Introduction

Open any encoder catalog and three numbers dominate the front page: resolution, accuracy, repeatability. Engineers routinely treat them as interchangeable — "the 23-bit one must be the accurate one" — and specify a sensor on the wrong figure. The result is a machine that looks precise on paper yet drifts in production, or one that pays for accuracy it never uses.

This article separates the three terms with plain definitions and real TIMSENS time-grating specifications, then shows how to read a datasheet so the number you optimize is the one your application actually needs.

The Three Numbers, Defined

Resolution is the smallest detectable change in position — the fineness of the measurement, not its correctness. Think of it as the ruler's smallest graduation. A TIMSENS angle encoder delivers 23-bit default resolution (up to 32-bit), meaning it can report position in steps as fine as 360° / 2²³ ≈ 0.00004° (about 0.15 arc-second equivalent). High resolution lets the servo loop subdivide motion smoothly and hold a commanded position tightly — but it does not guarantee that position is true.

Accuracy is how close the reported position is to the real position — the deviation from ground truth across the full measuring range. TIMSENS SEF high-precision angle encoders reach ±0.5″ absolute accuracy (standard ±3″ / ±5″); AEF integrated units are ±1″; UEF ultra-thin (UEF072) ±20″; TEF economy ±10″; linear LSA/LE scales ±3 μm/m. Accuracy is what determines whether a machined part lands within tolerance — independent of how finely you can subdivide a wrong number.

Repeatability is how consistently the encoder returns the same reading for the same physical position across repeated approaches — the tightest of the three, and the one that governs cycle-to-cycle consistency. TIMSENS SEF repeatability reaches ±0.3″ (standard ±1.5″); this is why a machine can hit the same pocket location shift after shift even if its absolute accuracy is a few arc-seconds off nominal.

The Key Insight: High Resolution ≠ High Accuracy

A sensor can have exquisite resolution and mediocre accuracy. Resolution is about fineness; accuracy is about truth. Two encoders at 23-bit resolution may differ by 10× in accuracy depending on calibration, scale quality, and error compensation. Optimizing resolution alone never fixes a systematic angular or linear error — only calibration and metrology-grade construction (and, for time grating, stable time-interval measurement electronics) close that gap.

Conversely, an encoder with modest resolution but excellent accuracy and repeatability can outperform a "finer" rival on real parts, because the part cares about where the tool actually is, repeatably, not about how many meaningless digits the readout shows.

How the Three Interact in a Servo Loop

• Resolution sets loop smoothness and minimum controllable step — too low and the axis jitters or quantizes; high enough and the servo is effectively continuous.
• Repeatability sets part-to-part consistency — the spread you see across 1,000 cycles.
• Accuracy sets the absolute offset from true — correctable in software only if it is stable and characterized, which is why calibrated accuracy matters more than raw resolution.

For most industrial axes, the priority order is: repeatability first (consistency), accuracy second (true position / tolerance), resolution third (loop quality) — yet catalogs lead with resolution because the bit count is the easiest number to print big.

Real Specifications: TIMSENS Time-Grating Families

China's National Institute of Metrology certified the nano time-grating sensor at 0.19″ accuracy / 0.05″ resolution — internationally leading level. Source: High-Precision Displacement Time Grating Sensor Product Manual (2025.04), TIMSENS.

Reading a Datasheet Without Being Fooled

1. Find the test condition. Is accuracy quoted per revolution, per meter, or full-range? A ±3 μm/m linear spec and a ±0.5″ angular spec describe different geometries.
2. Separate "standard" from "high-precision" lines. TIMSENS SEF lists both — the ±0.5″ figure is the high-precision variant, not the default.
3. Confirm absolute vs incremental. Accuracy of an absolute encoder is meaningful on power-up; an incremental encoder's "accuracy" still needs an index seek first.
4. Check repeatability separately. It is often the smallest, most consequential number for production consistency.
5. Watch units. Arc-seconds (″) for angle, μm/m for linear, nm for resolution — mixing them is the common spec trap.