Bands across single layer print
I have been applying knowledge gained from the Institute to improve my prints on my Ender3 V2. I have installed a BLTouch and Jyers Marlin firmware which has had me running in circles. I am presently using 3 x 3 Autobed leveling which has enabled me to print off a partially successful single layer test (200 x 200). A few artifacts to work on and a thin spot on one edge so needing more leveling work. My question however concerns the bands or striping across the print. I surmised that it would be due to a fault in the Y axis carriage and, sure enough, a quick dismantle and I could feel a notchiness in the carriage movement. This was cured by reducing the tension on the toothed belt so it ran smoothly but, as can be seen from the attached pic, the stripes are still there. Any suggestions as to what I should try next?
Thanks in advance
Do you see the banding on the sides of the print too (if you were to print a large cube or cylinder for example)? If not, then it suggests a problem with the bed (maybe uneven surface) or a problem in the levelling (check if the leadscrew moves while printing the bands - if it jerks up on the bands then it suggests an autolevelling glitch [bad points matrix for instance]).
If you do get it on the sides too, then an axis is to blame. Check the frame extrusions and plastic wheels for dirt and debris, then adjust the eccentric nuts on the wheels until they are snug on the extrusions, but not too tight. This will stop the moving parts from wobbling. If you don't know how to adjust the eccentric nuts, this video explains it - https://www.youtube.com/watch?v=P5HJgIrgGXc
If the artefacts continue to persist even after adjusting the nuts, you might have a damaged frame (bent bearing surface) in which case you'd have to look into a warranty repair or a replacement frame part or carriage or wheels.
Let me know the results of your testing!
Thanks for the prompt reply.
I have visited all of the points that you mentioned, to no avail. No banding on the side of prints, no movement of the leadscrew, (I have tried on both 3x3 ABL and 10x10 UBL) Frame is all tight and carriages tightened correctly and clean. I have printed off another single layer test but with an included 20mm high structure which is clean. The lines are in exactly the same place, at 64 mm centres which suggests a 20mm diameter wheel may cause it. The Y axis runs on 24mm diameter wheels in a groove so is that 20mm actual?
I believe my next course of action is to replace the wheels on the Y axis to see if that sorts things.
Since the walls are OK this is unlikely to be a wheel or X/Y carriage or belt or pulley issue - if that was the case, you'd be seeing a continuation of the line up the side of the wall, like so -
At this point I suspect either the bed, or the file you're using may also be the culprit. Can you send a photo showing the print on the bed (before removal) with enough of the printer visible to get an idea of position and orientation on the bed, and a photo of the bed surface after removal. Actually if it's not too much trouble, a youtube video showing a minute of the printer printer putting down this first layer (especially the part where it makes the line), would be even better to help more quickly zero-in on the cause.
Two more questions - if you run your finger over the bed in that area, do you feel a raised line that coincides with the band pattern? And does the printer make any particular noise while the nozzle is making those bands?
Could you also attach a copy of the gcode file you're printing?
Looking forward to your feedback.
Thank you for your time.
I have taken a close up of the wall as there is a slight distortion in the layers at that point. Also attached, the print on the bed, the bed without the print (two white marks on left hand edge to show where the bands occur) and a close up of the bed in that area.
To answer your questions, I don't feel anything on the bed in that area and likewise can't hear any change in sound as the nozzle crosses the band. As a further test, I have turned the glass bed plate through 90 degs and printed a Io3dp Centre.stl offset across the band. The distortion appears in the same position on the bed (not the glass).
I've never done a YouTube video so I hope it works.
Thank you for your continuing assistance.
Excellent troubleshooting- by spinning the glass 90 degrees you've ruled out a problem with the surface of the glass, which narrows it down to a problem in the Y axis (which moves the bed). A quick and simple test now is to remove the motor pulley so there's no resistance on the axis, then slide the bed back and forth and feel whether it snatches or catches in the location where the bands are. Without the force from the motor, any hiccups in the motion will be much more apparent.
Check if the bed wobbles up and down, then apply slight downward or upward pressure on the bed while sliding it back and forth - see if it suddenly sinks or rises at a position corresponding to the band.
Let me know what you find 🙂
OK, so here's what I have done. I removed the Y axis extrusion from the frame then removed the belt from the carriage. As you suggested, this made finding the sticking point(s) much easier. I then removed the Y carriage and mounted it to the bottom of the rail such that it was using a different portion of the 4040 extrusion. No change. I marked the rail where it caught, and also marked the wheels. I then slid the carriage along to the next point where it caught. The wheels had done exactly one revolution. I therefore switched all four wheels with some off the X axis. Still the same. I turned each wheel out of sequence in turn and isolated the one which seemed cause the problem. I changed that with one from the Z axis. A definite improvement but still there. I tried the carriage on the side of the rail. Still the same. I closely inspected the rail but could not see anything wrong.
