This will be the first in what I hope to be a series of tips for optimizing your 3D printed part design for mechanical performance.  An apology in advance: I have a masters degree in mechanical engineering and 15+ years of industry experience.  I'm used to speaking to people in 'engineeringese' with all of the made up technical words that entails.  I am going to do my best to express things in layman's terms as much as possible.  One of the awesome things about 3D printing is that anyone can do it; you certainly don't need to be a mechanical engineer to run a 3D printer or to design parts to send to a bureau.  So, I'm going to try to structure this writeup in a way that anyone can understand it.  Inevitably, I'll end up throwing in some jargon without realizing it.  Please call me on it and I will revise.


Many times, people do not associate 3D Printed parts with strength, but this does not need to be the case.  Paying attention to a few fundamental design / mechanics of materials principles can greatly increase the strength of your printed part.  Combining these principles with advancements in available materials will allow you to produce parts that can meet the structural requirements of more and more demanding applications.

Stress Concentrations

In this first installment, we'll talk about stress concentrations.  When you put load on a part, the way that the load is distributed through the part is referred to as stress.  Materials are rated based on the amount of stress that they can carry.  The stress generated in the part is generally proportional to the cross sectional area of the part.  I think that as 3D Printer people, cross sectional area should be easy to understand; it is the area of the part when you slice it against the direction of loading.  Of course the stress exists in the full 3 dimensions, but thinking of the two dimensional space first is adequate to get the point here.  (Making things more complicated:  what exactly is a cross sectional area on something that has X perimeters and Y% infill?  We'll ignore that for now and pretend that the prints are solid since the concept still basically holds, but this makes things significantly more complex if we really try performing stress calculations.)
Think of a diving board.  The amount of stress in the board is higher with a large adult jumping on it than with a small child jumping on it.  Let's say that the board is rated to handle a 400lb person.  Now drill some holes in the board.  You will find that the load capacity is reduced and that it is reduced by more than the small percentage of material you have drilled out.  Now instead of drilling out a hole, cut two sharp notches in the side.  The capacity will be reduced even more significantly.  In both cases, the point of failure will move to the feature you've added.
This principle where the ability of an object to carry load is reduced by adding sharp features is due to stress concentrations.  Both the hole and the notches are called stress risers.  They change the way that the stress is distributed in a part and increase it dramatically at the location of the stress riser.  Any change in geometry is a stress concentration.

Removing Stress Risers

So what?  How can you take this knowledge and use it to improve your designs?
Three words for you.
Fillets, Fillets, Fillets.
Fillets are features that you add to the geometry to round out sharp corners.
Red piece on the right has a sharp corner. This is a stress riser. By smoothing out that sharp corner with a fillet as in the part on the left, strength will be more than doubled.
In many traditional manufacturing methods, fillets become necessary simply because features may be oriented in such a way that it is not practical to get a sharp tool in there.   Even when you can, sharp mill bits typically don't do better than about a 0.005" radius, so you naturally get fillets unless you pay $$ for more exotic manufacturing processes.
With 3D Printing, no such limitation exists.  You can make perfect sharp transitions that are terribly stress risers quite easily.
Here's another example.  I was at an event and another attendee showed me this part.  He said, "the design geometry suits our needs perfectly, but it keeps breaking."  To which I replied, 'oh, is it breaking right here?' and pointed to the sharp points at the ends of the semicircle.  At which point, as if on queue, the part in his hand actually broke at those points as he handed it to me.  The ends of the semicircles are pretty significant stress risers.
The corners of the D shaped cutouts present nasty stress risers. It doesn't help that the material around it is relatively thin. Expect sharp features like this to be a potential failure site and avoid if possible.
Now that we've learned about stress concentrations, what would you do to fix this part?  Fillets?  That could work.  You could move the flat surface back and add fillets.  You could also adopt a simpler solution and simply make the semi-circle into a full circle.  This may come across as a surprise to some.  Removing material will actually make the part stronger!  Yes, it will still be weakest at that point, but the amount of load it can carry will be significantly increased by removing that stress concentration.
I hope that this article will help you improve your 3D Printed designs.  Our next article will elaborate a little bit on how properties of the filament you print with affect stress concentrations (spoiler:  stress concentrations are much more significant in PLA than they are in Ninja Flex).
This post contains assembly instructions for the Robo3D flat packed enclosure kit that we sell here:
It is a repost of a writeup done by Mike Kelly on the Robo3D forum with some edits to account for changes made since his original post.  
This mod allows you to enclose your printer so as to avoid injury from the hot and moving parts, while still being able to watch the print.

