actually change the program. I figure we are not changing a table as I have been with by Eprom chip burning equipment. So how does one change the basic program???

If example, how would I go about change the max flow limit for each of the 6 MAF tables in $32b???

 

Click on this great post from RBob on Code Patching to get your mind working overtime.

John

 

You can't use software to cure a hardware problem.
As the rates get higher thru the MAF the values compress to where there's not much resolution.

To do what you want with the MAF requires no patches. ie to some degree expand the range of the stock sensor.

Once you figure out what to do, I strongly suggest you use a WB to set it up.
And if your thinking about doing anything with the code you'll need to put together an ecm bench so that you can run the code on something other then a car and make sure it works. It's not as easy as it may sound for some items.

 

The maf returns voltage, and you have 1 register (8 bit) to convert what it responds to Gramps persecond, two grams per second, 1.5 grams per second, or whatever the hell you want it to reference from 0 - 254. (or is it signed? i can't remember)

Anyway.. If you want to make the scope. Wider, you need to consolidate the load it calculates based on this, fuel, etc.

So 255 grams per second, could really be something like 382 grams per second. It just won't be scaled as finely.


Its kinda-sorta how a 2 bar map works on the 749 vs 730. Instead of 5 volts being 100kpa (730), now its 190.. heh


What Grumpy said is true. The limit is the hardware. (not that I've looked that deeply into the code but, merging registeres on 68HC11 is supported, so the 1 byte limit is a load of crap - it must be in hardware).


-- Joe

 

The problem with the MAF setup is the MAF reports the airflow based on a curve. The curve comes from the temp coef of heat loss across the wire/resistor.

Map is a linear strain guage. Kpa/bit resolution and errors will be spread out more evenly.

You only have 255 points on an 8 bit A/D converter. Placing a curve on that presents a problem because the resolution gets worse as the curve becomes tighter. It becomes a simple math formula of Grams/sec per bit.

There is a reason GM went to a frequency based MAF later on.
SAE paper 200-01-0546 explains in great detail the difference between MAF A/D conversion and MAF/frequency conversion.

One of the best pieces of information on that document is the differences in the loss/gain of resolution on either end of the scale based on whether you use a higher resolution A/D converter or stick with frequency conversion.



No need for me to post the math here. If you read a few documents posted by the guys who designed this stuff, you quickly realize that some projects should never get off the
ground

Happy reading,
Need4speed

 

Oh I dunno, Gramps is kinda old and slow. You might get one, possibly 2 Gramps a second, but I wouldn't expect much more from the old codger.

Sorry, I just COULDN'T resist.

 

You'll also find that opinions change, and once you get into execution of some ideas, all is not as well a theory would suggest.

The only way to actually figure out what works is to spend the time to actually test things. And this is car specific, what works on my turbo car doesn't mean it will work on a small turbo'd car, and much of the data is useless for N/A.

heck read some of the papers on EGR and you'd swear it was the best thing for mileage since ultrasonic generators <g>.

 

I do spend a lot of time testing things. I do it with a calculator first. If it does not work there, it won't work on my test bench, and it sure won't work in a car.
Then it's off to the dyno. Then it's off to the street.

Certain losses in resolution is fine. That's what interpolation is for. But straighten out too many points on a MAF curve and I can guarantee you driveability problems.

I have played with forced induction applications. Tiny,small and big block. Turbo and blower.

I can also tell you from experience that there is no difference in application between a N/A motor and forced induction as far as fueling goes.

Themodynamics does not change over time. Air molecule density
does not care that it belongs to a second atmoshpere. So the comment of it will work with a turbo but not N/A is B.S. You must be at the wrong temperature, or on the wrong end of the resolution scale.


I also beg to differ with your EGR comment. EGR is your friend. If you missed that, go back and read it again.

 

Sometimes yes, sometimes no. I equate it more like a "drunken buddy" who's 5 feet tall but acts like he's 6' 4" and gets a big mouth when he's drunk.

I've used it really successfully in obtaing high fuel economy and have gotten outstanding mileage out my L98 when I ran the stock TPI setup. But I've also found it to be the source of many driveability issues.

 

Then it wasn't working properly or you had a modified engine which the factory EGR wasn't calibrated for and couldn't compensate for.

 

On some cars I suspect it was "coking" of the passage ways as the EGR was replaced. My buddies' were not prepared to take off the intake to clean all the passage ways when disabling the EGR worked so well.

