Why is the MAF so innaccurate?

Why is it that when you get your VE in line and then reenable the MAF that it needs so much work to get the MAF table in line? I mean, assuming you don't have a vacuum leak somewhere behind the MAF, shouldn't the OEM calibration table still be accurate?

Is the GM MAF I have on my car not actually calibrated "properly" from the factory? What I mean is... if you could hook the MAF up to test equipment, and push a known air mass through the MAF -- then observe the output frequency of the MAF -- would the GM MAF table hold true for that? Or would we find that GM fudged the MAF tables in order to make it work well with an OEM VE table?

For example, my VE is now +/- around 2%, it's starting to get close. For giggles I reenabled the MAF only by setting the fail frequency back up to 15K and fail count back up to 18 (both were at 1 for AutoVE). Upon doing that my car wanted to barely idle... I found that the lower frequencies for the MAF were off by 30%!! As I progressed to the higher frequencies the variance was less and less until eventually it had gradually progressed from a BEN of 1.30 down to a BEN of .80 on the high cells.

It would just seem that if the air mass data in the OEM MAF table was accurate that you'd never have to change it... but clearly that's not the case, and I wanna know why :muahaha:

There is a lot of deep discussion on the MAF.

Short an sweet, the MAF is only as accurate as the tune.

Here's some good reading: http://www.ls1tech.com/forums/showthread.php?t=363392

the MAF curve is a 3rd order polynomial, so for the lower range of values, there isn't that much difference in output, but once you get going, the range changes quickly even for small changes in input. that's just how it works. GM improved the MAF by having it work against the frequency, not the voltage, like a lot of other MAFs (there's a SAE paper explaining the advantages, but it's a really long way of saying 'frequency based MAFs are more precise') so just take it for what it is ;)

there's a good reason why GM went for the strange part-time MAF/SD hybrid switching over to pure MAF. MAF is very good once you start pushing more air through it. until then, it just doesn't have enough resolution, thus it gets split duty with SD calcs. another benefit from running dual mode is that if the two really disagree, that means one of them failed, that's why you see so many people who tune just one mode, and end up with 'limp mode.'

This is great info -- I am learning so much here.

I guess I'll stick it out and take a few more cruises with the MAF on to try to get it in line. It's just mentally unrewarding to spend a lot of time tuning VE, get it nice and smooth, and then take the next step with the MAF and have a lot more work to do just to get back to smooth.

What do you guys do about the upper-left cells (high vacuum, low RPM) in VE? I think partly as a result of the large injectors I'm running (60#) and maybe other factors, the upper left cells really never want to get in line. It seems to read .85-.90 BEN even after those cells have been adjusted .85-.90 (which is what made me think it was the injectors not pulsing quickly enough).

When I was just VE I didn't worry about getting it perfect because it seemed to drive well, but I could see those values now introducing noise into my MAF tuning.

Is the best thing here to just put a filter on low RPM/high vacuum cells for MAF tuning?

For the VE areas I could never get to, I just hand smooth my table into them.

I hear you on the MAF thing. Hince my !MAFness

I love my MAF.

Well, I did about 2 hours of driving, logging, and tuning tonight. I have *most* of my MAF table clean, BUT... I am running into a couple of issues.

The worst one is that whereas my car would drive nearly PERFECT down to 500rpm in 1st gear and come smoothly back up to 1000+rpm with the throttle... now it drives like crap and damn near stalls. By doing a quick dashboard scan I could see that where it wants to stall, it's running real rich (10:1 - 11:1 ish). But the same MAF frequencies where it reads rich/wants to stall (usually under 4500Hz) at 500rpm it can read perfectly fine at say 1300rpm.

What gives? It drove fine without the MAF down there with just my VE table running the show. If there was some good way to turn the MAF off below 5000Hz, I would do it... to RHS' point above, it seems the resolution down there kinda sucks. In SD I was able to solve the car wanting to stall when I blipped the throttle from idle or went into neutral when coasting down... with the MAF, it wants to stall more often. All the really bad stuff is happening below 5000Hz, and I can't seem to get a good BEN reading for those cells -- it's fine under one condition and is too rich under another.

The other issue is that the car has now developed a misfire at light cruise around 3000rpm that it didn't have before with just VE. Any hints on running that down?

Oh, and just to be sure I'm doing this right... someone please confirm that for moving from VE tuning to MAF tuning all you should do is set the MAF Fail from 1 to 15,000 (I think stock is lower, but with my turbos the MAF was failing at that lower value) and set the fail counter from 1 back to 18 (stock).

