Motor calibration

The Motor calibration box on Settings, below Machine configuration, has two routines. Run them in order:

1. Drive and encoder check. Checks that Encoder (step/mm) and Servo (step/mm) match the hardware and that the encoder counts the way the carriage moves. It drives the carriage 5 mm forward and back at 1 mm/s open loop: the servo runs its usual motion profile but ignores the encoder, so the drive gets exactly the pulses the move calls for, and the encoder is only watched. These moves are sent as G124, a linear move with the servo feedback off.
2. Motion calibration. Runs a short test program closed loop. It checks the machine can deliver Acceleration Max, measures how closely the servo follows its encoder, and suggests servo gains (Kp and Ki). It unlocks only once the check has passed in this session on the steps/mm saved on the machine, because it moves on those, and it uses the drive lag the check measured.

Both are report only. Nothing is written to the machine: you change the fields (and Write to device) yourself.

The saved profile's motion limits must be right first

The motion calibration is built from the profile saved on the machine: Velocity Max, Acceleration Max and Position Max. Its longest move is a tenth of Position Max, swung half either side of where the carriage starts, at up to Velocity Max. The panel shows the distance and speeds before you run it. Leave that much room each way.

Before you run it

• Enable motion and clear any fault.
• Leave the carriage clear of both end stops, with the room the panel asks for.
• Don't start a test or jog while it runs.

Before each measured leg of the check, a 0.5 mm move in the same direction takes up any slack, so slack doesn't count against the ratio. The panel disables motion if the encoder moves more than three times the distance the profile expects.

The check judges the steps/mm you have typed. If those differ from the ones saved on the machine, Run motion calibration stays locked and the panel asks you to write them, restart the machine, and run the check again.

Writing the configuration offers Reboot now, which restarts the machine. You can also power-cycle it.

While a routine runs, its button shows the step it is on and the other routine is locked:

Reading the result

The drive and encoder check shows a pass, warning or error, then what the run measured:

Sent 32000 pulses each way → encoder moved 12000 counts (expected 12000)

The pulses are what the machine sent, and the counts are what the encoder saw come back (averaged over both directions). If the encoder counted only one way, or counted the same way in both directions, an average would point at the wrong fault, so the line shows each direction's counts instead (for example "forward 2400 · back 2400"). Expected is what your two steps/mm values predict. When they don't match, a Fix line gives the value to change. One run only measures the ratio of the two values, so it gives both options. Trust the one you can work out from datasheets (usually the encoder: counts per revolution ÷ mm per revolution).

It also reports:

• Lost motion on reversal: how far the carriage falls short on the move straight after a change of direction. This is slack, backlash or the motor settling against friction. Harsh reversals (a high Acceleration Max) make it larger.
• Drive lag: how far behind the pulses the encoder runs at a steady speed.

The motion calibration program ramps at Acceleration Max up to a quarter, half, three quarters and all of Velocity Max, each speed out, back and out again so both directions are covered. Every move the firmware runs accelerates at Acceleration Max, so ramping to rising speeds is how it tests that one value: a motor has less torque the faster it turns. Then come 1 mm, 0.1 mm and 0.01 mm moves in both directions.

Acceleration says whether the carriage kept up:

• pass: it delivered Acceleration Max at every speed up to the one shown. If a tenth of Position Max is too short to reach Velocity Max at that acceleration, the report shows the speed it did reach, and ramps to higher speeds are untested.
• too high: on at least one ramp the carriage delivered under 80% of Acceleration Max. A carriage that can't accelerate as fast as its trajectory falls behind it through the ramp, so the following error beyond what the drive lag explains, divided by the ramp's length, is the share it fell short by. Lower Acceleration Max (or Velocity Max), write, restart and run it again.
• not judged: the run stopped before every speed had been measured.

The table gives each speed's peak following error and the share of Acceleration Max delivered. The servo's corrections during a ramp hide part of a shortfall, so treat a marginal pass with care; a carriage that stalls fails clearly.

Tracking sums up the program: the largest stop error, the peak following error (the gap between where the trajectory is and where the encoder is), the largest overshoot, and the longest settle time. Each move expands to the per-move numbers. These are measured against the machine's own encoder: they show how well the servo follows it, not the carriage's true position, which needs an external reference.

Suggested servo gains gives a starting Kp and Ki, with the reasons:

• Kp is the largest that is still safe: at most 0.5 ÷ the drive lag (above that, corrections arrive too late and the carriage rings), and at most Acceleration Max ÷ Velocity Max (above that, an ordinary error asks for more than the acceleration limit).
• Ki is only suggested when moves end consistently short (or past) in their direction of travel, at Kp² ÷ 8. Otherwise it stays 0, because an integral with nothing to correct only adds overshoot.

To try them, enter Servo Kp (1/s) and Servo Ki (1/s^2), Write to device, restart the machine, and run the motion calibration again to compare.

What the findings mean

Finding	What it means	What to do
Calibration passed	The encoder follows the motor in the right direction, and the two steps/mm values match the hardware (within 2%)	Nothing
The encoder did not count in either direction	No encoder counts at all	Check that the encoder is coupled to the motor, that A and B reach the board, that it shares a ground with the board, and that its outputs have pull-ups
The encoder counted only one way	One channel is missing, or the carriage did not move one way	Check the encoder wiring and that the carriage moves both ways
The encoder counts the opposite way to the motion	The encoder is reversed relative to the drive. The servo loop would run away as wired	Swap encoder A and B, or reverse the drive's direction setting
The encoder counted the same way whichever way the carriage was driven	The carriage did not reverse, or the direction signal is wrong	Check the drive direction wiring
The steps/mm entered don't match the hardware	The counts that came back differ from the expected counts by more than 2%, so one of the two steps/mm values is wrong	Use the Fix line: change the value you trust less
Forward and back differ by n%	The two directions disagree by more than 2%	Look for slip, backlash or lost pulses

Runs that don't finish

These are for the drive and encoder check. The motion calibration stops with "the carriage went further than a move asked for", "motion was disabled", "a move never finished" or "the machine turned a move down".

Message	Why	What to do
Not run: enable motion (and clear any fault) first	Motion was off or the machine was faulted	Enable motion, clear the fault
Not run: the machine is busy with a move or a test	A test was running, or the machine turned the move down	Wait for it to finish
Not run: no live position from the machine	No samples are arriving	Check the connection
Stopped: motion was disabled	Motion was turned off mid-run, or a fault disabled it	Clear the cause, run it again
Stopped: the carriage was still moving long after the pulses should have finished	It hadn't settled after twice the move's time plus 3 seconds	Check the drive and the mechanics
Stopped: the encoder moved over three times further than the profile expects	The drive is set much finer than servo steps/mm says, or encoder steps/mm is far too low	Check both values