How Design Datums Affect Optical Axis Alignment in CNC Camera Parts?

 

 

 

When you hold a high-performance camera lens, you are holding an assembly of multiple glass elements. The optical axis is the imaginary centerline that must run perfectly straight through the exact center of every single lens element, all the way to the image sensor.

 

If every lens element, the housing bore, and the sensor mount share this exact line with micron-level agreement(datums), the resulting image is sharp corner-to-corner. If they do not, the system suffers from soft edges, one-sided blur, or a tilted focal plane. In this blog, we will dive into the optical axis datum importance, and how to keep datum correct for CNC camera parts, and finally share a case study of how our factory successfully solves this problem for our clients.

 

 

 

Why Is an Optical Axis Datum Important?

 

In CNC machining, the optical datum acts as the foundation of the entire assembly:

 

• On a lens barrel, it is typically the precision internal bore that seats the lens elements.
  
• On a camera body, it is the lens mount flange face combined with the sensor mounting plane.

 

If the datum surface is tilted or offset by just 5 microns during machining, that error compounds as more parts are stacked together. By the time the light reaches the sensor, it can become a 25-micron focal plane tilt. For high-resolution cameras (50MP to 100MP+), a 25-micron tilt is more than enough to ruin image clarity.

 

In practice, standard CNC shops often treat these critical surfaces like ordinary mechanical holes. As long as the hole diameter is correct and it looks round, they ship it. This misunderstanding is the root cause of massive failure rates during the final optical assembly. How to solve this? It is a must to pay attention to both dimensional precision and geometric precision.

 

 

 

 

Dimensional Precision vs. Geometric Precision: How CNC Parts Get Warped

 

 

It is not enough to achieve ±5μm dimensional precision, but also to pay attention to geometric precision. Here’ s an example:

 

A typical CNC shop receives your lens housing drawing. They see a bore with a tight diameter tolerance. They chuck the part, cut the bore, measure the diameter — it’s within spec. They ship it. You assemble the lens. The image is soft on the left side.

 

Why? Because the bore was cut with the part held on an OD surface that had a 15 μm runout relative to the flange face — which is your real datum. The bore is perfectly round and perfectly sized. But its axis is tilted 15 μm relative to the face that seats against the next component in your optical stack. The CNC shop hit the dimensional tolerance. They missed the geometric one — because your drawing didn’t make the datum relationship explicit.

 

This is the gap between “CNC precision” and “optical precision.” CNC precision means: I can cut this feature to this number. Optical precision means: this feature’s position and orientation relative to the datum are controlled to sub-micron levels, across multiple setups.

 

 

 

 

The Two Machining Methods — Mechanical Centering vs. Error Compensation

 

 

 

When a lens housing bore must align with the optical datum to within 1 to 2 microns, factories typically choose between two different manufacturing routes. They yield very different results at different price points.

 

 

Method A: Mechanical Centering (Alignment Turning)

 

 

The part is mounted onto a ultra-precision rotating spindle. An operator or an optical autocollimator checks the baseline surface, manually adjusting the part until any wobble or runout drops near zero (under 1 micron). Once perfectly centered, the lathe machines all internal optical bores and steps in a single setup.

 

• Advantage: Because the part is never unclamped or flipped, every single turned surface perfectly shares the exact same rotational center. Coaxiality stays under 1 micron, completely avoiding errors caused by re-clamping.
• Limitation: Requires specialized, expensive equipment (such as dedicated alignment turning stations), cycles are slower, and per-part costs are high.

 

 

Method B: Adaptive Turning (Error Compensation Machining)

 

 

This method uses a standard CNC lathe or mill-turn center. Before cutting, the machine uses a physical touch-probe to measure the part's misalignment. The machine's software then calculates an offset and commands the cutting tool to adjust its path to "compensate" for the error.

 

• Mechanical Limitation: This approach can only correct for lateral shifting (left/right or up/down offsets); it cannot effectively fix angular tilt. When a part is spinning rapidly on a spindle, standard CNC toolposts cannot move back and forth fast enough to track a tilted axis. Attempting to mill the bore with a rotating cutter instead of turning it avoids the speed issue but causes the hole to lose its perfect roundness.
• Advantage: Faster cycle times, utilizes standard workshop machinery, and costs much less.
• Other Limitations: In real production, centering accuracy usually drifts above 5 microns, and angular tilt remains uncorrected. It is only suitable for non-critical structural brackets or outer protective covers.
  

