Custom Acrylic Mahjong Tiles — Spec Guide for OEM Buyers

I started Wetop in 2008 because I was tired of losing bids to factories that disappeared mid-production. Eighteen years later, I sit on the other side of the table — the founder watching procurement teams ask the wrong questions about a category nobody warned them was harder than it looks.

Acrylic mahjong tiles are one of those categories. A buyer comes in with a CAD file, asks for “144 tiles, custom logo, acrylic,” and assumes the quote will look like a print job. It doesn’t. A 144-tile set in cast PMMA is a manufacturing scope, with four spec decisions that decide whether the buyer’s set ships premium or ships disappointing.

I’ve watched at least a dozen OEM mahjong projects come into our intake calls with the same gap: the brief says “polished acrylic with character engraving” and stops. The four questions below are the ones we ask before we quote, because I’d rather lose the deal at the brief stage than ship a buyer a set their customer ends up returning.

Cast vs extruded for sub-30 mm tile parts — why cast wins on edge clarity

For sub-30 mm tile parts, cast PMMA holds optical clarity at the edges and laser-etches with a clean recessed channel; extruded PMMA clouds at both. This is the spec call most shops skip when they quote a “polished acrylic” mahjong set, and it’s the one that determines whether the set still reads premium 18 months in.

Cast and extruded acrylic both come from the same monomer, but they’re polymerized differently. Cast sheets are polymerized between two glass plates over hours, producing a slightly higher molecular weight and a more uniform internal structure.¹ Extruded sheets are pushed through a die in minutes, leaving residual orientation stress in the polymer chains. At sheet-scale (1.2 m × 2.4 m display panels), the difference is visible only under polarized light. At tile-scale — sub-30 mm parts — that residual stress shows up where the laser cuts the character channel: the polymer remembers the extrusion direction and clouds along the channel walls.

I’ve watched this happen on a 144-tile prototype run we did in extruded PMMA back in 2014, when a buyer pushed for the cheaper material to hit a tighter unit-cost target. The first 60 days the set looked identical to a cast control set we ran in parallel. By month 8 the laser-etched character on the extruded tiles had visibly hazed at the channel edges; by month 18 the buyer’s distributor was returning sets for the “fogged characters” their players complained about. We replaced the run at our cost and switched our default to cast for any tile-scale work permanently.

The cost delta is real but smaller than buyers think — typically 18–24% per finished tile at OEM volumes, not the 50%+ that gets quoted on raw sheet pricing. The reason: cast laser-cuts cleaner, so the secondary diamond polish step that extruded tiles need to clear the channel haze gets dropped. By the time the tile ships, cast is roughly equivalent in labor cost to extruded-plus-polish, and the optical durability is the buyer’s pure upside.

Engraving depth — laser, UV print, and CNC: when each looks right

The right engraving method depends on what the buyer wants the player to feel when the tile lands. Laser is right for crisp character depth (萬, 條, 筒); UV print is right for suit colors that need register accuracy across 12 colorways; CNC is right for the tactile bevel that makes a premium tile feel premium. Most production-grade mahjong tile sets use all three on the same tile.

Laser engraving puts the character into the tile body as a recessed channel — typically 0.3 mm deep on a 6 mm cast PMMA body — using a CO2 laser at carefully tuned dwell time. Cast PMMA actually flame-polishes its own laser-cut edges, which means the character channel has a glossy interior at no extra step. The character can’t wear off because it’s recessed below the play surface; only the tile face around it can scratch. We run laser engraving as the default for the character + numeral set on every tile we ship.

UV print is the answer for suit colors and brand graphics that need to be visible under play lighting and consistent across colorways. The print stack is roughly 0.05 mm thick — five sheets of paper — and gets a clear-coat seal after curing.² Without the seal, the print starts to soften at 18 months under normal indoor light; with it, the tile holds appearance past five years. The clear coat adds about 6% to per-tile cost and is the single highest-yield investment any boutique-edition mahjong set buyer can make.

CNC bevel is the spec most acrylic shops will leave out of their quote unless the buyer asks. It’s a 0.3 mm chamfer on all four tile edges, machined after laser cutting and before final polish. The chamfer catches studio light at 45 degrees — which is what gives a premium tile that subtle edge-glow when it lies on the table — and equally importantly, it removes the sharp edge that buyers report as “feels cheap” or “feels sharp” in the hand. A CNC bevel adds about 4 minutes of machine time per tile, or roughly $0.10 at our internal cost. Skipping it saves the supplier 24 cents on a 144-tile set; including it is what separates a tile that buyers describe as “feels right” from one they describe as “off.”

Edge tolerance and bevel — why 0.3 mm changes the table feel

A 0.3 mm chamfered edge changes how the tile reads at the table because acrylic edges, polished or unpolished, are visually noisy at sub-millimeter scale. The chamfer creates a small but consistent light-catching surface that resolves that noise and signals “premium” to the player without anyone consciously parsing why.

I learned this on a run we did in 2019 for a hospitality-tech brand opening their first mahjong lounge. The first prototype came back with diamond-polished but un-chamfered edges. The brand’s creative director held a tile, set it on the table, picked it up, set it down — and said “this feels like a sample, not a real tile.” We added the 0.3 mm chamfer on iteration two. Same material, same engraving, same finish on the face. He set the new tile on the table, picked it up — and said “this is it.” That experience told me the bevel is doing perceptual work the buyer can feel before they can name.

