Custom Dice ManufacturerVacuum Degassing vs Pressure Curing: How Bubble-Free Resin Dice Are Actually Made
A bubble is the only defect in a resin die that a buyer can find from a phone photograph. Bad number alignment needs a second glance, a half-degree of geometric distortion needs a caliper, and an under-cured face needs a fingernail — but a 0.6mm void sitting behind a translucent D20 face acts like a lens, catches the light and announces itself in every product shot the customer ever takes. That is why the two most misunderstood steps on our resin lines are vacuum degassing and pressure curing. They are not alternatives, they are not the same process at different pressures, and neither one is a cure-all. This is how both work on our floor, in what order, on which line, and what each of them genuinely cannot fix.
1. Why a bubble decides a resin order
Every mixed resin contains air. Some of it is dissolved in the resin as supplied; more is folded in when you stir the two components together; more again arrives with pigment paste, mica, glitter or a dye that has been sitting in a jar since the last order. In an opaque 16mm D6 most of that air is invisible and irrelevant. In a clear or translucent die it is the whole ball game, because a polyhedral die is essentially a cluster of flat lenses. Light entering a D20 face refracts, hits an internal void, scatters, and comes back to the eye as a bright silver dot. The die does not need many of those to look cheap.
There is a second, less obvious reason bubbles matter, and it is structural rather than cosmetic. A void that sits just below a number channel becomes an open pit the moment we grind, lap or tumble down to it — and then it fills with paint during the number fill, holds polish wax, or simply chips at the edge. A die can pass a visual check as a raw casting and fail after finishing because the finishing exposed a bubble that was always there. So bubble control is not a single inspection at the end; it is a chain of decisions from the weighing scale to the demold, and it is the single biggest reason two factories quoting the same "clear resin 7-piece set" can be 30% apart on price.
YSDICE — also known as Dongguan Yushun Hardware Co., Ltd. — is a registered dice manufacturer in Chang'an, Dongguan, and we run two separate resin lines with two different anti-bubble strategies: a sharp-edge line in epoxy, and a rounded line in unsaturated polyester. Both use vacuum. Only one uses pressure. Understanding why is most of what a buyer needs to know about resin quality.
2. Vacuum degassing: taking the air out before it is trapped
Vacuum degassing is a removal process, and it happens before the resin ever touches the mold. Once the A and B components are weighed on calibrated scales and the colour is dispersed, the whole container goes into a vacuum chamber. Dropping the pressure around the mix makes every trapped bubble expand — a bubble that was 0.3mm at atmospheric pressure grows several times in diameter — and an expanded bubble is buoyant enough to climb through viscous resin and burst at the surface. Dissolved gas comes out of solution at the same time. What you are watching is the resin rising into a foam head, holding, then collapsing back as the air leaves. That collapse is the signal the batch is done.
The practical detail that catches new operators is headroom. The container must be far larger than the volume of resin in it, because the mix genuinely climbs — it can triple in height at the peak of the foam. A cup filled to the shoulder does not degas, it overflows into the chamber. Then you have lost the batch, you have lost the pigment ratio you carefully weighed, and you have contaminated the chamber for the next run. On our process sheets the headroom requirement is written into the step itself rather than left to judgement, and it appears identically on both routings: step 5 on the sharp-edge epoxy line ("colour + vacuum degassing, container needs headroom for expansion") and step 4 on the rounded polyester line.
Degassing has a time cost that buyers rarely see quoted separately but always pay for. It is dead time in the middle of a working pot life — the resin is curing while it sits in the chamber, so the operator is trading working time against clarity, and a batch that gets degassed too long can gel before it is poured. That trade-off is one reason experienced resin benches run smaller batches than beginners expect.
3. Pressure curing: compressing what is left until it is invisible
Pressure curing is a concealment process, and it happens after the pour. The filled mold goes into a pressure pot and the whole assembly cures under compressed air. The physics is straightforward: bubble volume falls roughly in proportion to absolute pressure, so a bubble held at four times atmospheric pressure shrinks to about a quarter of its volume — well under half its original diameter. Crucially, the resin then hardens around it at that size. When the pot is vented, the bubble cannot re-expand, because it is now locked inside a solid. It has not been removed. It has been made too small for the eye or the camera to resolve.
