When you scale an optical design from standard-sized components down to micro-optics, something counterintuitive happens with centration tolerances. The same edge thickness runout that barely matters on a 25 mm lens can produce ten times the angular deviation on a 2mm lens. Understanding why—and how to specify tolerances that actually work at the micro scale—is the difference between a successful build and an expensive lesson.

This isn't about tightening every tolerance. It's about knowing which tolerances matter more as the diameter shrinks, and specifying them in ways that are both achievable and measurable.

*Note: Our online catalog refers to the centration tolerance specification as centering. Centering and centration are often interchangeable terms in the industry.

What Centration Tolerances Describe

Centration defines the alignment between a lens's optical axis and its mechanical axis. The optical axis runs through the centers of curvature of the lens surfaces—it's where light wants to go. The mechanical axis is defined by the lens edge—it's how the lens sits in your mount.

When these axes don't coincide, you have centration error. This error has two components:

• Decenter is the lateral displacement between the axes—the optical axis is shifted sideways from where it should be.
• Tilt (also called wedge) is the angular misalignment—the optical axis is tilted relative to the mechanical reference.

Both errors cause a light ray passing through the lens to deviate from its intended path. This deviation introduces aberrations, shifts images, and degrades system performance. In multi-element assemblies, deviations from each element combine vectorially, meaning small errors accumulate into system-level problems.

For standard-sized optics, centration tolerances are well-understood. The industry has decades of experience specifying and measuring them. But micro-optics—components with diameters from a few millimeters down to sub-millimeters—present challenges that aren't obvious until you encounter them.

Why Micro-Optics Are Different

The physics of centration doesn't change at small scale. What changes is the relationship between manufacturing realities and optical consequences.

Manufacturing equipment, measurement instruments, and assembly fixturing all have practical accuracy limits. For example: 

• Spindles have finite runout.
• Indicators have finite resolution.
• Lens seats have finite parallelism. 

Whatever those limits are for a given equipment set, they don't shrink with the workpiece—they apply regardless of diameter. But throughout the tolerance chain, these equipment floors appear in the numerator of geometric relationships while lens diameter D appears in the denominator. The same capability produces proportionally larger optical errors as lenses get smaller.

This "D in the denominator" pattern shows up everywhere.

In Manufacturing

When edging a lens, TIR (Total Indicated Runout) measures edge thickness variation. The wedge angle α = TIR / D, and beam deviation δ =TIR × (n−1) / D. If a particular spindle and indicator combination floors at12.5 μm TIR, that same floor produces 0.5 mrad wedge on a 25mm lens but 6.25mrad on a 2mm lens—12× larger for the same equipment.

The indicator resolution matters independently. If an indicator resolves to 2.5 μm, the minimum detectable wedge is ε / D: about 0.1 mrad at 25mm, but 1.25 mrad (4.3 arcminutes) at 2mm. Below that threshold, you cannot detect the error, which means you cannot verify the lens, regardless of how well you actually made it.

In Assembly

The pattern continues through mounting. A precision fit allowing 20 μm worst-case diametral clearance between lens OD and barrel ID permits 10 μm lateral decenter—fixed regardless of lens size. But that 10 μm is0.04% of a 25mm lens and 0.5% of a 2mm lens.

Same clearance, 12× larger relative error.

Lens seat parallelism creates a subtler problem. A surface with TIR_seat across contact diameter D_contact tilts the lens by θ = TIR_seat/ D_contact, shifting the center of curvature by Δ = R × θ, where R is the lens surface radius of curvature. A spacer with 5 μm parallelism over a 2mm contact, seating a lens with R = 2mm, shifts the center of curvature Δ by 5 μm. Two lenses on opposite faces of that spacer see a 10 μm relative shift. A "tight" spacer spec can consume your entire centering budget.

Focal length compounds the problem. Beam deviation from lateral decenter is δ = Δ / f, and micro-optics typically have short focal lengths.

At f/2, a 10 μm decenter produces 0.2 mrad deviation on a 25mm lens (f = 50mm) but 2.5 mrad on a 2mm lens (f = 4mm). Same decenter, 12× worse deviation.

The Pattern

Equipment capability in the numerator, lens diameter in the denominator. This is why specifying cost-effective micro-optics requires meticulous consideration of element manufacturing, mounting component tolerances, and assembly methods together—more so than for larger optics, where margin absorbs small mismatches.

Where to Spend Your Tolerance Budget

At the micro scale, the margin disappears, and allocation becomes explicit. Consider two 2mm plano-convex lenses (R = 2mm), convex surfaces facing, requiring ≤10 μm mutual centering.

You could constrain by OD: sort lenses to ≤7 μm deviation each to hit 10 μm RSS. Or you could constrain by surfaces: seat one plano on a precision shoulder, use a tight-tolerance spacer to reference the convex radii to each other, and tighten the spacer—not the lens—to close the budget.

The difference matters. A 50% yield hit on a $5 spacer costs $5. A 50% yield hit on a $200 lens costs $200.

Push tight tolerances onto cheap, easy-to-measure components; protect yield on expensive ones.

