Originally Posted by
RobCat030
Mapped points aren't telling the cams what to do, they're explaining what the cams are doing. There's infinite variability, so this is how it gets reduced into a tangible model.
You're correct about the vertical dependency of the calculator. It's calculating a series of regressions for each RPM point on the axis. So in other words, each RPM (1000, 2000, etc) gets its own y=ax^2 + bx +c. Y being MAP and X being cylinder airmass, which the calculator estimates to load. The horizontal relationship is somewhat meaningless in terms of the calculation in the sense that you're interpolating between two separate equations.
For me, the calculator wasn't particularly useful until I really understood how it works (both Ford's SD system and how the calculator is generating its values). The critical piece of information is that it's operating on the least-squares principle. In practice you're going to have data that falls outside the curve, if you get obsessive with it, you will chase your tail endlessly. 5% is a perfectly acceptable margin of error, and probably better than what many tuners here would send out of their shops.
Yes, after unlocking the cams you would ideally move through each angle pair and calculate each. Most people leave cylair anticipation disabled and don't touch SD, mainly to avoid the headache you're currently facing. Smoothing is somewhat counteractive to a least-squares approach, but so is using histograms with averaged data (footnote murfie, lol). Unfortunately to do it right there just isn't really a shortcut. You can monitor the active mapped points and only focus on recalculating the areas that are used, that's about it (ie. don't worry about MP0 at 7krpm and peak load if its not used there typically).
If you can attach a log it might help.