We use pressure curing on the sharp-edge epoxy line, as step 7 of the fourteen — immediately after the pour and before demold and post-cure. It is the right tool for that product because sharp-edge dice are optically demanding by definition: the buyer is paying for razor-flat faces and a clear window into whatever is suspended inside. It is also the right tool because epoxy is comparatively forgiving of the working time a pressure cycle consumes.
The rounded unsaturated-polyester line does not use it. That line cures at room temperature with the mold kept level, and the whole cure is managed by controlling ambient temperature, batch size and exotherm rather than by external pressure. It works because the product forgives it: on a rounded die, the soft edges are created by multi-stage tumbling afterwards, and the surface layer that would carry the most visible micro-bubbles is partly ground away in the coarse tumbling stage anyway. Different product, different physics, different step list — and any factory that tells you it "pressure cures everything" is either describing a single line or describing a brochure.
The rounded die's soft edges are formed mainly by multi-stage tumbling; the sharp die's corners and edge lines are formed mainly by the high-precision mold, and post-processing must be face-by-face flat lapping plus fine polishing.— from our internal process document, verbatim
4. What neither technique can fix
Here is the part that gets skipped in most factory tours. Vacuum handles air that exists before the pour. Pressure handles air that exists during and after the pour. Neither of them touches chemistry, moisture or geometry, and a surprising share of the bubbles we reject at QC were never air at all — they were gas generated by a reaction that should not have happened.
| Problem | Vacuum degassing | Pressure curing | What actually fixes it |
|---|---|---|---|
| Air folded in during mixing | Removes it | Compresses it | Slow mixing, correct blade, degas |
| Air entrained during the pour | No — it happens after | Compresses it | Pour from the lowest point |
| Moisture in mold or inclusion | No — gas is generated later | Partly masks, does not stop | Dry the mold, dry the inclusion |
| Exotherm from an oversized batch | No | No | Smaller batches, ambient control |
| Wrong A/B ratio, poor mixing | No | No | Calibrated scales, full dispersion |
| Demolded too early | No | No | Full cure plus post-cure |
Read that table as a warning about sequencing. A factory that owns a pressure pot but pours carelessly is compressing bubbles it did not need to create. A factory that degasses beautifully and then demolds an hour early is destroying good castings at the last step. The equipment is the easy part of bubble control; the discipline around it is the expensive part, and it is what you are really buying when unit prices differ.
5. Pouring from the lowest point — the technique that costs nothing
If we could only keep one anti-bubble practice, it would not be either machine. It would be the pour. Our sharp-edge routing specifies it in plain language at step 6: pour slowly from the lowest point of the mold so the resin rises naturally, never poured from height.
The reason is what happens to a falling stream. Resin dropped from even 10cm thins as it falls, folds over on impact and drags a sleeve of air down with it — and because the stream is now landing in a pool it has already disturbed, it keeps stirring air in for as long as the pour lasts. You degassed for eight minutes and re-aerated in twelve seconds. Worse, in a polyhedral cavity the trapped air does not simply rise back out; it gets caught under a face that is now facing downward, or in the apex of a D4 or D8 corner where the cavity narrows to a point and surface tension holds it.
The alternative is laminar and boring. Bring the cup lip or nozzle to the deepest point of the cavity, start the flow slowly, and let the resin level climb the mold with the pour point always submerged. Air is pushed ahead of the rising front rather than folded into it. On multi-cavity dice molds this also means accepting that the pour takes longer than it looks like it should — you are filling seven cavities at the pace of the slowest one, not racing across them.
The same principle governs the more decorative work. Two-tone and gradient dice on the rounded line are made with split or layered pours, each layer going in at the right viscosity so it neither mixes with the layer below nor traps a film of air against it. Inclusion pieces are set once the resin reaches the right viscosity — thick enough to hold the piece in position, thin enough to close around it without leaving a shadow of air behind the shoulders of the object — and then topped up.
6. The other five bubble sources we hunt
Vacuum, pressure and pour technique cover the air you brought with you. These five cover the air you make.
Where bubbles come from after the pour
- Moisture in the mold or the inclusion. A silicone mold that was washed and not fully dried, or an inclusion that has absorbed humidity, releases water vapour into curing resin. Dried botanicals, paper, wood and porous printed cores are the usual suspects. We record inclusion moisture content as a batch field for exactly this reason.
