04-05-2011, 11:32 AM
Join Date: Jul 2005
Total Cats: 68
VE table "plus" tuning: my suggestion for MS3
Instead of merely having a 3D VE table for fuel:
1) An entry for fuel pressure
2) An tickbox for manifold vs. atmo referenced FPR (because this changes the flow curve vs MAP)
3) An entry for injector flowrate and an entry for the FPR that was rated at
4) a 2D table for an RPM VE curve: i.e. VE vs. RPM - a '100' signifies 100% VE
5) a 2D table for MAP vs. VE curve: i.e. VE vs MAP - a '100' signifies 100% VE
Then, in order to calculate the fuel pulse width, the sofware will take:
1) the AFR table
2) the 3D VE table - a table filled with 100's will be the "nominal" values
The AFR table will be used as a multiplier for the fuel delivery.
With such a setup, VE vs. RPM curve will be shaped like the typical torque curve of a naturally aspirated motor. This is what we typically think of when we think "VE".
One would typically start tuning a new setup by keeping the 3D VE table and the 2D VE vs. MAP table flat, filled with 100's. Provided the injector and fuel pressure specs are entered correctly, overall fueling should be close. One would then do a few "pulls" at say, close to 100 kPa, and examine if the VE vs RPM table needs tweaking. And then do a few "MAP sweeps" at a fixed RPM such as 3000 or 4000 RPM, to tweak the VE vs MAP table. This should bring the whole map much closer.
After that, then the user fine tunes the 3D VE table to tweak the remaining few areas that need correction, such as the increase in VE associated with the turbo entering its sweet spot.
The software must modify the injector duty according to the AFR table. This way if the AFR target table says say, 10% richer at 200 kPa than 150 kPa, then the 3D VE table should still remain relatively flat with 100's entered. The 3D VE table should only reflect VE changes, not fuel demand changes due to changes in the AFR target.
When does one tweak the 3D VE table vs. the AFR target table?
If the AFR targets are being met with near zero WBO2 corrections or with O2 feedback off, then the VE table entries are correct. If one then wanted to change the AFR's, then the user would change the AFR target table.
If the AFR targets are not being hit, or the O2 FB loop is performing corrections to hit the targets, then the VE tables need to be tweaked.
Right now my AEM has a 2D VE table vs RPM, and a 2D VE table vs MAP. The VE table vs MAP compensates for the effect of a MAP-referenced FPR. I had to generate it using Excel. (this is where my suggestion for injector and FPR characteristics comes in)
My 3D VE table is near flat.
If I want to change a part of my AFR target table, I have to change BOTH the AFR target table, *and* the corresponding 2D or 3D VE table entries. It would be cool if only the AFR target table needed changing and the software would automatically adjust the injector pulse widths accordingly.
What's the advantage of this seeming complexity?
1) If the injector and FPR entries are correct, no need to make the VE vs. MAP table non-flat to compensate
2) With a VE vs RPM table, the 3D VE table doesn't need to compensate for this in every RPM column
3) With a VE vs MAP table, ditto
4) After tuning the above, with a flat 3D VE table filled with 100's, your fueling will be close. It's easier to initially tune 2 x 2D tables, than a 3D table where you would have to tweak all the columns, then all the rows.
5) If the AFR target table influences the injector on time, then changing the AFR target doesn't require any corresponding changes to any of the other tables.
After tuning, VE vs RPM and VE vs MAP trends would be visible.
How then would autotuning be done? My suggestion is that it only tunes the 3D VE table; and then a separate routine, off-line, would take the autotuned 3D table, then look for VE vs RPM and VE vs MAP trends, "flatten" the 3D VE table, and then apply the corrections to the 2D RPM and MAP VE tables.