The Great 3D Printer Experiment: Results of the First Test

Continued from the Prelude of The Great 3D Printer Experiment.

As a Soldier, I’ve relied on inductive reasoning /causal thinking for the majority of my career, rarely questioning doctrinal application. Once I finished Company Command, I found myself thrust into a more cerebral, deductive realm of the service. Naturally, such a change was repellent; questioning things that were explicitly written for decades put me in a very uncomfortable state, and yet, post command jobs required me to do just that. So, ironically, I conformed to the mandate to dissent. The discomfort eventually subsided when I saw the value in challenging convention. Now I question most assertions that come my way, respectfully of course, and especially when the assertion is made without evidence. I now get to continue this skepticism within the maker movement, which hopefully, is reflected in the results below.

In the prelude of this 3 (now 4) part post, I developed a framework that would help me figure out what the optimum slicer settings for an Anycubic I3 3D printer were for a specific cuboid object. Since publishing that framework, I gathered the data for 9 print runs. For a recap, each test is listed below:

Part One: Single Factor
Independent Variables
Run # Layer Height Print Speed Print Temp Outer Shell speed
1 (base) 0.2 50 215 30
2 0.3 50 215 30
3 0.1 50 215 30
4 0.2 70 215 30
5 0.2 80 215 30
6 0.2 50 210 30
7 0.2 50 200 30
8 0.2 50 215 10
9 0.2 50 215 20

I wanted to isolate the settings in CURA (Layer Height, Print Speed, Print Temp, Outer Shell Speed) during this part of the experiment to not only learn more about my machine but also produce the optimal print of the cuboid test object. During the the 9 print runs, I encountered a few problems:

  1. Vertical loading of the filament was causing strain on the extruder resulting in strange anomalies that were consistent throughout the print. I to repeat half of the prints to get more consistent results by horizontally feeding the filament into the extruder.
  2. The bed temperature was too high (70 C) resulting in a combination of adhesion issues (not adhering to the blue 3M tape, and slight deformations on the initial layer of the print). To fix this, I simply turned off the bed temperature.
  3. I noticed that print speed was not changing the expected time for completion (within CURA). After some reading, I learned that print speed was being overridden by outer shell speed due to the properties of the cuboid test object. Since the print speed tests occurred before the revelation about vertical feeding the filament, I decided not to reprint them.

Once the problems were fixed, I gathered up the test prints and enlisted the help of my son. We lined up the results:

initial-test-prints

We then developed a simple system for determining the quality of a print. Each sample was evaluated based on 3 unweighted subjective criteria: sharpness of corner, number of anomalies, and smoothness of face. Each criteria was assigned a number from 1-5; 1 being the best quality and 5 being the poorest. The sum was taken of all three combined criteria to determine an overall score (3-15). The best possible score is 3 while the worst possible score is 15. Jack, did an excellent job of assigning numbers to each of the criteria, though, I had to translate his findings by reversing the continuum. Results are below.

Independent Variables
Run # Layer Height Print Speed Print Temp Outer Shell speed  Score
1 (base) 0.2 50 215 30  8
2 0.3 50 215 30  12
3 0.1 50 215 30  4
4 0.2 70 215 30 N/A
5 0.2 80 215 30  N/A
6 0.2 50 210 30  6
7 0.2 50 200 30  5
8 0.2 50 215 10  10
9 0.2 50 215 20  5

The ranking is depicted in the picture below. The top of the stack reflects the best score and the bottom represents the worst/null. (jack wanted me to point out that the results look like an awesome sword).

ordinalrank

A layer height of 0.1 produced the highest quality print while a layer height of .3 produced the worst. Interestingly, the base print which is derived from the Anycubic recommended print settings produced a sample that rated 5th overall out of the tests. Closeups of the two extremes are posted below:

point1layer

0.1 Layer Height setting

point3layer

0.3 Layer Height Settings

So what does this all mean? As chunkystevo pointed out in the thingiverse forum, these findings will help determine the optimum settings to print out small, hollow, cuboid objects and nothing more. Even so, I still have to complete the taguchi set of runs before I can form any reasonable conclusions. I can say that  I am inclined to agree with chunkysteveo, not only because his reasoning is sound, but because his methods were already confirmed in other tests outside of this experiment (The boy wants a suit of Iron Man armor.. how could I say no?). Taking his advice, I am going to expand the experiment to include changing the cuboid to a more spherical object, after the taguchi series of runs (of course). Despite the setback of having to do some more tests, replicating the same series of tests on a more spheroid object will hopefully help me produce a more definitive conclusion at the end of this experiment. There is much more to discover and more will be revealed in the next post. In the mean time, I’d like to leave you with one additional print that I made by reducing the layer height setting to .05 mm. The sample, I think, speaks for itself:

point05layer

For extra measure: .05 Layer Height Setting

To be continued in “The Great 3D Printer Experiment: Results of the Second Test.”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s