A lens-grinding machine is a precision spindle-and-tool system that removes glass from an optical blank to generate a defined radius of curvature on the lens surface. It solves the problem of producing repeatable spherical or aspheric surfaces to sub-micron accuracy, something hand-grinding cannot hold across a production batch. The machine couples a rotating workpiece spindle to a tilted, rotating diamond cup wheel or lap, with the angle between the two axes setting the radius. Modern CNC curve generators like the Schneider HSC 100 CNC routinely hold radius tolerances of ±5 µm on 100 mm blanks.
Lense-grinding Machine Interactive Calculator
Vary cup wheel diameter and spindle tilt angle to see the generated lens radius, curvature, and angle sensitivity.
Equation Used
The calculator uses the lens-grinding geometry equation R = D / (2 sin(theta)). D is the diamond cup wheel diameter and theta is the spindle tilt angle. Increasing theta increases sin(theta), so the generated spherical radius becomes shorter.
- Cup wheel diameter and spindle axes define the spherical generating geometry.
- Tilt angle theta is measured between the workpiece axis and tool axis.
- Wheel runout, centering error, infeed, grit size, and polishing allowance are not included.
Operating Principle of the Lense-grinding Machine
The core of every lens-grinding machine is two intersecting rotational axes — the workpiece spindle holding the glass blank, and the tool spindle holding either a bonded diamond cup wheel for rough generation or a pitch lap for fine grinding and polishing. You set the angle between those two axes, and that angle directly determines the radius of curvature on the finished surface. Tilt the tool 18° relative to a 60 mm cup wheel and you get roughly a 97 mm radius. Tilt it 45° and you generate a much shorter radius. The math is geometric, not magic — the cup wheel sweeps a cone whose intersection with the rotating blank traces a sphere.
Why design it this way? Because grinding by feed-controlled plunge would chase the radius rather than define it. With axis-angle generation, the radius is locked by the geometry of the setup. The spindle then just removes stock — depth of cut, infeed rate, and coolant flow control surface roughness and sub-surface damage, but the curve itself comes from the angle. A diamond cup wheel running at 6,000 RPM against a BK7 optical glass blank rotating at 200 RPM is a typical curve-generation setup, removing roughly 0.3 mm of glass per pass at 50 µm/s infeed.
Get the tolerances wrong and the consequences show up fast. If the blank centre is offset from the spindle axis by more than 10 µm, you generate an off-axis lens — the optical axis of the finished part will not coincide with the geometric centre, and the lens will produce coma in any system it goes into. If the cup wheel runs out by 20 µm, you'll see chatter marks across the surface that no amount of polishing removes — they go deeper than the polishing layer can reach. And if you starve coolant flow below about 4 L/min, the diamond bond heats up, glazes over, and starts burnishing rather than cutting, which leaves sub-surface damage 30 µm deep that has to come off in lapping and polishing. Every one of those failures is preventable with proper setup.
Key Components
- Workpiece Spindle: Holds the glass blank on a vacuum or pitch chuck and rotates it at 100-400 RPM. Runout must stay under 5 µm TIR or the generated surface picks up a periodic figure error that shows up as astigmatism on the interferometer.
- Tool Spindle (Diamond Cup Wheel): Carries a metal-bonded or resin-bonded diamond cup wheel, typically 40-80 mm in diameter, rotating at 4,000-10,000 RPM. The wheel concentration is usually 75 (18.75% diamond by volume) for BK7 and crown glasses, dropping to 50 for softer phosphate glasses.
- Axis-Tilt Mechanism: Sets the angle θ between the workpiece axis and the tool axis. Resolution must be better than 0.001° on a CNC machine to hold radius tolerances of ±10 µm on a 100 mm blank — at this scale, 0.01° of angle error shifts the radius by roughly 50 µm.
- Coolant Delivery: Flood coolant — usually a synthetic emulsion at 3-5% concentration — delivered at 4-8 L/min directly into the cut zone. Inadequate flow lets the diamond wheel glaze, which doubles sub-surface damage depth and forces longer lapping cycles.
- Lap or Polishing Tool: For fine grinding and polishing, the cup wheel is swapped for a pitch or polyurethane lap charged with cerium oxide or aluminium oxide slurry. Lap pressure of 50-150 g/cm² over a 30-90 minute cycle brings surface roughness from Ra 200 nm down to Ra 2 nm.
- Blank Centring System: Edges the blank to the optical axis using either a bell chuck or a centring camera. Centring error above 10 µm shows up directly as wedge in the finished lens, which the assembly stage cannot correct.
Where the Lense-grinding Machine Is Used
Lens grinding sits at the start of every optical fabrication process — every camera lens, telescope objective, and laser optic begins as a sawn glass blank that gets curve-generated, fine-ground, lapped, and polished. The machine type changes with batch size and tolerance, but the geometry of axis-angle generation is the same whether you're making a 6 mm phone-camera element or a 1.2 m telescope mirror blank. Tolerance demands, blank diameter, and glass type drive whether the shop runs a CNC curve generator like the Schneider HSC, a manual Strasbaugh, or a pad-polishing machine like a Satisloh SPM.
