Run an OrcaSlicer G-code file, or a built-in pattern, through a Klipper-style planner and see the torque, belt tension, heat, and ringing each motor and driver goes through. Lost steps or a belt jumping a tooth end the run at the moment they would happen.
Technically viable
What this motor, belt tension, and heat can sustain, separate from the max_velocity and max_accel you set. Hover a card to see what limits it.
Viable speed500 mm/sstill has torque margin
Torque margin runs out near 500 mm/s. max_velocity is 300.
24 V is what holds that torque as speed rises. This board is rated to 29 V, so a higher supply needs a board rated above that. Current and the shaper do not raise this speed.
The typed limit is below this, so speed is not what is holding the printer back.
Viable accel5200 mm/s²limited by input shaper
Viable accel is 5200 mm/s² because of the input shaper. The motor itself reaches 38500 mm/s².
Smoothing near 0.12 mm is why the shaper cap is 5200 mm/s². A narrower shaper such as ZV raises that cap and leaves a thinner notch, so it only helps when the peak is sharp.
Torque ceiling38500 mm/s²worst move uses all of the motor's torque
On CoreXY the diagonal loads one motor with a share of both masses: 480 g and 1050 g, plus 2 N on each moving rail, and the rotor. One motor carries the whole belt.
The ceiling is where that move uses all of the standstill torque, 38500 mm/s². max_accel is 5000.
Run current is 1.41 A, 80% of the RMS nameplate. Heat still allows about 1.50 A, and torque rises with current up to that.
A second motor on the belt carries half the force. That is the largest mechanical raise.
Tension ceiling13100 mm/s²slack side keeps a little preload
At 9.0 N the slack side keeps a little preload up to 13100 mm/s². This span plucks at 110 Hz.
This printer's starting pluck is about 110 Hz.
A resonances CSV with the span tone in it puts a measured tension on this card.
Tension is below the torque ceiling. Raising it is what moves this card toward 38500 mm/s², until the belt or the 34.5 N pull is the limit.
Heat ceiling38500 mm/s²continuous current under 85% of RMS nameplate
Heat is not reducing the torque ceiling. Continuous current is under 85% of the RMS nameplate, so this card matches the motor.
A 4010 or a heatsink changes the settled temperature. It does not raise this number until run current sits above that share.
Shaper ceiling5200 mm/s²smoothing stays near 0.12 mm
MZV at 58 Hz on X, MZV at 42 Hz on Y. The 0.12 mm smoothing target is why this cap is 5200 mm/s².
The notch sits on the resonance fields.
Load the resonances CSVs to see which shaper the spectrum supports.
Recommendations
A 3010 on this driver would help
Still air settles the driver near 94 °C. A 3010 brings it to about 61 °C.
How the simulation works
Torque at speed comes from back-EMF (derived from holding torque and rated current) and the coil impedance: resistance and reactance at the electrical frequency, combined as a magnitude. Current is the supply voltage left after back-EMF, divided by that impedance, capped at the commanded current. Torque scales with that current, for one coil.
Moves go through the same planner Klipper uses: junction deviation from square_corner_velocity, the centripetal limit, the extruder corner limit, minimum_cruise_ratio, and lookahead to a stop at the end. CoreXY splits each force between the A and B belts. A Cartesian move keeps X on one belt and Y on the other. 2WD carries the whole belt force. Synced AWD carries half. A desync adds the hold the loaded motor spends fighting its partner, so an unsynced AWD run is not shown as easier than 2WD. Torque use at or above 100% ends the run as lost steps.
Belts split the pull around their preload. A slack side at zero ends the run, because the belt can jump a tooth. Ringing is a damped spring between motor and toolhead, driven by the input-shaped motion, including the speed jump at every corner. Shaper definitions and corner smoothing match what Klipper reports. Heat uses a first-order model for the winding (rated near 80 K rise at nameplate current) and for the driver (switch resistance and a typical heatsink). Hysteresis below what the coil needs, a slow chopper at high coil frequency, and stealthChop reduce the current the coil holds. Those factors are estimates, labelled in the recommendations.