Why should we use the smoothing after tuning VE ?

[wesam]

I noticed that the smoothing feature decreasing the VE in some places and increase it in some places
and i think why we could not just leave it with out smoothing after tuning it even if its look spiky ?

[Chevy366]

Personally I leave mine the way it logged spikes and all , I will do a little blending here and there but nothing over the entire VE table .
At first years ago I did the smoothing thing , but realized as you have it alters the VE (not in a good way) and skews the data you just spent time gathering .
If you look not all VE tables are smooth , and in fact doesn't need to be either .
My $.02 .

[joecar]

I wouldn't smooth... but I would manually adjust a few cells that didn't look right.

[5.7ute]

I keep mine reasonably smooth. Mainly hand smoothing out spikes & dips. This I do for a couple of reasons. One is burst knock can become active with a big jump between cells. The other is it does feel like it drives smoother. But that could be all in my head.

[swingtan]

I tend to smooth as well, but not over the entire VE table. I'll select any area's of extreme high or low values and smooth the peaks down to be a closer match with the surrounding cells. I can't see an engine actually requiring such a huge jump between adjacent cells and put this down to "insufficient log data". My thought is that if you logged for long enough, the correction would result in a fairly smooth VE table anyway.

Simon

[joecar]

...
I will do a little blending here and there but nothing over the entire VE table
...

I wouldn't smooth... but I would manually adjust a few cells that didn't look right.

...
Mainly hand smoothing out spikes & dips. This I do for a couple of reasons. One is burst knock can become active with a big jump between cells.
...

...
I'll select any area's of extreme high or low values and smooth the peaks down to be a closer match with the surrounding cells. I can't see an engine actually requiring such a huge jump between adjacent cells
...

Good answers...

Wesam,

What is physically possible in the real world...? The physical actual airflow/airmass can not suddenly step up/down... but it can have resonances and other effects, so the table will not be smooth, it will have troughs/ripples/dips/bumps...

if you try global smoothing and do another AutoVE iteration you will see the same troughs/ripples...

a localized big spike up/down in some cell is obviously bad data and should be removed...

and as Simon said, good data (progressive steady throttle) will result in less data having to be filtered out, resulting in the big spikes not showing up anymore.

[macca_779]

Good answers...

Wesam,

What is physically possible in the real world...? The physical actual airflow/airmass can not suddenly step up/down... but it can have resonances and other effects, so the table will not be smooth, it will have troughs/ripples/dips/bumps...

if you try global smoothing and do another AutoVE iteration you will see the same troughs/ripples...

a localized big spike up/down in some cell is obviously bad data and should be removed...

and as Simon said, good data (progressive steady throttle) will yield good data, resulting in less data having to be filtered out, resulting in the big spikes not showing up anymore.

I agree. Infact I go to the lengths of locking the cells I know I have alot of data for so I don't skew them when smoothing.

[Doc]

Interesting question, here's my thirty cents...

It would be great to be able to do some airflow modeling on an EFI ic engine in a vaccum. I bet you would find a beautiful painting of airflow models that had seemless transitions that were repeatable 24/7/365.

Back to our regular scheduled programming in the real world that as much as the Green Facisits would love to regulate, (to some unknown perfect reality) thru higher taxation and regulation, the cold hard reality is that the world is in a constant state of change, uncontrollable by man.

For the most part, GM's stock calibrations for airflow modeling are pretty darn good and repeatable with stock parts and fuel trims.

Ok, if you are registered on this forum your'e not really interested in stock cars...

With all that said, knowing every aftermarket setup is a little different, the one thing I have noticed over time and many vehicles is that the more data you have the better chances of a solid trend materializing.

When I say "more data," I don't mean just a few more trips out with the laptop or BBL. I mean, a whole year's worth of data for a particular profile/setup of a vehicle. All the seasons, all the conditions, not sexy, very boring, logging.

What will this yeild? Well, if you are chasing the magical, mythical +/-lowest possible fuel trim value possible, that so many on the internet describe as easy as 1-2-3, you will be in tuning nirvina.

I routinely see "new, stock" vehicles with as much as 5,6, even 10% fuel trim values come thru the shop. Are they crap? Shit? No. Just baselines that work pretty darn good but, can be better and allow me to make better with EFI Live.

I have been around these forums for a while now and I have always found it entertaining to see people obsessing with what airflow models (MAF, MAP, Virtual VE) "should" look like.

I believe with all transients properly filtered, you should arrive at a somewhat reasonably, from a global view, "smoothness".

I know, that was a vauge answer...sorry.

1) Give the vehicle what it wants as determined by evalutating data from proper instrumentation.

2) Be ever mindful that you cannot ever, ever, fix a car with a tune.

3) Don't ignore #2

4) Should you obsess with smoothness?- NO! Give the car what it wants.

As you collect more filtered, corrected data in OL your airflow models will naturally smooth out.

Are some setups radically "different?" YES.

I think this was .32- I would love to hear from other "senior" members about this subject in an unadualterated way.

Doc

www.efialchemy.com

[WeathermanShawn]

Good topic..

I do some manually smoothing..problem I have with no smoothing is that you can get differing DYNCYLAIR values for different gears (at times). Especially street tuning. A M6 further adds to a spiked VE Table, as 5th and 6th gear can alter it also.

Wesam, another technique is to change the VE Table units to g/s or g/cyl. You will have to pick a uni-charge temperature. You will see that when you get it out of VE % units, the 'need' for smoothing is not quite as drastic.

I will probably get an argument (civil of course) going here, but I think it should look very similar to the exponential slope of the MAF Calibration Curve..especially if you use g/s as your units. Based on that theory, I manually smooth to show a constant rise in airflow with RPM & MAP.

The bottom line is that I have a hard time letting drastic VE table spikes exist..especially if it does not match my understanding of airflow.

Just a theory, but my O2..