- Mixing too fast. A high-speed blade whips air in faster than any chamber can take it out. On the rounded line the MEKP curing agent goes in last and at low speed; in non-pre-accelerated systems the accelerator must be fully dispersed first.
- A single batch poured too large. Resin curing is exothermic and the heat is proportional to mass. Oversize the batch and the centre runs hot, dissolved gas comes out of solution, viscosity drops then gels abruptly, and you get hot spots, shrinkage, cracking and colour drift alongside the bubbles.
- Demolding too early. A die pulled before full cure is still soft. Corners deform, faces bow, and internal voids that were held closed under pressure can open up. On the sharp-edge line we release the D4, D8 and D20 sharp corners first, then lift the whole die — never lever it out by a corner.
- Wrong ratio or incomplete mixing. A/B epoxy is weighed on calibrated scales, and you can never add extra hardener to speed curing. Off-ratio or unmixed resin gives tackiness, cloudiness, yellowing and a soft zone that reads as a haze rather than a discrete bubble — and low-temperature cure does the same.
Those five, plus uneven lapping, are the complete sharp-edge failure list from our process sheet: wrong A/B ratio, incomplete mixing, low-temperature cure, moisture in mold or inclusion, single batch poured too large, demolding too early, uneven lapping — producing tackiness, bubbles, cracking, yellowing and geometric distortion. It is a short list, which is the point. There are not fifty ways to ruin a resin die. There are seven, and a factory either controls all of them or it prices for the ones it does not.
7. Two routings, two places bubbles hide
Both of our resin routings run to fourteen steps, and comparing them shows exactly where each product is vulnerable.
The sharp-edge epoxy line goes: design confirmation; a high-precision master by 3D print or CNC, polished face by face; a silicone mold cast from that master; A/B epoxy weighed on calibrated scales; colour plus vacuum degassing; the slow low-point pour; pressure curing; demold and post-cure; sprue and flash removal with local flatting at the gate face; face-by-face flat lapping on a levelling plate; ultra-fine polishing with plastic polishing compound; number paint fill; QC; and 7-piece set matching and packing. The critical craft step is the lapping. Each face has to lose the same amount of material — take more off one and the die goes lopsided, the edge lines bend, the corners blunt and the centre of mass shifts. Aggressive tumbling is forbidden on this line for the same reason: it would round exactly the corners the customer is paying for.
The rounded unsaturated polyester line goes: design confirmation; silicone mold prep with dust, moisture and old resin removed; resin colouring with MEKP added last and at low speed; vacuum degassing; casting and inclusion placement, using split or layered pours for two-tone and gradient work; room-temperature cure with the mold level and ambient temperature, batch size and exotherm all controlled; demold and post-cure; sprue removal and screening out chips, cracks, severe bubbles and off-centre inclusions; coarse tumbling; medium and fine grinding, stepping down media; fine polishing with a wash between every stage so a coarse grain does not scratch a finished face; number paint fill; clean and QC; and 7-piece set matching and packing.
Note where the rounded profile comes from — the tumbling, not the mold. That single fact explains most of the differences between the lines, including why over-grinding is a real risk on the rounded line: push the media too far and the dimensions shrink and the engraved numbers thin out. On the sharp-edge line, by contrast, the mold does the geometry and the bench does the surface, which is why the number channels must be completely clear of grinding dust and polish wax before any paint goes in.
8. The MEKP rule we never bend
Unsaturated polyester cures with MEKP — methyl ethyl ketone peroxide — as the initiator, usually alongside a cobalt accelerator. The rule on our floor is absolute and it is the first thing any new operator is taught: MEKP must never be mixed directly with cobalt accelerator. Direct contact between the two, undiluted, can cause violent decomposition. They only ever meet dispersed in resin, which is why the sequence is fixed — in a system that is not pre-accelerated, the accelerator goes in first and is fully dispersed through the resin, and only then does the MEKP go in, last and at low speed.
