Ringing and ghosting after sharp corners: input shaping, acceleration and belt checks
Short answer
Ringing, also called ghosting or echoing, is a set of ripples that repeat a sharp edge along the wall after it. Klipper's documentation traces it to vibrations from quick direction changes and lists a weak frame, loose or springy belts, misalignment and heavy moving mass as the mechanical origins to fix first [1]. Input shaping then cancels much of what remains; Klipper, Marlin and Prusa firmware support it, and Bambu Lab printers calibrate it with a frequency sweep [3, 5, 8].
Prusa recognizes ghosting as several waves that follow a sharp edge of the model [4]. Klipper calls the same defect ringing, echoing or rippling: edges repeat themselves on the surface as a faint echo because the printer keeps vibrating after a quick change of direction [1]. Because the waves come from the motion system, the fix splits into mechanical work, which raises or cleans up the resonance, and input shaping, which cancels what is left.
Confirm it is ringing
Ringing runs along the wall right after a corner or other sharp feature [4]. That separates it from two look-alikes:
- Z-banding shows as horizontal stripes that repeat with height regardless of corners. See Z-banding.
- Ripples that depend on wall speed rather than on corners. OrcaSlicer’s Resonance avoidance option slows outer walls out of the printer’s resonance speed range, and the wiki pairs it with a separate VFA speed test [12].
Klipper adds a direct test: on its ringing tower, true ringing follows the pattern of the curved notches. If it does not, the defect has a different mechanical or extruder origin and should be fixed before tuning input shaping [1].
The ripple spacing also tells you the frequency. Klipper’s formula is f = V × N / D, where V is the outer-wall speed in mm/s and N is the number of oscillations counted over a distance D in mm. Its example, 6 oscillations over 12.14 mm at 100 mm/s, gives about 49.4 Hz [1].
Symptom, likely cause, what to check
| Symptom | Likely cause | What to check |
|---|---|---|
| Ripples after corners on walls along one axis only | Resonance or loose belt on that axis | That axis’s belt tension; tune that axis’s shaper frequency [4, 1] |
| Ripples grow when speed or acceleration goes up | Vibrations increase with print speed | Lower speed or acceleration as a test [4] |
| Ringing appeared after a new hotend, fan duct, bed or belt work | Moving mass or stiffness changed the frequency | Re-measure the frequency [1] |
| Ringing returned after moving the printer | Support surface changed the response | Firm surface; recalibrate [4, 11] |
| Ribbing on X1 or P1 walls after a lot of ABS or ASA | Residue on carbon rods | Clean the rods [10] |
| Measured frequency below about 20–25 Hz | Flexible frame or heavy toolhead | Stiffen the printer or reduce moving mass [1] |
| Rounded corners or widening gaps after enabling a shaper | Smoothing from too much acceleration | Lower max_accel [1] |
Mechanical checks before tuning
- Belts. Prusa ties loose belts to ghosting, layer shifts and out-of-round circles [6]. Its ghosting article suggests plucking the belt for a deep bass tone and, on the MK3 family, reading Belt Status values, where 240–300 is fine and a lower value means more tension [4]. Bambu Lab models the belt as the spring and the gantry as the mass, with a natural frequency of f = √(k/m); a loose belt lowers k and the frequency, which affects vibration compensation [8].
- Guides and rods. Bambu Lab recommends cleaning X1/P1 carbon rods when ribbing or vibration artifacts appear on outer walls, and routinely after every 5 kg of ABS or ASA [10].
- Support surface. Prusa notes that ghosting can appear even with correct belts, grows with print speed, and that a firm surface helps [4].
- Loose parts. Prusa’s vibration article for the MK3S+ lists loose heatbed screws, loose LCD screws and misaligned Y rods among the sources of rattling and vibration during moves [7].
After belt or rod work, recalibrate: Bambu Lab says both change the toolhead’s resonance characteristics [8], and Klipper lists belt tightening among the changes that shift ringing frequencies [1].
Input shaping by firmware
| Firmware or printer | Where it is set | How it is tuned |
|---|---|---|
| Klipper | [input_shaper] with shaper_freq_x, shaper_freq_y, shaper_type |
Ringing tower with TUNING_TOWER stepping acceleration from 1500 mm/s² by 500 every 5 mm, or SHAPER_CALIBRATE with an accelerometer [1, 2] |
| Marlin (2.1.2 and later) | M593 X F<Hz> D<damping>, saved with M500 |
Frequency sweep patterns or a ringing tower; no accelerometer tuning [3] |
| Prusa Buddy firmware | M593 with F, D, and T for type (0 ZV to 5 3HUMP_EI); F0 disables it |
Factory calibrated; optional accelerometer, built into the CORE One L and XL [5] |
| Bambu Lab | Printer calibration | Frequency sweep during calibration; the A1 mini also flags belts that need tension [8, 9] |
Shaper choice. Klipper recommends MZV or EI for most printers. EI tolerates frequency changes better, which suits bed-slingers whose bed mass grows during a print and delta printers, but it smooths more [1]. Marlin’s built-in shaper is ZV [3]. OrcaSlicer notes that firmware defaults differ: MZV for Klipper and ZV for Marlin [12].
