PDA

View Full Version : A0014



SSpdDmon
June 20th, 2007, 06:32 AM
In the COS's where you can skew fueling based on IAT data, does this setting only work with VE? In other words, if I run a straight MAF setup (B0120=400rpm), will A0014 alter my commanded AFR to richen me up or lean me out?

Doc
June 20th, 2007, 07:45 AM
Without overstating the obvious, the definition of the table says the correction is applied to the VE table. Moreover, are you getting stable enough readings down that low for the MAF to work exclusively? Interesting expirement nonetheless.

SSpdDmon
June 21st, 2007, 12:46 AM
Without overstating the obvious, the definition of the table says the correction is applied to the VE table. Moreover, are you getting stable enough readings down that low for the MAF to work exclusively? Interesting expirement nonetheless.
Oddly enough, it's fairly stable. I does go about a half point rich down low. But, above 2K it's pretty consistent.

Just to clarify, this is an IFR tune now. B0120 is set to 400rpm, VE/MAF tables are stock, and IAT is relocated to an accurate location. As much as I've been pro-VE/MAF tuning, I now believe in this method (after seeing real world results for when it's done right) and it will be added to my repertoire.

Doc
June 21st, 2007, 01:26 AM
Can you please elaborate on the benefits to this method and or the faults/problems you have been encountering with VE/MAF tuning. I don't doubt your results, I am just curious as to what was missing from the tune that "was so good yo momma can drive it."

:)

SSpdDmon
June 21st, 2007, 01:50 AM
Can you please elaborate on the benefits to this method and or the faults/problems you have been encountering with VE/MAF tuning. I don't doubt your results, I am just curious as to what was missing from the tune that "was so good yo momma can drive it."

:)
The differences are minimal IMO - mainly time related. I've just been experimenting with alternatives lately. Curiosity got the best of me I guess. If I can find a faster way to do something and end up with similar or better results, I'm all about it (Supply Chain mentality).

VE tune - Worked pretty darn good. I was happy with it. I may have to go back and observe once or twice more now that the IAT has been relocated. But, I felt I was constantly tweaking here and there with the 19 columns to play with (one for every 5kPa of MAP). Time invested...well, I can get it close in a matter of an hour or two. Getting it dialed in required more time and I found myself constantly tweaking.

VE & MAF tune - I could get this close, but never as close as the SD tune. It seemed there was some unseen factor that made tuning the MAF troublesome for me. I think with the VE's influence, the swings or variance in AFR was greater. Time invested: VE tune plus a couple of hours to get the MAF close. Again, there were numerous days of tweaking afterwards to try and get it closer.

MAF only MAF curve tune - Results were almost as good as the SD tune. I could get consistent AFR values fairly easily. But, WOT proved to be a challenge as the MAF Frequency bounces all over the place (e.g. 8124Hz~8576Hz~8213Hz~etc.). It's like I needed the final 'smoothed' number the PCM was using to make accurate changes for WOT fueling or something. Basing changes off of the stock curve helped prevent a 1% correction from sitting next to a 9% correction (smooth is good). Time invested: about the same as above. A couple of hours and it was close. But, the tweaking afterwards went on for days.

MAF only IFR tune - Results are currently just as good as the SD tune when the SD tune was 'on'. I'm currently monitoring day-to-day variance and don't have a whole lot to go on. But so far, it's looking good. The pros of doing this method: VE and MAF curves are stock - no time required. IFR curve is 2D (only one column needs tweaking). I've invested less than 6 hours into this so far and my BEN's are looking good. Yes, commanded AFR and actual AFR are targeted to be equal....no fudging. Cons: IFR table isn't straight as an arrow anymore. But, does that really effect anything else? Calculated injector duty cycle simply looks at engine RPM and the commanded pulse width, which is measured in ms. Whether the IFR table is right or wrong, the same AFR for a VE tune or an IFR tune at the same RPM should have the same pulse width, and therefore the same calculated injector duty cycle. What tables are skewed by changing IFR? :nixweiss: People say that when it comes time to swap injectors, the VE tune is better because you can just change the IFR values appropriate for the new injectors. So, I guess the only con I can really think of is...if I change injectors in the future, then I may have to invest a couple of hours to dial things back in again (invest time now vs. invest time later). The way I figure it is, if I mod the car enough to require more than SVO 30's, it's going to need a retune anyway. :lol:

voda1
June 21st, 2007, 02:50 AM
Is the IFR table your referring to B4001, Manvac and gr/sec?

