Bones232

I'm trying to understand this MAF fueling thing. I just recently switched to a 7427 ECM with the MAF $OD code. After setting the injector flow rate, the car runs ok. Its still rich just about everywhere except one spot.

From 2800rpm to 3200rpm (Open loop, no PE and light engine load) the car goes ridiculously lean. It basically pegs my WB. In hot weather the MAF signal reads in the 5k to 5.5k hz and cool weather 6.6k hz. Above or below is pig rich.

As i've read in other threads, a MAF sensor reports the airflow into the engine and is independant of rpm and manifold pressure. So in theory if my engine flows 40 gms/sec (or 5500hz) at 1600 rpm @ 60kpa and the same amount at 2800 rpm @ 35kpa you should deliver the same amount of fuel in both instances, Correct?

After scouring my datalogs i can safely say that my BPW do not match when i'm experiencing the lean condition. Typically i see a higher BPW with the lower rpm opposed to the higher (ie. 3.67 vs. 1.67). I've made sure to examine only the data free of AE, DE, DFCO, PE, etc. influences.

As i understand it you can change the fueling by either changing the injector flow rate or modifying the MAF gms/sec settings. I've fiddled with both with limited sucess. If i dump a ton of fuel in to cover the 2800rpm lean spot, every other rpm spot is sick rich.

I've used this exact engine setup with an SD ecm with no issues. As a matter of fact, one day it ran on the SD setup and the next day it ran with the MAF setup. I'm confident the engine is not likely the problem.

I'm a little confused as to where to head next...

Mods in my sig.

RBob

iTrader: (1)

Hot wire MAFs measure airflow both into and out of the plenum...

RBob.

Bones232

Can you elaborate a little more? Am i missing a component?

By the way, I'm using a MAF off of a 5.3L truck (delphi 25168491)

dimented24x7

iTrader: (2)

He is referring to reversion. The MAF will see the airflow in two directions, which will alter the airflow it reports. But you have a realtively mild cam. I wrote the MAF code, BTW, and have run MAF with larger cams with lots of overlap. Basically what happens is that the engine will run rich at lower RPMs, but be fine everywhere else since the MAF sees the airflow that is scavanged out of the cylinder. The PCM adds additional fuel for air that isnt there, causing the rich condition.

This sounds like something isnt hooked up right, or the MAF is bad. The way to adjust the fuel is by using the MAF table. Set the injector constant for what the injectors flow, and then adjust the table. The table should always have a parabolic type shape to it, with no spikes or anything like that when your done tuning. Ill attach a pic of what it should look like. I can also send you a MAF table for your specific MAF if you wish. I have a few of them on hand.

Also, how is the MAF ducted? Are you using any long straight runs of pipe? MAF is great in that it measures airflow, but it requires careful consideration of intake design. There should be no long straight runs, and the MAF should be, at the very least, run with a cone air element on the end of it. Also, you need to make sure that your PCV air comes THROUGH the MAF. IOW, you need to hook your air inlet for your PCV into the intake ducting behind the MAF and air filter. Typically the intakes that work best with MAF are short and have variable cross sections that dont promote resonance. The air can resonate in a long straight run of duct, and this will screw up the MAF readings.

dimented24x7

iTrader: (2)

Also, can you send me a copy of your bin (PM me, and Ill give you my email offline)? I want to check to be sure your using the most up to date code. The MAF does work, but it requires a little care to make sure that all the air the engine ingests goes through the sensor. The MAF cant handle vacuum leaks very well.

Also, if you have any datalogs, Id like to take a look at those as well. Sounds like something is really out of whack by the way you describe it.

dimented24x7

iTrader: (2)

Heres what the curve should look like.

dimented24x7

iTrader: (2)

FWIW, the pulsewidth is correct for both instances when you have 40 grams/sec of airflow. The PW at 2800 RPM will be (1600/2800) x 3.67 = 1.84 msec, which is pretty short. The higher the RPM for a given airflow, the lower the airmass in each individual cylinder, so the PW will be shorter.

