3D Printer Belt Tensioning Guide: CoreXY and Cartesian Tension, Symptoms, and Tuning Tools — 2026

Your Ender 3 prints perfect calibration cubes but circles come out as ovals. Or your Voron rings like a bell on direction changes at 200mm/s. Both are belt tension problems — too loose in the first case, too tight in the second. Belt tension is the cheapest tuning variable in 3D printing and the one most builders ignore entirely.

Step-by-Step Belt Tensioning

1. Understand What Correct Tension Feels Like

On a Cartesian printer (Ender 3, Prusa i3), the X and Y belts should feel like a low-tension guitar string — a twang that sustains for about half a second when plucked. If it goes “thud,” it’s too loose. If it rings like a piano wire for 2+ seconds, it’s too tight.

On a CoreXY machine (Voron, RatRig), the belts are much longer and should ring at 110-140Hz when plucked. You can measure this with a guitar tuner app on your phone — hold the phone’s microphone 10cm from the belt at the midpoint, pluck the belt sharply, and read the frequency.

Belt tension at 6mm GT2 pitch translates to frequency based on the belt’s free span length:

Frequency formula: f = (1 / (2 × L)) × √(T / μ)

Where L is the free span length (meters), T is tension (Newtons), and μ is linear mass density (0.0083 kg/m for 6mm GT2). But in practice, you don’t need the math:

  • Cartesian X/Y (300-400mm span): Target 70-90Hz. Below 60Hz = loose, rings and dimensional errors. Above 110Hz = tight, accelerated bearing wear, motor current increase.
  • CoreXY A/B (600-900mm span): Target 110-140Hz. Below 90Hz = loose, severe ringing on Y direction changes. Above 160Hz = tight, stepper motor skipping on high-accel moves.
  • Z-axis leadscrew-based: Not applicable. If you have belt-driven Z, target 50-70Hz — Z moves slowly and doesn’t need high tension.

3. Tension Adjustment Procedure

Cartesian (Ender 3 style adjustable idler):
1. Loosen the two screws holding the tensioner bracket to the extrusion
2. Use the adjustment knob to tighten or loosen the belt
3. Pluck the belt at its midpoint and measure frequency
4. Tighten bracket screws once at target frequency
5. Home the axis and re-check — homing can shift the tension slightly

CoreXY (Voron style sliding tensioner):
1. Loosen the four bolts holding the motor mount to the frame
2. Slide the motor mount outward to increase tension, inward to decrease
3. Tension must be EQUAL across both A and B belts — a 10Hz difference causes skewed prints
4. Tighten motor mount bolts to spec (6-8Nm on M5 bolts into aluminum extrusion)
5. Re-run input shaping calibration after any belt tension change

4. Verify With a Test Print

Print a 40mm calibration cube at your normal speeds. Measure X and Y dimensions with calipers. If both axes are within 0.1mm of 40.0mm, tension is balanced. If X reads 40.0mm but Y reads 40.3mm, the Y belt is looser than X. The looser belt stretches slightly under acceleration, producing a slight undershoot — but paradoxically, the part dimension reads larger because the belt springs back at the direction change, overshooting the intended stop point.

Belt Tension Reference Table

Printer Type Belt Span Target Frequency Symptoms of Too Loose Symptoms of Too Tight
Cartesian X (Ender 3) 300-350mm 75-90Hz Oval circles, dimensional error >0.2mm Motor clicking, hot steppers
Cartesian Y (Ender 3) 350-400mm 70-85Hz Layer shifts on fast Y moves Bearing whine, increased VFA
CoreXY A/B (Voron 2.4) 700-900mm 110-140Hz Severe ghosting, skewed prints Stepper skips at >10k accel
CoreXY (Voron 0) 350-450mm 100-130Hz Ringing on small perimeters Motor overcurrent on travel moves
Belted Z (Voron Trident) 300-500mm 50-70Hz Z-banding, inconsistent layer height Z motor stall on fast Z-hops

Belt Tensioning Mistakes

Mistake 1: Tensioning by Feel Without a Reference
Every builder’s “feels tight enough” is different. After 100 builds, you develop a reference. Before that, you’re guessing. Use a frequency measurement — it costs nothing (guitar tuner app) and is objective. I’ve measured “feels tight” belts at 40Hz and “feels perfect” belts at 180Hz from the same builder on different days.

Mistake 2: Unequal CoreXY Belt Tension
CoreXY kinematics sum the motion of two belts. If belt A is at 140Hz and belt B is at 110Hz, the gantry pulls harder in one diagonal direction than the other. The print shows skewed geometry — a square comes out as a parallelogram. As we discussed in our 3D Printer Layer Shifting Fix guide, mechanical asymmetry is the root cause of most geometric print defects.

Mistake 3: Not Re-Tensioning After the First 50 Print Hours
GT2 belts stretch permanently in the first 10-50 hours of printing as the fiberglass cords settle into the rubber jacket. A belt tensioned to 85Hz on day one will be at 65Hz after a month of printing. Check tension monthly as part of regular maintenance — it’s a 5-minute adjustment that prevents hours of chasing print quality ghosts.

Mistake 4: Over-Tightening to “Eliminate Ringing”
Ringing is caused by the entire motion system resonating at its natural frequency. Tightening the belt raises that frequency but doesn’t eliminate the resonance — it just shifts it to a different speed. Input shaping (Klipper) or junction deviation (Marlin) is the correct fix for ringing. Over-tightened belts accelerate idler bearing wear and increase motor load without solving the problem they were tightened to fix. Our Klipper vs Marlin firmware comparison details the software tools that actually address resonance.

⚠️ Safety Notice: Belt tensioning involves mechanical adjustment of moving parts. Always power off the printer before adjusting belt tension to prevent accidental motor movement. Over-tensioned belts can damage stepper motor shafts, idler bearings, and aluminum extrusion slots. If you hear a metallic “ping” during tensioning, the belt is dangerously over-tightened — back off immediately. Replace belts showing frayed edges, missing teeth, or visible fiberglass cords — a snapped belt mid-print on a heated bed is a fire risk.

The PRUSA belt tension gauge is a 3D-printable tool that clips onto 6mm GT2 belts and gives a visual tension reading — no phone app needed. We carry genuine Gates GT2 belts in 6mm and 9mm widths at uavmodel.com, pre-cut to Voron and Ender 3 lengths.

Leave a Comment

Scroll to Top