3d Model Resources

Where to Find 3-Models for Printing

Thingiverse – When you say 3-D printing files, this is the site that most people think of it. Easy to search, and almost every file posted is designed and optimized for fabrication via 3-D printing.

GrabCAD – GrabCAD is Thingiverse’s older, PhD holding brother. Designed as a site for engineers to share CAD files, GrabCAD models are generally capable of being used for 3-D printing, but additional optimization or repair might be needed.

Modification and Repair

-If you’ve got a file, and it’s not in a shape that you can use, there’s a few options.

NetFABB – A handy tool for chopping apart, inverting and resizing models. It also can convert files from one format to another, which is useful if you need to get a .STL file from another format. It’s available as freeware. Paid versions provide even more usability.

Microsoft STL Repair – If you have a file that won’t slice properly, or isn’t watertight, and causes issues during printing, give it a run through this free tool that NetFabb and MS developed together. This program is super useful since a lot of freely available 3d models aren’t always designed with printing in mind.

Thermal Conductivity

As part of my interest in printing with POM, I’ve noted before that the largest issue with the material is the incredibly rapid rate at which it cools down. It dumps all of its heat into the environment around it so quickly that it audibly pops off the bed.

Thermal conductivity, is a measure of how quickly something absorbs and transmits heat. If you look at a simple table (such as this one), you can see at a glance, the values for different materials.

Looking at PLA, thermal conductivity is a mere .13 W/m-K, as borne out from the experience any of us have had grabbing a fresh print off the bed without allowing it to cool. It absorbs heat slowly, but also releases it slowly.

Looking at Acetal/POM, it’s only a mere .23, Nylon is .26 and something like Phenolic Resin (a common choice for high temp applications) is only .15

Air itself is only .024. Which, by the way, is the reason insulating gear/clothing for the outdoors is built around having fibers that trap air. Your body heat gets the material “up to temp” and then it takes a very long time for it to cool back down.

So, given all of that, it raises the question of where the heat present in the POM that’s being printed is going. The answer to that is in the bed. Aluminum has a thermal conductivity of 205!

What that means then, is that the aluminum bed allows prints to rapidly cool down, by serving as a heat-sink for the print that you’re making. Now, normally, this wouldn’t be much of an issue. PLA works just fine afterall, and aluminum printbeds are a standard for the ease at which they can be made into heated beds.

However, my hypothesis is, that by using a bed substrate which has low thermal conductivity, it might be able to just barely push the needle to make printing in POM work without adding a heated bed.

Looking at the list, Balsa wood stands out. It’s cheap, it’s the lowest thermal conductivity of any wood, and it’s easy to laser-cut. Acrylic also stands out for similar reasons. The advantage that I believe Balsa wood, or any wood presents is that wood has greater porosity, which means more sites and locations for the extruded plastic to form mechanical bonds with the substrate.

Which means that my next steps are:

  1. Create a pattern based on the current Printrbot Aluminum Bed I have.
  2. Purchase Acrylic and Balsa Sheet at a size near to the current bed.
  3. Have the materials laser cut per the pattern I create. (Seattle has just such a space, and I’ve been meaning to check it out anyways)

Expect to see a post in a week or two outlining how this all goes for me!

Experiments with POM

As mentioned before, the standard white masking tape I’ve been using lately has been a nice improvement over my prior tape of choice. It’s bond is so strong, and the filament adheres so well to it, that the slight amount of warping PLA creates has been more or less eliminated for me, without requiring a heated bed.

Given such great results, I thought it’d be interesting to try a run with some of the Acetal/POM filament I have on hand, and see how that works out.

Unfortunately, results were significantly less positive. These runs are admittedly a bit nicer than prior attempts, and at least the pieces are solid.

I’m inclined to believe that there isn’t really a way for me to get around the necessity of a heated bed for this one. Unlike nylon, which I’ve had minimal success in keeping adhered to the print-bed, the POM continues to pop off.

Polyoxymethylene, commonly reformed to as POM, or Acetal or Delrin has about the same shrinkage factor as PLA (approx. 2% give or take the filler content) so on its surface, you wouldn’t expect such issues, compared to a higher shrinkage rate polymer like ABS. Especially when PLA has to be heated to only about 185 C for good results, whereas POM requires 215 C minimum.

However, there’s a second factor coming in to play. As many of us have experienced, PLA retains a degree of heat after printing, often remaining relatively pliable and soft for some period after printing. This is because it has a relatively low crystallinity; it melts sooner, and it retains heat longer, because it is an amorphous material. It doesn’t set into an organized structure.