- Camera lens manufacturing: Canon's Utsunomiya plant runs Schneider HSC 100 CNC curve generators producing aspheric elements for the RF 50mm f/1.2L USM, holding radius tolerance to ±3 µm before MRF figuring.
- Astronomical optics: The University of Arizona's Richard F. Caris Mirror Lab grinds 8.4 m borosilicate honeycomb mirror blanks for the Giant Magellan Telescope on a custom large-tool generator before stressed-lap polishing.
- Ophthalmic lenses: Essilor's Thailand plant uses Schneider HSC ALS generators to produce free-form progressive lens surfaces, generating individual prescription geometries on CR-39 and Trivex blanks at roughly 90 seconds per surface.
- Laser and photonics: Edmund Optics in Barrington NJ grinds fused silica laser windows on OptoTech SPM curve generators, achieving λ/10 surface flatness after subsequent CCP polishing.
- Microscope objectives: Carl Zeiss in Oberkochen grinds high-NA objective elements on Loh Optikmaschinen SPM machines, holding centring error below 5 µm to maintain diffraction-limited performance.
- Defence and EO/IR systems: L3Harris in Rochester NY grinds germanium and silicon IR lens elements on Moore Nanotech 250UPL machines for thermal imaging modules used in the AN/AAQ-37 DAS.
The Formula Behind the Lense-grinding Machine
The heart of curve generation is the relationship between the cup wheel diameter, the tilt angle, and the resulting radius of curvature. You set up the geometry, the geometry sets the radius — there is no feedback loop on the radius itself during the cut. At the low end of the typical operating range, around a 5° tilt, the cup wheel is nearly parallel to the blank and you generate very long radii suitable for almost-flat ophthalmic lenses. At the nominal mid-range, 15-25°, you cover most camera and telescope objective work. Push beyond 50° and you start running into clearance problems where the back of the cup wheel fouls on the chuck or the workpiece itself. The sweet spot for most precision optics work sits between 10° and 30° — accurate, repeatable, and clear of mechanical interference.
Variables
| Symbol | Meaning | Unit (SI) | Unit (Imperial) |
|---|---|---|---|
| R | Radius of curvature generated on the lens surface | mm | in |
| Dcup | Outer diameter of the diamond cup wheel cutting edge | mm | in |
| θ | Tilt angle between the workpiece spindle axis and the tool spindle axis | degrees | degrees |
| sag | Sagitta — depth of the curved surface at the blank edge, used to set initial Z infeed | mm | in |
Worked Example: Lense-grinding Machine in an optics shop grinding a BK7 plano-convex lens
A precision optics shop in Rochester New York is curve-generating the convex side of a 100 mm diameter BK7 plano-convex lens with a target radius of 250 mm on a Schneider HSC 100 CNC curve generator using a 60 mm metal-bonded diamond cup wheel at 6,000 RPM. They need to know what tilt angle to set, and what radius they actually get if the angle is off by ±0.05° due to setup error.
Given
- Dcup = 60 mm
- Rtarget = 250 mm
- Dblank = 100 mm
- Ntool = 6000 RPM
Solution
Step 1 — rearrange the radius equation to solve for the nominal tilt angle θ at the target 250 mm radius:
Step 2 — at the low end of the practical setup tolerance, θ is 0.05° below nominal, so 6.842°. Recompute the actual radius generated:
That is 1.85 mm long compared to the 250 mm target — well outside any precision optical tolerance. On a 250 mm radius lens, a 1.85 mm radius error shifts the back focal length by several millimetres, which a camera lens assembly cannot accommodate.
Step 3 — at the high end of the same tolerance window, θ is 0.05° above nominal at 6.942°:
Step 4 — compute the sagitta to set the Z infeed depth at the nominal radius, using sag ≈ r2 / (2 × R) where r is half the blank diameter:
So the cup wheel must remove 5.00 mm of glass at the optical axis on a 100 mm blank to reach the full 250 mm spherical surface. At 50 µm/s infeed that is a 100-second cut, which lines up with typical Schneider HSC cycle times for this class of part.
Result
The nominal tilt angle is 6. 892° and the sagitta is 5.00 mm. In practical terms, that 6.892° is a shallow tilt — the cup wheel is almost parallel to the blank, which is normal for long-radius camera and telescope objectives. The ±0.05° setup error swings the actual radius from 248.18 mm at the high end to 251.85 mm at the low end, a 3.67 mm spread that no precision optic will tolerate, which is why CNC curve generators resolve angle to 0.001° rather than 0.01°. If your measured radius differs from prediction by more than 50 µm, check three things in order: (1) cup wheel diameter wear — a worn wheel that has dropped from 60 mm to 59.8 mm shifts the radius by roughly 800 µm at this geometry, so wheel diameter must be re-measured after every dressing; (2) thermal growth in the spindle column — a 2 °C swing on a cast-iron column moves the tool axis by 5-10 µm, enough to matter; (3) blank seating on the chuck — if the pitch pad has a high spot, the blank tilts and the actual cut angle differs from the commanded angle.