Laser cut parts:
1x Front Top
2x Front Sides
1x Front Front
1x Front Bottom

1x Rear Top
2x Rear Sides
1x Rear Rear
1x Rear Bottom


Weld-on 3 or Weld-on 4

Recommended Hardware / Printer parts = See "Mounting the Latch"

Build the Enclosure

While not necessary to build the enclosure, these tools were used during my assembly to help ensure a square shape. Basic masking tape is all that is truly needed as the pieces will interlock together.

2x 24in or 36in Ratchet bar
2x Corner Clamp (cheaper alternative)

I began by securing the Top, Bottom and Front panel in the corner clamps, verifying the pieces were in the correct positions.


I glued the pieces together in this order:

Bottom to Front

Side 1 to Bottom and Front

I would then remove the 3 glued part, as they no longer require the brace


Side 2 to Bottom and Front is next. Using the ratchet clamps to ensure the part doesn't move during gluing.


The Final step is to secure the top part to the other 4 pieces.

Completed front cover:


Repeat this process for the rear:

Note: I use a piano hinge on my front cover.

Mounting the Latch

We advise using this excellent printable mount system designed by Thingiverse user Leed3.  There are a number of other options available, but we believe that this is the easiest option.  

You can find the files on thingiverse here.

Be careful when installing the case during a print so as to not hit the moving bed

Why this shape?

In order to maintain the full build volume it needed to be this large. There are some places that could be reduced in size, but wouldn't work well with a laser cutter.

Internally there's 5mm of clearance on both ends for the drawer slide model. The R1 Y rails have more room because they require less space. Overall it only adds about 15 mm of length end to end.

Do I have to mount it like this?

I leave it open to each user to decide just how they want to secure it. I personally prefer a hinged method, though easy removal is what a lot of users requested so that's our suggested method. I'm interested to see what other users come up with for mounting to fit their needs.

I have an LCD, will it fit with the enclosure?​
More than likely not, I have designed a swivel mount design that uses commonly used lifter feet to accommodate the XXL LCD. If you have an alternative I can design something for you.​
The alternative solution is to cut off a portion of your enclosure, but I leave that choice to the user​

One of the most simple, yet frustratingly difficult things to get figured out when you are learning about 3D Printing is how to get and keep your prints stuck to the bed. People have all kinds of tricks.  Bare clean glass, borosilicate glass, plate glass, PEI, buildTak, aluminum plate, circuit board perf material,  Painters Tape, Kapton Tape, Glue stick, hair spray, PVC film, higher temperature, lower temperature, etc.  The list goes on.  There is science behind this and there are reasons why some materials work better than others.  There are also factors associated with the environment you are operating in and your technique that are difficult to control.  As things often are on the internet, everyone thinks their method is the absolute best and that all of the others don't work.  Don't worry about them too much.  If you've got a build surface that works well for you, stick with it (pun intended).  If you're curious, we typically use either painter's tape or glue stick/PVA on glass for most materials with a bed temp of 50-70C or Room Temp if I'm using a printer without a heated bed.  On the rare occasion when we run ABS, we use hairspray on Kapton at 100C.  We've played with bare acrylic and PEI but haven't been too impressed. However, of all of these build surfaces, none of them will work if you don't get that first layer right. Over on the Robo3D forum, we've got two amusing terms that have become part of the vernacular. 'Check your Skirt' and 'Throw your nuts'.  The second term is pretty specific to the design of the first version of that machine, but the first is particularly useful for anyone new to printing who is trying to figure out how to get that first layer to stick. Here's the idea. Get your bed cleaned / primed and leveled to the best of your ability. For all prints, run a skirt that is at least a few nozzle widths wide. Run the print speed relatively low during the skirt printing so that you can clearly see what is happening. If possible with your machine, you can adjust your bed level / tram and nozzle gap during the skirt printing to achieve the desired criteria. If you get a nice clean skirt that meets the criteria I am going to describe, then let the print continue.  If not, stop the print and restart. Here's the criteria.  Your skirt should look pretty similar to the one in this picture:

Photo Courtesy of Jeff Janes.  This picture shows a good first layer.  The multiple skirt lines have cleanly merged together and appear to be a single line.  Note the changes in height towards the center are due to the operator adjusting extrusion ratio as a demonstration. 

