Oh man that brings me back to my delta printer several years ago. I was just throwing stuff at the wall and seeing what stuck. There were so many variables it was overwhelming.
I’ll do what I can to help you through this, stick with it, and hopefully on the other side when we get it all working you can look back with your new experience and knowledge and go “that was pretty straightforward”.
Check out this thread I posted about getting the machine tuned up from scratch:
I’d recommend you go through the Teaching Tech calibration settings in order and start knocking out some of these variables.
On the first layer especially this is important, and you can gauge how it’s printing through the rest of the model as well:
Here’s some of my calibration cubes, hopefully you can start to recognize the signs of being over or underextruded with these:
This is the bottom, so the first layer. This was a bit underextruded because the strings of filament are still rounded and not smushed together enough in the middle:
Here’s the bottom of a better print for comparison:
So that underextruded picture, here’s the top of that. Those gaps in the corners should not be there:
This specific print was after calibrating E-Steps and with an extrusion multiplier of 1.0, and it was putting down the correct amount of filament at the perimeters. The perimeters are nicely tied together like they are supposed to be. But the skin (surface) is thinning in the middle. The cause of that was S3D’s infill overlap was too low for me. If I raised the extrusion multiplier the perimeters would be too bulgy, because they weren’t over extruded, it was just the infill and skin that was slightly under.
After changing the infill overlap, up from 15% default, to 30% (I just doubled it as a guess) this happened:
You can see the top surface of this is clearly overextruded because the nozzle is dragging through plastic where the filament had no where to go and was shoved upwards. Again, the perimeter outlines are all good, so this is just infill and skin issue, so I dropped it to 23% (I picked a number halfway between 15 and 30, so 22.5, rounded to 23) and got this:
This looks pretty good to me. After more test prints with other models I ended up settling on 21%. Clearly 15% to 21% seems like a small change, but it has big consequences.
The test cube here: XYZ 20mm Calibration Cube by iDig3Dprinting - Thingiverse. The picture shows clear underextrusion of the top layer because you can see through it.
This is the test cube I was using later on to fine tune things. It’s a fairly difficult cube to print perfectly, which is why I like it later in the process, early on I just used the Teach Tech models with simple cubes. Small, fast, uses minimal filament, and can test many things simultaneously, but on separate parts so it’s easy enough to keep all the variables separate.
Many MANY times this happened, and I’m sure it will to you too if you print this:
What ended up being the cause here is insufficient cooling, and putting a desk fan next to the printer helps. What happens is the thin wall can’t take much force (by design, it’s a test cube) and there’s overhang as the layers go up and the circle top is closing up. Without enough cooling the PLA curls up, and then the nozzle hits it on the next layer, breaking it.
OK on to bed levelling. That’s great, I’m actually shopping for an aluminum plate right now. I tested the inductive sensor with a small piece of aluminum foil and it picked it up, so thanks for the heads up. I was looking at gluing a PEI sheet also. How thick is the sheet you’re using? If it’s more than 1mm or so you could try adjusting the height of the sensor, there’s a slot and a screw on the back of the module. Just make sure there’s clearance that it doesn’t hit the bed while printing.
I’ve coated a bed in aluminum foil for levelling, that’s actually how I level my delta - I clip onto the nozzle and foil and use the nozzle’s electrical contact with the foil as a probe. Personally for this printer I was really hoping to now have to do that, which is what the aluminum buildplate is supposed to be for…
If you are going to do manual levelling, in my opinion you must have backlash compensation enabled because otherwise it’s incredibly frustrating, and also use the Luban machine control panel with a custom step size of 0.01mm. The smallest on the controller is 0.05mm which is too big in my opinion. I believe the touchscreen isn’t doing anything for calibration except sending commands to the main controller that you would talk to directly with a serial cable, so moving the Z height up and down with gcode or the control panel is Kosher as far as I know.
Something I’m working towards is using shims to flatten the plate as perfectly as possible. Hopefully auto bed levelling will become redundant at that point, and you only need the Z height offset, which there’s a firmware command to update that (M206 Z__)
Best of luck, hope you can make some more progress. Let me know how it’s going.
Brent