Choosing the Lense-grinding Machine: Pros and Cons
Lens curve generation is not the only way to put a radius on a glass blank, and shops pick between three main approaches based on tolerance, batch size, and glass type. The CNC diamond curve generator dominates precision optical production today, but loose-abrasive grinding on a spherical lap and single-point diamond turning still hold their corners of the market.
| Property | CNC Diamond Curve Generator | Loose-Abrasive Lap Grinding | Single-Point Diamond Turning |
|---|---|---|---|
| Radius accuracy on a 100 mm blank | ±5 µm | ±20 µm | ±1 µm |
| Surface roughness Ra after grinding | 150-300 nm | 400-800 nm | 5-20 nm |
| Cycle time per surface | 60-180 s | 20-60 min | 10-40 min |
| Glass type compatibility | All silicate, fluoride, IR materials | All glasses, slowest on hard glass | Soft IR materials only — Ge, ZnSe, CaF2 — not BK7 |
| Capital cost (typical 2024 USD) | $280k-$650k | $8k-$40k | $400k-$1.2M |
| Sub-surface damage depth | 20-40 µm | 10-25 µm | 0.5-2 µm |
| Best fit application | High-volume precision optics, ophthalmic, camera | Prototype, hobbyist, large astronomical mirrors | IR lenses, ultra-precision asphere prototyping |
Frequently Asked Questions About Lense-grinding Machine
This is almost always cup wheel wear that has not been compensated. A bonded diamond cup wheel loses 50-200 µm of diameter per shift on hard crown glass, and because R is proportional to Dcup, every 100 µm the wheel shrinks pushes the generated radius long by roughly 400 µm at a 7° tilt.
The fix is to probe the cup wheel with the machine's tool-setting probe at the start of every job and update Dcup in the controller. Schneider, OptoTech and Satisloh machines all have this routine built in — the issue is operators skipping it because the wheel "looks fine."
You've run into the geometric clearance limit. To generate R = 20 mm with a 60 mm cup wheel you need θ = arcsin(60/40) which is mathematically impossible because the argument exceeds 1.0 — meaning a 60 mm wheel physically cannot generate any radius shorter than 30 mm.
Switch to a smaller cup wheel. A 25 mm cup wheel can reach down to a 12.5 mm radius. For very short radii like high-NA microscope objective elements, shops use 10-15 mm cup wheels at 15,000+ RPM. The trade-off is shorter wheel life and higher run-out sensitivity, but it's the only way to clear the chuck geometry.
Sub-surface damage in glass propagates ahead of the cutting grit as a network of conchoidal cracks. The visible roughness is the residual surface texture, but the cracks underneath extend roughly 3-6× the average grit size into the bulk material. With a typical 76 µm (200 grit) diamond, that puts crack damage at 25-45 µm deep.
This is why optical shops always plan for a fine-grind stage with a 9-15 µm pellet tool followed by polishing — you have to remove enough stock to get below the damage layer from the previous stage. Skipping it leaves sleeks and digs that show up under dark-field inspection after polish.
For curve generation on BK7, metal-bonded wheels at 75 concentration are the standard choice because they hold form much longer than resin and BK7 is hard enough to keep the bond self-dressing. You'll get 2,000-4,000 surfaces per wheel before re-truing.
Switch to resin-bonded only when you move to fine-grinding stages where surface finish matters more than form retention, or when you're working softer glasses like phosphate laser glass or ZERODUR where metal bonds glaze quickly. Resin gives you Ra 100-150 nm versus metal's Ra 200-300 nm, but you'll re-true the wheel 5× more often.
Periodic chatter at fixed spacing is almost always a frequency match between the tool spindle and the workpiece spindle. At 6,000 RPM tool and 200 RPM work (a 30:1 ratio), every 30 cup-wheel revolutions the same point on the wheel hits the same point on the blank, and any single grit defect or wheel runout writes a regular pattern.
Two fixes: first, change the tool/work RPM ratio to a non-integer ratio — try 6,150 / 197 instead of 6,000 / 200, which breaks the synchronicity. Second, true the cup wheel — runout above 5 µm guarantees this pattern no matter what RPMs you choose.
The cut-off is roughly the radius tolerance and the batch size. If you need ±25 µm or tighter on radius, or you're making more than about 50 surfaces a week, the CNC pays for itself in cycle time and consistency. Loose-abrasive lap grinding will get you there on accuracy with patience, but each surface takes 30-60 minutes versus 2-3 minutes on the CNC.
For one-off prototypes, large astronomical mirrors above 500 mm, or any job where the radius is allowed to land within ±100 µm and figure is corrected later by polishing, lap grinding is still the right answer. The Caris Mirror Lab grinds 8.4 m mirrors this way — no curve generator exists at that scale.
References & Further Reading
- Wikipedia contributors. Lens (optics) — Manufacturing. Wikipedia
Building or designing a mechanism like this?
Explore the precision-engineered motion control hardware used by mechanical engineers, makers, and product designers.
