Just to set the record straight, and be somewhat informative about the subject, I'm going to do some splaining about what goes on, and try to once again clarify what I've been saying all along.

From what I've been able to see, or find, the fastest oem MAF'd car is Mark Jackson's GN. It ran ~9.7's in the 1/4 in 03. That car using the same basic code but MAFless did a reported 9.3 in testing (in 04). All the related chips were done by Steve Yaklin of Max Effort. As a favor to Steve, and acknowledged by him on his web site, I spun up each, and every chip for Steve on my ecm bench. I used DirectScan, logged everything, and then set the files back to Steve for analysis, and then Mark would run the chips. Per my argeement with Steve, I've never looked at his coding, or saved any of the files longer then to just copy them to a chip, and then run them on the bench. However, I've done enough recording and running of the chips on the bench to have a couple clues about what's needed to run 9s with the oem ecm, and MAF. So for those of you that have tried to impress me with your numbers, or infer that I haven't been to the strip (enough), excuse me if I continue be not impressed. For those of you that have run good numbers, and have reported the results, I salute you for your work, it's the guys playing the my dog is smarter then your dog, nonsense that make me ill.

Obviously, you can run fast with MAFs, unless you don't consider 9s as being fast.

The problems are (as far as the C3 ecms go) is the lack of granularity. While that makes tuning, somewhat easier, it also means you just can't get things really close, and for that throttle response, and mileage will suffer. It's the attention to detail that makes drivibility all it can be, and get all the mileage possible. While, running closed loop will crutch the MAFs into being decent, that's the best they'll ever be. At WOT, there isn't the critical need for that level of granularity, so they'll always work pretty well, there. Also, once you get into heavy mods, and change the stall speed of the converter, that changes the whole picture tremendiously, it allows means working on the LV8 scaler, something of which I've yet to see any of the alledged MAF gurus work with (BTW, BTDT, on working with LV8 scaler).

There is also the issue of over reporting the airflow during AE transitions, but like all things, once you identify a problem, there's probably a software cure for it. For those that missed it, some of the airflow measured by the MAF is actually used in changing the plenum vacuum, rather then being consumed by the engine.

For all the hints I've dropped, it seems no one's bother to look at the DIY archives, about how to deal with large airflows and MAFs. I'll give ya'll another hint, you might do a search there for the name Mark Romans, and see what you find.

While for some reason I've gotten pegged as being anti-MAF, that's only been by those that haven't bothered trying to understand what I've been talking about. All I've ever done, was point out the inherent flaws with MAFs, my intent was to get people thinking about what they're doing, so that they could look at what changes would be needed to improve the code in the C3s, or tune around the problems.

Now if someone has any questions, you might ask them, about what I've posted, rather then assume, or misquote what I've said. I'm not trying to be/ sound arrogant, it's just that I'm tired of arguing with people that have made some assumptions about what I've been saying and hinting at.

Now, you can write this in stone, at this time, IMO, the best system would be a hybreed of MAF, and MAP, thou who knows where the future may find us.

HTH

 

MAF-Hater!

:0

Must've missed something. Did you get a nasty-gram in your email or PM?

I thought it was already fairly obvious (even to the factory) that a MAF/MAP combination allows the best of both worlds. Short of that, a P4 and MAP are probably more controllable for most non-stock applications at stable altitudes for all but the lowest RPMs.

 

Nope, just consolidating some misc writtings.

The new OBDII systems use both for a variety of reasons. Obvious, is in the eyes of the beholders...

Might even include individual cyl timing control, and fuel as being something that's obvious needed since the latest PCMs offer it, but the gets to be a time/ stage, of deminishing returns.

Even a C3 with fully optimized code is one slick setup. The only downside to them is the slow ALDL. But, yes, a P4, is more then enough processing power for most any street car.

 

I'm not fast by any means, and I keep getting flak for using the stock 2.8 V6 MAF stuff, but hey - it's working. I just need to find a non-collapsable 5/8th" hose!

 

In my 8063_MAF.ASM on craigs site there is a simple routine in there to use one of the quad driver outputs as a 1280 bps high speed bit bang serial routine if anyone is interested.The C3 is awsome. I could practially program that thing with my eyes closed. Its a real shame the hardware sucks so much, cause thats a great computer IMO.

