Interesting problem

My neighbor has an older Nissan with 150K miles on it. He and a friend
plan to rebuild it this summer. They located a turbo chrager and plan to
add this as part of the rebuild. As you can tell he is in love with his
car.
He asked me if I could fabricate the flanges for the exhaust header and
the turbo inlet and outlet flanges. He had new gaskets so I thought I
could scan the gaskets into my flatbed scanner and then do a raster to
vector conversion and then converting the dxf file to G code. So I
thought this was an interesting project and agreed to make the three
complex flanges out of 1/2" steel plate.
Scanning the gaskets in using a setting of 200 DPI proved to me more
than adequate. The first problem I ran into was that my Scad2Cad program
did like the file types and so I tried several other raster to vector
programs. Later I found out that if I had used a .tif file type it would
have worked. I then used Tracedit ( shareware as I recall) and did a
centerline raster to vector conversion. It came out fine with just a few
small holes and only need a slight touchup. Next it was loaded into
bobcad 17 where I had to make other smoothing operations like replacing
the slightly out of round scanned holes with true circles. I also had to
create an off set for cutter comp. It is here that I found that using
the chaining command to select the outline of the complex manifold
flange had some gaps. My neighbor also wanted me to make the outside of
the flange about 1/16 wider all the way around and to change a few other
contours. This was easily accomplished and after a while the offset
command in bobcad drew the shape he wanted.
Then using bobcad I used the CAM module to create the G code.
Off to my 8X38 knee mill equipped with my servo system I started to
machine the exhaust manifold flange. It bored the 7 bolt holes and the
exhaust ports with no problem. As it was starting to machine the first
cut of the outline I noticed that the cutter was starting to make a
small circular cut where there was not suppose to be one. I paused the
machine and looked at the G code. It looked fine but I noticed that in
Mach1 it was showing several other circles that weren't on my dxf file.
Ah-Ha! I said, a bug in Mach1. To prove it was bug I loaded the same G
code file into another program that displays the tool path. Sure enough
the same unwanted circles showed. I couldn't figure out why I should be
getting these circles as the dxf file was perfect. It was also too early
in the year for mysterious crop circles to appear in my G code.
Knowing that there wasn't a bug in Mach1 I returned to bobcad. I
reloaded the G code and in the CAM program Selected All. Then I selected
Geometry from code.
To my shock the mysterious circles appeared on my drawing.
I zoomed in on one such circle and deleted the lines in the area and
then did a View Refresh. I noticed that there were two tiny specs
remaining after I deleted the lines in the area. Last I undeleted the
lines and then using select entitity located one of the two specs and
deleted it. I had to delete one of the lines in order to select a spec
under the line then redrew the line. I went back and created new code
using the Auto Machine mode in bobcam and and then slected all, then
selected geometry from G code and sure enough the offending circle
disappeared. I went through and had to correct about 6 more unwanted
circles all have small lines causing the problem.
After correcting the drawing and making a new G code file it worked
fine.
So the lesson I learnt was that using a scanner and using a raster to
vector software and using bobcad17 one should also review the G code
before machining the part either using the path mode in Mach1 or your
favorite backplotter.

Dan