How much error is typically introduced into the VE tables when changing exhaust/intake? How does that introduced error play out with closed loop control? How much error in the VE tables can the ECM correct for? What problems arise if the VE tables are not recalibrated? How does adaptive learning factor in?

 

Depends on what you changed from to what you changed to. IME it usually isn't rich or lean across the board. Some areas are lean and some areas are rich. 18% either way is what sensors can safely take care of.


Also depends. By remapping your VE tables to what you have. Ensures that your short term and long term (adaptive) is zeroed out. This helps the other tables that are dependent on the VE tables.

 

Know the VE's are used in calculating fuel delivery. What other tables/uses are they used for? IOW's, what other tables are dependent on them?

It's often stated that if you change either the exhaust or the intake that a "tune" isn't required. Change out both, then it is. Just pondering that...

First thought might be that - well you have closed loop control... But pondering further...

So changing out either will likely introduce more error in the VE tables. Changing both will likely result in greater error....

Speculating...

Even with closed loop control, as you transition through the cells and a cell is being entered, it's error (now potentially larger) is initially going to be seen. If the transition are happening too rapidly - transients - you'd probably drop out of closed loop control momentarily. Whatever error was seen as you transitioned through the cell was what it was. If the transition is long enough you'd have time for some error correction provided by the feedback (or at least perhaps enough time to determine what the error was). If nothing else was going on, every time you transition though a cell, it's error would be fully seen initially. However, there is adaptive learning - long and short term trims. I presume they mitigate that.

Is my speculating roughly correct... Or am I missing the boat in respects? The answer to that will likely determine my next question...

 

Acceleration Enrichment table is based on VE tables. It is a multiplying table. Meaning the base pulse width gets figured by the ECM. If accel is active. This table gets multiplied to that to add fuel. Warmup Enrichment is another. This is compared to tables such as Cranking Fuel which has direct control over PW.


Things are often stated that I don't agree with. One or the other? What if the one thing was a 90 degree heavy breather, or just install straight pipes? Or maybe just install a stack and ditch the air cleaner? There are some crazy exhaust out there made for looks more than performance. IME it depends on how much a bolt on whatever changes the air flow through the engine. This also includes how the air flows and not just how much.


Since tuning is basically trying to narrow the tolerance of error, and centering trims around the combination at hand. I've never been a fan of letting it tune itself in.


IF you are saying. Steady state conditions need to be held long enough in a cell to get feed back control so the ECM can trim in the correct amount of fuel, and that transition through a cell doesn't mean it was enough time to go into closed lop mode. Correct.


In order for short term fuel trims to take place. Enough closed loop data must be seen by ECM. This is only going to change it so much. Say the amount of that change is over 9%. The integrator is going to zero and that change is going to be made in the long term or adaptive. Now the short term is zeroed and that correction in the long term is saved and will always have an effect on that cell no matter if the ECM is in closed loop mode or open loop mode. It is learned. No learning is done with just short term. Things are remembered with short term but not through a key cycle.

 

Understand/agreed...

Hmmm... Might be germane to my next question...

Yes... that's essentially what I was saying.

So say adaptive learning was turned off and the VE tables had some non trivial error in them. How would this manifest (symptoms) when running with closed loop control?

Ignoring other objectives of performance tuning... So theoretically, might it be a fair statement that over time adaptive learning will correct for a lot of the error introduced by changing pipes and intake?

 

Adaptive turned off equates to no remembered learning from key cycle to key cycle. if 18% change can be made with both ST and LTFT's. A system that shuts off adaptive. Short term is expanded to cover all of the 18%, but with no learning. Target Tune does this and the main reason why it is so important to start w/ a mapped VE calibration.


Over time? This is dependent on a very large CL area in the tune. This might work great on stock bikes with restrictive exhaust. Think better sampling. Doesn't always work the best with more free flowing exhaust. Also something to think about. Stock bikes don't get each bike VE's mapped. Injector tolerance in one thing to consider. The other is engine tolerance. Ask what stock 103 heads CC at. You get a tolerance for an answer. Degree a stock chain driven cam. Just getting TDC lift only shows a tolerance here as well. Point is. A stock bike will allow the ST/ LTFT's to trim everything in, but we are already not starting with a bike that have these set to zero to start with.


So, will it adapt? Depends on the starting point and where the finish point needs to be.

 

I'm aware...

Understand that if adaptive is turned off then short term trims are only in play. Long term (remembered) trims are not.

As far as I know, Target Tune can also run with adaptive learning. No? DynoJet's sight states:

If memory serves I asked this question to the folks over at Fuel Moto. I think I recall that the Target Tune enabled tune that they sent me had adaptive learning turned off. I asked about it, and I think the outcome was that you could run with it either way. I've had it turned on on my bike.

For the purposes of my questions, a large CL area was assumed. Essentially what you get with a stock setup.

That's one of my fundamental questions... Better sampling? Why would a more free flowing exhaust be different? You mentioned 18%. Did you intend that to mean that the system could compensate for up to 18% error? So more free flowing exhaust/intake could result in error that exceeds the systems ability to compensate?

Figured error correction from closed loop control would compensate for injector tolerance. Same vein though, the sensors are also going to have tolerances.

 

Something one needs to understand is that Short term fuel correction is only active at that point in time, forget key cycle to key cycle. So let's say your calibration at 2000 RPM is off by +20%. IF you go there and stay steady there, the short term will adjust that 20% out. The problem is the moment you change the throttle or RPM, it moves to the new position and continues to adjust. So lets say that new position is off -10%. The short term has to adjust down by 30% to get it correct and this takes time. So as long as you stay steady again at the new point it will get there. Now you keep running and go back to the first point at 2000 RPM and it has to adjust back the other direction from the -10% it's at, back to +20%, again it takes time. So it basically takes a good base calibration or it doesn't work worth a damn.

People are slowly learning this and there is nothing one can do other than get the base tune correct and once you do that, the Target tune or HD's new one isn't worth having!

 

Yes... But this is assumes there is no adaptive learning and saved long term trims... No?

 

Thanks for the correction, Steve.