This one puzzles me. Hopefully someone here can shed light on the
situation.
I’m designing two interlocking parts with very close tolerance on
Rhino. One piece is a little bit thicker than the other.
On my computer they fit together perfectly with a little (although
very little) wiggle room. After I have them grown in extreme
resolution perfactory and cast, I get them back in both resin and
tumbled silver.
Here’s the question: They don’t fit together like they do on my
computer screen. I would expect that if there is any shrinkage it
would be during casting, but that’s not the case. Both resin and
silver fit the same.
Now in my years of casting experience I’ve found that the heavier
the piece, or section, the greater the shrinkage. Just the opposite
is true in this case. The heavier piece is larger, and will not
interlock with the other. It’s not hard to file it down a bit and get
it to work, and it’s easy to tweak my Rhino file, which I’m doing.
It just baffles me, and I wonder if some of the Orchid brains can
explain this one.
A friend has just completed some puzzle pieces and the magic number
seems to be a tenth of a millimeter clearance allowance and the
pieces fit together perfectly.
In general the tolerances you are talking about are probably too
fine for commercial prototyping processes. Since you are presumably
going to have to do some finishing anyway it would seem preferable to
have an imperfect fit up front so you can finish to it rather than
lose a perfect fit in trying to put a finished surface on the
material.
I know of one company in my city that produces (or used to) high
precision rifle parts via SolidScape rapid prototyping. Even though
they use precision CAD modeling and the highest resolution in the
prototyping they still machine the castings to tolerance.
Ultimately, however, I would urge you to contact your service bureau
and discuss the issue with them. If anybody’s going know they will.
I sent your question to Steven Adler, the guy with the most
knowledge about Perfactory machines (and growing models in general) I
know of. Here is his reply:
This does not surprise me... I had a Perfactory for 3 years and
tried to use it for dimensionally sensitive applications.
Endoscopic Jaws, telecom connectors, and even an aerospace
application.
The output of this process is accurate enough for jewelry "one
of a kind" patterns but, I was never able to use it for stone in
place, or master model production which is one of the reasons I
sold it....
The "high resolution mode" on Perfactory will in fact make a
very smooth model but, not necessarily a highly accurate
model.....add to this issues with both hygroscopic and thermal
expansion, you might want to try another technology like Sanders
or Solidscape where dimensional stability in more predictable for
direct casting.
BTW, Many people who use Perfactory will make a silicone mold
and then cast the subsequent injection wax. We all know what that
variable might do to dimensions.....might want to speak with the
vendor and find out exactly how he uses this technology
Steven Adler
A3DM
www.a3dm.com
I think that what you are experiencing is about the "stair stepping"
that extends outside the virtual model to ensure that the model is
printed in entirety. This excess needs to be cleaned off of the
plastic modelif interlocking parts are to fit. It is not shrinkage or
swelling.
I think that what you are experiencing is about the "stair
stepping" that extends outside the virtual model to ensure that the
model is printed in entirety. This excess needs to be cleaned off
of the plastic modelif interlocking parts are to fit. It is not
shrinkage or swelling.
No offense Bruce but, I think you are subscribing to an urban myth
about RP technology. While there is a stair step effect on all 3D
printing machines, the process of slicing data does not extend
beyond the CAD design. There is no dithering or interpolation in this
process.
No offense Bruce but, I think you are subscribing to an urban myth
about RP technology. While there is a stair step effect on all 3D
printing machines, the process of slicing data does not extend
beyond the CAD design. There is no dithering or interpolation in
this process.
Dunno about an urban myth, it has just been my experience. There is
an error between the virtual model and the physical model. The
virtual model has a much high degree of precision than the physical
model and will err. The engineers have opted to ensure that
finishing does not cut into the desired model. I don’t know if this
is the error that the originator of the post is addressing, but the
“stair stepping” is not required material, except in the production
of the physical model in that it is the minimum of excess material in
any given system. Finish off the “stair stepping” and the parts may
fit.
I have to agree with Bruce about the urban myth thing. What you see
on the screen is often at 6 decimal points, what I hold in my hand is
maybe accurate to 2 decimal points on a really good day. I mill stuff
rather than build but I think the rules are close to the same.
Rule ONE - don’t cut into the model EVER !!!
The engineers and software developer are doing it right when they
protect the model at all costs. Otherwise I would and have screamed
bloody murder.
There are a lot of ways to screw up, from idiot keyboard operators
to program bugs (but I like playing on the edge with the latest beta
version of MeshCam :-). But while I can live with funny looking
cutting paths, extra material, or needlessly long run times I want a
full model with no gouges. No undo key, when it’s gone it’s gone. A
file or burr or graver can fix an excess with ease but there isn’t a
laser in my basement
Jeff
PS: Standard stock disclaimer… While I am a moderator on the
MeshCam forum I have no financial connections. A bit of voluntary
work for a very good tool which fits most of my needs is a small
price to pay. Excellent and quick support.