Its entirely possible that the problem is with the injectors. With SD, you would end up compensating for this unknowingly by adding in VE to make up for the time that the pintle takes to come off its seat. With MAF, you need to fix it by adding in PW to the injector PW offset table. If it consistantly does it at low PWs, then you add in PW to the table at the commanded PW where its going lean to increase the actual time the injectors are fired for. This will keep them open long enough to deliver the fuel.

What size are the injectors? From your data it looks like they are ~55 pph as set in the calibration. Also, what pressure are you running them at?

Bones232

Ya threw a bunch at me there didn't ya....

If i try and tune out the lean spot, i'll end up with a spike in the 5k - 6.5k hz range of the MAF table and everything else will be pig rich.

I'll post a picture of my intake setup when i get home tonight. I have accounted for the PCV air by welding a bung in my "carb hat" to pull air just before the TB. A cone filter sits within 2" of the MAF inlet.

Injectors @ ~35 psi.

FWIW i can drowned the engine at WOT and 5k rpm.

I also unplugged my VAFPR to try and isolate this issue.

Also when i get home, I'll PM you my bin, xdf and a log for you to look at.

dimented24x7

iTrader: (2)

Assuming the bin looks OK, the most likely suspect is the injector offset at low PWs not being correct. Thats the only thing that will make it behave like that, aside from something not being set up right. Take a look at the pulsewidth instead when it goes lean. Im willing to bet that itll be less than 2 msec when it happens. The MAF output is actually fairly high for the load involved, so the MAF looks like its reporting the flow, just that the injectors arent delivering it.

You can tell for sure by running the engine at a lower RPM and a higher load, and trying to get the same MAF reading. If its rich there, and lean at higher RPM, low load, then its the injectors lagging at low PWs.

Bones232

Looks like you may be on to something. I had to add the BPW constants to my xdf and i found the following;

PW threshold for short BPW (L4963) was set to 0
Min allowed BPW (L4965) was set to 0
BPW offset (L4967) was set to 0

That would explain why my DE would act like my DFCO was active when it was disabled. It would actually zero out my BPW.

Actually the 'red flag' was sent up, and the motivation to begin this thread, when i noticed varing BPW with the same MAF reading and HUGE differences in WBafr. PW at the lower rpm was ~3.xx and the higher rpm was ~1.67.

I'll switch the settings and see what happens.

Here are a couple of pics of my setup.

dimented24x7

iTrader: (2)

I have mixed feelings about the intake. The tubing is pretty long. I used a similar setup with mine, and had resonance issues. On a quick throttle snap, the MAF readings would actually show a decaying sine wave when the airflow changed. You can tell if its a problem by how teh engine acts during sudden acceleration. Itll chug for a little bit like its lean until the standing waves die down, but the WB readings will still be OK.

Also, as far as the BPW compensation goes, you can tell what it should be by the WB AFR readings. Tune the MAF under normal conditions where the BPW is reasonable, and then datalog the BPW when its real short. The actual PW thats needed will be: (desired PW) x (actual AFR/desired AFR). The additional PW offset that needs to be entered into the table (new XDF that I sent) is then the actual calculated PW - desired PW. Calculate these for a number of samples, and the added PW's should all cluster around one value (ie .6 msec or something similar). Then add this into the existing values in the table. You can probably just use the same values in all the cells as the delay should be uniform across the board.

I suspect that the delay in the pintle opening is due to the extra fuel pressure and large injectors, but if you get some large value for the correction, then it might be an indication of a problem like failing injector drivers or low voltage at the fuel injectors.

Bones232

What is the setpoint for the "PW threshold for short BPW" and "BPW Offset"? I didn't see those scalars in the xdf you sent.

