Bands across single layer print
Oops, forgot something!
Now that I see the top of your cube, I can confirm that you are over-extruding. I base this on the texture of these parts of the cube -
You can tell that the plastic was a little too high after it was placed by the nozzle, and then the nozzle flattened it a bit on subsequent passes, leaving a "valley" of plastic in every nozzle path.
So first off, you will need to lower your extrusion multiplier. Start with a 3% reduction from where you are now (so if the one in the pic was printed at 95%, go down to 92%), and re-print the top of the calibration cube. If you wish to optimise it fully, you'll need to do it empirically with several cubes. To start with, keep lowering EM until you see at least one small gap appear between the diagonal lines anywhere on the top surface, then bump it up by 1-2% until your prints reliably come out without any gaps on the top surface.
The pattern you're seeing in your one-layer print is caused by a similar mechanism. The extra plastic has nowhere to go under the nozzle, so it goes sideways out of the nozzle, then gets flattened in the next pass and gets a rough surface. You should however double-check your nozzle-to-bed distance, because this can be exacerbated by printing too close to the bed. So, after tuning the extrusion multiplier via the iterative process described above, make sure your first layer height is set to 0.2mm then print some bed levelling squares (eg. https://www.thingiverse.com/thing:2789086 ). Measure their thickness at their perimeters (not the center, to avoid any effect due to over/under extrusion), and report back your thickness measurements.
Based on those measurements, we will know precisely how your levelling needs to be adjusted and we'll take it from there. Don't worry too much about the single-walled cube, if the STL had walls under 0.8mm thick you'd be entering the slicer's "thin wall" algorithms which are a whole other section. Best way to print a hollow cube is to take a regular solid calibration cube and set infill to 0% and top layers to 0. Then your walls are printed with normal settings and the slicer doesn't try to print a 0.35mm thick line with a 0.4mm diameter nozzle 😆 . But like I said earlier, a hollow cube is a very tricky way to try to tune extrusion multiplier because wall measurements tend to be erratic - the method of looking at the top layer of a calibration cube is much, much more robust.
Looking forward to your feedback 🙂
Sorry that my reply has been a bit delayed, I have had a number of dark days to cope with. Back to the land of the living hopefully now.
I have followed through your instructions above and believe that I am getting close. The calibration cube is looking good (at least to me) and I have managed a nice clean single layer print now that I have fitted the linear rail on the Y axis. I am still having some issues with the small squares although I have managed a few sufficiently to measure them. Layer thickness varies between 0.17 and 0.20 so not too bad.
I believe that I now need to work on the UBL as the Z height does not seem to be reliable. I have found Marlin's instructions for UBL and realised that installation is not as simple as I have done, so I will now reinstall it by their instructions and see if it works better
The one remaining question with the cube is the foot which I have been unable to prevent the spread on. I've tried the bed down to 40c, I've tried under-extruding, I've even tried printing with zero bottom layers in the cube, all to no avail. The cube measure 20 x 20 x 19.3 mm. Anything else I should try?
Thanks again for all the help.
Nice progress! That first layer is looking great! 🙂
For the elephant foot, you've tried all the common fixes (lowering bed temperature and extrusion), so here's what I propose:
1. Make sure that the first layer height and width are 100% and 100% respectively.
2. Make sure that the mechanical levelling is as good as you can possibly make it before adding UBL to the mix. This is because of something called fade height. Check your printer's LCD menu (generally under bed levelling settings) to see what this is. Basically, the way it works is that the amount of auto-bed-levelling-correction starts at 100% for the first layer, then is gradually reduced as the Z-height increases, such that it reaches 0% by the fade height. Past that point, the z-axis no longer makes any movement for autocorrection - you'll be relying exclusively on the printer's mechanical straightness to make sure the print doesn't come out lopsided. Of course you could make the fade height 0 to turn off the feature (so autocorrection continues at 100% strength all the way up the print) but really when you're way up and away from the bed in terms of z-height, it's not worth the extra wear on the z-axis components to keep them autocorrecting like that.
This image shows how it works -
As you can see, in layer 1 the nozzle tracks the imperfections of the bed; it slowly straightens out, reaching straightness by the fade height (in this case 10mm). The image shows why it's important to have good mechanical levelling regardless. If your build plate isn't slanted, you won't end up with a lopsided print as illustrated in that image.
3. If all else fails, there's elephant foot compensation built into most slicers (tell me which one you're using for more specific guidance). This is a feature that masks the problem by shrinking the first layer of the print (i.e. scaling it down in the X and Y directions). The idea is that by setting a correction of say 0.2mm the first layer (or several layers) is/are printed smaller, the factors that are making your first layer expand will still expand it, making it balloon to the perfect size, and ultimately the cube ultimately comes out perfectly straight.
This is however a bit of a hack. Before relying on that, I would wait until you've got the UBL set up completely correctly since the correction could be messing with your first layer (in fact, if it is making the nozzle too low you might experience that problem).
You said that you're questioning the accuracy/repeatability of your probe measurements. Fortunately Marlin has a routine that will test this for you. It's called a probe accuracy test and it's started by sending M48 to the printer over serial. See syntax here - https://marlinfw.org/docs/gcode/M048.html . You have to have "verbose mode" turned on. The reply from the printer will give you the mean and the standard deviation of several taps of the probe. If there's a wide variation in the touch height between subsequent probing cycles, your deviation will be a high value. This 1-minute clip shows it in action - https://www.youtube.com/watch?v=IlZxrBeEGVI
Your printer might also have the probe accuracy test in the LCD menu - in that case it looks something like this - https://www.youtube.com/watch?v=fqQGqukg_jU .
Keep up the good progress! Looking forward to your feedback 🙂
We’re a bunch of 3D printing and design nuts. We just want to make great 3D design available to everyone.
Good design takes time, great design takes a process. You can learn this with us, to build your skill set in this rapidly expanding market.