Your prints show diagonal ringing on one face but not the other. Or circles come out as slight ovals. Both symptoms point to belt tension — one axis too loose, the other too tight, or the gantry racked out of square. Belt tension is the most underdiagnosed print quality problem because it’s invisible: the belts look fine, but the print tells the truth.
Step-by-Step Belt Tensioning
Step 1: Understand Target Tension for GT2 Belts
GT2 belts (2 mm pitch, the standard on virtually all consumer FDM printers) have a recommended tension range of 4-6 pounds of static deflection force at mid-span for 6 mm wide belts. For 9 mm wide belts (Voron, Rat Rig), target 6-8 pounds. Too loose: backlash, skipped steps, ringing. Too tight: accelerated bearing wear, stepper motor strain, belt stretch. The practical measurement method: pluck the belt like a guitar string and measure the frequency.
Step 2: Measure Belt Frequency with a Spectrum Analyzer
Install a free spectrum analyzer app on your phone (Spectroid for Android, SpectrumView for iOS). Pluck the longest span of the belt — the section between the motor pulley and the idler. The app shows a frequency peak. For a GT2 belt with a 200 mm free span, the target frequency is roughly 80-100 Hz. Shorter spans yield higher frequencies for the same tension. The formula: Tension (N) = 4 × mass_per_unit_length × (span_length × frequency)². For 6 mm GT2 belts: mass = 0.0083 kg/m.
Step 3: Adjust Tension Evenly on Both Sides
If your printer uses a single belt loop for each axis (Prusa/Ender style), tension adjusts at the idler end. Loosen the idler mounting screws, pull the idler to increase tension, re-tighten. Measure frequency on both the upper and lower belt runs — they should match within 5 Hz. If they differ by more than 10 Hz, your idler bearings are binding or the belt isn’t tracking straight.
Step 4: Verify Gantry Squareness
A racked gantry (X-axis not perpendicular to Y-axis) causes dimensionally inaccurate prints even with perfect belt tension. On a Cartesian printer, measure the distance from the X-axis extrusion to the top of the frame at both ends — they must match within 0.5 mm. On CoreXY (Voron, Rat Rig), the A and B belt loops must be tensioned identically. Unequal A/B tension pulls the gantry out of square. Adjust by loosening one motor pulley, rotating the motor shaft slightly to even out the belt, then retightening.
Step 5: Validate with a Test Print
Print a 30 mm calibration cube at 100 mm/s. Measure X and Y faces with calipers. If X measures 30.0 mm but Y measures 29.8 mm, your Y-axis is undersized — the Y belt is likely too loose (backlash) or too tight (stepper skipping). A test pattern of concentric circles makes belt issues visible: loose belts create flats on the circle where direction changes.
Belt Tension Reference Table
| Belt Type | Width (mm) | Free Span (mm) | Target Frequency (Hz) | Target Tension (N) | Symptom if Too Loose | Symptom if Too Tight |
|---|---|---|---|---|---|---|
| GT2 | 6 | 200 | 80-100 | 21-33 | Layer shift, ringing, backlash | Bearing wear, stepper noise |
| GT2 | 6 | 300 | 55-70 | 21-33 | Direction-change artifacts | Vertical banding |
| GT2 | 9 | 300 | 45-55 | 30-45 | Corner overshoot | Motor overheating |
| GT2 | 9 | 400 | 35-45 | 30-45 | Diagonal ringing | Belt stretch over time |
| GT2 (CoreXY) | 6 | 350 | 50-65 | 21-33 | Gantry racking | Motor current spike |
| GT2 (CoreXY) | 9 | 350 | 40-50 | 30-45 | Unequal X/Y scaling | Idler bearing noise |
Common Mistakes & What Most Makers Get Wrong
Mistake 1: Tensioning by feel alone. “Feels tight enough” is the leading cause of mystery print quality issues. Human fingers can’t distinguish 15 N from 30 N on a short belt span. A free spectrum analyzer app on the phone you already own removes the guesswork entirely. Five minutes of measuring beats weeks of chasing print artifacts.
Mistake 2: Overtightening CoreXY belts to “eliminate ringing.” CoreXY printers route belts through multiple idlers. Excessive tension doesn’t fix ringing — it masks the root cause (usually acceleration/jerk or input shaper misconfiguration) while accelerating bearing wear on every idler in the path. Proper tension is precise, not “as tight as you can pull.”
Mistake 3: Forgetting to re-tension after break-in. New belts stretch 2-5% in the first 20-30 print hours. That fresh “perfectly tensioned” build from last month is now 10 Hz low. Re-measure belt frequency after the first 50 hours, then every 200 hours or whenever you notice print quality declining.
Mistake 4: Tensioning one belt while the other is locked. On CoreXY, tensioning the A belt changes B belt effective length because they share the gantry. Adjust both belts in small increments, alternating sides, and verify gantry square after each adjustment cycle.
⚠️ Safety Notice: 3D printers involve high temperatures, moving mechanical parts, and electrical components. Belt tension adjustments should be performed with the printer powered off. After adjustments, verify that all fasteners are tight and that no wiring is pinched or rubbing against moving components. During test prints, stay present to observe for unusual noises — screeching or grinding indicates a tension or alignment problem that can damage hardware.
Related Reading
Belt tension is one piece of the mechanical precision puzzle. Combine it with a complete 3D printer maintenance schedule covering rail lubrication and nozzle wear. And after tensioning, run through our input shaping calibration guide to tune out the remaining resonance.
Product recommendation: The Gates GT2 (genuine, not generic) 6 mm belt kit for Ender 3 and similar Cartesian printers includes pre-crimped steel-core belts that don’t stretch during break-in — one less variable to chase. Available at uavmodel.com.