Discussion:
Anelok: ABS meets acetone, part 1
Werner Almesberger
2017-02-02 01:07:26 UTC
Permalink
I did a first acetone (*) experiment on the Anelok case. For this,
I printed a version of the case "sideways", i.e., not laying flat
on the top surface but standing on its side. Printing it this way
takes longer but produces smoother curves, since - in this case -
they then all lie in the XY plane of the printer.

(*) That's post-truth acetone, i.e., the acetone-free nail polish
remover that still preserves some properties of acetone. I
don't know how real acetone would perform in comparison.


1) The print
------------

For slicing, I used a recent version of Cura (from git). The material
printed is ABS. The result looks like this:

Loading Image...

Print direction (bottom to top layer) was from right to left. One
can clearly see the print layers running in the Y direction. The
size of this model was about 58 x 40 mm.

The strange banded structure in the middle is support, to keep the
overhang (i.e., left side) from sagging. I should point out that my
printer has a 0.4 mm nozzle but Cura only lets me choose 0.3 mm.
This may cause some anomalies that wouldn't exist when using the
correct nozzle size.

There is the bottom side, nicely showing the path of the filament:

Loading Image...

There are a number of printing parameters one can tweak, so it's
likely that I could obtain smoother results. But, one experiment at
a time.

SolveSpace approximates curves with polygons and currently doesn't
like it if one chooses too good an approximation, so the rounded
parts of the Anelok case aren't as round as one may expect.


2) Cleaning up
--------------

After removing the support material, we get this:

Loading Image...

The shape is pretty good, but there are lots of grooves. This is the
inside:

Loading Image...


3) Sanding
----------

I sanded the case with 180 grit sandpaper. That got rid of some of the
grooves but I would need to sand off more (with coarser paper) to also
get at the deeper ones:

Loading Image...
Loading Image...

What looks like a little bulge near a deep groove at the bottom was a
drop of water.

A round of 250 grit paper made already smooth areas even smoother, making
the deeper grooves stand out more clearly:

Loading Image...


4) Acetone vapour
-----------------

There are two methods for smoothing ABS parts with acetone vapour: hot
and cold. In both cases, the ABS part and acetone are placed in an
air-tight container. Then acetone vapour dissolves and smoothens the
surface of the ABS part.

With the hot method, the container is heated, to accelerate the
vaporization process. The cold method uses a paper towel soaked with
acetone and patience. Keeping things simple, I chose the cold method.
I also added a cap full of acetone to act as buffer.

Loading Image...

Three hours later, at an ambient temperature of about 25 C:

Loading Image...

The surface is partially dissolved and thus sticky, so it's important
to let the part "dry", and to avoid touching it until the acetone has
evaporated.


5) The result
-------------

The top side looks pretty smooth but also a little blurry:

Loading Image...

Most of the grooves are gone and the one that were too deep to
completely disappear stick out clearly. Also some larger "wave"
patterns begin to appear. I would expect them to be less visible
when printing in black.

The effect on the sides was much less pronounced:

Loading Image...

If anything, the acetone treatment only made the smudging from
the sandpaper stick out more.

The inside, where no sanding too place, looks like this:

Loading Image...

The display panel didn't go in very easily, probably because the
case bulged a little due to the "blurring".


6) Conclusions
--------------

Alas, one can't just let the acetone do all the work. In fact,
sanding seems to accomplish a lot more than the acetone, so we'll
have to see how this goes with black plastic, e.g., if it causes
discoloration.

This experiment showed that acetone vapour smoothing can make properly
sanded surfaces smoother, especially if they face upward. In order to
avoid excessive blurring, the treatment time should probably be even
shorter than three hours (if using the cold vapour approach).

Also, adding a fan (with a brushless motor) to the setup may help
distribute the vapour more evenly, and improve the effect on vertical
surfaces.


To be continued.