This where my attempt at logic deserted me. The catch points on the rail were always at the same place, whichever face I used. The catch points were exactly one wheel circumference apart. At this point my brain blew a fuse and gave up.
Good going! If the catch-points repeat one wheel revolution apart, the things to look for would be the wheel circumference, and the bearings in the middle. The wheel itself might be oval or out-of-round (eg. flat spot on the wheels), or there could be dust particles stuck on the wheel or the extrusion. Or the extrusion itself could be bent slightly in that area. Strong light will help in making defects more visible (even if it's just the flash built into the phone). In terms of bearings, they could have a sticking point or be out of round, or the nuts could be loose and spinning inside the holes in the brackets. I'm sure you're familiar with how the eccentric nuts work on the printer (they have a hole drilled off-center, like so -
so they cause the plastic wheel to press into or away from the extrusion when turned). If loose, the nut would be free to rotate and would cause a shift that repeats once per revolution.
When you replaced the plastic wheels, did you replace the whole wheel including bearing, or just the plastic part? If just the plastic part, look closely at the metal part (the bearing and nut). Make sure the underlying self-locking (nylock) nut is tightened and not wobbling, and make sure the eccentric nut (between the bracket and the wheel) is secure and not sliding.
Also, aside from the wheels, make sure no part of the bed is rubbing with anything else under the printer. Things like cables, parts enclosures etc - just to cover all the bases.
Try and find any problems listed above, and if nothing is apparent, another video demonstrating the catch point might help - try and show also the wheels and the extrusion.
Did you encounter any up/down looseness or repeating changes in feeling in that direction? It seems as though the filament stops sticking to the bed on those bands so an up/down deflection might be partly to blame.
After doing all the above, reassemble the printer using the combination of wheels that feels the best, then tighten all the eccentric nuts on each axis as recommended in the video I sent earlier, level the bed, and try printing a 20mm calibration cube and measure its dimensions. With luck the act of disassembling, upsetting and reassembling might have removed the effect of the defect. If not, the dimensions of the cube will reveal whether one of the axes is compressing or expanding dimensions.
If nothing else helps, next thing I would suggest is to buy some wheels and a new frame part to rule it out definitively. This person had a similar problem that was traced to the rail, and posted an instructable showing how to replace it - https://www.instructables.com/Fixing-an-Ender-3-Pro-Y-axis-Rail-Issue/
As a retired engineer, I have examined all the mechanical options and agree with you that there is a vertical deflection at the point of the bands. This can be clearly felt whilst moving the carriage along the rail. So the fault is definitely in the wheel / rail setup. The bands always being in the same position of the rail confirms that the rail is at fault. However, the fact that the bands are exactly 1 revolution of the wheels apart confirms that it is the wheels at fault. At this point, logic deserts me as I cannot ascertain which item is to blame although it suggests that I would need to replace both. After trawling the web, both UK and China suppliers, for hours last night I have decided that the ideal solution would be to top the existing Y axis extrusion with a linear rail to alleviate both issues. What do you reckon?
What a coincidence, I'm a mechanical engineer too! That might explain why you tried all the different orientation/wheel combinations I was thinking about before I actually wrote them down - engineer instinct kicked in hehe 😆 .
Since the situation is that the grooves are always on the same spot relative to the rail, I would say the rail is the culprit. I would further suggest that the distance being one wheel rotation apart might be incidental. Perhaps some foreign object got stuck to the plastic wheel at some point, and it got crushed between wheel and extrusion, causing damage the rail over the course of 2 or 3 rotations.
If you'd really like to rule it out, see whether you can disassemble that frame member and rotate it 180° (front becomes back and vice versa) so that the suspected damaged part shifts to a different part of the bed - then confirm the bands still correspond to the old location on the rail. I.e. if the bands are an inch from the front of the bed now, after rotating the rail they should migrate to a position an inch from the back of the rail (assuming perfect symmetry).
Linear rails would be a massive improvement, they don't require periodic adjustment of eccentric nuts like wheels do, they tend to wear out a lot slower than the plastic wheels, and they tend to have better accuracy. Only thing is the kind of rails that are affordable for a 3D printer tend to be Chinese made products of medium to low quality. That means if you're unlucky you might get one that's not quite straight or that has sections that don't slide as smoothly. On the whole though that shouldn't significantly affect your print quality. A genuine high-quality linear rail might easily cost more than the whole printer hehe.
Disadvantages of rails are that they're slightly noisier (you can hear the ball bearings click faintly if you have really quiet motors/drivers), and they require periodic lubrication with grease (every several months though, about the same periodicity as you would grease your leadscrew). And of course cost, which is why they weren't on the printer to begin with ;).
