Welcome to the community.
I am not going to address your filament calibration settings directly, but I can offer some broader troubleshooting guidance that applies not just here, but to filament tuning in general. In short, you’re on the right track but there are some techniques that may help which I will address later.
I often say: “If you can master filament calibration, every $10 spool can perform like a $35 spool.” The point is that when filaments are evaluated under controlled, repeatable conditions, the differences within a material class are usually smaller than marketing claims suggest. As I noted in a post here last year, many filaments share effectively identical base formulations, with differences largely limited to branding and packaging.
For example, Bambu does not operate its own filament foundry. Prusa, by contrast, produces its own filament and includes a factory process ticket with each spool that can be scanned to see its production history. That does not automatically make it better filament, but it is an interesting level of transparency.
Spaghetti
Based on what you shared, spaghetti in cases like this is almost always caused by separation from the build plate. That’s relatively easy to troubleshoot: either the filament separated because the build plate temperature/filament temperature was too low, or the build plate was dirty. Since you are printing in the correct temperature ranges within ±20 °C, my money would be on the build plate not having been cleaned properly. The most reliable method is to use very hot water and clean it with dishwashing detergent like Dawn (US/Canada) or Fairy Liquid (UK/EU) - it’s the same soap under different names. That method will completely degrease and remove material that is interfering with first layer adhesion.
But don’t take my word for it, try it for yourself. Right-click in the slicer and select cube primitive and size it to 200x200x0.2mm. This will produce a 1 layer filament over the bulk of the plate - size to preference if you want to test just a portion of the plate. If you don’t get a complete coverage and there are voids in the material, there’s your smoking gun. Here’s an example of what that looks like where I deliberately contaminated both a textured and smooth plate. Texture does not reveal fingerprints but smooth does, however, the first layer test reveals the problem immediately:
Troubleshooting technique and calibration.
Forget the models you download unless it can prove that it has the correct Gcode setting. That would be either a GCode or 3MF file - STLs only contain geometry. Orca generates those on the fly. Forget about K factor tests in Bambu Studio, that is a blunt instrument.
Here’s an example of what I mean. A temp tower generated by Orca where you can see the changes in temp VS just an STL model called a temp tower that was downloaded. Once you flip on the Temp indicator from the drop-down preview menu, if you don’t see a temp gradient, all you have was a pretty lump of one temp plastic which tells you nothing.
Similar indicators are available for max flow rate too. There you would use the speed view.
Here’s a good set of tutorials on proper calibration - they are the Orca Tutorial, which is mostly up to date and Butter Pocket Prints.
Although the Butter Pocket Prints videos lineked below is a bit older, it’s still relevant and I found that his methodology is easy to follow. There are other YouTubers who also cover calibration but I think his videos are clearer and shorter. Note: Bambu has cut off access to Orca if you are running firmware after 1.08, however, you can still output to GCode file and move it over manually via an SD card. Pain in the butt, yes but still better than Bambu Studio.
The forum software appears to be acting wonky when I posted this so if the YouTube links don’t come across correctly, here are the manual text for copy and paste just in case.
https://www.youtube.com/watch?v=5CVq6DycUOE
https://www.youtube.com/watch?v=ymywqch6R8w
https://www.youtube.com/watch?v=02mLDrxEpwQ