- Werner

_______________________________________________
Qi Hardware Discussion List
Mail to list (members only): ***@lists.en.qi-hardware.com
Subscribe or Unsubscribe: http://lists.en.qi-hardwar
Dave Ball
2017-02-03 00:02:13 UTC
Permalink
Post by Werner Almesberger
I did a first acetone (*) experiment on the Anelok case.
Really interesting Werner - Not bad for a first try!

What printer did you go for?


Dave



_______________________________________________
Qi Hardware Discussion List
Mail to list (members only): ***@lists.en.qi-hardware.com
Subscribe or Unsubscribe: http://lists.en.qi-hardware.com/m
Werner Almesberger
2017-02-03 00:56:26 UTC
Permalink
Post by Dave Ball
Really interesting Werner - Not bad for a first try!
Thanks ! :) By the way, I'm evaluating designing the Anelok case
using SolveSpace [1]. This is in the following repository:
https://gitlab.com/anelok/mexp/, directory alt13/

[1] http://solvespace.com/index.pl

SolveSpace is really nice for designing 2D geometry. For example,
this is a side view of Anelok, with the geometry of display panel,
PCB, and the battery clearly visible:
Loading Image...

The dimensions of the display panel are imported from a separate
model I made of just the panel.

I'm still not sure how it will pan out in the end. As complexity
and size of the design increase, I'm running into some issues,
but so far Peter "whitequark" and Alexey "Evil-Spirit" have been
good at leading me back to sanity when I tried something stupid,
and at fixing the bugs I bumped into :)

The case shown in the krampelok picture [2] is from the FreeCAD-based
design while the case for the ABS experiment is already done with
SolveSpace.

[2] Loading Image...
Post by Dave Ball
What printer did you go for?
It's a locally made Prusa i3 variant:
http://www.kuttercraft.com/impresora-3d-armanda

- Werner

_______________________________________________
Qi Hardware Discussion List
Mail to list (members only): ***@lists.en.qi-hardware.com
Subscribe or Unsubscribe: http://lists.en.qi-hardware.com/mailman/lis
p***@kristianpaul.org
2017-02-12 02:13:29 UTC
Permalink
Could try to add more slices and thinner layers to your builds? (0.3mm
nozzle etc..)
Post by Werner Almesberger
http://www.kuttercraft.com/impresora-3d-armanda
Werner Almesberger
2017-02-12 03:51:10 UTC
Permalink
Post by p***@kristianpaul.org
Could try to add more slices and thinner layers to your builds?
I did try that, but only with PLA so far, not ABS. Here are a few
side-by-side comparisons. The cases on the left were printed with
a layer thickness of 50 um, the ones on the right with 100 um.

Loading Image...

One can see that the steps on the top of the side wall are smoother
with 50 um. The side looks smoother, too, if somewhat chaotic, but
that's hardly noticeable when handling the device.

Both have the problem of unwanted deposits at the edges of the
curved surfaces. In Cura there's an experimental feature called
"coasting" that may help.

The large bottom plate of the 100 um case has some patterns after
the screws. These were caused by me not making the piece lay flat
on the table (or, rather, Cura not doing as it was told, and me
not checking - trust is good but control yields better prints) and
are not related to the resolution.

Speaking of chaotic, this is what happens with a flat surface held
by support (this is a work-in-progress part, so the geometry is a
bit confusing):

Loading Image...

100 um actually looks tidier. 50 um also produces "noise" at
strange place, e.g., inside the battery contact area.

That the flat surface after the side wall is very smooth at 50 um,
but has a diagonal pattern at 100 um, is caused by imperfect build
plate calibration (these prints were made without a raft). Again,
not related to the resolution.

A last one:

Loading Image...

Here, the 100 um side feels much smoother than the 50 um side.
Also the curvature looks smoother.

So I think there's no clear winner here. 50 um improves some
details, but many aren't really noticeable in real life. And it
trades some problems for others. I still have to do some more
testing with ABS, though.