I'd definitely opt for that option, it's an upgrade over the stock plastic wheels and should in theory address the band problem since it replaces all affected components.
I have decided to go the linear route so just quickly reassembled the printer whilst I await delivery. I am using UBL on a 10 x 10 grid, created mesh, saved and loaded, then a quick 200x200 single layer print.
My brain is now officially blown! What the heck?
Interesting pattern! You seem to be extruding too far from the bed - your entire first layer has barely stuck to the bed, as evidenced by the adjacent diagonal threads that haven't fused together to form a coherent layer. Meanwhile you may be overextruding also as the filament is bunching up in places. On the bright side the bands seem to have gone (I was half-expecting this after such an overhaul - sometimes things tend to work themselves out after tightening all the screws one more time).
You'll have to mechanically level the bed, calibrate the probe Z-touch height, redo the auto levelling, verify your e-steps, and print a calibration cube and look at the top layer to estimate whether your extrusion multiplier needs tweaking. Let me know if you need any help with any of these.
For the first layer, I like to use 100% extrusion width (0.4mm with a 0.4mm nozzle), 100% layer height (0.2mm with a main layer height of 0.2mm), and a speed of 50% for PLA (which works out to about 30mm/s) or 30% for PETG (which works out to about 15mm/s).
The nozzle height should be adjusted so that after autolevel and the rest, you can slide a piece of paper between the nozzle and the bed when you set Z=0.1mm but it grabs the paper if you lower to Z=0.0mm.
I've found one problem which is that the Z Offset was wrong. originally 2.72, now 3.23. I didn't think that that was a figure that changed after first install of the probe. So looking better now.
My extrusion rate is at about 95% so I left it at that. First calibration cube had a big foot as I had a bed temp of 70c so I reduced it to 55c and tried again. I would now appreciate your comments on the various elements of the cube.
Still spread a bit on the first layer, strange patterning on the inner bottom and more patterning up the back of the cube. Not sure what to look for in the top layer?
Dimension are 19.98 across in both directions and 19.4 in height.
And thank you for the continuing help, it is much appreciated.
Your X and Y axes seem to be quite well calibrated, so we can stop focusing on those for the time being.
I actually meant for you to print the classic solid calibration cube 😊 , and show a closeup of the texture of the "Z" surface (a lighter-coloured filament would help to see the details better). With the hollow-walled one you printed, you can still evaluate extrusion multiplier by taking some accurate calipers or a micrometer and measuring the wall thickness of the top three layers or so. Compare to what you have set in the slicer (for example if it's 0.4mm in the slicer, you're measuring a thickness of 2 walls so the reading should be 0.80mm). Then in theory you calibrate the flowrate depending on the reading - eg if you measured 0.78mm with your 95% flowrate then your new corrected flowrate would be 0.80/0.78*95 = 97.4%.
That's the theory. In practice I've found wall thickness measurements quite finicky and a little variation in pressure on the calipers can swing the reading quite significantly and throw your corrected extrusion multiplier completely off. Plus the fact that layers are never placed perfectly upon one another in perfect alignment means you'll always measure the wall thicker than it really is (hence why I wrote above to only measure the top few layers, to minimize the influence of the misalignment). This diagram illustrates a cross-section of a print:
The red lines indicate the jaws of the caliper. You've got to click on it and see it in full scale to see what I mean - the jaws will be pushed apart by the one layer that's a little more offset than the others. In essence, the little misalignments in the layers would make the jaws settle at effectively the maximum wall thickness of the layers sampled rather than the average wall thickness (which is what we want), so the measurement tends to be inflated, and the fewer layers you sample the more accurate your measurement, to an extent. This is why I personally dispense with the hollow single- or double-walled calibration cube and just print a solid one and visually inspect the top to see whether the extrusion multiplier is too high or too low. Here's the texture you're looking for -
Looking at your print, the worrying thing is that the bottom layer has two different textures - one half is practically perfect and one half looks underextruded. Your sidewalls also show the pattern characteristic of underextrusion. The way it's intermittent suggests a problem with the extruder. Could you confirm that the extruder gear isn't slipping and that it's not clicking and bouncing back during printing? That would cause intermittent underextrusion as the gear slips or clicks or grinds occasionally.
Let me know how the extruder fares and how your solid cube looks 🙂
The walls of the cube measure 0.75mm so some under extrusion? I had printed off two of the cubes and both had the same texture difference on the inner bottom.
I have now printed the full cube and had no clicking or jumping from the extruder at all. The cube measures 20.04 across width and length with 19.3mm height along with the elephant foot syndrome. It is printed in PLA+ at a bed temp of 55c.
I have also run off another single layer test which, as you can see, still has the bands and also has the weird patterning and roughness. Your advice on this would be appreciated. Incidently, I have ordered a linear rail so will fit that on Thursday.
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.