 

 

Table of Mechanical Centering vs. Error Compensation

 

 

 

 

        

How Your Datum Strategy Determines Alignment Precision

 

 

The datum you choose on your drawing directly determines how your supplier will fixture the part — and therefore where the errors end up.

 

 

Datum Type Matters

 

• Planar datum (a flat face): Controls one translation (axial position) and two tilts. Good as the primary datum for a lens mount flange — it sets the plane perpendicular to the optical axis. Machine this face first, then reference it for all subsequent bore cuts.
• Cylindrical datum (a bore): Controls two translations (radial position) but not axial. Good as the secondary datum for a lens barrel. But it only works if the bore is accessible for the fixture — and if the supplier actually indicates on it, not on the OD.
• V-groove datum (a pair of angled faces): Kinematic. Six points of contact. Repeatable to sub-micron levels when used with precision balls. Common in optical bench setups and interchangeable lens mounts. If your part needs to be removed and re-mounted while maintaining alignment, this is the only option that works.

 

 

The Datum Hierarchy Rule

 

Your drawing should establish a clear datum precedence:

 

• Primary datum (A): The surface that locates the part in the optical assembly. Usually the lens mount flange face.
  
• Secondary datum (B): The feature that centers the part. Usually the largest optical bore.
  
• Tertiary datum (C): The feature that clocks rotation. A dowel pin hole, a flat, a keyway.
  

 

This A-B-C sequence isn’t just GD&T formalism. It tells the machinist: “Indicate on A first to set the plane, then center on B, then clock on C.” In that order. Every time. That process discipline is what produces repeatable optical alignment.

 

 

 

 

Tolerance Stack-Up in Optical Assemblies — A Worked Example

 

 

 

Here’s how datum errors compound through a real optical assembly. Take a camera with a lens mount flange, a spacer ring, and a sensor PCB.

 

 

The Components

 

• Lens mount flange — machined from 6061-T6 aluminum. Datum A = flange face. Optical bore Datum B, concentricity to A ≤ 3 μm.
  
• Spacer ring — ground stainless steel, 12.000 ±0.005 mm thickness. Parallelism ≤ 2 μm.
  
• Sensor PCB mount — machined aluminum plate. Datum A = sensor seating face. Flatness ≤ 3 μm.
  

 

 

The Stack-Up

 

 

 

For a 24MP APS-C sensor with 3.9 μm pixels, a 4.3 μm focal plane tilt across the diagonal means one corner is more than 1 pixel out of focus. At 100MP on a full-frame sensor with 2.9 μm pixels, that same tilt puts you nearly 1.5 pixels out.

 

This is why manufacturers of high-resolution cameras — scientific, cinema, surveillance — work with suppliers who understand datum structure. It’s not magic. It’s stack-up arithmetic and process discipline.

 

 

 

Where Can You Tighten?

 

The spacer ring is the biggest controllable error source. A ground shim can hold ±1 μm parallelism and ±1 μm thickness — but costs 3-5x more than a standard ground ring. The sensor mount perpendicularity is process-dependent: single-setup machining (all features cut without re-chucking) holds ≤3 μm. Multi-setup (part flipped and re-fixtured) typically drifts to 10+ μm.

 

 

 

 

Key Material Choices for Optical Datum Parts

 

 

The material you choose affects machinability, stability, and how well it holds a datum over temperature and time.

 

 

 

 

 

The Heat Problem

 

Machining generates heat. Heat expands the part. You cut a bore to 50.000 mm. The part cools. The bore is now 49.997 mm — and the datum face has relaxed by 2 μm.

 

This is why optical datum parts require:

 

• Coolant flood, not mist — temperature stability during the cut
  
• Roughing + stress-relief + finishing sequence — not rough-to-finish in one go
  
• Inspection at thermal equilibrium (20°C per ISO 1) — not fresh-off-the-machine
  

 

A supplier who measures bores immediately after machining is reporting dimensions that do not represent the part you’ll receive.

 

 

 

How to Specify Optical Datum Requirements in Your RFQ

 

When preparing a drawing package for an optical component supplier, verify these five process capabilities to ensure they can hold tight optical tolerances.