My printing time for all the case parts is currently about 4.5
hours at 100 um, and about 11 hours at 50 um.
Post by p***@kristianpaul.org
(0.3mm nozzle etc..)
Might be worth a try. Some problems seem to be caused by insufficient
flow, especially "under-printed" layers. So these may get worse. But
having more lines may help elsewhere.


By the way, all the design files are in the directory alt13/ of
https://gitlab.com/anelok/mexp.git

If you still have a 3D printer, you could give it a try yourself ;)

The most important parts:

- an-top.slvs is the main structure (top surface, side walls,
battery compartment with contact pockets)

- an-mid.slvs is a middle part that goes between PCB and the bottom
cover. It also has two "shoulders" that cover the battery cables,
though that may be a misfeature, given that they turned out to be
fairly fragile.

- an-bot.slvs is the bottom cover

- an-but.slvs are the buttons

To use these files, you need SolveSpace:
http://solvespace.com/index.pl
https://github.com/solvespace/solvespace.git

- Werner
Werner Almesberger
2017-02-12 14:22:16 UTC
Permalink
More comparisons, with parts where only the support has been removed,
but where I didn't trim any other problems. Note that I've changed
the color mapping to produce fairly strong contrast. (In the photos
I post, I always adjust the color mapping a little, but normally not
as much as in this case.)

PLA, 100 um layer thickness:

Loading Image...

Issues marked in yellow:
- top: plastic protrudes at the edge of the large curvature
- top: transition from top surface to curvature is not smooth
(this may be partially also be a build plate calibration problem,
i.e., the build plate may be a bit too close to the extruder)
- top: we also see a light case of "elephant's foot"
http://support.3dverkstan.se/article/23-a-visual-ultimaker-troubleshooting-guide#elephant
(Thanks to Neil for pointing me to this great guide !)
- top: in the corners, there are very strong support residues that can't
be broken off (most of the support is easy to break off) but need
cutting.
- top: there is occasionally material on top of the rim on which the
cover rests. This should be as level as possible. (Note: this is
printed on support, so it can't be perfect. But I'd hope that it
can be better.)
- middle: small plastic deposits on the side of the curved shoulder.
This may be cause by the plastic bending sideways when printing the
next layer, either due to friction or maybe it bends a little upward
when cooling and then collides with the extruder. I've observed such
sideways bending.

Loading Image...

- bottom: again, sharp edge in what should be a smooth transition.
- top: mysterious little gaps between the rear wall and the structures
holding the battery contacts. This is strange. We should have a
continuous solid there. Maybe there are some mesh anomalies that
confuse the slicer. In any case, this happens only occasionally.
- middle: again, poor definition of the shoulders.
- middle: the "flat" surfaces look rather untidy.

PLA, 50 um layer thickness:

Loading Image...

- top: edge is a bit smoother but still far from clean.
- top: in the distance, we can see that the other side has become
rough, too.
- top: plenty of "fuzz" on the rims that support cover and panel.
It's very hard to trim such things, given that only a chisel-like
tool can reach them properly, and there isn't much material to
guide the tool.
- middle: intense fuzz around the shoulders.

Loading Image...

- bottom: strange outgrowth on the curved edge.
- top: random hairy outgrowths. Probably of this kind:
http://support.3dverkstan.se/article/23-a-visual-ultimaker-troubleshooting-guide#stringing
- middle: lots of fuzz on the shoulders
- middle: also the surfaces are very "fuzzy". (Note: also here,
printed on support.)

So while going to 50 um generally improves the basic accuracy, it adds
fuzz and random hairs, which can make things even worse.

The middle part is still evolving, so it's hard to say how much of a
problem it will be in the end. I may also be able to bring some order
into the chaos by increasing the support density. But this comes at the
cost of making the support more difficult to remove. We'll see.

- Werner

_______________________________________________
Qi Hardware Discussion List
Mail to list (members only): ***@lists.en.qi-hardware.com
Subscribe or Unsubscribe: http://

Loading...