On the Original Prusa XL, adjusting the belt to the optimal tension is crucial to achieving good quality on the prints. A loose belt can cause Layer shifting, Ghosting, or other print abnormalities, like getting an irregular shape instead of a perfect circle. An over-tightened belt can cause an irregular movement in the X and Y axes.
The Original Prusa XL geometry is CoreXY: it has two belts for the XY-axes.
Alignment
It is crucial to have perfect alignment between the right and left sides. The X-axis should reach both sides of the front printed parts at the same time.
For the procedure, ensure the stepper motors are disabled. This can be done by switching the printer off, or through the LCD Menu -> Control -> Disable Motors.
By hand, move the toolchanger or Nextruder to the frontmost Y-axis position, and to the center of the X-axis.
Test the alignment by squeezing the CoreXY gantry and the corner printed part, first on one side and then on the other. There should be no play between the parts. With perfect alignment, the squeezing test on both sides feels like pressing two solid objects together.
If there is misalignment, then the squeezing test reveals play in the front corner. In that case, loosen the belt on both sides slightly until the gantry is aligned. Loosen the tensioning screw of the corner with more play less than the other.
Repeat the procedure and squeezing tests until perfect alignment is achieved.
The right side image shows the mechanism behind the tensioning screw. Tightening the screw pulls the tensioner towards the front, resulting in a more tensioned belt on that side.
![]() | ![]() |
Belt tuner
We have our own belt tuner. The application determines the belt tension by measuring the frequency of the vibration generated by strumming it. Follow the instructions in the video below to adjust the belt.
Homing tower
We prepared a specific homing tower, to visualize possible issues with the belt tension and have the possibility to correct them on the spot.

In case the belt is loose, the tower will have an inconsistent finish, with scattered material on the walls instead of a smooth finish. Without interrupting the print, tighten the screws on both Y-axis rails by half turn.
If the print finish does not fully improve, repeat the procedure once again, until the homing tower has a smooth, consistent finish.
An over-tightened belt can result in layer shifting.
Once the tower has a smooth finish, loosen the same screws very slightly, and observe if this further improves the tower. Repeat the procedure until the tower has an ideal finish.

30 comments
Also, my phone doesn't hear the frequency at all. I had to put the bottom edge of the phone onto the printer arm above the belt, blocking the microphone, but enabling the microphone to physically "feel" with the vibration. It's still not consistent but does occasionally give a reading.
Finally, I bought the Belter tool to have a more reasonable way to measure the belt tension in the future. The wierd thing is if I set both belts to the same tension it pulls the gantry out of alignment. If I use frequency to get them the same and aligned, the Belter gives different readings per belt. Very strange and not intuitive. For anyone curious and wanting to compare readings, my belts are at 78-80Hz, gantry is aligned, and Belter reads 7-7.05 on the left belt and 7.2-7.25 on the right belt. This is with the printed end piece on the Belter.
Interesting on the readings between the left and right. I will try to share my settings when done.
I have -6.25 on the left at 82Hz and -6.24 on the right at 81Hz
this is truly a bad joke -- it's like asking a painter to be a metronome
Zero accuracy. If it's so important there should be a more scientific way.
In the few minutes it took me to read this bad joke I came up with an idea of a 3D printable tension checker that might not be accurate to the T, but at least it would be more reliable and consistent... than a phone mic placed in an absolutely random point.
Come on guys - be serious, a $5K+ printer cannot be this badly supported.