Storage follows the same logic. Resin and organic peroxides live in separate zones per their safety data sheets, not on adjacent shelves in the same cabinet, and the work area is ventilated. This is not a bubble control measure in itself, but it belongs in the same conversation for two reasons. First, an uncontrolled reaction is an exotherm event, and exotherm is the parent of half the void defects on this page. Second, and more honestly: a factory that treats peroxide handling casually is not a factory that will be meticulous about container headroom, mold drying or scale calibration either. The safety practice and the quality practice are the same habit wearing different clothes.
9. What bubbles cost: our real defect rates
Very few factories publish defect rates, which is a shame, because the number is the most honest description of a process that exists. Ours, given directly by the factory owner rather than reconstructed from a spreadsheet, are these.
Rounded resin dice without an inclusion or core run at about a 5% defect rate. That covers bubbles, chips, cracks, colour drift and dimensional rejects across the whole routing. It is a normal, healthy figure for hand-poured resin and it is why plain rounded sets are the cheapest resin product we make.
Rounded resin dice with an inclusion — animals, figures, machined cores — run at 20–30%. The reason is not the core itself. We machine our cores in-house to tight tolerances, so the fit inside the cavity is fine. The reason is placement: the piece has to be set at exactly the right moment in the viscosity curve, centred in three axes, at the right depth, in a cavity the operator is looking into at an angle, and then topped up without disturbing it. It depends heavily on the bench worker's experience, and experience is not evenly distributed across a shift.
Do the arithmetic and the commercial consequence is unavoidable: for every 100 good inclusion sets, we may cast 125 to 140. That is the honest reason inclusion dice cost more per unit and need longer lead times than a plain rounded set of the same size and colour. It is not a premium for novelty; it is the yield. Any supplier quoting inclusion dice at plain-resin prices with plain-resin timelines is either not doing the screening step, planning to ship the off-centre ones, or has not run the product often enough to know.
10. Proving bubble control: what a buyer should actually ask
"Do you degas?" is a useless question, because the answer is always yes. These are the questions that separate a real process from a claimed one, and none of them require you to be a chemist.
Seven questions that get real answers
- "Which of your lines gets pressure cured, and at which step?" A factory that runs both processes will answer with a routing position. One that does not will answer with an adjective.
- "How much headroom do you leave in the degassing container?" The answer should be about expansion, not about volume in litres.
- "Where does the resin enter the mold?" You want to hear lowest point, slow, rising — not "we pour it in".
- "What is your defect rate on this product, by type?" Two numbers, plain versus inclusion, is the sign of a factory that counts. A single low number across all products is the sign of one that does not.
- "Can you send a sample cut in half, or photographed backlit?" Backlighting a translucent die shows internal voids that a top-lit product photo hides completely.
- "What batch fields do you record, and can I see a filled-in sheet?" Covered in the next section — a blank template is not a record.
- "How are MEKP and accelerator stored?" Separate zones per SDS, ventilated area. Any hesitation here is informative.
Add one physical test to that list, because it costs you nothing beyond a sample fee. Take the sample dice to a window, hold each one up against daylight, and rotate it slowly. Bubbles that are invisible under showroom lighting appear immediately when the die is backlit and turning. Then run your fingernail across each face and each number channel: a bubble that was exposed by finishing feels like a pinprick before it looks like one. Do this on the sample and you will never need to do it on 500 sets.
11. What a batch record actually contains
Bubble control is only reproducible if it is written down, and the record is what turns a good week into a standard. On every resin batch, Dongguan Yushun Hardware Co., Ltd. records nine fields:
| Recorded field | What it protects against |
|---|---|
| Raw-material lot | Traces a bad batch back to the drum, not the operator |
| Ambient temperature and humidity | Explains low-temperature cure and moisture-driven voids |
| Mix ratio | Proves the A/B weighing, catches off-ratio tackiness |
| Degassing result | Confirms the foam rose and collapsed, not that a timer ran |
| Mold cleanliness | Dust, moisture and old resin removal before casting |
| Inclusion moisture content | The single biggest cause of gas voids around a core |
| Cure state | Stops early demolding and the distortion that follows |
| Appearance consistency | Colour drift between batches within one order |
| Finished dimensions | Catches over-grinding and thinned number channels |
The value of these records shows up months later, not on the day. When a customer emails about three cloudy dice from a reorder, the useful reply is not an apology — it is checking the humidity and material lot for that specific pour, comparing them against the run that shipped clean, and being able to say which variable moved. Roughly once a season that trail lets us fix a real cause instead of replacing product and hoping. It is also the difference between a factory that improves and a factory that just re-runs.