Slicer-driven calibration. OrcaSlicer’s input shaping test prints a ringing tower across a frequency range, 15–110 Hz by default, with damping usually started at 0.1–0.2. It asks for the highest acceleration and speed the printer can run without losing steps, and an opaque glossy filament makes the ripples easier to read [11]. The full walkthrough is on the input shaping calibration page.
Acceleration, smoothing and max_accel
Input shaping does not make acceleration free. Klipper explains that higher acceleration with a shaper adds smoothing, visible as a widening of the 0.15 mm gap in its test model, and recommends using the lower of two values: the acceleration where ringing is acceptable and the one where smoothing starts. In its example the gap grows from 3500 mm/s², so max_accel is set to 3000 [1]. SHAPER_CALIBRATE prints a suggested max_accel for each shaper, but it does not change the [printer] setting by itself [2].
Klipper also advises keeping square_corner_velocity at its default of 5 mm/s when using input shaping, because raising it adds smoothing [1]. Marlin’s jerk and junction deviation settings play the same cornering role; the cornering settings guide compares them.
If ringing measurements are unstable, with oscillation spacing that varies along one axis, Klipper says the printer may have several resonances on that axis and points to a separate procedure for that case [1].
Order of operations
- Confirm the pattern follows corners, not layer height.
- Tension and inspect belts, clean rods, tighten anything loose, and put the printer on a firm surface.
- Measure the frequency per axis with a ringing tower or accelerometer.
- Enable the shaper and choose a type.
- Set
max_accelfrom the ringing and smoothing results. - Re-check pressure advance; Marlin notes its frequency sweep runs with Linear Advance off and the K factor may need recalibrating [3].
Loose belts also produce layer shifts, so a printer showing both needs belt work before any tuning.
Questions
Do I need an accelerometer to tune input shaping?
No. Klipper's ringing tower method and Marlin's frequency sweep and ringing tower methods use only a printed test [1, 3]. Marlin does not provide accelerometer-based tuning; Klipper supports ADXL345, MPU-9250, LIS2DW and LIS3DH sensors for automatic tuning [2].
How often should input shaping be recalibrated?
OrcaSlicer suggests about once a year or after mechanical changes such as moving the printer, a new support surface, new belts or motors [11]. Klipper lists toolhead mass changes, belt tightening, frame stiffening and bed changes as reasons to re-measure [1].
Why are my corners rounded after enabling input shaping?
Input shaping smooths motion, and smoothing grows with acceleration. Klipper recommends choosing max_accel low enough to keep the test model's 0.15 mm gap from widening [1].
Sources
- Resonance Compensation. Klipper documentation. Firmware documentation. Causes of ringing, ringing tower method, frequency formula, shaper choice, max_accel selection. Accessed October 12, 2026.
- Measuring Resonances. Klipper documentation. Firmware documentation. Supported accelerometers and SHAPER_CALIBRATE output. Accessed October 12, 2026.
- M593 - ZV Input Shaping. Marlin Firmware. Firmware documentation. ZV shaper parameters and print-based tuning methods. Accessed October 12, 2026.
- Ghosting. Prusa Knowledge Base. Manufacturer documentation. Appearance, belt checks, speed, firm surface. Accessed October 12, 2026.
- Input Shaper (CORE One, MK4/S, MK3.9/S, MK3.5/S, XL, MINI/+). Prusa Knowledge Base. Manufacturer documentation. Supported printers, M593 parameters, factory calibration, accelerometer. Accessed October 12, 2026.
- Adjusting belt tension (MK4/S, MK3.9/S, MK3.5/S, MK3/S/+). Prusa Knowledge Base. Manufacturer documentation. Loose belts cause ghosting and layer shifts. Accessed October 12, 2026.
- Vibration when printing (MK3S+/MK2.5S). Prusa Knowledge Base. Manufacturer documentation. Loose screws and rod alignment as vibration sources. Accessed October 12, 2026.
- X1/P1 Series Belt tensioning procedure. Bambu Lab Wiki. Manufacturer documentation. Belt as spring, natural frequency, frequency sweep, recalibration. Accessed October 12, 2026.
- A1 mini Belt Tensioning. Bambu Lab Wiki. Manufacturer documentation. Frequency scan during calibration flags belts needing tension. Accessed October 12, 2026.
- How to clean the carbon rods for Bambu Lab Printers. Bambu Lab Wiki. Manufacturer documentation. Ribbing or vibration artifacts as a reason to clean the rods. Accessed October 12, 2026.
- Input Shaping. OrcaSlicer Wiki. Slicer documentation. Ringing tower test settings and recalibration interval. Accessed October 12, 2026.
- Motion Ability. OrcaSlicer Wiki. Slicer documentation. Resonance avoidance and default shaper per firmware. Accessed October 12, 2026.