:bash: (the guy on the left is me... on the right, is my MAF... mocking me)

diynoob, please email me your 'almost perfect' log, i'm working on a generic cleanup filter, i'd like to see if it will fix your problem or not

Quote:

---

Originally Posted by redhardsupra

diynoob, please email me your 'almost perfect' log, i'm working on a generic cleanup filter, i'd like to see if it will fix your problem or not

---

I didn't save any logs while I was doing the MAF tuning, I ejected the data after each paste and multiply.

Just let me know which PIDs your cleanup filter needs though, and I'll log the car on the way to work tomorrow. :cheers:

BTW -- anyone know for sure if the MAF will be considered "failed" if it goes below the minimum frequency, but is under the min engine speed?

I'd like to fail the MAF under 5000Hz but not set a DTC, so I was thinking about maybe doing something like a 2500RPM min engine speed and a 5000Hz fail frequency. I am just not sure if setting the DTC and failing the MAF/reverting to SD are coupled.

maffreq and mafflow are enough, but all the other usual suspects are always useful.

Log file attached to this message. Some BENs in the MAF curve are a bit high or low -- but everything around them makes sense and is close. So I didn't tweak just one cell in the MAF table down before it was showing rich or lean. My assumption is if the MAF curve would get out of whack to dial in air/fuel then that means that there is something else wrong in my tune. I couldn't feel it so I just left it alone.

Any help you can offer is appreciated. I have to work to keep my car above 1K RPM because now that I have the MAF back on, the car wants to stall a lot more.

I had a similiar phase were I could not MAF dialing to make sense. I found that I had to go back and make sure all was closed up from the AutoVE mode then change ONLY this:

Step by step:

0. Assuming a calibrated VE using speed density mode and "AutoVE"...
1. Put the tune in closed loop and all normal PE, etc.
2. Disable LTFT {B3801} and STFT idle trims {B4108}.
<!--[if gte vml 1]><v:shapetype id="_x0000_t75" coordsize="21600,21600" o:spt="75" o:preferrelative="t" path="m@4@5l@4@11@9@11@9@5xe" filled="f" stroked="f"> <v:stroke joinstyle="miter"/> <v:formulas> <v:f eqn="if lineDrawn pixelLineWidth 0"/> <v:f eqn="sum @0 1 0"/> <v:f eqn="sum 0 0 @1"/> <v:f eqn="prod @2 1 2"/> <v:f eqn="prod @3 21600 pixelWidth"/> <v:f eqn="prod @3 21600 pixelHeight"/> <v:f eqn="sum @0 0 1"/> <v:f eqn="prod @6 1 2"/> <v:f eqn="prod @7 21600 pixelWidth"/> <v:f eqn="sum @8 21600 0"/> <v:f eqn="prod @7 21600 pixelHeight"/> <v:f eqn="sum @10 21600 0"/> </v:formulas> <v:path o:extrusionok="f" gradientshapeok="t" o:connecttype="rect"/> <o:lock v:ext="edit" aspectratio="t"/> </v:shapetype><v:shape id="Picture_x0020_1" o:spid="_x0000_i1025" type="#_x0000_t75" alt="http://efiforum.iqd.co.nz/images/smilies/icon_wink.gif" style='width:11.35pt; height:11.35pt;visibility:visible;mso-wrap-style:square'> <v:imagedata src="file:///C:\DOCUME~1\Bruce\LOCALS~1\Temp\msohtmlclip1\01\cl ip_image001.gif" o:title="icon_wink"/> </v:shape><![endif]--><!--[if !vml]--><!--[endif]--> 3. Log a bunch of data - normal driving, wide open (to hit the high Hz, do some bonzai 4th gear 100+ mph runs
4. Save the log file.
5. Filter the throttle transients out using the BEN filter.
6. Copy and paste the MAF map as a factor into your MAF table in the tune.
7. Repeat steps 3 through 6 and you'll have the MAF dialed.

My MAF is dialed in to within 1% from 2375 Hz to 11k+.

Then it all came together as it should. Make sure you have the right MAF table for your car.

