Can someone give me the logic behind this function. I know .9 provides very little smoothing and a .1 will be very aggressive at smoothing.
What mathematical function in happening when you use this feature?
How do you implement it? Specifically with VE tables. Tried it this weekend, and ended up with a very smooth curve, but after flashing, VE learns wanted to make some significant adjustments. So, I reverted back to my previous tune and used less aggressive smoothing. Still have a bit of a jagged curve.

While we are discussing VE tables, from VE learn to VE learn what is the expected noise? +/-1, +/-2% or is greater OK?
I had done a series of learns and all was well from drive to drive I never saw VE corrections above 2%, most were 0's or 1's. When starting to data log and work on WOT, I noticed 100%+ DC on the injectors. Bumped base FP up to 50PSI, and started a new set of VE learns. I am at the point where all seems happy again, but have a bit of a rough VE: the ECM seems to be happy about it.

 

I guess I wouldn't get too wrapped around the axle on what the table looks like if the engine is happy with it.\

Does the $6E have separate WOT Power Enrichment tables like $8D? If so, and if the VE's are good, I'd probably start working on the PE tables.

BTW, if you're hitting 100% duty cycle on the injectors, I'd probably start looking at getting bigger injectors. I'd probably try to stay at 85% or below.

 

Yes $6e does use a separate pe table percent change.
.4 works ok as a smooth value

 

Remember - what is *visually* smooth to you, may not be what the engine wants. Give the engine what it wants. And usually if the fuel map looks like marble madness then there's issues with the scalars and setup (such as injector latency, engine displacement, etc) that are creating problems with the math on the back end.

GD

 

I’ve had tunes with spikes that work better than smooth

 

I am not, spent the weekend enjoying tops out motoring

I was seeing a few ticks above 100% on 22lbs@43.5psi Injectors. I am going to install 24lbs@43.5psi and run them at 50psi. I don't have my calculations with me, but I am sure that is enough for where I want to do.

By the way, this is on a Supercharged car and it never shows high duty cycle except under boost. Vortech Kit, 6PSI. Nothing special.

 

Not Marble madness, but bumpy. No Major spikes or depressions.

So what about VE noise? In the main operating areas of the table, from drive to drive I will see corrections of 1-2%, both +/- and if I save and flash, I just get it again in the opposite direction. I am inclined to think this is OK and normal, but would like a second opinion.

 

I have been curious about this function for a long time. Never used it however because I didn’t understand the relationship of factor to how much change is made. I just finished doing some manual smoothing on my tables by plotting where the VE Learn affected the table and bringing segments of the VE table into alignment with the general direction the learn was pointing to. For the most oart I figured either spikes or depressions in the VE table was due to that cell being “missed” by Learn, so it remains the same while cells surrounding it may be adjusted up or down.

 

I had been doing some manual blending as well. A friend suggested I tried using this function, but I hate to do something like that when not understanding what it does.

 

That's very typical. Once you get within a few percent that's just what it's going to do.

For SD tuning - anything +/- 5% is plenty close. I usually aim for my maps to be subtracting some fuel rather than on the lean side though.

GD

 

Same here. I shoot for my BLMs to be a little >128 that way the computer is only adding fuel and not subtracting.

 

ULT,

I look to be ever so slightly on the lean side as well. I find that the exhaust note is mellower and tighter. On the rich side it tends to be blubbery especially on inclines. And I agree, when you start to “seesaw” from rich to lean and back again between Learns, you’re probably about as close as you’re going to get. In the last several learns, I have taken to modifying the increase/decrease in the new VE value and cutting the delta by 50%. That seems to be helping w the overshooting of the correction.

 

FWIW. The TunerPro smoothing algorithm appears to "normalize" the selected items, bringing them closer to a mean or median value. I generally use 0.9 and execute it 5 times. This will provide gradual smoothing. Try it. But as TunedP said, sometimes having slight spikes is OK because that's what the engine is reporting. I generally don't like large spikes and will smooth them and the surrounding areas and repeat logging.

Regarding VE and BLMs and acceptable noise: Generally a BLM +/- 4% is considered optimal. Now, why not 128 with no variation = optimal? It's because the PW calculation and eventual BLMs are extremely sensitive to temperature (coolant always, and depending upon the bin mask, MAT possibly). $8d code simulations verify that PW, which results in INT and BLM, varies with very slight coolant/MAT changes.

DC%: Don't know how you are determining DC% with $6e, but if you're using the ADX-reported $6e PW for that calculation, don't believe it's the PW the ECM and the injectors are seeing. All OBDI code hacks and ADX files I've examined report PW BEFORE injector voltage compensation. So to get proper DC%, need to add the PW from the injector voltage compensation table to the ADX-reported PW to get the true PW the injectors are seeing. S_AUJP v6 now reports "Final PW" which is the o2 sensor adjusted PW + injector voltage PW correction. This ensures correct DC% calculation.

HTH, Elky

 

Yes helped a lot. I am using EBL with their XDF file and mask. I am not sure which TBI base they build off of. So, how the final DC% is calculated, I am not sure. I am using their DC% output in my log.

To speak to the sensitivity, when I first installed the system, I left my intake tube together (including MAF) until it was up and running well. As an engineer, I try my best to avoid tossing multiple variables into an equation if I don't have to. Recently, I ordered silicon tube and elbows and made a custom intake to get the MAF out of the tube and pretty things up a bit. In doing this, I changed my recirculation valve to vent to atmosphere.
No you would think this wouldn't change much. However, on the first VE learn with the new tube I am seeing corrections of 1s, 2, 3s and 4s all over the place on my VE table. Got to comparing an old log and I saw why, apparently at low speed the heat soak of the previous system continually recirculating air had a nominal 20*F increase in IAC temp. This started to disappear in the higher RPM parts of the curve as I don't believe there was as much heat soak with higher CFM consumption. So, I can certainly see the sensitivity.

 

The PW shown is the PW that is programmed into the hardware, it includes the compensation values. As such the DC% calculation also includes them.

RBob.

 

I love using the 3D map table and manually adjusting. With enough known data from a good log you can adjust those specific cells to where they need go be and manually bring everything around it up closer to it to smooth it out. Rarely need to use the smooth function. For me i find most ve tables are relatively smooth looking when everything is tuned right. Even on my holley with wideband correction, there is little correction across the board once everything is dialed in, and that ve table looks smooth with no real spikes anywhere

however there can be times when induction and cam start working together at a certain load and rpm that starts really filling the cylinder and need more fuel so you get more of a jump in ve but i dont see too many spikes.

its funny cuz i have seen some cars seem to run well with very rugged looking ve table. I just cant help but think it be better if smooth

 

From a mix of smoothing function, manual blending on the graphs, and come calculations between cells and this is where I am:
88 L98, stock heads, cam, intake with headers, 1.6 Rockers, blower venting to ATM, and the all important air foil. I am pretty pleased at this point with how the curves have came together. Most drives I get less than +/-2% so I am not going to re-flash. Soon I will start the process over with larger injectors and AS&M runners.


Current Low Speed

Current High Speed

Low speed differences as compared to 3005.bin

High speed differences as compared to 3005.bin