The following tutorial is a detailed guide with charts, graphs made by EOS GMBH as an "Investigation of detail resolution on basic shapes and the development of design rules." The work is by and based on a dissertation by Dominik Sippel.
A lot of hard work went into that dissertation and we are proud that we have been given permission to share this work with you. A machine manufacturer would like to share the basic design rules of its process and machines with you: the Shapeways community. A machine manufacturer, in the highly competitive 3D printing machines business: wants you to learn the limitations of their process so you can learn to design complex mechanical parts. A machine manufacturer is giving you direct access to its own R&D. It is putting a portion of its intellectual property online, not on its own website, but on another site, Shapeways. It is not sharing this information with a select number of its own customers but rather a customer's community.
This is huge. In IT and online we have gotten used to getting API's, open source libraries and acess to the code. So used to this in fact that we do not realise how incredibly rare this is outside "the web" and open source. We hope that you reaslize the significance of this. Because we would like to encourage EOS to keep allowing us to use this information and because we realize that this is an important step for them and technology in general we ask you to respect this material.
The information and images below are the Copyright of EOS Gmbh. Feel free to use the information in these images for your own use. Please do not download or otherwise distribute the images on this page.
If something is unclear or not explained well please email william (at) shapeways (dot) com.
The data below has been compiled by testing EOS SLS systems. EOSINT systems such as the p100, p390, p800 etc. will tend to obey these design rules and adhere to the data. The data however will not be useful for use with other selective laser sintering systems from other manufacturers. Other 3D printing processes will also require very different data. The PA 2200 material tested is known on Shapeways as White, Strong & Flexible.
The EOS Selective Laser Sintering machine used for these tests was an EOSINT p390. The layer thickness of this machine is 0.15mm. The material used is PA2200(polyamide 12). On Shapeways we call this material White, Strong & Flexible.
The beam offset used was 0.33mm.
Since Selective Laser Sintering(SLS) works by a laser beam that selectively hardens a bed of powder the beam offset is especially important. If the beam offset were changed everything that came out of the machine would be smaller or bigger depending in which direction it was changed. Beam offset is not the diametre of the laser beam. Beam offset is the distance from the outer edge of the layer that is to be sintered to the middle of the laser beam.
One of the purposes of this research was to test the detail resolution of the SLS process. The geometries used to test this were: font, walls & pins and cutouts such as gaps & holes.
A special test piece was constructed to check the font accuracy(labeling quality) of SLS. They used Arial in sizes ranging from 8pt to 14pt. There was also variation in the letter height ranging from 0.2 to 1.2mm.
The test text was printed in the vertical, horizontal UpSkin and horizontal DownSkin directions. UpSkin means that the detail being printed was up. So the UpSkin is the top of the part when the part is in the SLS machine. Upskin details are directly sintered by the laser. The accuracy of upskin part therefor reflects more on the accuracy of the laser and this type of detail in this position.
VerticalSkin is on the side of the part when it is in the machine. This side is built up layer by layer. The accuracy on this side has to do more with the layer height of the machine.
DownSkin means that the detail being printed was at the bottom of the part. There are cute little logos that will make it all clear.
Because this test consisted of printing out a part and then visually inspecting it, the test was subjective.
I told you the logos would explain everything. In the VerticalSkin side we can see a difference between cut out letters and ones that are prominent. The conclusion was that, "VerticalSkin offers very detailed labeling quality."
For outside or prominent letters do experience a degradation at low heights of 0.2mm. If you want to have upright or outside detail more prominent and taller detail would tend to work better.
In contrast to the good labelling quality of VerticalSkin in the Selective Laser Sintering proces, "front UpSkin orientation shows rather low labeling quality." Only the 14pt font size was good. This was good across the board.
Since you can not choose the orientation of your Shapeways model we would therefore recommend using 14pt font of at least 0.4mm high if you were to label something.
The font shows good quality if it is placed in a downskin orientation.
As you might be able to tell from this collection of tables, the only font size you can consistenly produce well is 14pt in heights(depths) of more than 0.4mm.
The next test was meant to create fundamental design information from analysing basic shapes. The object pictured above was printed to test the wall thickness accuracy in SLS. The walls varried from 0.5mm to 2mm in 0.05mm increments. The orientation of the part was vertical when it was in the SLS machine. A series of these parts were printed and then tested with high end machinery.
"The investigated test parts showed a max. deviation of ±0.06mm." We recommend wall thicknesses of 1mm or larger in part because the results show that the deviation could be much higher. But, in addition to this one has to take into account how strong the part needs to be in order to get it out of the machine to you. As explained in the wall thickness tutorial, do not use wall thicknesses of less than 1mm and do not consider sub 2mm wall thicknesses for larger parts of your model.
A "switch from edge to contour exposure increase deviation up to 0.1mm." So, for your 3D printing needs you should accept this number as a guide. The reliable minimum value is approximately 0.8mm and the maximum deviation from this is approximately 0.06mm. However, these values apply to edges. If we have a contour in your model then a deviation of 0.1mm could occur.
The next step was to analyze anothe basic shape: pins. The pins have a diameter of between 0.8 and 2mm. Again the 0.05 increment is chosen and the part is printed vertically.
"The investigated pins showed an average deviation of -0.2mm" I can not stress enough how important this is. The average deviation is 0.2mm but the maximum deviation could be 0.3mm. At very thin diameters there can be even more deviation. Please think about this when you are designing anything that clicks together or has to move such as an axle.
In the following test the tested parts the gaps had a length of 10mm and the width of 0.5mm to 2mm. The wall thickness was varried from between 0.3 and 6mm. The part was printed horizontally.
Less thick walls means that you get higher resolution of details. The quality of a gap depends on the wall thickness and the size of the gap.
"The design chart shows the usable area in the relation gap size/ wall thickness." The parts of the chart that are in red show where loose powder is still trapped in the gap.
In this test the accuracy of holes was tested. The diameter of the holes was between 0.5-2mm. The increments were 0.1mm. The wall thickness was in increments of 0.3mm and between 0.3mm and 6mm. The part was printed horizontally.
"Hole accuracy depends heavily on the wall thickness."
"The design chart shows the usable area in the relation wall thickness / hole size." If you would like a more detailed hole: then either increase the diameter or decrease the wall thickness.