POM is highly crystalline by comparison; it naturally falls into a more “organized” state, which is part of why it has so many great properties (low friction, high durability) but is also why it is so hard to work with. It sheds heat quickly, cooling down into a solid within seconds of printing. When it is being printed, it becomes amorphous, and clear; upon cooling it becomes an opaque white as you can see in the photos above.

Because of this, it cools off so rapidly, that all 2% of its shrinkage occurs at once, instead of over a period of minutes like PLA. In turn, the material shears off the build plate, and curls inward due to contraction along the perimeter of the print.

As a result, there are likely only 2 ways to get solid reliable prints with PLA.

  1. Utilize a heated bed to provide enough heat to the print to slow the cooling rate down (likely somewhere around 80 to 140 C)
  2. Utilize a heated build chamber to slow the print’s cooling rate down.

The latter is what the $10,000 plus Stratasys machines do, and they have quite the nice patent on it. It’s also a lot harder to design. So, I think my shopping requirements for the next few weeks are pretty sorted out.

In the meantime, I will probably continue to see how it performs on other substrates. I’ve heard some mentions that it prints extremely well onto particle boards like OSB, so I might have a small wooden build plate laser cut for the purpose. I haven’t had a good opportunity yet to go up to the Metrix Create Space in town.

Funnels and Slic3r

At the beginning of the year, it’s often the time for projects, big or small. I decided that this was as good a time as any to make a few useful things for my car. I already have all the fluids I need to clean up my car. I just don’t have a funnel.

Rather than run across town to an auto parts store for a funnel to fill up the fluids in my car, I just made one, using this wonderful design I found on Thingiverse. Hat tip to bluesroq, for the great design.

You might also notice the tiny little funnel. Needed something to fill the flask from NYE. Hat tip to KKHausman for that design.

I think the more interesting thing I’ve observed, after extensively testing these funnels for their ability to be water-tight, is that it’s definitely a function of two factors, layer height (relative to the model size) and vertical/horizontal shell-count.

Bearing in mind that this is all slic3r specific, and so users of Skeinforge/Cura/Etc might not find this as handy. Shells are simply a representation of how many solid “walls” exist in the object when it’s being printed. In slic3r, these shells are represented as being either horizontal (perpendicular to the build-axis) or vertical.

If the object your printing is a cube, you could say, set your slic3r to have one horizontal shell and one vertical shell. So you would get a solid bottom layer, and a solid top layer, and 1 perimeter “wall” in which the gap between the edge of the walls is filled with your infill of choice.

When printing simple objects, that do not have lots of vertical angles, this is often an acceptable setting to reduce printing time, and increase printing speed. However, when printing a complex object like a funnel, you run into a difficult problem. Since each layer is set “back” slightly from the other to form the slope of the piece, with one vertical shell, you only get one point for each layer to bond to. Which means that if your printer has a half second hiccup, or your piece has too steep of an angle, there will be a hole. 

The solution is to add more vertical shells then. Because now, the interface of the two layers isn’t one outer wall offset from the outer wall below it, but instead 2 or 3. So far, I’ve found 3 shells to be the magic number for forming waterproof containers.

However, there is another complicating factor. Layer height.

If you haven’t noticed by now, there’s actually some nice overlap with calculus in this whole thing. At least a few people might be having flashbacks to Riemann Sums. Layer height has an immediate and noticeable impact on the quality of prints such as these. When I initially printed the tiny funnel, I printed it at .2 MM height I used with the flask.

The layers didn’t work. It’s such a small piece that the printer can only lay down 1 shell, with each layer barely touching the other. The solution then was to go down to .1 MM to improve the print quality. At .1MM the printer could put down multiple shells, and maintain the slope of the piece without issue.

On the other hand, the larger funnel? .4 MM. I decided to take a risk and it worked out well. At .4 MM the funnel prints quickly, and works well. It’s completely watertight.

So, to keep it simple. When setting up your prints in slic3r, think about the geometry of the piece you are making.

The best pieces are going to utilize a layer height and shell requirement that optimizes printing speed while still maintaining the integrity of the piece. Take some time while setting things up to look at how the piece is layed out. When slic3r completes, spend a few minutes checking out the preview in Repetier.

The last thing you want to see is a preview where it looks like single strands of filament have been carefully stacked on each other while barely touching. Those pieces rarely come out well.

Happy Printing in 2015!