re: resolution needed? was Re: Timing Belts?
Posted by
ballendo@y...
on 2001-01-08 23:51:18 UTC
Tom wrote:
Drafting plotters (older, vector style) generally have a step size
of .025mm, which is about 1016 steps per inch. Or 40 steps per mm.
Look at some 3mm text from one of these if you can... (more
suggestions for visualising below)
There are more than a few commercail engraving machines with only
1600 steps per inch. A LOT of commercial engraving machines have
steps/inch of 2000 to 3200. This is .0005" (about .013mm)to .003125"
(about .008mm).
Using the coarser step/mm of .013, there will be 3mm/.013= 230
steps in each letter(full height and CAPS. The 'circular parts'
(lower case o, a, etc.) will be about 100 steps.
To see what this will mean:
1) get some graph paper (suggest engineering type (fine ruled)).
Using a compass, draw a circle of 100 unit diameter. Now using the
vertices(where the graph lines cross) of the paper, "re-draw" the
BEST circle you can using short lines between the vertices CLOSEST to
the compass drawn circle. You can go up/ down/ sideways/ and
diagonal.
Draw some slanted lines with a ruler, also (like the capital letter
A) . Try to "fit" these to the vertices as you did with the circle.
also.
This exercise will show you the potential "bumps" in the tool path.
Now that reality is staring you in the face; don't get worried, it
gets better! The toolpaths you have just drawn are only valid for a
REALLY FINE!! point tool. Any tool with a diameter to its' cutting
path will tend to 'smooth' these bumps...
The .75mm line width you mentioned will be about 59 'squares' of your
graph paper. Anyway, if you again use the compass, set at about 29.5
squares, at each point(vertex) in your toolpath, you will see the
REAL accuracy of your machine in a "magnified" way. Don't forget
the 'tangent' lines connecting the circles representing your tool
(parallel to the original 'toolpath'), if you want to see the
TRUE 'smoothness'.
2) Use a pixel program (like MS paint) to do the same thing as
described above. I prefer the graph paper, but it's fun (and eye-
opening) to look at the non-straight/perfect lines of "real-world"
CNC, no matter how you do it!
3) You can use a CAD program to see this also. If you scale the view
correctly, you can get a pretty good view of the REAL toolpath.(since
the SCREEN is pixel based, and of a given RESOLUTION)
For most people I've worked with, going through this exercise
releases them from the "resolution" trap... They see that a circle
can become an octogon, or even a square, but that their parts still
work. A few have gotten MORE concerned, but only a few...
Hope this helps.
Ballendo
P.S. I first used this technique many years ago to check
the 'accuracy' of various line and arc interpolation algorithms...
>The reason I am obsessing about fine resolution is because if I everTom,
>get a mill built it will be used for making jewelry; in other words,
>very small parts.i really want the ability to engrave very small
>text (~3 mm high,0.75 mm line thickness or less) and the ability to
>make very smooth surfaces for molds. I'm finding it hard to get an
>idea of where the cut off point is in terms of "useful resolution"
>if you know what I mean.
Drafting plotters (older, vector style) generally have a step size
of .025mm, which is about 1016 steps per inch. Or 40 steps per mm.
Look at some 3mm text from one of these if you can... (more
suggestions for visualising below)
There are more than a few commercail engraving machines with only
1600 steps per inch. A LOT of commercial engraving machines have
steps/inch of 2000 to 3200. This is .0005" (about .013mm)to .003125"
(about .008mm).
Using the coarser step/mm of .013, there will be 3mm/.013= 230
steps in each letter(full height and CAPS. The 'circular parts'
(lower case o, a, etc.) will be about 100 steps.
To see what this will mean:
1) get some graph paper (suggest engineering type (fine ruled)).
Using a compass, draw a circle of 100 unit diameter. Now using the
vertices(where the graph lines cross) of the paper, "re-draw" the
BEST circle you can using short lines between the vertices CLOSEST to
the compass drawn circle. You can go up/ down/ sideways/ and
diagonal.
Draw some slanted lines with a ruler, also (like the capital letter
A) . Try to "fit" these to the vertices as you did with the circle.
also.
This exercise will show you the potential "bumps" in the tool path.
Now that reality is staring you in the face; don't get worried, it
gets better! The toolpaths you have just drawn are only valid for a
REALLY FINE!! point tool. Any tool with a diameter to its' cutting
path will tend to 'smooth' these bumps...
The .75mm line width you mentioned will be about 59 'squares' of your
graph paper. Anyway, if you again use the compass, set at about 29.5
squares, at each point(vertex) in your toolpath, you will see the
REAL accuracy of your machine in a "magnified" way. Don't forget
the 'tangent' lines connecting the circles representing your tool
(parallel to the original 'toolpath'), if you want to see the
TRUE 'smoothness'.
2) Use a pixel program (like MS paint) to do the same thing as
described above. I prefer the graph paper, but it's fun (and eye-
opening) to look at the non-straight/perfect lines of "real-world"
CNC, no matter how you do it!
3) You can use a CAD program to see this also. If you scale the view
correctly, you can get a pretty good view of the REAL toolpath.(since
the SCREEN is pixel based, and of a given RESOLUTION)
For most people I've worked with, going through this exercise
releases them from the "resolution" trap... They see that a circle
can become an octogon, or even a square, but that their parts still
work. A few have gotten MORE concerned, but only a few...
Hope this helps.
Ballendo
P.S. I first used this technique many years ago to check
the 'accuracy' of various line and arc interpolation algorithms...