Purpose of smoothing

[99FormulaWS6]

I had this discussion with another tuner friend of mine, why do people smooth? yeah, it makes the table look nice and neet, but you are taking data directly from a logger and then skewing your data. It's like paying a 1.57 for something and then the cashier rounds the price up to an even 5.00 so you don't have change in your pocket? Am I seeing this wrong?

[02sierraz71]

I had this discussion with another tuner friend of mine, why do people smooth? yeah, it makes the table look nice and neet, but you are taking data directly from a logger and then skewing your data. It's like paying a 1.57 for something and then the cashier rounds the price up to an even 5.00 so you don't have change in your pocket? Am I seeing this wrong?

yes, I personnaly rarely rely on polynomial smoothing and if I do I make sure that 5 decimals are selected and the table is allready very close.

The reason to have a smoothe ve or maf graph is because it is a representation of how air flows throught the engine and engines dont act sporadically they behave in a general more linear fashion. When you mash on the gas you dont immediately get a huge amount of air in the cylinders it occurs gradually over time. This is the reason the ve and maf tables thought the maf hz relates a huge immediate air increase that isnt actually what is going on in the cylinders. Smooth gradual graphs are a good thing and make for better et's and mileage. Im sure jimmy blue has some good philosophies on this hopefully he will lay some tidbits on us.

[LosLS2]

I can say that I was scanning and tuning and scanning and tuning and I kept getting readings that were off. Finally, I went on Horist's site and downloaded some other peoples tunes and started looking at them in the 3d view. I found that "most" had a more wavy look than the peaks and valleys that I was seeing. I smoothed things out and now my scans are more consistant and I have been able to get the car running alot smoother. As stated above, there is a difference between the poly smoothing and hand smoothing. The computer does not know that one cell is more accurate than its neighbor and therefore will try to achieve a balance even if it leans heavier towards the incorrect cells.

[jimmyblue]

Poly smoothing has no intelligence. It only
tries to make things, well, "smooth". But it
will as happily mess around good, fitted
cells to blend them against a freak from
bad data. You have a bit more understanding
of which cells are to be believed, and which
ones might want tamped down.

I use the poly smoothing as the "finish coat"
but all of the shaping wants craftsmanship
that automation can't provide.

The motor is in general a pretty continuous,
not-lumpy airflow machine and any graph
that looks like you wouldn't want to run a
tire over it, probably has something to tell
you (like, don't leave nails on the driveway
even if you arrange them symmetrically).

[EC_Tune]

Although there is nothing *truly* linear about an engine, if your VE table has a mountain next to a mole hole there's something wrong.
Following the MAP line, if your VE is steadily increasing with manifold pressure or is nearly flat and not dropping as the MAP rises, you're OK. If you have wierd peaks in the VE table as MAP rises, it's probably bogus data.
There is a ridge that shows up on most N/A LSx engines at about 2800 RPM from low MAP to mid MAP that dips down at 3200 RPM and then rises rapidly from 3600 to 4000 RPM (toward peak torque). This is a normal occurrance.

EC

[PurplePiss]

I remember back questioning SE's VE table because he had a spot in his that looked like clevage. I don't remember anyone able to explain why it was there. Since he's in SD, it can't be corrected with a hump the other way in the MAF curve right?

[ninobrn99]

i think the cleavage look is about the ve's matching the torque curve... or something like that. thats what i read sonewhere on one of these forums.

[Scooter70]

What if we could get a feature for smoothing whereby we could lock certain cells and smooth around them? As Jimmy eluded to, some cells are believable and you don't want them touched but smooth will adjust the entire table or selected area. If we could select some cells ahead of time to lock them and then apply a smooth around them, I would think that smoothing would be more accurate.

[barkingspud]

What if we could get a feature for smoothing whereby we could lock certain cells and smooth around them? As Jimmy eluded to, some cells are believable and you don't want them touched but smooth will adjust the entire table or selected area. If we could select some cells ahead of time to lock them and then apply a smooth around them, I would think that smoothing would be more accurate.

You can.....Just select the cells you want to smooth, switch to 3D (or just stay in the table view if you want to) view and click on smooth selection....

[ninobrn99]

or highlight all cells and click the ones you dont want smoothed while holding the cntrl key.

[Scooter70]

Cool. I just learned something new. Thanks guys.