On others, I was making the GM ECM do things it was never intended and the EGR DC needed to be reworked.

Taking Highway Mode fuel & spark to the limits will require a lot of work/testing in "phasing" when working in conjunction with the EGR's DC. It all depends on what your are doing.

On my Firebird, I've now gone Miniram, so the EGR is a non-issue.

 

Agreed. You hit the nail on the head.

I don't want to start a fight, but there is a ton of information on EGR flow properties. Air is to be considered the dilutent with EGR. Not the other way around.

You start getting into cam profiles with self EGR, and you have a whole now range of parameters to correct and rpm ranges to keep the EGR out of.


Need4speed.

 

Your kidding right?.
You test with a calculator?. Sorry but that's not testing that's more often then not just ego justification. If your idea of testing was even close to accurate Detroit could save millions a month in research.

Since your explaination involves chest beating with dismissing other then using, a calulator as BS. Your on way too much of an ego trip to be worth trying to communicate with.

And you can beg all you want <your EGR comment>, I've done the testing and know what happens when it gets pass theory and gets to execution.

 

You never learn.
Since you and I have gotten into it in the past before, I guess it's time again. Every time somebody with knowledge about automotive comes around you, there is a fight.


When it comes to things like rescaling a maf, or rewriting code, you had better use a calculator. All of you B.S. ecm bench guessing is a waste of time. Get it right the first time. Then prove it in the real world. I never said all you need is a calculator. You need to learn how to read.

If it was not for the fact that somebody gave you guys some internal documentation from GM about the 7747, you would still be stuck on it.

Quit B.s'ing Bruce. I know all about it.

"General Motors (unpub.)
GMCM MARCO CROSS-ASSEMBLER"

Should ring a bell. If not, you were B.S'ed by somebody else.

Look Bruce, you are not the only one who has used lab grade
emissions equipment, airflow bench, engine and chassis dynos. Been there, done that. I have made a lot of good friends over the years and need to thank them again for all of their help and insight.

I also have a good friend who does emissions testing for a living. E.O. numbers to be exact. So keep your play-date routines to yourself. EGR on a street machine is warranted and works.


I'm not here to make my self look good. I have better things to do. I also don't pass out garbage info either. If I write it here it's because I have done it, or know somebody who does it for a living and has given me the answers. I like to pass good info along.


Again, this thread was about trying to rescale a MAF to work with an 8 bit A/D converter. It won't. The numbers won't allow it. The margin of error is to great.

Maybe you should call GM and let them know you have their problem solved.

Talk to me with facts and figures that disprove this. Or just be quiet.

 

you cant work outside of the 255 limit to the ad conversion. the loss of resolution is livable on a modded streetcar . itll require a bit of bsing to work and the fuel trim will be a bit sloppy for sure. its still better then a carb and from my standpoint alot more drievable then speed density which is prone to lossing as much if not more resolution as the vacum at idle goes down. there always a comprimise. thats thet end story. now the other factor no ones been looking at is the fact that the 68hc11 can do 16 bit fucntions but it would require a major code overhual and the physical a/d converter might be an issue even with a FQ based maf. im not enough of a hardware guru to even dive into how to make it work. ive been investigting but nothing bright and shiny pops up. i think its still better to eat the compressed resolution and mopve on with life. or we could expand the maf table to the full 255 entrys and ust work from there taking full advanatge of what the a/d converter has left to offer. there might be some fuel trim left just hiding. there some options on the plate but none are the end all be all.

 

If I'd had the **documentation** I would have sure as heck not made the errors that are/were in what I printed.

Since you seem to know so much about internal documention, care to share any?.

Do you think that I'd publicly release an internal GM document?. If some one had given me such material, do you think I'd risk getting them fired over it?. Much less risk the wrath of GM's lawyers?. Please put any thoughts of your's into the prospective that at the time, noone had publiclly tried hac'ing. There were some guys that had reverse engineered parts of the code, but the amount of info., publicly domained at the time was very limited.

Your claim, just doesn't make sense.

 

If you were to actually look at the fueling in the MAF type systems, you'd have enough clues to figure it out for yourself.
Why should I feel obligated to do your homework for you?.
Do some data logging and look at what happens.

As for your misunderstandings about MAFs, you might spend some time investigating what happens with a MAF and the code at high flow numbers. If you were to look at the GNs and what they've found out, you might be better informed about what's going on.