This picture shows a good first layer. The multiple skirt lines have cleanly merged together and appear to be a single line. Note the changes in height towards the center are due to the operator adjusting extrusion ratio as a demonstration. Photo courtesy of Jeff Janes


Here's a little more explanation if you like the details.  Note that some of the details here (specifically the wagon track) vary with nozzle geometry, but it's good general guidance.   _filamentcross  

We highly recommend giving this method a shot if you're struggling with getting your prints to stick and/or getting a good first layer.  This is an excellent quick and dirty, but very effective method.  Advanced users going for high z-axis precision may prefer to us a more technically sophisticated method of setting nozzle gap with precision shims or a dial indicator. Full credit for the 'check your skirt' idea and the pictures used go to Jeff Janes.  Check out some of his models here:  Tesseract

Our friend Mike Kelly on the Robo3D Forum put together an awesome upgrade and set of instructions for the robo3D. Check it out! [divider scroll_text="SCROLL_TEXT"]   [IMG] Click for Thingiverse What is this?: Converts the stock RoBo 3D printer with a single direct feed extruder and replaces it with dual bowden extruder with no loss in x travel. This is just the x-carriage as the cold end will be released separately [IMG] Background: I have designed and fabricated, with the help of tessarect, a dual extruder upgrade to the RoBo 3D printer. As many RoBo 3D owners make the upgrade to a E3D all metal hot end, it makes sense that instead of going from stock hot end to kraken, many users will want to go stock hot end > single E3D > dual E3D to save cost on parts. Note: With that in mind this mod is directed at the latter group of people. If you are on the stock robo 3d hot end and want to upgrade to multi-extruders, I HIGHLY recommend not doing this mod but rather wait for me to release the kraken version. The kraken is smaller, lighter weight, easier to adjust, and provides more options later on. This mod is just inteded for people with an existing E3D looking to add a second. BOM: Printed:
  • 1x X carriage
  • 1x Fan Shroud
  • 1x Flow Director
  • 1x Fan Holder
  • 2x J-Head Saddle
  • 2x J-Head Mount
Hardware: Build Instructions Note: Animations may show the E3D as being fully assembled. Please check instructions to verify when to fully assemble the E3D as it will increase build complexity to have it fully assembled from the start Phase 1: Remove original X-Carriage Watch the original install video to understand how to properly assemble/disassemble the x axis Phase 2: Build Dual X-Carriage Step 1: Assemble J-head mount [IMG] During this step you'll only be using the bowden heatsink from the E3D. Attach the J-head mount and saddle around the heatsink. Press in the 2 M3 nuts in the open slots, this may require a tool to get them fully seated. Once the nuts are in position use 2x M3-25mm screws to secure the saddle to the mount. Step 2: Securing the E3D [IMG] The x carriage is designed to allow the E3D heatsinks to pass through from the top. The height is controlled by two screw/nut/spring combos per side. Only one side needs the spring, though it doesn't hurt to have both sides spring controlled The principle behind this is by tightening/loosening one dual extruder plate you'll be able to lower/raise the nozzle height respectively. You should only need to adjust a small fraction of a mm but this gives you that option. Cut away view of height control: [IMG] The order goes:
  • Push in the M4 nut
  • Screw in the M4-25mm to create a post, ensure on one side the part fan holder is in place depending on if you want it front or rear facing
  • Place the spring into the housing
  • Insert E3D and j-head plate over assemble
  • Secure almost fully tight using the M4 Lock nut
Step 3: Push in bearings You'll need to push the bearings into the holder. I used a bearing press but you could use a vise. ABS is recommended to help prevent splitting. Step 4: Install x end stop screw. Remove from the stock assemble and install in the new one. Step 5: Assemble Fan Shroud [IMG] Ideally you'll want to install your fans in the shroud before installing it on the E3D heatsink. You should only do 3 screws on the outer sides of the fan as it can be difficult to remove the middle ones later. Run the fan wires towards the middle and through the front hole in the case Step 6: Finish assembly of E3D The E3D can now be assembled. The heater cartridges can be fed through the hole in the plate from either direction. Step 7: Assemble Parts fan [IMG] You'll need 4 16mm m3 screws for this. I had to cut 2 of mine down to fit, but since it's self threading it's not a big deal to do. The parts fan and airflow director will need to be tight against the backing otherwise there will be potential for interference with the print. Phase 3: With the X carriage now assembled you'll need to install it as if you were installing the stock x-carriage. I will follow up later with how to calibrate this nozzle. For now please let me know if there's any other steps needed. You can view my entire build photo log here: Click Here to Download!