12. Lead times and the paperwork behind them
Bubble control shows up on the calendar as much as on the invoice, and rounded polyester is the slowest resin product we make precisely because the cure cannot be rushed and the tumbling stages cannot be skipped. Here is the full picture across materials, split into light customisation (your colours and logo on existing molds) and full custom (new mold required).
| Material | Light custom: sample / bulk | Full custom: mold / sample / bulk |
|---|---|---|
| Metal | 3–5 d / 10–15 d | 20–25 d / 3–5 d / 20–25 d |
| Sharp edge epoxy | 3–5 d / 10–15 d | 20–25 d / 3–5 d / 15–20 d |
| Rounded unsaturated polyester | 7–10 d / 25–30 d | 15–20 d / 7–10 d / 25–30 d |
| Acrylic | 5–7 d / 20–25 d | 20–25 d mold |
| Silicone | 5–7 d / 20–25 d | 20–25 d mold |
| Gemstone | 5–7 d / 15–20 d | No mold required |
| PU dice tray | 5–7 d / 15–20 d | Light customisation only |
Notice that the rounded line's sample stage is 7–10 days against 3–5 for sharp-edge epoxy. That gap is the cure and the multi-stage tumbling, and it is not padding — it is the same time the bulk run will take per piece, which is exactly why the sample is worth waiting for. If a supplier offers you a rounded-resin sample in two days, ask what they left out.
On compliance, we hold two documents and describe them precisely, because over-claiming here is common and easy to check. Our material supplier holds an SGS report on the unsaturated polyester raw material tested to EN 71-3:2019+A2:2024 Category III, with all 19 regulated elements reported as not detected. A separate RoHS report covering 31 substances exists for the number-fill paint. Both are supplier documents on the raw materials, held by the material suppliers and available by email on request — they are not a product-level certification of a finished die, and we will not describe them as one. If your market needs finished-product testing, that is a separate arrangement with a third-party lab, and we will tell you so before you order rather than after.
Bubble control FAQ
Is a pressure pot better than a vacuum chamber? They solve different problems. Vacuum removes air before the pour; pressure shrinks whatever survives the pour until it is invisible. On our sharp-edge epoxy line we use both, in that order. On the rounded polyester line we use vacuum and control the cure instead.
Can any resin die be genuinely 100% bubble-free? Free of visible bubbles, yes — that is the standard we ship to. Free of every microscopic void, no, and a supplier promising that is describing marketing rather than physics. Ask about visible-defect standards and inspection method instead.
Why do my inclusion dice cost more than plain ones of the same size? Yield. Plain rounded resin runs about 5% defects; inclusion dice run 20–30%, so 100 good sets means casting 125–140. You are paying for the sets that did not make it past screening.
Does a bubble affect how the die rolls? A small void is far too light to bias a die in normal play. The much bigger balance risk is uneven lapping on a sharp-edge die, which shifts the centre of mass — which is why equal removal per face is a rule, not a preference.
Can you fix bubbles after casting? No. A void inside cured resin is permanent; the die is scrapped at screening. This is why every control on this page sits before the resin hardens.
What should I send you to get a realistic quote? A sketch or reference image, your material preference, quantity, and whether you want an inclusion. We will tell you which line it runs on, whether it gets pressure cured, and what the yield realistically means for your unit price — YSDICE quotes the defect rate into the price rather than discovering it at shipment.
Planning a clear or inclusion resin run? Send the design — even a rough sketch — to queenie@ysdice.com or message Queenie on WhatsApp. You'll get a free 3D proof of every face, the line it will run on, and an honest unit price with the yield already in it.
Queenie has cast, poured and shipped custom dice from Chang'an, Dongguan since 2018 — and still answers every project inquiry herself. Questions about your run? Write to queenie@ysdice.com or message us on WhatsApp.
From sketch to a free 3D proof in 3 days
Send an idea — even a napkin drawing. Queenie, your English-speaking project manager, replies with a free 3D design of every face, unit price and timeline.
Get a free design