3d scanning of fan parts

[David Yurko]

Using a 3d scanner to try and help with the collar gear reproductions. Its going rather well, having to learn as I go, hopefully this will prove helpful and will lead to more parts being available.

[Louis Luu]

Looks real nice.

[Louis Luu]

When you get a chance...I need to verify my measurements/positions for the nose cone.

[David Yurko]

Im still scanning the first gear, but will let you know when i get that scanned, it takes a while for the detailed parts, also learning the machine 

[David Yurko]

I'm learning a lot,  what normals and point meshes are,  and the limitations and how the 3d scanner works,    3d scanning is kinda the same increase in complexity from 2d printing to 3d printing.   The scanner uses two lasers and a sensor in the middle to grab a series of distances from where the laser contacts the surface,   Parts of the piece can shade each other,  or things like gear teeth can reflect the laser off of the surface of one tooth, against the other tooth and cause 'Noise' that cause erroneous points,

You then have to take multiple scans that can take hours and put them into a program to assemble them into the part, 

You have to crop the parts you use to hold the original part then you have to smooth out the "normals"  normals are vectors that show which side of the point is up and down. and Meshlab uses these to help render the point field,  and to help align the parts.

Once you align the different scans,  you start erasing parts of them that aren't accurate,  these will make the model inaccurate.

Then you merge the layers into a merged layer and keep doing this until you have the part rebuilt as a point mesh.

  I found a holder was essential to position the parts in all sorts of angles to make sure all areas are covered,     you want the part to to be as perpendicular with the laser path to give the best detail,   IE you want the surface you are scanning to be a right angle between the laser and the surface,    if the surface is too close to parallel with the laser,  the results won't be good.

Another tip is that you can manually set the rotation steps on my scanner,   setting this to the minimum value produces a much more accurate scan,  also moving the part closer to the edge of the bed and just doing a partial scan also produces a better result.   You can also set the number to the minimum and make a wire frame of the part, to check to see if it will scan the way you are expecting it to.

[Louis Luu]

Thank you for your effort.  I will not be able to use those data...too many fixings to do.  I have the nose cone done.  I did it the old fashion way...lots of measurements and lots of double checking.

 

[David Yurko]

Yes,  that is far from the final product,   I would probably use this to get the profile and rebuild in Cad,   I wouldn't have to get so many details doing it in that method,   This would be better for things that have a hand carved or very complex curves

Though I am still learning doing this box,  and I plan to complete it,  just for my own learning sake.