How to find your 3D printer's max acceleration

The test depends on your printer. On Klipper, run SHAPER_CALIBRATE with an accelerometer or print the ringing tower without one; on Marlin, set the M201 and M204 limits by hand. Bambu Lab, Creality K1 and Prusa MK4S or CORE One printers come tuned: use the acceleration in their slicer profiles.

How do you find your 3D printer’s max acceleration?

Print a test that raises acceleration in steps and note where quality or motion breaks down. On Klipper, the ringing tower raises it 500 mm/s² every 5 mm; set max_accel to the highest band with small ringing and no widened test gap. With an accelerometer, SHAPER_CALIBRATE reports a suggested maximum for each input shaper.

Your printer	Test	What it gives you
Klipper with an accelerometer	SHAPER_CALIBRATE	A suggested max_accel per input shaper
Klipper without one	Ringing tower with TUNING_TOWER	The band where ringing or smoothing starts
Klipper, motor limit	TEST_SPEED macro	The highest acceleration before steps skip
Marlin	M593 ringing tower, then M201 and M204	Input shaper frequency; no acceleration test in Marlin’s docs
Bambu Lab, Prusa MK4S and CORE One, Creality K1	Built-in or factory calibration	Use the profile’s acceleration

With an accelerometer: SHAPER_CALIBRATE

1. Add the sensor to printer.cfg (an [adxl345] section plus [resonance_tester]) and check it with ACCELEROMETER_QUERY.
2. Turn the fans off and run SHAPER_CALIBRATE. It shakes each axis hard, so watch the first run and stop it with M112 if it gets violent. On a bed slinger, run SHAPER_CALIBRATE AXIS=X with the sensor on the toolhead, move it to the bed, then run SHAPER_CALIBRATE AXIS=Y.
3. Read the suggested max_accel lines. Klipper’s documentation shows this result for one axis:

Input shaper	Fitted frequency	Suggested max_accel
ZV	34.4 Hz	4,500 mm/s²
MZV (recommended)	34.6 Hz	3,500 mm/s²
EI	41.4 Hz	3,200 mm/s²
2HUMP_EI	51.8 Hz	3,000 mm/s²
3HUMP_EI	61.8 Hz	2,800 mm/s²

4. SAVE_CONFIG stores the shaper but leaves max_accel alone. Set it yourself under [printer], below the X and Y suggestions for your shaper: Klipper calls them “only a theoretical maximum”, and stepper torque can cap the printer lower. Klipper’s example:

[input_shaper]
shaper_freq_x: ...
shaper_type_x: ...
shaper_freq_y: 34.6
shaper_type_y: mzv

[printer]
max_accel: 3000  # should not exceed the estimated max_accel for X and Y axes

Without an accelerometer: the ringing tower

1. Slice ringing_tower.stl from Klipper’s docs/prints folder without rotating it. Use 0.2 or 0.25 mm layers, no infill or top layers, 1 to 2 perimeters (or vase mode with a 1 to 2 mm base), 80 to 100 mm/s external perimeters and a minimum layer time of 3 s at most. Turn off any dynamic acceleration control.
2. Print it once with input shaping off to measure the ringing frequency: speed × ripples counted ÷ distance measured. Klipper’s example is 100 × 6 ÷ 12.14 = 49.4 Hz. Put that in [input_shaper].
3. Run these commands and print the tower again. Stop early if the printer shakes hard or skips steps.

SET_VELOCITY_LIMIT MINIMUM_CRUISE_RATIO=0
SET_PRESSURE_ADVANCE ADVANCE=0
SET_INPUT_SHAPER SHAPER_TYPE=MZV
TUNING_TOWER COMMAND=SET_VELOCITY_LIMIT PARAMETER=ACCEL START=1500 STEP_DELTA=500 STEP_HEIGHT=5

Acceleration at any height is 1,500 + 500 × floor(height ÷ 5):

Band	Height	Acceleration
1	0 to 5 mm	1,500 mm/s²
2	5 to 10 mm	2,000 mm/s²
3	10 to 15 mm	2,500 mm/s²
4	15 to 20 mm	3,000 mm/s²
5	20 to 25 mm	3,500 mm/s²
6	25 to 30 mm	4,000 mm/s²
7	30 to 35 mm	4,500 mm/s²
8	35 to 40 mm	5,000 mm/s²
9	40 to 45 mm	5,500 mm/s²
10	45 to 50 mm	6,000 mm/s²
11	50 to 55 mm	6,500 mm/s²
12	55 to 60 mm	7,000 mm/s²

Then read the tower twice and put the lower of the two values in [printer] max_accel:

• Ringing: find the highest band where the ripples after the notches still look acceptable.
• Smoothing: find where the wall’s 0.15 mm gap starts to widen. In Klipper’s example it grows from 3,500 mm/s² (band 5), so max_accel is 3,000.

The max acceleration test for skipped steps

To find where the motors lose steps, run the TEST_SPEED macro from Ellis’ Print Tuning Guide, a community tuning guide for Klipper printers.