As far as the P4s go, if interrupts for the timer input compares are available, it looks like at least one of the TIC's could be used for a pseudo-16 bit MAF frequency input. In the 68HC11 based PCMs, on the surface of things, it looks like the rate that the free running counter runs at could also be changed, allowing for more precise period measurement of a high frequency maf output Downside is that it would really f other things up. Or maybe take start/end measurements over several periods to help reduce some of the uncertainty. Even better is if the pulse accumulator is available for use.

 

If I read this right, when running a MAF setup aka 165 ECM, I can only hope to achieve close to a 128 BLM. I will never get it "perfect" across the board?

JOhn

 

I don't want to respond with yes but there are things that are causing the issue that have to be considered.
Like Grumpy stated above and in many other posts, the MAF is not good at reading fast changes due to vacuum air flow. Meaning that during steady state flow and rising/falling flow is read quite well (except for the resolution issues in the code)
The problems begin when you apply AE to it and now have to "try" and guess what types of delays are built into the system physically.
The volume of air that is taken in during a stomp of the throttle is initially (for a very short time) under read due to the fact that the pressure changed and not just the flow of air. then as the steady rise of flow is being done all is better.
When the throttle is closed another delay or misread occurrs due to the volume of air required to generate the higher vacuum.
Those transitions are not compensated for over the time period they occur. The calibrations in the code can be made close but it is definatly on a setup by setup basis to get the tune "close" enough.
I interpet the matter as these conditions and similar things.
If I've gone off base please correct me. I'd like to get a better understanding if that's the case.

 

there will always be some resolution error, especially at low loads. Typically something like 3-4% or so. This could equate to BLMs of 120-135 for when the system is 'tuned'.

 

Talking about Percentages... You state 120-135 when things are "Tuned" and 3% to 4% can cause that kind of range.

With that stated, is a 4% change equal to approximately 8 on the BLM scale? The reason I am asking is my BLM's range from 114 to 128. If I decrease my grams reading by 4% then in theory I should be closer to ideal.

When you say "Low Loads" are you referring to a light acceleration, or constant speed driving? Both conditions should have light loads.

I am not trying to hijack a thread... It just helps me understand BLM's in relationship to percentages is all.

John

 

No, you can get 128s, across the board, the problem is how many data points, the board has. The more *nitpicky* the calibration is, the more finely tuned it can be, that is what gets you the ability to split hairs and give the engine exactly what the engine wants, when it wants it.

128s just means the ecm isn't making any corrections to the table values to achieve a Stoich AFR. It's largely in part due to the actual calibration of the MAF. A MAF while it is in the air stream, still relies on mathmatical modeling to interupt that sensor's output into a air consumption value.

Not to mention that the LV8 calculation adds another *rounding error* to the table.

 

Pretty much. The maf error will appear right in the outputted pulsewidth since the actual and indicated airflow will be off. This will, in theory, be corrected for by the ecm.

By low loads I mean low vac/low rpm operation where the airflow is low. With a voltage based system, the resolution declines as teh airflow decreases.

 

I think I see what your saying, for example With the MAF table there are gaps between the Grams/sec numbers. For example on one of my MAF tables it does not climb in small increments, rather it skips some numbers and the ECM fills in the gaps. Whereas a MAP table would have more numbers to play with. MAP is more absolute, MAF is more general.

John

 

I have also noticed that under light acceleration my BLM's tend drop off a few points. I assume that is normal since it is adding fuel to accelerate... Or can I compensate for that?

John

 

Don't forget the timing is also changing, and it's using an LV8 calculation, which is going to be different then what the fuel is doing, since it's *direct reading*.

If the BLs are moving around a *little* don't sweat it. If you can keep them within +-5, you doing really well. +-10 is even acceptable, IMO.

 

On thing that hasn't been mentioned is that in the 31T code for the Turbo TRs, they use both a % enrichment by RPM, and TPS.

Just another item that kinda shoots down the *perfection* in the way an early MAFs *worked*.