Then you'd also understand the *whys* of the various different fueling strategies from system to system.

 

The problem with a voltage maf on an 8 bit A/D converter is trying to recover the curve. When the ECM is between the points it can read it uses linear interpolation to determine what the value should be.

Linear means a straight line. When you lose too many of your curve plot points, you start to straighten out that curve.

16 bit math won't be much of a help to recover the curve. It still boils down to the input value.

You are going to lose a ton of resolution on the lower end of the ariflow scale.

Frequency based MAF's use the pulse accumlator and 3 sample times to determine the frequency as a funtion of time.

 

Once again, you know nothing and seem to be willing to prove it on a regular basis.
Tust me, I don't need you to do my homework for me.

Here is a snippet on how the frequency based MAF determines
input value.


I don't suppose you would be willing to tell me which one of the above is the pulse accumulator and which is the timer capture?

 

OK, I'll sit down, and let you demonstrate what you claim to know.


So just for openers, what do the faster processing rates of the latest PCMs allow?.

 

more diagnostic features and thats about it. aside from things like C.O.P. and sequential fueling they arent much different from there older obd1 batch fire cuousins.

math is math.

back onto this whole maf cruve drop out. lets say we take the interpolation out of the quation for a moment.

if you only have roughly 35 table entrys in 4 2d tables. then yopu have a much larger linear scale. with larger distances between plot points.


now say you have 70 table entrys in the same 2d tables. you now have double the resolution to look at the same plot line so you can show a more correct curve.


now lets say we have 255 entry points to convert from

we now have a higly detailed plot line. in fact the newer pcm maf tables ive seen are roughyl 150 entrys. why do that ?? easy if you can create a more graduated table conversion then you can toos your resolution issue out the door.


these 68hc11 ecms can do 16 bit math. so lerts talk about making that work.

if we can trunciate to a 16 bit multiplier then we can beef up the lower flow resolution losses. switch back to 8 bit math in the top say 155 entrys. have 100 entrys for low flow and 155 for high flow. and even with a more compressed linear model it still works becuase the begining and end voltage are the same.

if the maf is 0-5 volts.

all you need to do is add tables to fix the issue. and for giggles i dont think its a big issue.

as for gm going back to FQ based mafs.


lets have this one hit home.

THERE CHEAPER.

end of story.

 

Amazing, for all the talk about this no one's actually done a thing about it.

Instead of just writting about it folks you might actually do some testing and see what you actually need.

You might just be amazed,
well at least I was.

 

Funstick, have you read the Delphi SAE article on
Air Meter Interface Stategies # 2000-01-0546 ?


Need4speed

 

Also sheds some light on why MAFs leave something to be desired on high performance Turbo cars.

Or the strategy they used with them in the past to make up for the poor resolution at high loads.

I believe I answered both some time ago in one of the Final Answer posts.

 

I'm almost tempted to say something about later model
ECM/PCM's as they relate to sample rates and signal pre-
processing.

But I'll just be quiet. It seems you guys already know the
answers.

 

Well, while ya'll be wanting to moan and groan about trivia.

Someone tossed me some info., so as make generating my own source code for another ecm, almost easy.

To bad, no one wants to *just do it*, and folks would rather hassle about stuff that has nothing to do with getting the job done.

So with my romulator, source code, and a scanner, I have a system better then any of the aftermarkets for a fraction of the cost.

If you read this as smuggness, it's not, it's just another wake call about *just doing it*.
*Just Doing It* means leaving the feature creep to the folks that just want excuses for not getting anything done.

Oh well, I got some code to study..

 

Well, got an even bigger gift and hand.
I now have Source code that assemblies and generates a valid .bin file. It is now partcially commented, and I have more info to add to it.

So it can be done, with *Just Doing It*.

 

im not worried about sampling rates they have little to due with actual sample size. it doesnt matter how fast you can read one inch. it does matter if you can read less than one inch. my question is can the 165 read less then one inch ?

as for source code. ive been bruning my eyes out learning assembler. i final wrote 5 lines of code in a 68hc11 simulator that actually add things together. so yeah im learning it but its painfully slow. i dont have enough time to devot to it. i wish i did.

as for the maf resoultion fall otu read above. and those improved sampling rates are only for emissions purposes. and obd2 diagnostics.