Re: I Wish...

From: David Flynn (Plynetics Express  Corporation)
To: RP-ML
Date: Friday, January 20, 1995
Subject: Re: I Wish...
Rob Connelly writes...
>(1) I would hope that ALL of the RP vendors and users out there would start looking ...
     The way I see it, this problem is similar in nature to the line width
compensation problem. The bottom surface of any "flat down-facing feature",
of "FDF" in 3D parlance, is subject to a systematic error which should be
compensatable (did I just make up a new word?). These FDF features include
not only large horizontal flats, but any cross-sectional geometry which
results from triangles in the STL file which possess negative Z components
to their unit normals. (BTW, SLC won't make this any better) 
Energy delivered to the surface of the resin is attenuated as it passes into
the liquid. The bottom of the cured material is actually just the boundary
between super- and sub-critical exposure. The next layer drawn has a defined
"thickness" based on controlled exposure, but in fact, not all of the
incident energy is used to cure this layer - some of it is transmitted
deeper into the bath, through the previous layer (increasing its degree of
conversion) (Are you with me here? This would be easier to draw...) , and
finally arriving at the bottom surface of the FDF feature. This "print
through" exposure now effectively lowers the boundary between the super- and
sub-critical exposures, adding material to the bottom of the FDF feature. I
haven't crunched the numbers on this one, but I would expect this
print-through buildup effect (or perhaps build-down in this case) to
diminish rapidly, perhaps four or five layers into the part above the FDF
feature.
     If the laser power, Ec, and Dp are known accurately (and are truly
controllable), this is indeed a systematic error. It should be possible to
apply a vertical compensation analogous to LW comp. There will be special
case problems, of course, such as feature that is only two or three layers
high. But this is like the horizontal case of a wall which is thinner than
two LW's. It looks to me like the compensation should be applied during the
slicing process. What are your thoughts?
     BTW, we've observed that up-facing to up-facing vertical dimensions are
reliably +/-LT, which is exactly what you would expect from layer rounding
alone. This seems to indicate that shrinkage in the vertical axis and
z-stage error are negligible. What have you seen?

>(2) I wish 3D's field service would update their calibration techniques to keep ...
     I concur. This points up the need for everyone to determine process
parameters for their own machines, and use the published values for
reference only.

Dave Flynn