The bevel does mechanical work too. A square edge is a stress concentration; on a tile-on-tile click, the impact load focuses at the corner and over thousands of plays will produce micro-chips along the edge. A chamfered edge distributes that load across a wider contact area, dropping the peak stress at any one point. Our internal 6-month tabletop play simulation across three sets shows zero edge chipping on chamfered tiles vs visible micro-chip patterns on un-chamfered controls by month 4.

For OEM buyers spec’ing this, the language to put on the drawing is: “0.3 mm chamfer on all four edges, top and bottom face, ± 0.05 mm tolerance, CNC machined.” Without the tolerance call, suppliers will sometimes try to substitute a hand-deburred edge (which looks similar in photos but doesn’t behave the same in play). With the tolerance call, the spec is auditable and the buyer gets the experience they paid for.

Color register across a 12-colorway production run — the spectrophotometer protocol

Color register is where a 12-colorway production run separates manufacturers from middlemen. The spec is a Pantone-matched mill-cert on cast PMMA pellets at the front of production, plus a spectrophotometer reading at four points on each batch with a ΔE ≤ 2 acceptance threshold.³ Without the protocol, batch-2 visibly drifts from batch-1 and the buyer’s QA flag every set after the first.

Here’s what this looks like in practice. A buyer specs 12 boutique-edition colorways for a tournament gift program — say, six tournament-themed accent colors and six brand-color variations. Each colorway needs 50 sets (50 × 144 = 7,200 tiles per colorway, 86,400 tiles total). A typical acrylic shop will pull from in-stock cast PMMA pellets, which means colorway 1 might come from a batch produced last month and colorway 2 from a batch produced two months earlier. Cast PMMA holds color extremely well within a single mill batch but drifts measurably batch-to-batch — typically ΔE 2.5–4.5 — because of small monomer + pigment ratio variations.

What we do instead: 14 days before production, we order a custom mill batch sized to the entire 86,400-tile order, with the pigment ratios verified to a single Pantone reference card per colorway. The mill ships a certificate (mill cert) documenting the actual ΔE of the batch against the Pantone reference. Once tiles are cut, our QC team runs a spectrophotometer at four points on every 50-set batch — top-left, top-right, bottom-left, bottom-right — and rejects any reading above ΔE 2 from the batch median. The 14-day mill-batch lead time on the front end is invisible to the buyer’s downstream timeline because production runs in parallel for the other components.

The cost for a custom mill batch is non-trivial — typically $400–$1,200 per colorway depending on volume — but it amortizes across the full run. On an 86,400-tile order across 12 colorways, that’s roughly $0.10–$0.17 per tile in mill-batch overhead, which is well below the cost of a single QA rejection at the buyer’s distribution warehouse.

Packaging + travel tolerance — why mahjong sets fail in transit

Mahjong sets fail in transit before they fail in play. Silicone-cradle tile inserts, lid latches that survive 200 G handling drops, and ISTA 3A drop-test verification are what separate a 0.5% defect rate at the buyer’s warehouse from the 5% rate that sinks margin on the second container.⁴

The failure modes are predictable. Tile-on-tile rattle inside the case during sea freight produces face-edge wear on tiles that have no rattle protection. Lid latches that aren’t engineered for repeated impact will bounce open under freight handling, allowing tiles to migrate and chip. Inadequate corner protection on the outer carton means the case itself takes the drop load and transmits it directly to the tiles inside. Across the category, returned-shipment damage rates from poorly-spec’d insert and packaging combinations land in the low-to-mid single-digit percentage range — entirely preventable with the production-grade spec below.

What works: a silicone-cradle tile insert tray that holds each tile in its own recess with no lateral movement, a lid latch with a mechanical interlock (not just magnets — magnets fail under freight vibration), and an outer carton with foam corner protection. We run every production run through a single ISTA 3A drop test on a random sample case before approving the shipment, which costs us about $300 per run and catches the small percentage of cases where the latch or insert spec drifted in production.

For OEM buyers spec’ing this, the line items to put on the production drawing are: “silicone-cradle insert, individual tile recess, ± 0.2 mm fit tolerance,” “lid latch with mechanical interlock, 200 G impact-rated,” and “ISTA 3A drop-test verification, single random sample per production run.” If the supplier balks at any of these — and most will, because they add roughly 8–12% to per-set cost — the right answer is to push back. The cost of rejecting an entire colorway at the buyer’s distribution warehouse is far higher than the protection these line items provide.

If a buyer wants to talk through any of these spec calls — or just see a sample tile in cast PMMA with the engraving method they’re considering — send the brief over to our team. For the broader fabrication context behind these decisions, see our published guides on diamond vs flame polishing and cast vs extruded acrylic. And for buyers who want the manufacturing-side lens on supplier vetting before they commit to any acrylic OEM partner, the 8-point manufacturer audit checklist is the framework we run on every supplier we qualify ourselves.

Footnotes

1. ASTM International. ASTM D4802-21 — Standard Specification for Poly(Methyl Methacrylate) Acrylic Plastic Sheet. https://www.astm.org/d4802-21.html ↩

2. ASTM International. ASTM D5946-17 — Standard Test Method for Corona-Treated Polymer Films Using Water Contact Angle Measurements. (Surface energy reference for ink adhesion on PMMA.) https://www.astm.org/d5946-17.html ↩

3. International Commission on Illumination. CIE 2000 ΔE Color Difference Formula. Standard reference for color-difference perception thresholds. https://cie.co.at/publications/colorimetry-part-6-ciede2000-colour-difference-formula ↩

4. International Safe Transit Association. ISTA 3A — Procedure for Packaged-Products for Parcel Delivery System Shipments. https://ista.org/test_procedures.php ↩