PowerFC cell interpolation

[Wildcard]

This may be a 'best guess' question but does the PowerFC interpolate between map cells [do other comparible ECUs?]? I've been trying to monitor this via the hand controller but my tests have been inconclusive.

For instance if you have a 35 in one cell, and 39 in the next would you expect to see a transitioning value of 37?

I ask because I believe if not when tuning for such an ECU you would [generally] want to be at the upper limits of a cell before setting fueling and ignition timing approprately - is that correct? My thinking, for instance is if a cell has an 4500-5000rpm range you would [generally again] want to set your tuning near 4900rpm to accomodate for the 'worst case scenario' which can occur in that cell?

Thoughts?

[fivebob]

Assuming that they work the same as most ECUs then it would involve some form of fuzzy logic, utilising the values of the four surrounding cells.

To make sure that adjacent cells don't have an effect, when the X & Y values are close to an actual cell, there will be some point at which that cells value is used as the sole point of reference. Typically that would be around 10% of the difference between adjacent cells. Outside that area the calculations would involve determining the weighting to be applied to the various cell values.

e.g. If you had RPM columns at 4500 & 5000 rpm and MAP rows at 150 & 180kpa, then the four cells involved would be (4500,150), (4500,180), (5000,150), & (5000,180).

If the engine was at 4750rpm and 165kpa then all cells would be weighted 25% each.

At 4750rpm and 150kpa then the (4500,150) and (5000,150) cells would receive 50% weighting each.

At 5000rpm and 165kpa then the (5000,150) and (5000,180) cells would receive 50% weighting each.

At 4550rpm & 153kpa the (4500,150) cell value would be used.

And for a more complicated point, like say 4800rpm & 160kpa ,the (5000,150) cell would get the greatest weighting, then the (4500,150) & (5000,180) cells, with the (4500,180) cell having the lowest weighting.

Hope that isn't too confusing, and bear in mind that it's just a best guess based on how the Motec does it... BTW the next release of the Motec software will have 4D tables (e.g. RPM, MAP & TPS dependent) so the algorithms will be even more complicated then

[Akane]

If the PFC works like the S-AFC (and I don't see why not since they're quite similar in software layout), then yes, you'll see a transition from one cell to another. But how great the interpoloatoin is very questionable. IIRC it's about 100rpm steps.

[fivebob]

If the PFC works like the S-AFC (and I don't see why not since they're quite similar in software layout), then yes, you'll see a transition from one cell to another. But how great the interpoloatoin is very questionable. IIRC it's about 100rpm steps.

I doubt the PFC has such crude interpolation, because a) it's a 3D map, not 2D like the SAFC, and b) the RPM/MAP values are not fixed like the SAFC. They may seem fixed when using the Commander, but with the right software you can change them

Given these facts, and some understanding of software alogorithms, I think it would be harder to implement fixed RPM or MAP steps. Therefore the interpolation is likely to be more fluid than that and would probably use a weighting algorithm.

If I had some datalogit logs on this computer I'd be able to determine the exact algorithmn, but alas they're back in NZ on another computer