 

 

1."What is your procedure for establishing the primary datum during setup?"

 

• Acceptable: "We indicate the primary datum face on our fixture using a sub-micron dial indicator or autocollimator to verify Total Indicated Runout (TIR) before clamping and cutting."
  
• Unacceptable: "We drop the raw stock into a standard vise and rely on the machine's default alignment."

 

 

2."Are all critical optical features machined within a single setup?"

 

• Acceptable: "Yes, all internal bores, registers, and seating faces sharing the optical axis are cut sequentially without unclamping or re-chucking the part."
  
• Unacceptable: "We machine side A, then flip the part over to finish side B on a separate fixture."

 

 

3."What specific metrology equipment do you use to verify geometric alignment?"

 

• Acceptable: Sub-micron coordinate measuring machines (CMMs), optical alignment telescopes, laser interferometers, or specialized alignment turning metrology.
  
• Unacceptable: Relying entirely on manual vernier calipers, micrometers, or low-resolution inspection probes.

 

 

4."What is your environmental control and thermal stabilization protocol?"

 

• Acceptable: "Finished parts are held in a climate-controlled room at 20℃±1℃ for at least 4 to 6 hours to achieve thermal equilibrium before final inspection."
  
• Unacceptable: Parts are measured directly on the shop floor immediately following machining.

 

 

5."Can you provide a First Article Inspection Report (FAIR) validating all datum-related GD&T callouts?"

 

This is not optional for optical parts. The FAIR proves they can hit your tolerances on the first run. If they won’t provide one, they’re not confident they can.

 

 

 

 

VMT CNC Machining Factory Case Study

 

 

The Problem: Lens Mount Alignment Across a Multi-Vendor Supply Chain

 

 

 

A European scientific camera manufacturer approached VMT in 2025. Their product — a cooled CCD astronomy camera — required the sensor plane to be parallel to the lens mount flange within 5 μm across the full sensor diagonal (43 mm). The assembly had three precision-machined parts from three different suppliers:

 

• A lens mount flange (aluminum 6061-T6, with a C-mount thread and a locating bore)
  
• A sensor PCB mounting plate (aluminum 6061-T6)
  
• A set of invar spacer shims between the flange and the sensor plate
  
• All three suppliers were machining to print tolerances. But assembled cameras were showing 15-25 μm focal plane tilt — 3-5x the specification. The manufacturer was rejecting 30% of assembled units. Rework was costing them.

 

The Root Cause

 

When VMT engineers analyzed the drawing package and incoming parts, three failures emerged:

 

• Supplier 1 (flange): The drawing specified the C-mount thread as Datum B, concentric to Datum A (the flange face) within 10 μm. The supplier cut the thread in a second setup — flipping the part after machining the face. The re-chucking introduced 8-12 μm of position error. The thread bore axis was tilted relative to the flange face.
  
• Supplier 2 (sensor plate): The supplier achieved flatness and parallelism specs. But the part was measured at machine temperature (~30°C after machining). At 20°C, the sensor mounting face relaxed by 4 μm — exceeding the 3 μm flatness requirement.
  
• Supplier 3 (invar shims): The shims were correct in thickness. But they were delivered with burrs from stamping. The burrs — only 5-8 μm tall — acted as microscopic springs between the flange and the sensor plate. Each burr contact point introduced a local tilt. Across three shims, the cumulative tilt was unpredictable.

 

VMT’s Solution

 

The manufacturer moved all three parts to VMT for single-source production:

 

• Single-setup machining: All optical bores and datum faces on the flange were cut in one setup on a mill-turn center. No re-chucking. Concentricity of Datum B to Datum A measured at 1.8 μm average — well below the 10 μm print tolerance.
  
• Thermal protocol: All optical parts were stabilized at 20°C ±1° for 6 hours before final CMM inspection. FAIRs reported dimensions at thermal equilibrium.
  
• Datum-first fixturing: The fixture located on Datum A (flange face) as the primary reference. Bore alignment was verified with an autocollimator before cutting started.
  
• Burr-free shims: Invar spacers were precision-ground rather than stamped. Edge breaks controlled to <2 μm burr height. Each spacer measured for parallelism individually.

