Significant distortion on first layer - skew, straight lines print wonky.
I've made a test project which extrudes a single line to-and-fro, like this:
I print it, then scan the finished (single layer) print in a (high quality) flatbed scanner. I then squash the image in only one axis (no shearing, rotating, distortion or any other transformation) so that it's easier to see how straight/parallel/etc the printed lines are over their length. This squashing does not exaggerate the actual amount of deviation in each printed line, it just makes it look like it happens over a shorter line. Here's one section of that squashed scan. The full image is in the attached zip. Note that there was poor adhesion in a couple of places (at the ends of lines) but these are not relevant and can be ignored:
The actual length of the vertical lines as printed is about 200mm. The extrusion width is nominally 0.5mm. Belt tension and alignment are adjusted correctly as per https://help.prusa3d.com/article/adjusting-belt-tension-xl_401793, and the printer produces a clean homing tower.
I would consider this result to be unexpectedly bad. There appears to be a general underlying shear or skew, and then a modulating systematic pattern of deviation away from straight, with deviations exceeding one extrusion width in places. Any ideas as to why this is happening?
The source 3mf is included in the attached zip, I'd be interested to see this test repeated on other printers. You can easily see the bendy lines by holding the print sheet so you are looking at it from an edge along the lines - like you would when checking if a piece of timber is straight (this has a similar visual effect to squashing the scanned image). The 3mf has a couple of settings adjusted to improve test reliability (extrusion multiplier set to 1.05, no elephant foot compensation, bed temp to only 50C to avoid "absorbing heat").
RE: Significant distortion on first layer - skew, straight lines print wonky.
NB: This was done on a 5-tool, but I'm posting on the single tool forum as this unlikely to be multi-tool specific.
RE: Significant distortion on first layer - skew, straight lines print wonky.
Update: live chat support does not have any immediate answer or suggestion for this behaviour, hopefully they will try to reproduce the issue this coming week. It would be very helpful if other users can repeat the test and share their results. I'm using firmware 5.1.2, please include the FW version used if you repeat the test.
For interest, attached is the full-bed version - although this is too big to scan unless you have a huge scanner.
RE:
I've done several further identical test prints after adjusting different things. In all cases listed below, the test print is essentially identical - the exact same particular pattern. So I suggest that it's safe to conclude that these things are not the cause of the distortion:
- Having a approximately 1mm tilt on the bed (by misaligned lead screws) versus correctly calibrated z alignment.
- Creating a raised area under the print sheet so that mesh bed levelling is certainly changed (I used a roughly 8x5cm stack of folded paper about 0.3mm thick).
- Tensioning both XY belts to 80hz (but with non-ideal end stop alignment as per https://help.prusa3d.com/article/adjusting-belt-tension-xl_401793) versus approx 10hz difference between the belts but with ideal end stop alignment.
- With/without input shaper.
- Using phase stepping with the 6.0.0 alpha firmware.
Another observation: while moving the x-axis to and fro (via the control screen), I can see with the naked eye small movements in the Y axis.
Searching for this behaviour more generally as regards corexy kinematics, various comments suggest that this is caused by subtle changes to belt path lengths as the toolhead moves. For example, a repeating pattern could be caused by an eccentric idler or pulley. One thing I have noticed is that the belts that run along the x gantry are not perfectly coplanar with the rest of the belt path: the belts are clamped a little too high or low on the back of the toolhead. This might account for some unexpected Y movement during X moves. I plan to investigate this further.
RE: Significant distortion on first layer - skew, straight lines print wonky.
Following further discussions with Prusa support, we suspected an issue with the corexy motion system - belts, pulleys, motors etc. At this point Prusa felt that I had taken more than enough burden of troubleshooting, and did the right thing by offering warranty return to base repair, or a replacement corexy assembly. But at my request I asked that we investigate the belts further as swapping belts is much easier than swapping the whole corexy assembly. To check the "bad belt" theory, I swapped the upper and lower corexy belts - and finally there is a change to the distortion pattern. So Prusa are sending replacement belts as a next step. It's worth noting that Prusa have not seen this problem before.
Here's an animated gif before/after swapping the belts. Open the image in a new tab to see the animation. Ignore the width-scaling difference - the images were hand cropped.
RE: Significant distortion on first layer - skew, straight lines print wonky.
tl;dr: replacing the belts and one of the timing pulleys has fixed the issue 🙂
I received the replacement belts and timing pulleys earlier this week, and finally had time to do the swap. The timing pulleys were included as a precaution as logically it was not possible to determine which of the belt(s) or pulley(s) were the cause (or indeed, it could be both). While removing the original belts, and with better lighting/positioning of the printer than before, I noticed black deposits in the grooves of the right hand timing pulley. The left hand pulley was clean. So I decided to replace the right hand pulley and both belts. I reassembled and re-tensioned the belts. Got 79 hz and 84 hz with perfect racking alignment.
I did the manual test of moving the x gantry back and forth slowly by hand (without touching the belts or toolhead). When I did this test with the old belts, the resistance to movement was not smooth and the toolhead would start moving left or right of its own accord at certain repeatable regions in the back and forth travel of the gantry. Doing the test now - the movement feels smooth and the toolhead does not move left/right.
I repeated the parallel lines test print - and the result shows... no noticeable distortion 🙂
So I conclude that the original belts and/or the right hand timing pulley were out of spec. Perhaps regions of the belt had teeth with an out of spec profile than caused binding with the pulley, leading to the black deposits. I think I (and Prusa) were pretty unlucky to hit this issue, as they use quality components (gates belts etc). I've documented all this with Prusa and encouraged them to add additional tests for this defect to their production QA.
Here's the comparable test image after replacing the belts and pulley:
The deposits on the right hand timing pulley:
