stmcculloch/arc-overhang

Tests to find optimal print settings

Opened this issue · 1 comments

stmcculloch commented

Initial comment by @nicolai-wachenschwan. Moved here to keep things on topic.

I thought about the data gathering today.
How do we understand the adhesion process and its needed conditions + limits?
=>If we do so we could taylor a specific algorithm that performs ideal in various (specific) usecases.

What I propose to find the Process Limits:
we could measure the maximum printable radius without any support aka the point where the support can't support its own weight. This would allow easy user warning if the print is to ambitios and is likely to fail.

To quantify the print quality we could measure the maximal successfully printable radius, depending to the other factors. like speed, width,...=>If we achive rMax>100mm we can save us a lot of algorithmic hassle in arc generation.
=>Supporting the arcs with a middle beam.
Suggested factors:

  1. extrusionMultiplier,
  2. ArcWidth(aka Overlap between individual arcs),
  3. nozzle_diameter ,
  4. width
  5. print speed
  6. print direction (I noticed failures at some specific angles, maybe because of cooling)
  7. Temperature
  8. ?

we can choose up to 8 factors to analyse with only 16 prints, executed as in the current idea each print will take 2-6 hours.
What we will get is a definitive model of the process, so we can tell for example: Temperature does not matter much, as the effect is very small.
Possible Test Geometry:

DOE_TestGeometry_v1

What I can do: prepare the gcode-files and do the numbercrunching to extract the effect strength etc. (my "Minitab" Licence expired but my Matlab should do the trick: Multivariante Analysis of Variance).
What I can't do:
Currently I dont have time to do the prints myself :/
Also I cant do the pretests to determine the range of the parameters (example:width:5-50mm).

What do you think of this approach?

PS: I modified the script so it can generate the gcode for these tests.

Originally posted by @nicolai-wachenschwan in #10 (comment)

stmcculloch commented

@nicolai-wachenschwan More testing is definitely needed. I am happy to do some test prints.

For a slicer implementation, we definitely need to figure out the process limits, and provide some good, conservative default settings so it will work for most people. I would also want to test:

  • min/max layer height
  • compatible materials. (currently only tested with PLA)