 

The Results

 

• Focal plane tilt reduced from 15-25 μm to 3-6 μm across the full sensor diagonal
  
• Assembly rejection rate dropped from 30% to under 2%
  
• The manufacturer eliminated incoming inspection for these three parts after 6 months of zero-defect deliveries
  
• Production stabilized at 200-300 camera units per month

 

 

 

 

Final Thought

 

 

Optical alignment is not a machining tolerance. It’s a process tolerance. The difference: a machining tolerance says “this bore is 50.000 ±0.005 mm.” A process tolerance says “this bore is concentric to the datum face within 2 μm, verified at 20°C, machined in one setup on an indicated fixture.”

 

Your drawing controls the first while supplier controls the second. Make sure both are right.

 

VMT machines optical datum parts for scientific imaging, cine lenses, surveillance cameras, and industrial vision systems — from single prototypes to 5,000-unit production runs. Single-setup datum machining. FAIRs on every first article. Thermal-stabilized inspection.

 

[ Upload your drawing → Get a quote with optical datum review within 24 hours. ]

 

[ For learning about more comprehensive information about camera lens parts manufacturing, welcome to click and read our technical white paper: The Ultimate Guide to High-Precision Camera Lens Parts Manufacturing. ]

 

 

 

 

FAQs

 

 

Q1: What is an optical datum in CNC machining?

 

An optical datum is the reference surface from which all optical alignment features are measured. On a lens housing, it’s typically the flange face (Datum A) and the primary bore (Datum B). The datum establishes the part’s position and orientation during both machining and assembly. If the datum is wrong, every downstream alignment is wrong.

 

 

Q2: What accuracy can CNC machining achieve for optical alignment?

 

Standard CNC with careful fixturing can hold 5-10 μm coaxiality on optical features. Alignment turning (mechanical centering on a dedicated station like TRIOPTICS ATS) achieves <1 μm centering and <0.2 arcmin tilt. The gap is process, not machine capability. A standard CNC with a disciplined operator can hit 3-5 μm. A standard CNC with a standard operator will hit 10-20 μm.

 

 

Q3: What is alignment turning and when is it necessary?

 

Alignment turning lathes the optical surfaces of a part while it is mechanically centered on the optical datum — no software compensation, no re-chucking. It is necessary when the coaxiality requirement is ≤3 μm or the tilt requirement is ≤0.5 arcmin. Typical applications: lens barrels for cinema lenses, prism mounts, interferometer reference housings.

 

 

Q4: How do materials affect optical datum stability?

 

Three factors: CTE (thermal expansion), internal stress relaxation, and machinability. Invar (CTE ~1.2) is the gold standard for athermalized optical benches — it barely moves. 6061-T6 aluminum (CTE 23.6) is stable enough for most applications if machined with stress-relief cycles. Brass (360 alloy) is excellent for small precision bores due to chip control and diamond-turnability. The key rule: rough-machine with coolant, stress-relieve, then finish-machine — regardless of material.

 

 

Q5: How do I specify optical datums on a CNC part drawing?

 

Use GD&T datum callouts with clear precedence: Datum A = primary locating face (flatness tolerance), Datum B = centering bore (concentricity or position tolerance to A), Datum C = clocking feature. Add a note: “Machine Datum A, B, and all optical bore features in a single setup without re-chucking. Dimensions apply at 20°C.” Ask for a First Article Inspection Report covering all datum-related GD&T.

 

 

Q6: What’s the biggest mistake designers make with optical part drawings?

 

Placing the primary datum on a non-functional surface — typically the part OD — instead of the surface that actually locates the part in the optical assembly. The OD is convenient for fixturing but irrelevant to optical performance. If the flange face is what seats against the next optical component, make it Datum A. The bores align to the face, not the OD. Get the datum hierarchy right and half the alignment problems disappear.

 

 

 

Disclaimer

 

The technical information and manufacturing advice shared on the VMT website are for general guidance only. While we strive for accuracy, VMT does not guarantee that the processes, tolerances, or material properties mentioned are applicable to every specific project. Any reliance you place on such information is strictly at your own risk. It is the buyer's responsibility to provide definitive engineering specifications for any production orders. Final specifications and service terms shall be subject to the formal contract or quotation confirmed by both parties.