This is the micro-optics tolerancing challenge: understanding where D appears in the denominator throughout your tolerance chain, and deliberately choosing which components absorb the tightening.

Looking for a stock or custom micro lens to fit your optical system?

Explore our catalog to find the perfect fit for you.

Specification Language vs. Shop Floor Reality 

When you write a centration tolerance on a print, that specification must eventually be verified on the shop floor. But the language of specification and the language of measurement don't always align.

Formal drawing specifications per ISO 10110-6 express centration as surface tilt (σ) in arcminutes or arcseconds, or as beam deviation. These are optical parameters. On the shop floor, the centering machine measures TIR—the total indicator movement as the lens rotates on the spindle. TIR is a mechanical measurement that correlates to wedge, but isn't the same quantity as the angular specification on your print.

This disconnect matters for micro-optics because the conversion between TIR and angular deviation depends on diameter. A shop achieving a consistent 12.5 μm TIR produces very different angular results at different scales. When your specification calls for angular deviation but the process controls TIR, someone has to translate between the two—and that translation must account for diameter.

For incoming inspection, the same question arises: can the tolerance on the print actually be measured on the finished part?

Angular Deviation as Specification Language

Given this scale dependency, how should you specify centration for micro-optics?

Angular deviation is often the most useful metric because it directly describes the optical consequence—how much the lens bends a ray from its intended path. Rather than specifying TIR or decenter in linear units (which mean different things at different scales), angular deviation normalizes the specification to what matters for system performance.

Decades of industry experience provide rough capability benchmarks:

• 5 arcminutes: Achievable without special effort on standard equipment
• 1 arcminute: Readily achievable with good equipment and practices
• 15 arcseconds: Requires very special care, precision equipment, and experienced operators

But notice: they're expressed in angular units, not linear units, precisely because angular deviation is what the optical system experiences.

For micro-optics, you need to work backward from your system's tolerance budget. How much angular deviation can your system accept from each element? What's the cumulative budget across all elements?

Only then can you determine whether a given centration tolerance is appropriate—and achievable—for a micro-lens at your diameter.

Common Specification Mistakes

Engineers new to micro-optics procurement often make predictable specification errors. Recognizing these patterns can save time and money.

• Over-specifying: Copying tolerances from a similar but larger lens, not realizing that the same angular deviation tolerance is much harder to achieve at a smaller diameter. The lens might be manufacturable, but you’ll pay for unnecessary capabilities — or the supplier may reject unachievable specs.
• Under-specifying: Using "standard" tolerances without calculating whether they provide adequate angular deviation control at the micro scale. The lens meets its specification, but your system doesn't perform because the allowable deviation was too large.
• Mis-specifying: Expressing centration in units or conventions that don't translate well to micro-optics manufacturing.
• Specifying what can't be measured: Requesting tolerances tighter than what can be reliably verified on the finished part. If you can't measure it, you can't confirm it, and you're relying on process control rather than inspection. For critical applications, the tolerance and the measurement capability need to align.

Each of these mistakes stems from the same root cause: treating micro-optics specifications as a direct scale-down of standard optics specifications, without accounting for how the physics and manufacturing realities shift.

How Ross Optical Helps

Ross Optical supplies precision optics across a broad range—spherical and aspherical lenses, prisms, windows, filters, and custom assemblies—with micro-optics as one specialized segment of that portfolio. This breadth matters because experience across the full size spectrum provides a clear view of where challenges change as lenses get smaller.

We see the transition points where customers encounter trouble. A tolerance that's routine at 15mm diameter might be achievable but difficult at 5mm, and approaching equipment limits at 2mm.

Knowing where these thresholds fall—and how they vary by lens geometry and material—is knowledge built from manufacturing thousands of optics across the size range.

What we offer for micro-optics procurement:

• Tolerance guidance from manufacturing experience. We can review your specification and identify tolerances that may be problematic at your diameter—before you commit to a design that's difficult or expensive to build.
• Translation from system requirements to lens specifications. If you know how much angular deviation your system can tolerate, we can work backward to appropriate centration tolerances that account for your specific lens geometry, diameter, and material.
• Measurement capability and data. We can measure centration on individual micro-optics and provide data for your incoming quality assurance. If a tolerance is on your print, we can verify it—or tell you if it's beyond practical measurement limits.
• A partnership approach. We flag problems before they become your problems. If a tolerance doesn't make sense—too tight for the application, too loose for the system requirement, or specified in a way that creates ambiguity—we'll say so. The goal is lenses that work in your application, not just lenses that meet a print.

Getting It Right

Centration tolerances are essential to micro-optic performance, but they don't exist in isolation. At the micro scale, three effects compound: wedge angle amplification, diameter tolerance play, and focal length sensitivity—all with diameter in the denominator. Defining tolerances correctly requires understanding how the physics, the manufacturing, and the measurement all shift together as lenses get smaller.

Ross Optical brings perspective from the full optics size range to every micro-optics project. We can help define tolerances that are realistic, manufacturable, and right for your application.

If you're specifying micro-optics and want to ensure your tolerances make sense, contact our engineering team. We're ready to help.