Klipper Input Shaper Calibration: Eliminate Ringing in 2026

What Input Shaper Actually Does and Why It Fixes Ringing

Input shaper is the single highest-impact feature in Klipper firmware, and it is the reason a Klipper-converted machine can outprint its original firmware on the same hardware. The function is mechanical: input shaper modifies the acceleration commands sent to the motors so that the natural resonant frequencies of the printer frame and toolhead are not excited by the motion. The visible result is the elimination of ringing — those repeating ghosted edges that appear after sharp direction changes — and the practical result is that you can print 30-50 percent faster at the same visual quality.

This guide covers the calibration procedure end to end: how to capture the resonance signature of your printer with or without an accelerometer, how to pick the right shaper type for your machine, how to apply the result, and what to do when the calibration produces unexpected numbers. Klipper input shaper is well documented, but the documentation assumes you already know what you are looking at — this guide does not.

The Two Calibration Methods: Accelerometer vs Visual

The accelerometer method uses an ADXL345 (or LIS2DW, MPU-6050) mounted to the toolhead, which measures the actual mechanical vibration during a sweep of frequencies. Klipper runs the input_shaper_calibrate macro, the toolhead executes a motion pattern at increasing frequencies, the accelerometer captures the resonance peaks, and the algorithm computes the best shaper parameters. This method is fast (5-10 minutes per axis), accurate, and reproducible. The hardware cost is $5-15 for the accelerometer plus a small mounting fixture; many users solder one to a temporary cable for calibration and remove it afterward.

The visual method (also called “ringing tower” or “vibration tower” method) involves printing a test print with deliberate motion patterns at varying speeds, measuring the position of ringing artifacts on the printed part, and computing the resonance frequency from the artifact spacing. This method works without any added hardware but is less accurate, more time-consuming (20-40 minutes including print time and measurement), and harder to reproduce. The visual method’s published frequency accuracy is plus or minus 5 Hz, while accelerometer calibration is plus or minus 1 Hz.

Use the accelerometer method if you are doing input shaper for the first time. The hardware investment is small and the calibration quality difference is large.

Mounting the Accelerometer Correctly

The accelerometer needs to be mounted rigidly to the toolhead — the part whose motion you are trying to characterize. Loose mounting (tape, ziptie, double-sided foam tape) gives garbage data because the accelerometer itself resonates against its mount and the captured frequencies are the mount’s resonance, not the toolhead’s. A small 3D printed fixture screwed or hot-glued (rigid contact bond) to the toolhead is the right approach.

For X-axis calibration on bedslinger printers (Ender 3, Prusa Mini, most consumer printers), the accelerometer mounts to the toolhead and captures X-axis resonance during X-direction motion. For Y-axis calibration on those machines, the accelerometer should be moved to the bed, because Y motion is the bed moving. For CoreXY machines (Voron, Bambu, RatRig), both X and Y motion happen at the toolhead, so a single accelerometer position calibrates both axes.

The orientation of the accelerometer matters for axis assignment in the configuration file. Confirm that the X-direction motion produces accelerometer output on the X axis of the chip, and likewise for Y. Klipper’s config has axes_map parameters specifically to handle accelerometers mounted in non-standard orientations; check the captured data after a first calibration run to confirm axis assignments.

Running the Calibration Macros

Klipper provides two macros: SHAPER_CALIBRATE (which automatically finds and selects shaper parameters) and TEST_RESONANCES (which captures raw resonance data for manual analysis). Start with SHAPER_CALIBRATE — it does the right thing in most cases.

The procedure: home the printer, run SHAPER_CALIBRATE AXIS=X, wait for the motion pattern to complete (about 90 seconds), Klipper analyzes the data and outputs recommended shaper type and frequency, repeat for AXIS=Y. The output looks like “Fitted shaper ‘mzv’ frequency = 47.6 Hz” — write this down. The macro can save the result automatically with SAVE_CONFIG, but reviewing the captured data first is worthwhile.

The CSV files from calibration are saved to /tmp on the Klipper host. Copy them off the host and plot them with the calibrate_shaper.py script that ships with Klipper, which produces a PNG showing the resonance peaks and the modeled response under each shaper option. This visualization is what makes input shaper calibration go from “trust the recommendation” to “understand what the calibration found.”

Reading the Resonance Plot

A healthy resonance plot has one or two clear peaks in the 30-80 Hz range, with relatively low response above 100 Hz. The peak frequencies are the natural resonances of your printer frame plus toolhead — these are the frequencies that input shaper will cancel. For a typical bedslinger Ender 3 class machine, expect X peaks around 50-60 Hz and Y peaks around 35-45 Hz (Y is lower because of the bed mass). For a typical CoreXY in the Voron 2.4 / RatRig V-Core class, expect peaks in the 60-80 Hz range on both axes, because the rigid frame moves the resonance higher.

