E40 VE Tuning using MAF Numbers

I have always had a difficult time getting the GMVE table in the E40 nailed down. I always had to multiply by %-half which took too long and even then it would end up overshooting. Doug (EC_Tune) gave me the idea of logging VE airflow and MAF airflow and plotting the difference against the VE table. Since I have my MAF dialed in pretty good and I run with the dynamic airflow disable RPM set to zero which disabled VE lookups, I decided to try it. I added VE Airflow to my table and created a custom PID for "VE Error" which is sort of like AFR Error except it is a percentage of how far off from the MAF flow the VE flow is. It goes like this:

([SENS.40]-[PID.2311])/[PID.2311]*100

Where SENS.40 is the Mass Air Flow (SAE) and PID.2311 is the VE airflow.

Then I set up a histogram to plot this custom PID against the primary VE table. Then you just copy and paste special, multiply by % the errors displayed in the histo.

I got some pretty decent error since I hadn't touched my VE since I installed my 100mm MAF and scaled my IFR to stay under the 12.2k Hz and 512 g/sec limits of the MAF table. I didn't try to hit the entire VE table, it was a drive to work so it's mostly in the lower MAP/RPM range but the result for the cells that I did hit was a very nice smooth VE table. I'll flash this for my drive home and see how closely the VE airflow matches.

I'm not familiiar with Bluecat's program because I don't have a VE equation, but I would imagine you could use the % errors you get from this the same as fuel trim or AFR error numbers to plug into his program to derive your VE constants. I donno.

So does this mean your going to start running with your VE tables enabled again Bill?

Perhaps. Just to see if it improves anything. The log from the way home still showed very negative numbers so it may take a few iterations, not sure why. If I can get the VE airflow numbers to jive with the MAF numbers then I'll try re-enabling and see if it does anyhting.

I'm pretty sure they don't match because the numbers in the VE table (equations) don't have the density applied yet (MAT correction). Usually somthing like .5 in the low airflows and .7-.8 range in the high airflows. All being depandant on IAT, ECT, and your bias tables, Reguarless to do it right you'd have to make a pid that took the maf ve number and the estimated MAT and back into the ve table version of the ve numbers. Which first someone like RHS is going to have to reverse engineer exactly how the computer uses MAT to arrive at the final number.

I've supposed to have been making the guys over at LG a spread sheet or scanner config to do this very thing, but haven't had time or a car with a E40 or E38 here at the shop the last few weeks. If I just had a test mule handy I could make a log with the MAF off , then log GMVE and MAT. Compare the VE Table or (equation generated) to the GMVE and equate the difference to the corisponding logged MAT. Say for a specfic rpm and map the VE is 1000, but in the log the GMVE is 600. In the same frame the MAT is 60c. If what I think the computer is doing end up being right, every frame the MAT is same, the ratio between he VE table and GMVE would be the same. The it just comes down to sampling enough MAT's and VE ratios to back into the math the pcm is using to do the density correction. Then we can make a pid to estimate the tables version of the ve from the MAF generated GMVE. Make a histogram of that, fill in the blanks to get a full ve table and plug those numbers back in to the pcm. VE table populated from the MAF calibration.

I do this all the time (everytime I tune a car), but I'm going the other way. I always tune the SD first. Then before I turn the MAF back on, I build a starting point for the MAF table from the last log I made in speed density. If you log dynamic airflow (g/sec) while in sd mode you can see the end result of the SD calc in g/sec. Histogram that number vs maf hz and poof, you have your starting MAF table. You for some reason have to add about 3%, but you end up with a MAF tune that matches the SD tune pretty good. Good enough that one more log / fine tuning of the MAF table and its ready for the dyno.

Oh, and Bill, I just wanted to say I agree with you. When you turn off the sd calc and run MAF only, that the cars do run just fine. The very first big cammed c6 I tuned I had to that very way because my program didn't work right at the time. I will say altough the car ran and drove ok, it just felt like the transient fueling was off. It would never stuble or anything on a throttle blip, but it just didn't feel "right". I think that if you get a correct SD calibration running under your MAF calibration, you will find that the car just feels cleaner.

