CoreXY Geometry Issues in TwoTrees SK1 After Head Crash

[dmit2k]

Printer Model

TwoTrees SK1

I have a TwoTrees SK1 which I believe is a kind of factory-made Trident fork. The printer was properly calibrated, including belt tension and dimensional accuracy. However, I recently noticed that the corners of my prints are no longer square. A 100mm test pattern now shows that the diagonal along motor "A" is more than 1mm longer than the diagonal along motor "B," effectively forming a parallelogram.

When checking belt tension, I found a significant discrepancy between the acoustic method and an electronic tension meter. Previously, both methods gave consistent results, but now setting both belts to 110Hz results in different readings on the meter, and vice versa.

Neither tuning the belts "by sound" nor "by the meter" fully resolves the geometry issue—there's still a 1mm difference per 100mm between diagonals A and B. Further adjustments only create an excessive tension imbalance between the belts.

I suspect the root cause might be a recent hard head crash along the X-axis (left to right), where the head got stuck, and the motors kept pulling for a few seconds until I turned off the printer. Based on my understanding of CoreXY mechanics, I see two possible explanations:

1. Belt A may have stretched during the crash.
2. Belt A may have slipped by one tooth through its clamp.

Could more experienced users confirm if my reasoning is correct? What would be the best way to diagnose and fix this issue? Any advice would be greatly appreciated!

 

[Sanity Agathion]

CoreXY is just kinematics, VORON uses CoreXY, but not every CoreXY is VORON. SK1 is in no way, shape or form related to Trident or VORON project It's by no means a factory made Trident fork.

I can only offer generic advice like pull the belts out, make sure the gantry is not racked by whatever means SK1 can be adjusted, and put them back in matching exact number of teeth. Also check bearings idlers etc. if they are not dislodged or slanted. Mind you, I have absolutely no idea whatsoever how that printer is built and if user can service it at all. You may need to contact their customer support https://www.twotrees3dofficial.com/pages/help-center and consult with them what to do, they sell those printers and should provide support for them.

 

[dmit2k]

It's by no means a factory made Trident fork.

Thank you very much for your prompt reply!

Yes, I fully understand it has no direct relation to Voron development and is a "kinda fork" more or less inspired by Trident design.

But...The TwoTrees support is known for being absolutely useless. So the only place I could imagine where to ask for reasonable answers and further ideas was the Voron community forum, full of people who design, build and maintain similar machines

I am new to CoreXY cinematic and came from the Slinger world. And I was surprised there is not much info about dealing with dimensional accuracy of printed models on the CoreXY mechanics. So I'm just looking for direction where to start digging from and if there are any well known issues that could lead to such behaviour.

I measured the rack and chassis geometry and it seems to be pretty tight and square, except for the X gantry, which is not absolutely perpendicular to Y-rails. The error was about 3mm for 350mm width. I was able to make it almost square and now the error is only about 1mm for the whole width. But surprisingly, that gave me only 0.16mm improvement by A-diagonal on my calibration octagon model.

At the moment I was able to tension the belts the way that both produce 110-116 Hz audible response AND comparable values measured by electronic tensiometer. But still having two issues (most probably related to each other):

1) I can't tune one belt below 115 Hz even if its tension regulation bolts are fully released

2) one diagonal of a printed 100 mm octagon has perfectl 99.98 mm size, while another is 100.86 mm

This makes me think my first guesses are wrong and it could not be caused by one belt's tooth slipped out through its clamp...

SK1 itself has more or less standard belts design with stacked pulleys. It can be seen and understood from their short video about tuning the belts: https://wiki.twotrees3d.com/en/3DPr...cementAndAdjustment#sk1-adjust-belt-tightness

Now I feel that I have to remove both belts and compare their lengths - maybe those were incorrectly measured from factory... But then I'm still in doubt how I was getting good dimensional results before failure???

P.S.
Just for test, I added "gear_ratio: 0.9916:1" parameter to the problematic stepper config in Klipper, and now get absolutely perfect dimensions across the whole bed! Which is a good sign I think...

 

[Sanity Agathion]

Thank you very much for your prompt reply!

Yes, I fully understand it has no direct relation to Voron development and is a "kinda fork" more or less inspired by Trident design.

But...The TwoTrees support is known for being absolutely useless. So the only place I could imagine where to ask for reasonable answers and further ideas was the Voron community forum, full of people who design, build and maintain similar machines

I am new to CoreXY cinematic and came from the Slinger world. And I was surprised there is not much info about dealing with dimensional accuracy of printed models on the CoreXY mechanics. So I'm just looking for direction where to start digging from and if there are any well known issues that could lead to such behaviour.

I measured the rack and chassis geometry and it seems to be pretty tight and square, except for the X gantry, which is not absolutely perpendicular to Y-rails. The error was about 3mm for 350mm width. I was able to make it almost square and now the error is only about 1mm for the whole width. But surprisingly, that gave me only 0.16mm improvement by A-diagonal on my calibration octagon model.

At the moment I was able to tension the belts the way that both produce 110-116 Hz audible response AND comparable values measured by electronic tensiometer. But still having two issues (most probably related to each other):

1) I can't tune one belt below 115 Hz even if its tension regulation bolts are fully released

2) one diagonal of a printed 100 mm octagon has perfectl 99.98 mm size, while another is 100.86 mm

This makes me think my first guesses are wrong and it could not be caused by one belt's tooth slipped out through its clamp...

SK1 itself has more or less standard belts design with stacked pulleys. It can be seen and understood from their short video about tuning the belts: https://wiki.twotrees3d.com/en/3DPr...cementAndAdjustment#sk1-adjust-belt-tightness

Now I feel that I have to remove both belts and compare their lengths - maybe those were incorrectly measured from factory... But then I'm still in doubt how I was getting good dimensional results before failure???

Uneven belt length can cause racking with CoreXY. It is possible that belt slipped by tooth or two. It's also possible one has internal fibers torn and belt is stretched or even damaged. Try inspecting full belt path to see if there is anything obvious.

It's also possible XY joint or motor mount is damaged, bearing stack shifted or such. Really hard to say... I'd start with belts and checking length.

 

[Spydyr]

As someone that has a TwoTrees Sapphire Plus v1 with numerous upgrades and klipper conversation. It is closer to a VZ Bot. Also unless quality control has gotten a lot better the entire upper frame is warped because it is stamped steal. The Y rails are not parallel with each other. While you can get it somewhat close it will always be off.

To align the Y axis only take one side loose. Do not take both loose or you will pull your hair out. I know from experience. Look at skew_correction to help with the angular error.

I'm converting mine into a trident because over the past 5 years my accuracy has went from .05mm to .16 at best and is steadily getting worse. Because I can reuse all my electronics and motion just adding one motor and one rail I think it could be a straight forward build. I'll let you know in a few weeks my aluminum shows up today.