I am going to take a slightly contrarian stance on this contrarian
article. ;)
The article does underestimate the capabilities of the current printer
models somewhat. The positional accuracy of the Huxleys printhead is in
the order of ten microns - but the practical accuracy is limited by the
size of the extruder, misalignment of the frame and the vagaries of
printing material. That said I have produced parts with dimensions
within three hundredths of a millimitere of the expected value. The
last 20mm ABS test cube I printed has dimensions of 20.06mm x 19.9mm.
This I find quite impressive considering that ABS in *not* known for its
dimensional stability. Feature size of the printed parts are
significantly larger than for a mill though, since the nozzle is 0.5mm
across and you cannot extrude *part* of a stream.
An advantage of additive methods over subtractive is also the ability to
do complex internal structures. In subtractive manufacture you
generally start off with a solid piece and finish up with a smaller
solid piece. The plastic printed parts we produce are hollow (10-20%
infill) with a reinforcing structure to increase stiffness. This allows
parts to be very light while reducing material costs. This can lead to
some interesting issues about internal stresses and failure modes, but
again for most desktop printing users I don't expect too many issues.
The material costs he is quoting seem to be for a full on Resin
stereolithography printer which is a very different (and far more
accurate) beast than a Makerbot or RepRap filament printer. Getting
good filament can be tricky, but will tend to last a long time.
The comments about speed are dead on though. Printing is *slow*.
In the end it is horses for courses. Pick your technology for your
task.
J.