trims (that's what I experienced few times lately when tuning car that got a new big cam).

that's it for now, time for lunch please comment, this stuff is still not 100% understood and i'd really like to be done with it, as it's very important.

SmokingWS6

iTrader: (1)

I think its necessary to tune the maf down to about 3500hz, after that the MAP seems to be the determining factor in trims.

Idle trims. could this be due to exhaust gas temps and o2 temps? If I let my car warm up to 175* w/o driving the o2's are very lazy, they swing but its as slow as the hand of time. If I go for a drive and come to a stop they are swinging a bit faster, possibly reading a richer mix than when they were cold. Thats my best guess as to why it gets 'richer' after its been driven.

I agree with your VE table findings. I've rarley touched ve above 4800 rpm as i've narrowed it down with a wideband and pe editing at wot.. However if the the car is in PE but not WOT, say 50% tps at 5500rpm during a burnout i've seen cars respond very rich or lean depending on how the VE table responded in SD mode before tuning.

TAQuickness

iTrader: (28)

Still thinking about the rest of you post... But, number 2 really grabbed me.

I got a really good log of stagnant traffic. Litteraly took me 45 minutes to travel 2 miles. During this time my IAT reached a peak of 133*F (average during this time was 125*F). I would have expected my WB-AFR to go rich, however, it went lean.

more on this later - it's beer thirty