You will find dozens of comments about belt tension and related problems on the Prusa blog and elsewhere. It took me a while to understand the system, but it's really quite simple. I'll offer a couple fundamental concepts that will bring logic to the tensioning process.
First, let's say the belts are loose, the motors are off and the X-axis bearing rail moves freely forward and backward. You pull it forward, noting when the ends hit the stops. Both ends should hit the stops at the same time, indicating that the X-axis is perpendicular to the Y-axis. This is fundamental for the system to work right. Small errors can be corrected by simply forcing the rail a bit in the direction you need it to go, which will very slightly bend the brackets. If you checked the parts and built it correctly it should be very close. If you hold one end against its stop you can gently press the other end to check for any gap. Then hold the opposite end and check the other way. Do not use belt tension to correct an out-of-alignment X-axis.
The belt paths are mirror images of each other. Equal tension on the two belts should result in equal forces on the X-axis rail ends and no angular error. If the belt tensions are not equal, it will force the X-axis out of alignment, as it does have a bit of compliance or springiness. Thus, the change in angle is a fairly sensitive measurement of how equal the belt tensions are. We can make a rule:
Remember, we care about tension, not vibrational frequencies, though those are a useful tool that we'll use to tension the belts. The 6.4.0 firmware has a clever belt tension utility that strobes the illumination LEDs and oscillates the motors so you can see the resonant frequency of the belts. What the utility does not know is if the belt tensions are balanced and it proceeds as if they are. The front belt sections do not have the same length because one end of each is tied to the Nextruder (same location) and the other ends go around the pulleys at slightly different distances, 13.5 mm apart. The belt tension utility has an acceptance window of between 90 & 98 Hz.
We can make another rule, a bit less fundamental, but very necessary to our understanding:
Read Prusa's instructions and watch the video to see the utility in action. There are also several other videos you can watch. With that background and armed with the above knowledge, I'm going to give you the approach I used with good success:
I've probably gone on enough, but I think a big source of confusion is that there are two adjustments, one for each belt, and two belt frequencies. It just seems logical that you would adjust for the two frequencies. That idea derails quickly. First, the two tension adjustments interact so severely that you'd be hard pressed to set two specific frequencies even if you wanted to. Second, the requirement to keep the X-axis in alignment to the Y-axis is more important than hitting an arbitrary frequency. Fortunately, the two tension adjustments react almost identically. Thus, you can adjust for alignment, and then turn the tension adjustments by exactly the same amount to move the two frequencies up and down to the desired target window.
BTW, adjusted as I've described, my Core One homes with just a few bumps on the stops.
C. Hoffman
November 23, 2025
Last edit January 25, 2025