Saar

so that's why GM uses the "ribbed" looking rubber intake ducts... and i was told it was ideal to have a straight section of tubing 10 times as long as it was wide to eliminate some kind of airflow disturbance(can't remember the technical name for it at the moment).

dimented24x7

iTrader: (2)

I can add them in. IIRC, theyre mainly used in conjunction with the table, as well as the alternate firing strategy in the PFI code, which isnt used in TBI mode. The main thing you need is the offset table that I added. I can check, but Im pretty sure those constants dont do much in the current MAF code, which is sync at all times to keep the injectors in lockstep with the MAF updates.

Bones232

Just a reality check. I was under the impression that the PW offset is the main injector offset for the code, the PW bias Table is an adder table for the offset and the PW threshold for short BPW signifies when the adder should be included in the fueling. Is that not the case?

Since i've been using the new xdf you sent, my O2mv and WBo2 don't report correctly in my ALDL stream, any ideas? I have a dash mounted afr guage, so i'm not completely SOL.

Anyways, I've been playing with the injector PW bias table (currently around 1.1 msec in all cells). I've got the lean spot at 2900 rpm and light load pretty close to target. I've also been increasing the injector flowrate to lean out the higher PW areas and bring them closer to target, like mentioned above. Previously I had to set PE to 14.8 just to get the WBafr to show up on my gage. Now the commanded 14.8 will yield me high 11's or so. So i'm going in the right direction.

One other thing i noticed was at lower PW, especially at low rpms the BPW will zero out for a short period and then come back. I haven't had a chance to examine the log to see if DE or DFCO were enable, however it happen while at steady throttle (1-5%)

Bones232

Ok so I'm looking at my datalog and see the BPW dropping to zero. DE and DFCO are inactive at the time. See attached to view a snippet of the datalog.

What else could cause the BPW to zero out?

Fast355

iTrader: (2)

You are either dropping below the minimum pulsewidth and its storing the pulsewidth and adding it to the next pulse or perhaps the proportional gains are doing it. Have you tried turning closed loop off to see if the BPW still zero's??

dimented24x7

iTrader: (2)

The BPW in the datalog is actually the timer output compare (TOC) hardware value used for firing the injectors. It is the PW, but the value is cleared and updated as the PCM goes through the interrupts in the code to fire the injectors. Its possible that you may be accessing it right after its cleared and seeing "0" when there is still PW. I can update the ADX to use the commanded BPW instead of the hardware value.

Does the WB AFR shoot up to the free air value when it happens?

dimented24x7

iTrader: (2)

I think the code also still has async fire. If thats the case, than the TOC value reported in the datalog will go to zero as the PCM will switch from the software driven injector outputs to the asnyc accumulator. Its been a long time since I wrote it, so I cant remember for sure if it does this.

dimented24x7

iTrader: (2)

On second check, it does still use the async. The actual BPW commanded is shorter than the reported datalog value since a fairly large bias is being added in to compensate for the injector opening delay. The PCM is probably switching back and fourth between sync and async. That would explain the reported PW going to zero periodically at low load. I can change the datalog to report the commanded instead so it doesnt appear to zero out like that. I should also update the ADX and code to make everything report correctly. There are a few different ADS and ADX files out there for the TBI and PFI MAF code, so its hard to keep track of what is in each.

Bones232

It's definitely killing the fuel. When it happens the afr spikes.

I'll check the datalog to confirm if async is active or not. I've disabled it in every other ecm i've use, i'm thinking i might do that with this one too.

In open loop i don't experience this phenomenon, but i still have a pretty rough tune. Everything is pretty rich still.

Bones232

Async is definitely on. Attached snippet verifies it.

Is there an option bit to turn off Async firing or do i need to change the thresholds so they are never satisfied?

dimented24x7

iTrader: (2)

If your PWs are that short, you may want to keep it in async fuel. The AFR will be a little funky since it holds PW for several firings to build up enough firing time, but in sync, you run the risk of dumping the PW entirely when the required fueling is real low. The other option is to drop the fuel pressure to lower the flowrate, and get more PW at low load.