Quote:

---

Originally Posted by Bruce Melton

I had a similiar phase were I could not MAF dialing to make sense. I found that I had to go back and make sure all was closed up from the AutoVE mode then change ONLY this:

Step by step:

0. Assuming a calibrated VE using speed density mode and "AutoVE"...
1. Put the tune in closed loop and all normal PE, etc.
2. Disable LTFT {B3801} and STFT idle trims {B4108}.
<?xml:namespace prefix = v /><v:shapetype id=_x0000_t75 stroked="f" filled="f" path="m@4@5l@4@11@9@11@9@5xe" o:preferrelative="t" o:spt="75" coordsize="21600,21600"><v:stroke joinstyle="miter"></v:stroke><v:formulas><v:f eqn="if lineDrawn pixelLineWidth 0"></v:f><v:f eqn="sum @0 1 0"></v:f><v:f eqn="sum 0 0 @1"></v:f><v:f eqn="prod @2 1 2"></v:f><v:f eqn="prod @3 21600 pixelWidth"></v:f><v:f eqn="prod @3 21600 pixelHeight"></v:f><v:f eqn="sum @0 0 1"></v:f><v:f eqn="prod @6 1 2"></v:f><v:f eqn="prod @7 21600 pixelWidth"></v:f><v:f eqn="sum @8 21600 0"></v:f><v:f eqn="prod @7 21600 pixelHeight"></v:f><v:f eqn="sum @10 21600 0"></v:f></v:formulas><v:path o:connecttype="rect" gradientshapeok="t" o:extrusionok="f"></v:path><?xml:namespace prefix = o /><o:lock aspectratio="t" v:ext="edit"></o:lock></v:shapetype><v:shape id=Picture_x0020_1 style="VISIBILITY: visible; WIDTH: 11.35pt; HEIGHT: 11.35pt; mso-wrap-style: square" alt="http://efiforum.iqd.co.nz/images/smilies/icon_wink.gif" type="#_x0000_t75" o:spid="_x0000_i1025"><v:imagedata o:title="icon_wink" src="file:///C:\DOCUME~1\Bruce\LOCALS~1\Temp\msohtmlclip1\01\cl ip_image001.gif"></v:imagedata></v:shape><!--[if !vml]--><!--[endif]--> 3. Log a bunch of data - normal driving, wide open (to hit the high Hz, do some bonzai 4th gear 100+ mph runs
4. Save the log file.
5. Filter the throttle transients out using the BEN filter.
6. Copy and paste the MAF map as a factor into your MAF table in the tune.
7. Repeat steps 3 through 6 and you'll have the MAF dialed.

My MAF is dialed in to within 1% from 2375 Hz to 11k+.

Then it all came together as it should. Make sure you have the right MAF table for your car.

---

If you want to tune the MAF rapidly without having to worry about any input from the VE table, set parameter B0120 = 0 or 200 or some very low RPM. Any rpm point above that parameter value will result in using the MAF exclusively for airflow. :cheers:
Edit: Re-read your post. Closed loop with BEN's does not work, as the O2 sensors will adjust fueling to stoich. You need to remain in open loop and re-enable the MAF, then disable airflow calculations.

raw data, no filtering:

http://static.flickr.com/172/363548621_93f9af2ecf_d.jpg

Z-score based filtering, 2 sigma:

http://static.flickr.com/188/363548615_a0d4896828_d.jpg

Z-score based filtering, 1 sigma:

http://static.flickr.com/120/363548613_7a31161d6b_d.jpg

few things become apparent once you chart your dataset:
1. you need more data, especially in mid/upper range
2. even the simplest statistical filtering works, however, for it to work well, it needs lots of data--look at the topend where data is so scarce it doesn't know what to do with it.
3. you're using BEN's to adjust your MAF table right? that's one way to do this. i prefer to use dynair values mapped onto maffreq scale for this particular calibration. this way ve table and maf table are truly describing the same thing, just in two forms. if you do two tunes, one to adjust ve, and another to adjust maf, that's two separate tunes. they'll be close, but since they operate from two different datasets, they are describing two different tunes really. try my method sometime, maybe it will work better for you.

i'll try to port this method to calc pids, right now it's in excel, cause it's simple to prototype things this way. i can post the spreadsheet if you're interested, as this was actually much harder to do in excel than i originally thought, so there's some really useful tricks in there.

hope this helps,
Marcin

Marcin,

That's a good bit of research you have done... :cheers:

The raw data shows more scattering in the upper range...would I be right in saying that the MAF is not as consistent in the upper range, or is this due to something else...?

Yes, please post the spreadsheet, that would be great.

Cheers,
Joe
:cheers:

5.7 Hemi's run speed density only.
If its good enough for them .....

Quote:

---

Originally Posted by ringram

5.7 Hemi's run speed density only.
If its good enough for them .....

---

:muahaha: :D

Yep, never have to worry about a failed MAF... :D

You'd think it must answer the question (as does the HSV GTS) about the effects of weather and altitude change. Australia, like the US can experience a wide range of extremes...

Quote:

---

Originally Posted by ringram

5.7 Hemi's run speed density only.
If its good enough for them .....

---

So do the 4.2L I6 LL8s until late 05 when GM redesigned them and put a MAF on so they could use a V8 PCM instead.