Every step in building leads to something, that is fun to do, but not strictly required. I have been sanding wing leading edges by hand and eye for as long as I remember, and I must say, never an issue. But, since the CNC-Router is working again, the 3D printer is working too, it’s miserable weather outside, and Claude.ai has nothing to do…
Yes, it all leads to using the braincel again! How about this: I try to copy the wing-profile on a piece of graph-paper. Take a picture of said paper. use some image processing software to extract the shape. Then make a sanding block that matches the profile, and I will have the most beautiful nose!
So far, that’s the theory, and indeed, at this point I have created a proper Bellanca.dat file! I will try to document the steps how I got there, so next time I will be able to do it all again.
First step: make an drawing of the wing-root on graph paper. Do it in pencil and trace later with a good black marker. It is easier for the software to find the curve if it is well defined. I mostly used Gimp for this, I will need to retrace/reinvent the steps, because it was not that easy to get a decent image. (Most of these programs I use a few times a year, and never get really good at using them, it’s a lot of trial and error, and I am sure there are better ways to do it.) The goal was to get that shape, by any means possible.
Further down the line I wanted to use Profili and DevCad to do some of the work required. Profili to work on the Airfoil, DevCad to do the CNC stuff. Even though the image looks good, Profili did not want to know. Never mind, there are other ways to get the coordinates of the profile. Enter Claude.ai. Hey Claude, need some help here. Here is an image of an airfoil, I need that converted to a .dat file. (yes, sounds easy, but I skipped 6 hours tinkering here 😉 Anyways, code is here for the interested souls. Not clean yet, work in progress and since it worked once as needed, I might never expand it. Yes, it is the worst of programming. Welcome to the new AI-world, where me make code for single use. The days or large dedicated programs that do everything under all circumstances are gone. (Not my sentiment, but many a good programmer says so.) 
This is what the code contour detection produces. Looks pretty neat.
Now that I have a .dat file, I can play with Profili and XFoil. I’ll skip that part for now, since the results were exactly as I expected. The profile is not that clean, and especially the nose section, which is sort of like 100% the most important, is not that great. It caused jumps in the plots. (as in: pretty lousy performance.)
So, what I have now is roughly the basics in place, and now it is time to do some refinements. I’ll start with the original image, which is on raster paper. Camera’s by their nature distort images, but there are methods to sort that out. That will be the next step. Then due to distortion of the image, the profile thickness comes out 3% less then it should be. Then we most certainly need to employ some software tricks to get a well defined properly calculated nose radius.
In the meantime, I’ll cut some templates and start the rough work by hand. 😉