Evolution of a AutoVE Tune

[kwhiteside]

So I'm cycling thru the AutoVE process. Make a run, log it, get the BEN/Map recommended changes and multiply them to the VE table. Repeat the process.

I must admit, I barely unstand what I'm doing as I go through the motion of the tutorial from a engine standpoint, but now that I have some data, I'm starting to see what is happening.

The Tutorial starts by increasing the Kpa numbers by 15% across the table. Richening things up for protection. As you drive and log, the process makes real world changes to the table, bringing those numbers into line with reality, usually lower. Best case scenario, you would get real world number for every cell in your VE table. Seems that isn't going to ever happen, and this is the cause for the Rocky Mountains that many have complained about. Especially at the high RPM's where you car doesn't usually go there but for a quick moment.

In the image below, you see the graph of the mountains and valleys and the cells causing the affect. These cells started out with much higher values per the tutorial, the the logging process makes real world changes to only some of the cells.



In the image below you see on the top my original tune. The middle tune had the 15% tacked on, and then the real world changes from a spirited run around town. The bottom tun is after 4 iterations of this process. Note the highlighted cells in across the 3 tunes.

They start in the original tun with

83
85
91

Tutorial process pumps them up with some fuel to much higher values and they start working down

90
90
94

Now actually leaner than my original tune from many real world iterations.

79
81
86



My theory here is that the huge valleys created are actual real world reading from the cells that are getting hit. The peaks are not getting hit (enough anyway) so no adjustments are being made to them. So in theory, the valleys are correct, and the peaks should be leveled off to match the cells that had valid real world readings. Anybody disagree with that theory, please explain if you do.

[joecar]

You could level out the peaks to almost blend with the valleys, and then do another iteration.

[kwhiteside]

I flashed it a few minutes ago. If I get it started, I'll get a good log on the way home and see how close it is now.

[joecar]

See the other thread.

[waterbug1999]

How long did you log?

[mr.prick]

Open Loop tuning on the streets sucks.
You can log, filter, paste and smooth all you want but {B0101}
will not be consistent at all temperatures, altitude and even load in OLSD.
You can get a general shape for the VE table this way then enable LTFTs
and tune with them for non PE areas.
When PE is on and LTFTs are 0%, you can use WBO2 to adjust the WOT areas.
No smoothing has been done to this VE table and LTFTs are between 0% and -3%

[kwhiteside]

Well my VE table isn't as nice as your but after about 6 logs and tweaks, my AFR and commanded are dead on, and my Ben/Map barely recommends any changes to my VE table. This is what I ended up with, and I've turned the MAF back on. I'll have to figure out how to tune that now!



Oh, BTW, when doing street logging. I've discovered that during low volume traffic, you can find some good on ramps that have about the same function as a 1/4 mile dragstrip. Get off and on the highway a few times and you have some really nice full throttle red line logs!

[mr.prick]

How did you go from post #1 to post #7?
I seem to have missed the point of this thread.

[kwhiteside]

How did you go from post #1 to post #7?
I seem to have missed the point of this thread.

I followed the process and tested my theory over and over until I achieved the desired result as shown in my VE Graph progression from post 1 to post 7.

The thread was to show those people who don't like the AutoVE tuning process because it leaves the VE table looking like the rocky mountains that with a little more process, it can be smooth and a great baseline.

To repeat the theory. The tutorial has you pump up the gas in the VE table which creates the high mountains. Then real time logging and Ben/Map tweaking starts creating the valleys. The real time stuff is right on, so smooth it out to match the cells you do get. Use the interpolation spreadsheet, then go ahead and tweak manually the cells that you most likely will never hit. Do some more logging until you find the recommended changes from the Ben/Map are barely visible anymore.

[joecar]

Your VE table from post #7 is quite good (maybe a little too smooth, check your BEN's now).

...
The tutorial has you pump up the gas in the VE table
...

Some notes:
VE table is a normalized airmass table (normalized for temperature and pressure), its units are g*K/kPa (not %).
PCM uses VE table (and/or MAF table) to compute cylinder airmass;
PCM then looks up AFR from richer of OL fuel table or PE table (if PE enabled) in OL, or uses stoichiometric AFR in CL;
PCM then computes fuelmass required to meet that AFR;
PCM then looks up IFR table and computes the injector pulsewidth required to deliver that fuelmass.

So increasing the VE table (or the MAF table) increases the fuelmass (in an indirect manner), just be aware of this.