SSpdDmon
June 21st, 2007, 04:05 AM
Is the IFR table your referring to B4001, Manvac and gr/sec?
That would be it. :)

Oh...the other thing I noticed between SD and a 100% MAF tune is in the transitional throttle. The MAF tends to transition to the rich side whereas the SD tune tends to transition lean. I think this has to do with the fact that the MAF is reading the air before it enters the motor vs. the PCM calculating fueling requirements after it notices a pressure change with the MAP sensor.

Doc
June 21st, 2007, 07:44 AM
Thanks for sharing your experiments with the board. The results you posted up seemed to be what I expected. Prior to acquiring the Road Runner I had run into the same feelings / results as you have had with VE tuning. Kinda like trying to find your way in a dark room with only a match (your wideband) while a hurricane keeps blowing it out (having to stop the car and fiter out/make adjustments).

The RR w/ RTACS eliminated all of those concerns you expressed, especially the time factor. Simply put, the emulator board is fast and the active feedback loop with filtering is eye opening (especially how the pcm averages things).

To me, I embrace the expanded resoultion of the COS Commanded Fuel while in while in open loop {B3647} and VE table over the 2-D, low res, Stock OS {B3605} and IFR {4001} and PE {3618}tables. I like how the COS's tables can be used to address more load/temp conditions.

Granted, it is more to keep up with but, fortuneatly for EFI Live users we have the software and hardware that can keep up and make short work of it.

SSpdDmon
June 21st, 2007, 07:54 AM
Thanks for sharing your experiments with the board. The results you posted up seemed to be what I expected. Prior to acquiring the Road Runner I had run into the same feelings / results as you have had with VE tuning. Kinda like trying to find your way in a dark room with only a match (your wideband) while a hurricane keeps blowing it out (having to stop the car and fiter out/make adjustments).

The RR w/ RTACS eliminated all of those concerns you expressed, especially the time factor. Simply put, the emulator board is fast and the active feedback loop with filtering is eye opening (especially how the pcm averages things).

To me, I embrace the expanded resoultion of the COS Commanded Fuel while in while in open loop {B3647} and VE table over the 2-D, low res, Stock OS {B3605} and IFR {4001} and PE {3618}tables. I like how the COS's tables can be used to address more load/temp conditions.

Granted, it is more to keep up with but, fortuneatly for EFI Live users we have the software and hardware that can keep up and make short work of it.
If only I wasn't a cheap bastard and would fork up the money for a Road Runner PCM. :lol:

Doc
June 21st, 2007, 10:00 AM
Furthermore, as tenacious as you are Jeff, I can't wait for you to get a RR because I am sure you will teach me a few things. Call Jerry Lewis, let's get the telethon going for kids without a RR!

;)

SSpdDmon
June 21st, 2007, 11:53 PM
Furthermore, as tenacious as you are Jeff, I can't wait for you to get a RR because I am sure you will teach me a few things. Call Jerry Lewis, let's get the telethon going for kids without a RR!

;)
:lol: I'll think about it. If my latest round of 100%MAF IFR tuning doesn't pan out in a couple of weeks, I'll definitely be giving it some more thought. As for my MAF (screened SLP 85mm), it's stable at low RPM. Right now, it's richening up about a half point or so down low. But other than that, no real side effects noticed yet.