Bones232

How do you go about tuning the async portion? Right now async sends everything into chaos. its not a very smooth transition either.

dimented24x7

iTrader: (2)

How does it behave when it enters async? Does it sort of vacillate between rich and lean? Thats usually what happens, and with TBI, it can cause issues as there can be a number of cylinders that can go without fuel before the next async pulse is delivered.

Async is really a last ditch effort to deliver the fuel when the PW becomes really low. You can disable it by setting the async PW threshold to zero, and it will just stay in sync. But, I think the computer limits the min pulsewidth to 1 msec or so. A better solution is to adjust the fuel pressure to target a DC of about 80% at WOT. If your lucky, this will result in a lower fuel pressure than what your running now. The lower pressures will allow a longer PW, and make P/T fueling easier.

I was working on an updated XDF, but my desktop took a dump. In the next day or two, I will update the XDF to include all the async and injector stuff.

Bones232

Thats cool. I'll fiddle around with the async settings to see if i can get it to work. If not, I can plug my VAFPR back in and go from there. I had it disconnected so i could get the fueling straight without the pressure jumping all over the place.

With 55# injectors and 35psi I was at about 90% DC on the old setup. Reducing the pressure overall would most likely starve me at WOT without larger injectors.

dimented24x7

iTrader: (2)

The code has a table that can compensate for the VAFPR, if you know the relationship between the fuel and manifold pressure. You just enter the corrected injector flowrate at each cell.

Bones232

I've got the mandatory async disabled (rpm threshold maxed out), however it seems to async fire at low pulse widths. Is there a setting for min pulse width to enter async somewhere? Even with the the VAFPR plugged in i get low pulse widths on decel and light loads (<30 kpa).

I got a buckin' bronco when it jumps between sync and async...

dimented24x7

iTrader: (2)

You might want to consider using DFCO when your going into low MAP decel with a stick. I used the code with a stick as well, and DFCO was a must at low load, high RPM situations. If you have it disabled, you want to re-enable it. Its a must with TBI and high flow injectors. The other option is, if its happening under load, is not to run it up to a high RPM under light load. Even my SFI setup misfires sporatically if I let it run at high RPMs under light load. Easy solution is to simply upshift and run at a lower RPM. Its hard to deliver fuel there. With MAF, you cant fudge the fueling to lengthen the PW, so the larger injectors will pose a limit on the fueling.

I also have a copy of the code with all the addresses of each variable included. I'll forward that to you. I also still owe you an XDF. Ive been a bit buisy, so I got sidetracked, not to mention that I havent had my desktop working since it crashed. I should be able to finish it though, as promised this weekend. Do you have anything else you want to see besides the async stuff? I can add that in as well. Alternatively, you can create your own with the list that I sent that has all the addresses.

Bones232

I don't have a problem adding to the xdf. I just need a reliable hack or list of parameter locations.

I'm a fan of DFCO and activate it whenever possible. However, Async was also activating durning low rpm light loads. I ended up using one of your old hacks to locate the low pulse width thresholds to activate async. Once i figured out they were 16bit terms, i set them to zero to see what happens. It definately reduced the amount of time it went async and ran smoother also. I haven't been able to datalog yet though. Hopefully tonight.

I see you posted the '411 stuff the other day, so i know your busy. I appreciate any time you can spare.

helo

post up that XDF here if you can.

bones, what injector offset are you running? I'm using factory 5.7 tbi injectors @30psi with my offsets bumped to like 530 usec. (and that may even be a bit low).


I'm adding back-pressure EGR and tinkering with some things. I'll probably be testing EGR-MAF out next month (or two..) depending on amount of free time at my job for coding =).

Bones232

My injector offset is set to 1.1msec. Its a bit on the high side. However, going to extremes sometimes helps me understand the impact various terms have on the calibration.

I'm thinking i'll drop the offset to 900usec. I am also using the 55pph injectors at 34psi.