Optical Encoder Reference Mark and Datum Strategies for Reliable Axis Homing

Why the datum anchors everything downstream

An incremental scale reports motion, not location. Until the axis references, every commanded position is a guess dressed in micrometres. Tool offsets, pallet origins, measurement baselines and process zero all inherit the quality of the homing event, so a reference mark that repeates to a few counts on one approach and a different few counts on another injects scatter into every downstream number.

This is why datum strategy belongs in the axis specification. The properties that matter are not exotic: repeatability of the mark at the chosen approach speed, independence from approach direction or consistency when direction is fixed, and behaviour under the awkward cases that real machines encounter, such as stopping inside the reference window or coasting past it after a fault.

Single reference marks: simple and position-sensitive

The classic scheme places one reference mark beside the incremental track. The controller counts at high speed toward the mark, slows, re-approaches at low speed and latches position on the mark edge. Because the same mark is used every time, the mechanical quality of that single location defines the datum: mark edge quality, scanning quality at the approach speed and the electronics gating window all add their scatter.

The approach profile dominates the result. Homing at production speed onto a mark specified for a slow approach produces spread that grows with velocity, and approaching from opposite directions can land on different sides of the gating window. The remaining structural weakness is placement. A single mark concentrates all referencing capability at one location, so damage, contamination or a mechanical obstruction there disables the datum for the whole axis, and the mark must sit outside the work area yet remain reachable before any limit is hit.

• Simplest electronics and lowest scale cost
• Repeatability depends on fixed approach speed and direction
• One mark means one sensitive location for contamination and damage
• Placement must clear travel limits, cable extremes and the work area

Distance-coded marks: datum from anywhere

Distance-coded scales carry a series of reference marks whose spacing follows a known mathematical sequence. After power-up, the axis moves until the next mark is scanned; because the spacing pattern is unique along the scale, a single mark observation, or a short sequence of marks, is enough for the controller to compute absolute position within one code period. The homing traverse shrinks from full travel to the distance to the next mark, and after an uncommanded stop anywhere on the scale, recovery needs only the same short scan.

The scheme removes the single-point failure of the lone mark and makes long-travel machines far more tolerant of random stop positions, which is why it is common on long beds, gantries and measuring machines. The specification points shift accordingly: confirm the mark spacing, the maximum traverse needed to find the next mark, the minimum and maximum speed the scanning accepts during the recovery move, and whether the controller requires motion in a defined direction to decode.

Absolute codes and when they replace referencing

An absolute encoder answers the datum question at the source: position is read directly from the code at switch-on, no traverse needed. For machines where downtime is expensive, where axes stop at arbitrary positions or where a homing traverse is mechanically awkward, absolute feedback is less an upgrade than a different contract with failure modes. There is no homing to be non-repeatable, and recovery after power loss is immediate.

The choice is economic as much as technical. Incremental scales with distance-coded marks deliver most of the recovery benefit where homing time is tolerable; absolute code earns its premium where every restart counts or where the axis cannot safely traverse on power-up.

Specifying and verifying the homing routine

Whatever the mark scheme, the controller routine carries half the responsibility. Fix the approach speed, the direction and the final latching speed as parameters, not as whatever the commissioning engineer chose that day. Gate the reference capture against the enable state and the incremental signal quality so a degraded scan cannot produce a false datum, and log every homing event with its starting position and resulting datum so drift becomes visible in maintenance data instead of in process scrap.

Verification closes the loop. Home the axis twenty or thirty times from spread-out starting positions at production conditions, record the datum scatter and compare it against the encoder datasheet figure. A datum strategy that passes those checks is one the machine can rely on for its working life; one that was never tested is a fault waiting for the worst possible shift.

• Approach speed, direction and latching speed fixed as parameters
• Reference capture gated on signal quality and enable state
• Homing events logged with start position and resulting datum
• Acceptance test: repeated homing from varied positions plus power-loss recovery