Here is how I see the week's flow :
Table of content :
I started to look for a test 3D model to try our Ultimaker 2 here at the GreenFabLab.
Understanding the challenges of 3d printing (overhang, bridges, tolerances) I picked this one on Thingiverse:
We'll be able to try many different challenges that occur during 3D printing.
But first we need to be able to transform our model in something that the ultimaker can read.
We call this g-code. This is a set of basic instructions, like "move from here to there at this speed".
So the next step is to download a software that can generate a gcode from a model, usually in STL format.
These softwares are called "slicing software" because they split the model height-wise in many layers that will later be printed one by one.
We'll use Cura :
After we've installed Cura, gone through the doc and updated our Ultimaker's firmware, we load our test model and here it is in Cura, all sliced :
A nice feature to make the model stick to the heated plate of the 3D printer is the "Brim" feature.
It basically builds a very thin layer of the material we use (here PLA) around the base of the model that's easy to break off after the print is done.
There is many other parameters we can play with in expert/advanced mode such as the thickness of specific layers, the speed, how the model is attached onto the printer plate (brim), if it needs support underneath, etc.
But for a start, to print our test model, we'll just use the normal print from the basic mode :
So we generate the gcode and save it to the SD card that the Ultimaker 2 reads.
Right before printing the machine (nozzle) heats up :
Till melted plastic exits the nozzle smoothly :
When our print starts, I notice that the Brim is not working properly.
Every tracks should stick to each other, which is not the case :
To correct this I do two things : I make sure our plate is level and at the right "zero height". The Ultimaker has a nice tutorial to get it leveled. I followed the instructions indicated.
I also correct the buildplate temperature. It was at 40 °C. I put it at 60 °C :
Okay, second try :
Good, the brim is one single thin and continuous layer.
We contemplate our first test being printed :
Here it is finished :
Alright, so what's working and what's not ?
Using a caliper I notice a shy .5mm difference between the measurements in the CAD software and reality. So I'll have to take this in consideration for future designs.
It shows with the gaps in the model : Only two are actual gaps in reality.
Well, the overhangs are pretty good, up to 45°.
The bridges also (horizontal connections) :
So to conclude on the experimenting, the design rules are defined as follows :
So... we all this in mind, let's model something !
One thing that's very handy on construction sites is a long distance level.
Let's make a part that can easily turn two plastic bottles into a water level. Cause water sits flat right?
To help me, I cut in half a cap and reverse engineered it !
This is a part that cannot be manufactured substractively.
It's going to be difficult to achieve the water tightness as it is done with the cap material because PLA is harder but I'll try.
Using solidworks to design it :
I use the Helix feature for the thread :
But before I print the whole system, let's make some test on the thread.
Here is the part we'll use to test :
Ok so I print it, I put it on the top of the bottle, try to screw it .... and guess what... wrong way, the thread is reversed. Thanks Murphy.
So I change the design (super easy in Solidworks, two clicks with the counterclockwise/clockwise option of the helix). Again, that demonstrates here the value of designing with vector history based software.
It works but it's super tight !
I reduce the thickness of the thread and enlarge a little the size of the inner hole. It works
Let's go back to the full part :
I first tried to print it horizontally with supports :
It totally fails
I decide to print it vertically with custom made supports :
It prints okay but the supports are way too thick and hard to detach.
I then make small holes between support and the part :
Still too strong. Plus it leaves some matter on the part which is not good for water seal with the hose.
New design with tiny bridges :
This time it works :
It takes 2 good minutes for the water to leak, the sealing is not so so bad. But I think I'll still use a gasket and put a flat part instead of that lip inside :
So I designed and laser cut gaskets :
After having cut my gaskets, I put them into two new prints of my parts and here is my water level not leaking anymore !
Before we jump on something else, I noticed a weird noise when the ultimaker goes along one axis. It sounds like an over-centre lock noise :
Maybe some sewing oil would do it. We ordered some.
I also wanted to try to print a telsa valve :
This is also a part that cannot be manufactured substractively.
After 7 hours (the valve is 20 cm long) :
So it's true, it's easier to blow from one side compare to the other. But I was expecting better results.
Since I did this print overnight, I wanted to check inside. So I laser cut it :
Maybe the not so good results are coming from the fact that the sharper the edges the better the valve works and it's hard to have super sharp edges with 3d printing.
One thing that I might need for my final project is some sort of a fluidic diode.
I'm thinking of achieving this by having a ball allowing flow in only one direction :
Little support would be broken afterwards :
This is also obviously a part that couldn't be manufactured substractively.
In Cura :
The printing process :
We can see the ball inside :
The part printed :
We break the ball free :
We can hear it loose inside :
Trying to blow against the arrow is clearly harder than the other way, it is not water or air-tight though.
Something that I didn't plan but found too tempting to resist was to try to generate my own gcode from Grasshopper
This idea came to me because I wanted to find an efficient way of creating tiny little support for my plumbing part print
I wondered if it would be possible to print "in the air" kind of like the 3Dpen :
So I experimented with this idea and was successful to generate valid gcode but my spiral didn't really work
Here is the lame try :
I'm pretty sure I need to lower the speed and increase a lot the "E" (extrusion) value in the gcode.
Then I played with the Kinect to 3D scan.
I noticed that because it works with infrared, the Kinect doesn't really work in the sun or bright light (actually "warm light"). So I went for a rather dark place :
So with the help of my icelandic academy mate Guðjón Logi Haraldsson, we scanned the cellar
and here is what we got :
Lastly, I played with 123D catch.
It's a photogrammetry software from autodesk which uses the power of cloud computing to recreate a 3d model from a set of images taken with our camera :
It's super easy. We take pictures, upload them and receive the model.
Here are the three basic steps :
Here is what I wanted to scan :
The little pieces of paper on the floor are trackers that will help the image processing
After an hour of cloud computing, I got this :
Editing it in Max to remove some faces :
And then, rendering it !
Here you can play with it in 3D :