As you've seen in some of my previous posts, I've spent a fair amount of time experimenting with using an acetone vapor bath to smooth 3D Printed parts. Recently, I saw this cool make of a Moai head posted by thingiverse member Shapespeare.  He used acrylic cement to manually smooth the PLA printed part with a really great result.  This looks awesome, but I don't want to deal with a bunch of little jars of acrylic cement and I'm not sure about the full list of chemicals contained. That got me thinking back to vapor polishing PLA and other materials.  I've read about using THF as a vapor to smooth parts, but have also read concerns about storage of the chemical.  It's also quite expensive.   I've also read about people using DMC.  This is a standard industrial vapor polishing solvent, but is hard to come by and really should only be used with a vapor recovery system for safety. So, on a whim, I tried out some 'MEK Substitute'.  According to a closer read of the label, it is Ethyl Acetate, which is used in some acetone free nail polish removers, processing of some food products, and is generated in the wine fermentation process according to the always accurate wikipedia.  It's sold in the paint aisle of hardware stores as a replacement for MEK, which is harder to come by and more potentially harmful to your health.

I decided to use the Cute Octo model for this experiment.  My daughters really like carrying these around so I figured it wouldn't hurt to have a few more laying around.  It also has enough minimal detail to judge effectiveness of the polishing, but not so much that the detail is ruined. I printed up 5 Cute Octos in the following materials. 

Here is a picture of the 'as-printed' octos.  These were printed on an Ultimaker with a modified 0.8mm nozzle (future post coming about why I love my 0.8mm nozzle) using a printing profile that was primarily optimized for woodfill with adjustments made only to temperature and cooling for the different materials.  Layer height was a fairly beefy 200 um to make the smoothing more obvious.

2014-03-20 18.38.04

I then poured a little of the Ethyl Acetate into the bottom of the trusty vapor polishing machine (a deep fryer), set the octos in the basket, and turned up the heat.  As soon as I saw vapor condensed on the tops of the parts I turned off the heat and removed the basket. **Note:  If you are performing this process, use appropriate safety precautions.  Breathing mask, safety glasses / goggles, solvent gloves, and a well ventilated area are all a good idea.** Here are the results.  As you can see, most of the parts smoothed very well.  There are still visible printing lines, but just barely.  The parts do feel quite smooth when you run your fingernail across them.  I'd conclude that this process works, but more is needed to really nail down time and temperature process settings.   The XT definitely did NOT clarify like I have seen with my past acetone experiments, but it smoothed extremely well.   The woodfill seemed to just bleach out.  I wouldn't recommend doing this with woodfill.  

2014-03-21 18.16.28

2016 update:  Since writing this blog post, I've found the MEK substitute I purchased is no longer available.  I've heard from other users that they've tried different brands of MEK substitute as well as just straight Ethyl Acetate with varying levels of success.  This tells me that there was probably something else in that MEK substitute that helped with the smoothing.  If I learn more, I'll share.


