NefMoto

Still trying to learn as much as I can, forgive me for any dumb questions...

Why is KFMIRL set for such a higher specd load than LDRXN(in both stock form, and most tunes I've seen)? In stock form LDRXN's max point is 145, whereas KFMIRL is 191.

I understand you would want KFMIRL higher so you could control max load with ldrxn, but why so much higher?  Why not just a point or two?

As far as I understand it, KFMIRL limits aren't reached with the stock ecu.  However, if you factor in KFTARX and other "boost adders" to the stock tables, a much larger load can be observed.  I don't know exact details on this, but I do know this.  If you have K04's, keep a stock KFMIRL, and put on a RS4-like LDRXN with the maximum value scaled to 191, your car will drive amazingly.

I played with all of the new "torque maps" (KFMIOP), adjusted "boost maps" (KFMIRL), and always had KFTARX set to 1.0 across the board.  The car's velvety smooth with what I described above, like a tuned K03 car with a much bigger pull that ramps up fully around 3000 rpm.

Kfirml is a requested where ldrxn is the limit. From what I understand and as stated above it allows for other factors which may take the actual upto near the requested limit. If its set 1 or2 points above you will likey then go above requested with the actual and go into limp mode. Setting it above gives a degree of margin. I may be completely wrong here of course :-)

So using nehalem's experience, those of us using k03s still(am I the only one :P), a flat KFTARX, and a LDRXN that doesn't go above 191 it should be better if we left KFMIRL and IOP stock.  This is the reason I brought this up, since I was reading all about interpolating your IRL table to generate IOP, I thought why not leave both stock if I don't ever go over 191.

I'm also a big fan of a flat KFTARX, if not one that slopes down a bit more aggressively for really high IATs if running 91oct.

IMO nehalem is on the right track here.

Just to illuminate my points a little further, I've created some graphs in MATLAB to demonstrate what I was saying.

There are three steps to doing this properly.
1)  Compare the S4 and RS4 tables complete with their respective RPM axis values.
2)  Create a RS4 shape-interpolating spline model and recalculate the theoretical S4 table using the S4 RPM axis and interpolated RS4 load values.
3)  Scale the theoretical RS4 load values at the point its clear the turbos are in effect.  In other words, the spike in load needs to be scaled to 191.

This is definitely not the only way to do this, but I'm confident in the results based on how my car feels.  Its nice to also have a confidence in the theoretical side of things above and beyond the results feeling great.

If you have any more questions please let me know, but I'm not sure how much I can help for non K03 / K04 turbos.


As for modifying KFMIRL and KFMIOP, I think the only correct way to do it as this point is to use the RS4 maps.  I've tried all of the new spreadsheets and if you compare the values suggested by them, they generally don't seem consistent - and furthermore they don't seem to reproduce even the stock tables given stock input.  Thats a baseline necessity for anything I intend to use for the long-term anyways.

Anyways, here's my recommendation for a stock axis s4 with K04 turbos:
110.02   110.02   110.53   112.86   114.38   122.25   136.88   180.41   177.76   175.12   174.30   176.75   181.43   188.15   191.00   187.42

I just tried this on my car and attached is a log of the run.  My car has K04's, RS4 MAF and RS4 injectors (APR Stage 3 hardware).  I've come up with a correct KRKTE value, used the stock RS4 MLHFM and KFKHFM is set to all 1's.  KFTARX is also all 1's.  Does this look correct to you?

Can you help me understand where you derived those values from?

That definitely doesn't look right. You should pull up LDRLMX and log the available variables.

This still doesn't explain why KFMIRL is so high just to be capped by LDRXN. My guess is the curve was developed on the dyno and later capped by LDRXN.

In the function LDRLMX, max load is controlled based on many different things. It always follows the minimum path and finally LDRXN if it is lower.

I had a feeling it didn't seem right.  Am I missing something because I cant find LDRLMX?

Can you be specific about which variables to log?  I am still in the dark ages using ECUx and haven't got around to learning ME7Logger.  I've got it all setup and ready to go, I've just got no idea which values to enable in the config file.

FWIW, I used a completely stock M-box file and only modified the above mentioned items as well as upping the rev limit and removing the rear O2's.

Hmm, what about your wastegate duty cycles?
One tuning aspect that still humbles me is the boost PID, but that's what I immediately think of when looking at your logs.
It looks to me like the boost is tapering off too much, which I think might be the wastegate being too active.
(Nyet or phila_dot please confirm...)    My thought here is that we also need to look at KFLDRL (boost linearization map).

You will need to use ME7L. LDRLMX is a section in the FR. You will need to find the key variables from that section in your ME7L config file. I can post what I use next time I get on my laptop.


That is also possible. It would be easier to confirm with full logs (MarkP template).

You mean ECUxPlot

FYI - I put both KFMIRL and KFMIOP back to stock, and adjusted my LDRXN curve to peak at about 189...runs nice and smooth, I also notice less jaggedness in the logs at the top end.

PS: I'm running K03s still so YMMV