An unhealthy plot has many small peaks across a wide frequency range, no clear dominant peak, or a peak below 25 Hz. Multiple small peaks usually mean the toolhead or accelerometer mount has its own resonances corrupting the measurement — fix the mount and recalibrate. A peak below 25 Hz means something is mechanically wrong with your printer (loose belt, loose frame, missing fastener); fix the mechanical problem before trying input shaper. Input shaper cannot compensate for a loose printer; it can only optimize a tight one.

Picking the Right Shaper Type

Klipper offers six shaper types: ZV, MZV, ZVD, EI, 2HUMP_EI, and 3HUMP_EI. The differences are how aggressively they cancel resonance and how much speed penalty they introduce. ZV is the simplest and lowest-penalty, but only cancels a single sharp peak. MZV (modified ZV) is the default for most machines and handles a slightly broader peak. EI shapers handle even broader peaks at the cost of more speed penalty. 3HUMP_EI is the broadest-band shaper, used when the resonance is spread across a wide frequency range.

The SHAPER_CALIBRATE macro picks the shaper that gives the best speed-vibration trade-off. For most consumer printers, MZV is the answer for at least one axis. CoreXY printers with stiff frames sometimes end up with ZV for both axes, indicating very narrow resonance peaks. Older or less-rigid bedslinger machines sometimes need EI or 3HUMP_EI to handle their broader resonance.

Manually overriding the macro recommendation is rarely worthwhile unless you are debugging a specific quality problem. The macro is conservative — it will recommend a less aggressive shaper if the vibration data is borderline rather than a more aggressive one that might introduce its own artifacts.

Applying the Calibration and Validating

After SHAPER_CALIBRATE completes, run SAVE_CONFIG to write the shaper parameters into printer.cfg. Klipper restarts and the new shapers are active. Verify by printing a ringing test print — a model with multiple sharp direction changes at varying speeds — before and after. The before-print should show ringing as ghosted echoes after sharp corners; the after-print should be visibly cleaner at the same speed, or you should be able to increase speed substantially without ringing returning.

The standard validation print is the Klipper-supplied input shaper test STL, which has a series of perpendicular features at different positions. After tuning, sharp corners on this print should be crisp at the speed where the pre-tuning print shows obvious ringing. If ringing remains after tuning, the calibration captured something other than the dominant resonance — usually because the accelerometer mount was loose, or the printer has multiple distinct resonances and the chosen shaper handles only one of them.

Common Mistakes and Their Fixes

Calibrating with a loose belt: tighten belts to spec before calibrating. Loose belts introduce a low-frequency resonance below 30 Hz that contaminates the captured data. The right tension is one where the belt produces a clear musical note when plucked, not a dull thud.

Calibrating with a different toolhead than what you print with: if you swap nozzles, hot ends, or fans, recalibrate. Toolhead mass changes the resonance frequency; significant changes in tool weight (CHT nozzle, ceramic heater, dual-fan setup) shift resonance by 5-10 Hz, which is enough to make the previous calibration less effective.

Calibrating at a different bed temperature than typical print conditions: thermal expansion of the frame at typical print temperatures (60-110°C bed) changes resonance slightly. Calibrate at a bed temperature similar to your most common printing material — for PLA-focused users that means 60°C bed, for ABS users 100-110°C.

Skipping the mount step and using zip ties: the captured data will look weird, the recommended shaper will be aggressive, and the actual print quality might be worse than before tuning. Do the mount properly; it takes 10 minutes and changes everything downstream.

Beyond Initial Calibration: Pressure Advance and Smooth Time

Input shaper alone solves ringing. For overall print quality at the speeds input shaper unlocks, also tune pressure advance (which solves corner bulging and underextrusion at speed changes) and review the smooth_time parameter (which smooths the shaper’s own response over time). Pressure advance has its own calibration macro and is worth running after input shaper since the higher achievable speeds make pressure advance effects more visible.

For most users, the workflow is: input shaper calibration first, pressure advance second, validation print third. Together these two calibrations unlock the speed potential Klipper is famous for — printers that print 200 mm/s with quality matching their 60 mm/s configuration. Without input shaper, those speeds produce visible ringing on every sharp corner. With proper input shaper, the corners stay sharp and the print finishes faster than the original firmware could produce at any speed.