A few nights ago, I printed a thin walled part with colorFabb _XT.  It was very tall and unstable.  Essentially a 4 inch disk standing on end.  I used a brim, but that wasn't really enough.  As the part built up higher and higher, it became less and less stable.  As such, I ended up with a pretty low quality print and had a few layer delaminations. I tried brushing on a little MEK based plastic solvent weld (Plastruct Plastic Weld).  To my surprise, it worked quite well and the material clarified nicely where it was brushed. colorFabb XT disk with a clarification from a brushed line of MEK solvent. colorFabb XT disk with a clarification from a brushed line of MEK solvent.

That got me thinking.  If the MEK solvent works, then can I vapor polish it in acetone?

Turns out the answer is YES Before going further, I should mention that vapor polishing can be dangerous.  It is highly flammable and there are both acute and long term exposure risks.  Make sure you understand these risks and use appropriate safety equipment and ventilation if you vapor polish. I took a Stretchy Bracelet (emmett) / CC BY-SA 3.0 that I had made and tossed it into an acetone vapor bath for a very brief exposure.  The material clarified nicely.  It also became noticeably more flexible.  Here is an overall picture of the bracelet and a closeup of it pressed to my finger.  Notice that you can clearly see my finger print pattern through the bracelet without distortion.

Emmett Stretchy bracelet printed in colorFabb XT on an Ultimaker at 100micron step height.  Polished briefly in acetone vapor.  Emmett Stretchy bracelet printed in colorFabb XT on an Ultimaker at 100micron step height. Polished briefly in acetone vapor.

Closeup of Emmett Stretchy Bracelet link.  Notice that finger print lines are clearly visible through the link.  Closeup of Emmett Stretchy Bracelet link. Notice that finger print lines are clearly visible through the link.

WOW! So, acetone vapor polishing in a very brief exposure clarifies single shell wide colorFabb XT material.  What happens to thicker parts with longer exposures?

I chose the model Crystal Do Dad (cerberus333) / CC BY 3.0 to try it out.  I scaled it down to 80% to ensure that it easily fit into the vapor bath.
I printed up a model with 3 shells and with 1 shell then polished.  My objective here was to evaluate how well the surface actually smooths, but I also learned some other things.
First, here is a pic of the 3 shell model printed with XT on an Ultimaker at 200 micron layers.  As-printed, no vapor polishing.
Crystal Do Dad.  3 Shells.  200 micron.  As-printed. Crystal Do Dad. 3 Shells. 200 micron. As-printed.
The clarity of the material makes it somewhat difficult to see the surface well.  So, I made an impression with some play-doh.  You can see the layers fairly clearly.Impression of 200 micron printed part in play-doh. Impression of 200 micron printed part in play-doh.

Then I took a previous attempt at this print, same settings, and put it in the vapor polish for about a minute.  {As an aside, this print failed due to a break at the bottom of the model just near the brim.  With XT, I've found that if you don't get the nozzle fully purged of PLA prior to starting the print, you get really weak brittle material until the nozzle is purged.}

Vapor Polished failed print Vapor Polished failed print

You're probably thinking, that looks LESS clear than the original.  You're right.  At longer exposure to vapor, the part becomes cloudy and internal layering lines become more exaggerated. However, look what happens to the impression!

Impression of 200 micron vapor polished part in play-doh.  Look ma, no lines! Impression of 200 micron vapor polished part in play-doh. Look ma, no lines!

The lines are GONE.  GONE GONE GONE! This gets me really excited about using this material to print positives for investment casting and mold making! Of course, you can do this same thing with ABS, but the XT has the benefit of being tougher and printing with no warpage. Before I end this, I just wanted to highlight one additional effect.  After vapor polishing, the material becomes extremely flexible and durable.  I took a single layer XT printed part, vapor polished it for about a minute, and then shot the following video the next morning.


9/9/13 Update. Turns out you lose some of that flexibility after the part has had some time to sit. Five days after polishing, the bracelet is still extremely flexible, but there is no way that crystal do dad is going to be crunched into a ball again without breaking. Still, a pretty impressive trick to show off, no? 

2014 and beyond update:  Since this blog post was originally written, colorFabb has reformulated XT.  Sadly, the newer version doesn't polish quite as cleanly with acetone, but we've had some success with MEK Substitute.