1. Add the macro to printer.cfg and set max_velocity to a fast speed you know works. Heat soak the printer first if it is enclosed.
2. Run TEST_SPEED ACCEL=5000 ITERATIONS=2. Raise ACCEL until the printer skips, then lower it until the skipping stops.
3. Confirm with 50 iterations, for example TEST_SPEED ACCEL=5000 ITERATIONS=50.
4. Watch and listen: shuddering, erratic moves and loud noises are major skips. For minor ones, compare the X and Y stepper positions printed at the first and second homing: a change of more than one full step (32 at 32 microsteps) means that axis skipped.
5. Save about 15% lower, because synthetic tests can pass and still skip in real prints. A pass at 10,000 mm/s² becomes 10,000 × 0.85 = 8,500 mm/s².

The guide warns these maximums are often “way higher than is reasonable to daily drive” and suggests re-testing after you tune input shaping. Print at or below the ringing-tower value.

How to calibrate acceleration on Marlin

Marlin’s documentation has no acceleration test, so you tune input shaping and then set the limits by hand:

1. Tune ZV input shaping (Marlin 2.1.2 and later) with the ringing tower on Marlin’s M593 page. It gives a frequency, not an acceleration: 29.84 Hz in Marlin’s example, saved with M593 X F29.84 and M500.
2. Send M503. The M201 line shows each axis’s maximum acceleration; the M204 line shows the printing (P), retraction (R) and travel (T) accelerations.
3. To raise acceleration, raise both: the planner lowers M204 values so no motor exceeds its M201 limit.
4. Find the ceiling by hand, the way Klipper’s ringing tower does it: print the same small test at your current M204 P, then again after each 500 mm/s² step up (M204 P2000, M204 P2500 and so on). Keep the last value that printed with small ringing and no layer shift.
5. Save with M500 if your build has EEPROM_SETTINGS enabled.

If the motors skip steps, Marlin’s driver guide says to check for binding, lower acceleration and jerk, improve driver cooling or raise motor current.

Bambu Lab, Prusa and Creality K1: use the profile’s acceleration

Use the maximum acceleration in the slicer’s printer profile. These printers tune vibration compensation themselves or come calibrated from the factory.

Printer	What it calibrates	Where	Profile or spec maximum, X and Y
Bambu Lab X1 Carbon, P1S	Vibration compensation	X1: Control > Calibrate. P1: Settings > Calibration. Bambu Studio: Devices > Calibrate	20,000 mm/s²
Prusa MK4S	Input shaper, pre-calibrated at the factory	Optional accelerometer: LCD Menu > Settings > Hardware > Input Shaper > Calibration	4,000 mm/s²
Prusa CORE One	Input shaper, pre-calibrated at the factory	Same as MK4S	10,000 mm/s² (7,000 extruding and travel)
Creality K1	Input shaping with its G-sensor	Settings > Self-check, tick Input Shaping	20,000 mm/s²

When to rerun the calibration:

• Bambu Lab: after moving the printer, a firmware update, belt tensioning, a part change or a drop in print quality.
• Creality K1: after a firmware update.
• Prusa: after hardware modifications. On a stock printer it is “usually unnecessary”.

Do you need an accelerometer to tune acceleration?

No. Klipper’s input shaper guide tunes from a printed ringing tower, and its “Selecting max_accel” step reads max_accel off the same print; an accelerometer is the alternative that “requires some crimping and soldering”. Marlin’s documentation also calls a printed ringing tower the easiest way to tune input shaping.

An accelerometer helps when the ripples are uneven and hard to measure, which Klipper says can mean several resonances on one axis. Sensor options:

• Klipper: ADXL345, MPU-9250 family, LIS2DW and LIS3DH chips. An ADXL345 connects over SPI to a Raspberry Pi or the printer board, and some toolhead boards carry one, such as BigTreeTech’s EBB36 and EBB42 CAN boards.
• Marlin: no accelerometer input of its own. Its docs point to a browser-based USB resonance tester (an RP2040 board with an ADXL345) that returns a shaper type and frequency.
• Self-calibrating printers: built in on Bambu Lab printers, the Creality K1 (a G-sensor) and the Prusa CORE One L and XL. On the MK4S and CORE One it is optional.

Don’t run shaper calibration before every print or every day: Klipper warns the vibrations wear the printer and can shake parts loose.

Set the slicer’s acceleration at or below max_accel

Keep every per-feature acceleration in the slicer at or below the value you found. On Klipper the slicer has the last word: an M204 or SET_VELOCITY_LIMIT ACCEL= in the G-code replaces max_accel for the moves that follow, even above the printer.cfg value.

Leave Klipper’s square_corner_velocity at its default 5 mm/s: with input shaping, a higher corner speed adds smoothing. Jerk vs acceleration explains the difference.

Enter your max_accel in the acceleration calculator to see what it buys in speed: at 3,000 mm/s², a 20 mm wall set to 300 mm/s peaks at 244.9 mm/s. The print time calculator estimates the whole job.