You know 5L / Blue, I noticed the same post 5L mentions and have had similar ideas about turning on my low-speed calcs again. Maybe this weekend, if I can get comfortable with the way the E67 handles VE.

Well it hasn't been very successful. Yesterday's logs had the VE ~15% too rich. I applied those percentges and this morning the VE is ~15% too lean. So it's pretty much the same as tuning the VE table with AFR error or fuel trims, multiplying by %-1/2 is the best I can do (at least as far as I'm willing to go) and thats not accurate enough because of the things that Bluecat mentioned above. He is much more intimately familiar with the inner workings of the gen 4 VE tables/equations. For now I'm content to leave VE disabled.

Bluecat, your program doesn't work on the E40 VE tables right? Just the equation/constant based VE's? Also, did I read your last paragraph from post #4 correctly that I should be using dynamic airflow to compare against the MAF flow instead of the VE flow?

Quote:

---

Originally Posted by Bluecat

I do this all the time (everytime I tune a car), but I'm going the other way. I always tune the SD first. Then before I turn the MAF back on, I build a starting point for the MAF table from the last log I made in speed density. If you log dynamic airflow (g/sec) while in sd mode you can see the end result of the SD calc in g/sec. Histogram that number vs maf hz and poof, you have your starting MAF table. You for some reason have to add about 3%, but you end up with a MAF tune that matches the SD tune pretty good. Good enough that one more log / fine tuning of the MAF table and its ready for the dyno.

---

Yup, this is exactly what I used to do, until I realized there's problems with going this way, due to the temp estimators, this why right now I'm working on going backwards--from MAF to SD. MAF should account for all Temp based changes, allowing for more precise airmass calculation, which then in turn can be used to determine GMVE _IF AND ONLY IF_ Temp estimators are set right. No matter which way you're gonna go, you're gonna end up with one equation and two unknowns in the SD model. This is why the full simulation and modeling method is the only scientific solution. Anything else is forceful oversimplification resulting in a neverending sequence of retunes.

Quote:

---

Originally Posted by redhardsupra

Yup, this is exactly what I used to do, until I realized there's problems with going this way, due to the temp estimators, this why right now I'm working on going backwards--from MAF to SD. MAF should account for all Temp based changes, allowing for more precise airmass calculation, which then in turn can be used to determine GMVE _IF AND ONLY IF_ Temp estimators are set right. No matter which way you're gonna go, you're gonna end up with one equation and two unknowns in the SD model. This is why the full simulation and modeling method is the only scientific solution. Anything else is forceful oversimplification resulting in a neverending sequence of retunes.

---

You make life seem so dismal.. lol

heh, that's just a snippet, i got about 30 unpublished pages of other writeups about it, i titled it 'everything you know about ve tuning is wrong.'

Quote:

---

...forceful oversimplification resulting in a neverending sequence of retunes.

---

That was my experience with VE tuning on the E67 the one time I tried it. Yay MAF!

Quote:

---

Originally Posted by 5_Liter_Eater

Bluecat, your program doesn't work on the E40 VE tables right? Just the equation/constant based VE's? Also, did I read your last paragraph from post #4 correctly that I should be using dynamic airflow to compare against the MAF flow instead of the VE flow?

---

Yeah, I meant E38. I keep forgetting what year your car is. I fixed my previous post. Still the backing into a VE table from MAF numbers would be the same since both have the same feel on the GMVE numbers.

No, you probally need to keep working with the GMVE numbers. I was just sighting an example of how I go the other way, VE to MAF.

Schedual says I've got a 08 vette with a procharger coming in this morning. I'll take a good log with specific data towards the cause.

Alright, this is rough around the edges since most of my data came from one log, but it looks like (at least in an over simplified form) the computer uses this to go from ve to the final air calc:

g/cly = (ve / 345) * ((map * 100) / (287 * mat))

I found it easier to back into it via g/cyl instead of the GMVE pid after I seen it was probally just using the idea gas law to do the density. Reguardless to go the way we want it's:

ve = ( g/cyl / (( map * 100) / (287 * mat))) * 345

...or for lazy people who just want to cut to the case, creat a custom pid called "Estmated VE" and use:

( [PID.2321.MET] / (( [SENS.30.MET] * 100) / (287 * [PID.2126.MET]))) * 345

This was done on a 08 vette with equation based ve so I don't know if the math is the same in a ls2 car with a ve table, but its a start. You'll need log Manifold Air Temp, Manifold Absolute pressure, and Dynamic Cylinder Air to have everything needed for the calculation.

