Glass Encoder Disc Selection: Codes, Readout and Eccentricity

A glass code disc is a sensor component

A patterned glass encoder disc or linear scale modulates light at a readhead. The detector and electronics convert that modulation into incremental transitions or an absolute position word. The disc is part of a mechanical and optical loop, not an external test target.

Choose the encoding family

Incremental patterns use periodic tracks, commonly with A/B phase signals and an optional once-per-revolution index. Absolute patterns use multiple code tracks or another defined code sequence so the electronics can identify position without depending on a homing cycle. The released code table and readhead definition control compatibility.

Transmissive and reflective are optical choices

In a transmissive design, the source and detector are on opposite sides and the clear and opaque regions modulate transmitted light. In a reflective design, the source and detector share a side and the coating stack creates the required reflectance contrast. Neither geometry is universally more accurate; the mechanical envelope and readhead determine the choice.

• Specify wavelength, incidence angle, clear or opaque transmission, high and low reflectance, working gap and observed surface.
• Review rear-surface reflections and stray light when the glass is used in a compact reflective assembly.

Resolution and accuracy use different budgets

Resolution is the smallest reported increment, often affected by line pitch and electronic interpolation. Accuracy is the difference between reported and true position and includes pattern pitch, eccentricity, runout, readhead alignment, thermal expansion and electronics. More interpolation cannot correct a mechanical error.

Eccentricity creates a once-per-revolution error

For one readhead and an offset e small compared with active reading radius r, the first-order angular error is θ(α) ≈ (e/r) sin α, up to the selected phase origin and sign. The peak is approximately 206265e/r arcseconds when e and r use the same units. Use the active reading radius, not half of the glass outside diameter unless the track is actually located there. The approximation estimates sensitivity; installed errors require measurement.

Control the mechanical datum

Pattern-to-bore centring and the installed pattern-to-axis centring are different quantities. Hub fit, shaft alignment and radial motion can move the active centre even when the loose disc meets its drawing. Axial runout and tilt can change sensing gap and optical contrast. Measure the installed datum relationship and avoid clamp or bonding stresses that deform the glass.

Set pitch, line width and substrate together

A finer pitch increases feature density but leaves less margin for etch bias, line-width variation, contamination and readhead sampling. Glass grade and thickness affect thermal stability, stiffness and the optical path. Choose the material from temperature, wavelength, diameter and mounting needs rather than from a resolution number alone.

Validate the installed signal

Inspect A/B amplitude, phase relationship, duty cycle, index timing or absolute code transitions under the real readhead and speed range. Record the disc revision, readhead gap, wavelength, temperature and mechanical runout. A microscope image of the tracks cannot prove the complete encoder signal.

Use a release checklist

Before release, confirm the code table or line count, optical mode, material, diameter or scale length, bore and datum, pattern-to-datum accuracy, flatness, temperature range, surface condition and inspection record. Keep system accuracy separate from the glass component drawing.

Use harmonics as diagnostic clues

Represent a repeatable angle-error curve as ε(α) = a0 + Σ ak sin(kα + φk). A strong first harmonic is consistent with eccentricity but does not uniquely prove its cause. Higher orders can come from pattern placement, reader geometry or electronics; within-period interpolation error uses the signal-period scale. Compare error after re-centring, reader adjustment, remounting and direction reversal. Harmonic order alone cannot separate all mechanical and optical causes.

Understand multi-readhead averaging

For m equally spaced heads observing the same repeatable angular-error function, equal-weight averaging cancels ideal harmonic terms whose order is not a multiple of m. Two opposed heads cancel odd orders and retain even orders; four heads retain multiples of four. Unequal spacing, weighting, zero offsets and local reader errors leave residuals. This ideal geometric result is separate from four-field sensing inside one optical period.

Validate compensation outside the fitting data

Measure output against an angle reference with suitable uncertainty and sample density. Fit repeatable error or store a lookup table, then verify corrected results with independent positions and both directions. Sampling can alias error harmonics, so point count alone does not establish what has been captured. A static table cannot remove random noise, backlash that changes with direction, or changes caused by temperature and remounting. Record the conditions under which the correction remains valid.

Explore the related product ranges

Related technical guides