Just out of curiosity. What filament brand are you using? I ask because when I loaded your 3MF file it defaulted to Elegoo Rapid PLA+
I could not reproduce your plate examples because your profile was using the smooth cool plate which I don’t have. For the sake of consistency and to remove a variable from the test, have you tried the original PEI plate that came with the printer? Temperature is a huge factor in your results and I am wondering if that isn’t the culprit here which I will explain next, I have other theories but let’s stay focused on temperature.
Also, what advantage does the cool plate give you in a model like this? I am curious?
I’ve been trying to replicate your results and I’ve only been able to come close twice. One of the things I was able to determine so far is that the size of the model and orientation of the model have an effect. This leads me down the path to suspect temperature as a factor. The larger uncut model with greater mass was the only way I was able to come close to replicating your results.
What I can’t do is isolate the effect using portions of the model. That’s a problem because at 0.28, the fastest speed, it takes 1:30 minutes to print. That’s not a practical model for rapid experiments so as is my standard practice, I cut the model down to a size that should allow for reproducing the effect and I found that any reduction in size reduces, if not eliminates the problem. Again, this points to thermal mass as part of the issue.
Here are my findings so far. Note that I tried to reproduce your tests but the 3MF file only had a single model and did not have your test scenario, so I had to guess as far as what infill settings you had. When you reproduce your affect the next time, if you could do a Save-As and append the filename with “-exp #1”, “-exp #2” and so on, this would adhere to sound scientific method and allow us to freeze the settings and share exact test cases. But here is what I have so far.
Test conditions
- Unless noted, all layers were 0.28
- I used two different types of Textured PEI. Look in the 3MF file to see the plate temps used.
When I placed objects in the pattern your earliest post suggested(again, I need your actual 3MF file in order to faithfully reproduce the test scenario) I was only able to get marginal and inconclusive differences.
Here’s the file for inspection purposes. There are 3 plates creating scenarios that vary the test conditions. Again–it’s very important to get your “exact” 3MF file so that we are testing the same thing.
Vibration test test #1.3mf (649.7 KB)
This was using Elegoo Rapid PLA+ using the calibration settings in your 3MF file.
I repeated the experiment using the exact same file plate #1 but this time I used my own calibrations settings which I had dialed-in for my use. Note the “C” is my calibration and the “UC” is yours. I was able to reproduce some of the affect(Archimedean) but not exactly with my calibration. What does this prove? Nothing, but it at least allowed me to create a test case.
Note that these two examples were at 0.20 layer height. Note that these models were cut down to reduce the print time to 30 minutes.
For testing purposes, I then switched to 0.28 layer height to speed up the testing frequency. I also switched to 3DJoHor filament because I was running low on the last spool of Elegoo I had and wanted to have enough filament in reserve for multiple experiments without running out, plus the black filament produced better contrast for photos.
Only successful reproduction of the problem
This was the only successful test case where I was able to reproduce your results using 0.28 layer and black filament. The 1,1 is an indicator of it’s position on the plate. This was the full model size.
From there, I conducted the following experiments. The conclusion I will note are not scientific but the ALL point to thermal mass issues.
All objects were printed at the same location on the build plate but are arranged here so they can be seen together. Note that grid location 1,1 indicates center plate. I did this purely for the sake of later moving the model.
Sample #2 is the most notable. I was able to correct the problems with a minumum of changes. Here is what was changed that made all the difference.
- Orienting the model in a perfect 90 degree orientation. Your model was skewed.
- Cutting the model down to the minimum height so that it included just enough of the center knob as well as the entire width and length of the block mass just before the model filet started to curve.
Temporary conclusions
I believe the problem is related to thermal mass and can be eliminated simply by tuning or model orientation. I am unclear why Archimedean Cords are being used but they don’t lend themselves to this model.
Other possible theory.
As I was performing these tests, it occurred to me that I saw this exact same “Positional Plate” issue about a year ago with Bambu Silk. I noticed that when I created a cylinder primitive, I got very different results with cylinders that were just 20mm apart. Here is what that looked like:
I’ve posted these before. These two models were examples I used. They were printed at the front right corner near the door. Position 2,2 in my coordinate system above.
So how did I fix it you might ask?
I dried the PLA filament!!!
________________________________
I don’t know how much further you want to go in exploring positional difference but before I continue, for the sake of speeding up the process, I would ask that you create a much smaller model that takes between 20-30 minutes to print. I’m willing to assist but not at 1:30 minutes per model.
Also, I must insist that if we are to collaborate, you follow rule number #1 of troubleshooting theory; “change only one variable at a time”. That means if you are going to test an object, you must use only that object and clone it on the same plate. Also, to verify that nozzle movement is not an issue, repeat the experiment by turning off the models one at a time and printing them in isolation and together. Let me know if that makes sense or if I have to clarify.