Force it in high speed mode so it runs off the MAF, then get your MAF table good and dialed in. After you've got it where you want it, make a log using the custom pid shown above and compile the data into a histogram.... And you should get enough data to build a ve table without having to do any ve tuning.

Let me know how it works. We'll probally have to jack with the 2 constants a little bit, but the format of the equation should work enough to get the job done.

Bluecat, you got email ;)

Forget everything I just said. RHS informed me that its simply Cylinder Air Mass = GMVE*MAP/TEMP.

Beyond that, I really screwed up because I though I had seen that VE the numbers as they reside in the VE table or Equation didn't match the GMVE pid when it was logged. I checked the log I made the other day and they match just fine. I need to get more sleep. So about every post I've made in this thread is completely useless. Whats worse is that's what I've though for almost a year. :crazy:

So lets start over. For those who have been running with MAF only via forcing high speed mode, why isn't build a ve table from a MAF data a direct step?

If the GMVE pid reflects the MAF calc while in MAF mode, its a simple as building a histogram of the GMVE pid. If it dosen't, then it's still as simple as deriving the ve numbers from CAM, MAT, and MAP while in MAF mode. GMVE = MAT / MAP * CAM * 1000. So the PID formula would be:

GMVE = [PID.2126.MET] / [SENS.30.MET] * [PID.2321.MET] * 1000

the discrepancy of the numbers you're seeing is direct result of not converting everything to metric. remember that IPW is in milliseconds, and the numbers HPT displays for GMVE is is milligrams*K/kPa. Temps need to be in Kelvin.

No, it is not as simple as backcalculating GMVE from MAF-derived Airmass. All that would get you is GMVE you already have, as you just tuned for it with the current Temp estimator. The only way arrive at true GMVE is to simulate different bias and filter curves, based on that generate simulated Temp, and THEN figure out the GMVE values. Then using these GMVE values generate an airmass figure, which you then count up discrepancies against airmass figures derived from fuel usage and AFR. Now how you arrive at the bias and filter values efficiently ( you cant brute force it, I've tried ) is a whole different story. Also, making sure that GMVE values make sense and occur in smooth and continuous manner is yet another very difficult task. Enjoy ;)

Everything is in metric, I'm not having that type of descrepancy. All the numbers jive just fine. What I was talking about was from a year ago when I first noticed they added the GMVE pid in the scanner. On a 05 GTO I compared the GMVE pid with what was in the VE table and was under the impression that they didn't match. Apparently I was wrong because that was not the case the other day.

This isn't about trying to get the ve and bias tables dialed in perfect, I know as well as you do what that would entail. All this is about is trying to get a functional or at least a starting point ve table for these guys that have thiers completely disabled because its so far out of whack.

You guys keep saying your running totally off the MAF on a E40 LS2. How is that done? I thought no matter what, the VE table is always blended with the MAF? Please detail out what you are changing to run off maf.

Thanks

LS1 and LS2 engines use VE + MAP-sanity-check at low RPM (below 4000 on most, below 3600 on a few) and use MAP-only above said RPM. This is 'dynamic airflow mode" also called "low-airflow mode" and such.

If you change the value for "high-airflow mode" (the mode where dynamic airflow is disabled and thereby VE is ignored by the PCM) to below-idle values, then you only run off the MAF sensor for fuel calculations.

Ok, you dont really run off MAF only - fuel injector tables, OL multipliers, CL feedback, and tons of other things still affect fueling - but assuming your injectors are accurate (for example stock with stock settings) then you essentially run off the MAF sensor for fuel calculations.

What I do is remove the last zero so that I have an easy way to 'rememebr' the original values - just add a zero.