And thanks guys for keeping an open mind. The hardest part of all of this was fighting my urge to tell myself, "Dumbass...IFR tunes are the devil!" :lol:

redhardsupra
June 22nd, 2007, 03:37 AM
IFR tunes are the devil. ;)
for the same reason i could say that 'MAF tunes are the devil' or 'VE tunes are the devil'

i'm saying this only to make a point about tuning very large and complex systems with only 1 or 2 variables is totally ignorant of complexity and dimensionality. I know you're trying to simplify it to such a system, becuase that's something our brains can deal with, but it's just not gonna cut it. airflow depends on temperature, and temperature depends on airflow. that alone forces you to adjust both at the same time. and there's a lot of other stuff coming to play on top of that. until we figure out injector setup 100%, all the rest is complete and utter bullshit, since we figure out airflow from the fuel consumpion.

SSpdDmon
June 22nd, 2007, 04:14 AM
Post deleted on account of being dazed and confused at the time. :)

redhardsupra
June 23rd, 2007, 03:03 AM
HUH?! i have no idea how you conclude that, and i've been accused of Reverse Polish Logic many times ;)

Doc
June 24th, 2007, 12:37 AM
I have a HP RPN calculator too.

SSpdDmon
June 24th, 2007, 03:06 AM
HUH?! i have no idea how you conclude that, and i've been accused of Reverse Polish Logic many times ;)
My point was...

I'm running a MAF only tune. The MAF is measuring airflow at a fixed point, right? Once it goes past the MAF, it doesn't just disappear off in a dark corner of the engine. Once it's measured, it's consumed. So, why wouldn't the most important reading be IAT - the temp at which the MAF is measuring airflow?

Like I said earleir, I'm having a hard time making myself understand where my brain was going with that. Twas just a random thought.

Either way, I don't think the temp bias table has much of an effect on my MAF only tune. At least, not that I noticed. I'm thinking it's primarily there for VE calculations.

redhardsupra
June 24th, 2007, 03:14 AM
1. if IAT was all that's necessary then why would there be a BIAS table?
2. temp of aircharge inside of the cylinder is NOT the same as in the intake, even if you're moving quickly, it's still got some ECT component to it.

hit me up on IM sometime, i'll show you some stuff i've been working on, i just got my first round of testing on a E40 done, it's been extremely enlightening.

SSpdDmon
June 24th, 2007, 11:33 PM
1. if IAT was all that's necessary then why would there be a BIAS table?
2. temp of aircharge inside of the cylinder is NOT the same as in the intake, even if you're moving quickly, it's still got some ECT component to it.

hit me up on IM sometime, i'll show you some stuff i've been working on, i just got my first round of testing on a E40 done, it's been extremely enlightening.
IDK if the bias table effects more than the VE table. I've tried my current setup with it zero'd out, with it at stock, and with a custom curve. Neither one seemed to make much of a difference. I agree the air temp changes by the time it gets to the cylinder - how much depending on the rate of airflow. Maybe that's why my 100% MAF tune richens up a point under 1500rpms?? The odd thing is, as soon as I'm above 1600~1700, AFR is right back in line with commanded.

redhardsupra
June 25th, 2007, 03:44 AM
jeff, the fact that your model has errors with specific shapes and in specific spots means it's not a good model. read anything about estimation theory and you'll see it's a textbook example of a bad model. i can explain more if you'd like, but it should be fairly obvious if you look at the distribution of residuals.
send me some of your logs and i'll show you what i'm talking about.


you're right about the bias thing. it shouldnt matter for maf only tunes. however, i believe that the maf/sd hybrid model in the gm computers is there for a reason, and i'm going to work with it, not around it.

SSpdDmon
June 25th, 2007, 04:00 AM
jeff, the fact that your model has errors with specific shapes and in specific spots means it's not a good model. read anything about estimation theory and you'll see it's a textbook example of a bad model. i can explain more if you'd like, but it should be fairly obvious if you look at the distribution of residuals.
send me some of your logs and i'll show you what i'm talking about.


you're right about the bias thing. it shouldnt matter for maf only tunes. however, i believe that the maf/sd hybrid model in the gm computers is there for a reason, and i'm going to work with it, not around it.
I realize this. It's more thinking out loud. Like I said earlier, I was typing it, but not making sense of it. :lol: