Apr 26, 2015

Rotate It! Bed Position Matters in 3D Printing

Getting my 3D Prints to stick to the bed is probably one of my biggest challenges. It's mostly with small parts - and I've mostly had issues when trying to print many copies of the same part all at once. But recently, there was a specific part which was giving me issues where a little experiment uncovered something surprising.

The Problem

The part I was printing is a link for a bracelet - so it is only about 25mm square and has 4 protrusions starting at the base layer which are the receptacles for pins which hold each link of the bracelet to the next. Those protrusions - or two of them in particular - were peeling up from the bed after the first few layers of printing, and causing the printer extruder head to force it out of the way as it printed successive layers. This caused some mis-shaping - but luckily the model would eventually self-correct and continue printing to the end without forcing the whole object to come loose from the bed. You can see the start of the problem pretty clearly in the video embedded here - the protrusions on the right side are fine, but not on the left - notice how they are higher than the current print layer.
The ultimate impact of the peeling was that the protruding parts where the peeling occurred were mis-shaped and not smooth at all in the final object. The print head eventually re-melted and flatted the problem sections, but not in the way which the part was supposed to be printed. You can see this in the images here too.
Here's a close up of the final print - those protrusions which had the problems (now shown on the right) were consistently mis-shaped.
Original Position - the left protrusions were the ones experiencing peeling
For you technical 3D Printers - this is the bottom layer from CURA - in the original rotation position

Discovery of an Idea

I was using  a live video service - Periscope, on twitter (@periscopeco) - to show people this particular print, and someone commented that perhaps it was the position of the object's receptacle parts that was making this worse. I figured it was worth a simple experiment - since I had tried a few other fixes that didn't help.

Experiment - Rotate the Object

Without changing the model or the slicing parameters at all, I rotated the part 90 degrees clockwise so that the receptacle which had the worst effect was now not the first part of the print layer. I could tell right away that this small adjustment did have an impact, as the first few layers of printing were clean and flat, and there was no peeling off the bed. In the prior experiment, I had slowed the printing of the bottom layer to 15mm/sec (in advanced settings in CURA) - but that adjustment alone did not help - but I left it like that and did the rotation.

After 90 degree rotation, the receptacles are now printing smoothly and are shaped as expected in the model design.
NEW ROTATION POSITION, with the problem parts now pointing away from me on the print bed.

And again for the technical folks, the bottom layer shows hardly any pattern change - just the rotated position.

Results

In the end, the object printed much better, without peeling, with a simple 90 degree rotation! I was quite surprised - but pleased of course. Now I am analyzing the layers to see what may have changed in the slicing - so that I can get to the bottom of what the rotation changed in the printing commands, as I have a hard time believing that the rotation itself actually matters. But - TRY STUFF! When you have a problem, try lots of potential changes to the printing parameters and even the positioning on the bed. You never know what you might discover.


Apr 17, 2015

pin-pivots to improve snap-together 3D Printed parts

After lots of experimenting with snap-together parts, I decided to try another method of connecting multiple parts - a pivoting pin. The idea came from Paul Gross (thanks!) in a comment on my Google+ post - and I decided to start from scratch to design something.

The benefits I was hoping for compared to the snap-together model were mostly to get easier construction of multiple parts, a smooth pivot and less accidental separation when pressure was applied to the joint. The other snap-together part designs I came up with were pretty good, but far from strong-holding or easy-to-connect.

Design Highlights:

There were certain things that I learned in this design worth sharing, even if you don't care about the details (which are all TL;DR below)

  • The gap in the pin needed to be wide enough at the tip to allow it to compress enough to get the wider tip through the narrower receiving hole.
  • The whole bottom side of the pin needed to be flattened to provide a flat bottom for the bed (think about what the pin head would have done to that if it too were not flattened) and to help with bed adhesion while printing
  • The top side of the pin legs were flattened just to simplify the printing and reduce the surface area of the touching parts when inserted and pivoting.
  • The measurement from pin head to pin insertion tip needed to approximate, but not be less than, the space between the receiving holes (so there's not too much lateral motion of the pin).

Findings:

This design works well! It provides a smooth pivot, doesn't take much effort to insert, and doesn't separate too easily. That said, this is not a perfect connection. A bit of push on the tip of the pin, and it will unclip and start sliding out - but it takes effort - and in connections where the tip of the pin is mostly behind other parts, this is not likely to happen. I have not yet tried to print lots of pins in one print job, but hopefully that will work well. NOTE: This design is probably NOT safe for kids under 3 yrs old, as the pins are clearly small enough to swallow (and they are not delicious).

Soon I'll post some actual useful objects I plan to make with this design. First order of business, another name bracelet for my daughter ;)

Design Details:

The receiving side of the equation was simple - a couple of aligned holes to accept the pin. I started with a radius which would give plenty of room to receive the pin without binding so the parts would flex easily around the pin when connected. The measurements I used were 2mm Radius on the receiving hole with 1.5mm radius on the pin. That provides 0.5mm gap on all sides - plenty of room for printing precision issues. On the pin ends, I had to go larger than the hole to keep the pin in place once inserted. On the head which would never have to enter the receptacle, I went with 2.4mm radius. On the side of the pin which would be inserted, I went with 2.2mm - and cut a gap of approximately 1.2mm at the widest, which narrows to zero as you go toward the pin head side.

Apr 11, 2015

Experimenting with 3D Printed Linked Objects

All the experimental models.
After doing a bit of experimenting with #3Dprinting snap-together bracelets, I took on a new challenge to #3Dprint pre-linked objects - that is, objects that are linked at the time of printing. Printing some snap-together bracelets is what really prompted this - I was looking for an alternative design which might print well in metals (through Shapeways, for example), and didn't think snap-together parts would work in metal.

This round of experiments ended up being 6 models. I've outlined each below with my thinking at each stage. All of these designs focus not only on getting a strong-but flexible connection, but also on a structure which would print well the first, and every time on a 3D Printer. That meant parts which build up from the bed at an angle or lay flat. Although most printers can handle some "bridging" (parts which are suspended practically in mid-air across two other parts), I avoided these to make the printing simple, and, frankly, for the design challenge. I also didn't want to have support material to clean up (cut away) after printing. You'll notice in each model that the linkage parts always angled up from the platform, which helps to make sure it can print.

Model 1: Inter-lock

This first experiment was actually quite successful in some ways. The linkage parts of each individual object start out being printed separately from it's parent, but then slowly angle toward the parent object until they connect on the top layers. with just about 0.3mm between the interlocking parts, the printer handled this well and didn't bind the objects together at all. When removed from the bed after printing, the parts moved independently as designed. There was one major flaw - they only moved in one direction - up. there was no flex in the downward direction, since the links were horizontal and restrictive in the vertical direction. This would be an interesting design for links in which you WANT to restrict the motion.

Model 2: Two-Part

After that first experiment, I went back to thinking that independent parts might be better after all - as long as I could come up with a design which made it super simple to put together while still hard to get apart (so linked objects don't fall apart). This first attempt was mostly a complete failure. The parts would link together with lots of twisting effort, but the freedom of motion wasn't there and there was simply too much linkage bulk, and the links were sharp-cornered. I thought it was worth tweaking this one more time to see if it could work...

Model 3: Two-Part Curved

I took the prior model and used curved connectors rather than rigid right angle connectors. This one was super simple to connect, but also way too simple to take apart. The links were also much less likely to injure people with the curved smooth links, but that wasn't enough benefit to keep going (but a lesson for later). There's something interesting here for another whole set of experiments - but I also realized I went far afield of my initial goal - linked parts. I really didn't want independent parts that had to be connected. Back to the drawing board (literally).

Model 4: Interlock Curved

I took the idea from Model 1 and tweaked it to see if it could be made more flexible. This one has more of a chain-like feel, but with multiple connection points - sort of like a double chain. The linkage parts are oval shaped in an attempt to keep them more flat than circles would be. This unfortunately restricted the vertical movement more than I had hoped, but it was workable - and the horizontal "bend" (laterally to the left and right) was appropriately restricted. This was a good design for things like bracelets, which you want to bend vertically to wrap around your wrist but don't need/want them flexing laterally too much. A little tweaking on this one would yield a great result, but I wanted to try other basic designs which were less complex.

Model 5: Loose Link

To get a much more flexible, chain-like connection, I tried a single link design. The concept of avoiding bridging, and using angles to make printing more straightforward, very clearly influenced this design. This one was quite successful - gave me a very strong link and printed without binding at all. The main downside on this design was the "stickiness" of the right angles in the link. While the parts move freely as a chain would, the edges are quite sharp, a bit bulky, and they tend to get stuck in each other's hard corners and don't move as smoothly as desired. Fixing that main flaw was the focus of the next design.

Model 6: Loose Link Curved

This final model was an adaptation of the prior "Loose Link" model, but with two main changes. First, I curved all the angles - using the "Fillet" feature in Autodesk 123D Design to soften every sharp angle into a curve on both the vertical and horizontal loops. Second, I simplified the horizontal loop to start angling up directly at the base of the object rather than first coming out flat on both sides. You can see this clearly if you compare the images of the models from the prior model and this one below.

You can probably tell that all these models above are simply tests and not actually useful - but I expect to use that final model as a method to link parts for kids crafts, jewelry and other models.
If you want that final model in .STL or .123D format, or any of the other experimental models, ping me on twitter!