TL;DR: Embossing geometry that looks clean on a CAD file will tear, crush, or register off-spec on press unless your designer understands five die-to-substrate constraints before the first proof.
TL;DR: In our tooling, a 0.4mm minimum wall thickness between adjacent emboss cavities is the floor — below that, the die steel fatigues within 8,000–12,000 impressions on 350gsm SBS.
When the CAD File and the Die Disagree — What Actually Breaks #
A brand submitted artwork for a premium spirits gift box last year: a repeating geometric relief pattern, 48 separate raised elements per panel face, each element 3.2mm wide with 0.3mm inter-element spacing. The designer had modelled it in Illustrator with precise vector paths. The structural review in our DFM-02 checklist flagged it immediately — but the brand’s internal approver, looking at a rendered 3D mockup, pushed to proceed to tooling.
The male die was cut at 0.9mm relief height. On press at 120 tonnes of impression force against 350gsm coated board, the inter-element land areas began to crush. By sheet 300 of a 5,000-sheet run, the 0.3mm walls between cavities had cold-flowed enough that adjacent elements merged at their bases. The finished boxes had a smeared, undefined texture across roughly 40% of each panel. The tooling cost was non-recoverable and the run was scrapped.
The root cause was not a die-cutting error or a substrate defect. It was a tolerance stackup that had never been modelled: die relief height, inter-element clearance, substrate caliper under compression, and impression force all interact simultaneously. No single parameter was wrong in isolation. Together, they were catastrophic.
This is the central problem with embossing design engineering. CAD tools model geometry in static space. Embossing operates in a dynamic mechanical environment where substrate fibres compress, recover partially, and cold-flow under sustained pressure. Unless the designer has internalized the translation rules between vector geometry and physical die behaviour, the gap between screen and press is large.
The Five Parameters That Determine Whether a Relief Design Is Manufacturable #
Relief height is the most visible variable, but it is not the most important one. Our tooling specifications run from 0.3mm shallow-relief for texture-effect embossing to 1.5mm high-relief on heavyweight board. Beyond 1.5mm on boards below 450gsm, you risk fibre delamination on the reverse face, which shows as bubbling under gloss laminate.
Substrate caliper under compression is the parameter designers most frequently omit from CAD inputs. A 350gsm SBS board nominally measures 0.45–0.50mm caliper. Under 80–120 tonnes of emboss impression, that same board compresses to 0.36–0.41mm at peak load, then recovers to roughly 0.43mm after release. If your relief height is set for nominal caliper and the substrate is sourced from a different mill lot with 0.47mm caliper, your registered emboss-to-print alignment can shift by 0.08–0.12mm — enough to show at the edge of a foil panel on a luxury box.
Inter-element spacing is where the spirits box failed. Our DFM-02 minimum for adjacent cavity wall thickness in steel dies is 0.4mm for designs running fewer than 50,000 impressions, and 0.6mm for production tooling above 100,000 impressions. Below those thresholds, steel fatigue and substrate cold-flow cause the problem described above. For brass dies used in short-run or embossed label work, we increase the minimum to 0.7mm because brass is softer and deflects earlier.
Draw ratio — the ratio of relief height to element width — predicts fibre stress. A 1.2mm relief height on a 4mm-wide element gives a draw ratio of 0.30, which is well within the safe window for most coated boards. The same 1.2mm relief on a 2mm-wide element gives 0.60, which requires either a reduced impression force or a pre-moistened substrate to avoid fibre fracture at the shoulder. We use the ISO 12647-2 substrate classification as a starting reference for paper fibre response, then adjust draw ratio limits based on internal test data from our sample archive.
Registration tolerance stackup is the fifth parameter, and the one most relevant to multi-process designs where embossing is combined with foil stamping or spot UV. On our flatbed embossing press, our hold tolerance is ±0.25mm in both axes. If your design places a 0.5mm-wide deboss trench immediately adjacent to a 0.4mm foil stripe, the combined tolerance of the two processes leaves a window of roughly 0.1mm for the finished result to remain clean. That is a spec the process cannot reliably hold in production. The design needs to be opened up to at least 0.8mm spacing between the two effects.
| Design Parameter | Our Minimum (Short Run < 50k) | Production Minimum (> 100k) | Common Designer Error |
|---|---|---|---|
| Inter-element wall, steel die | 0.4mm | 0.6mm | 0.2–0.3mm vectors from Illustrator |
| Relief height on ≤ 350gsm board | 0.3mm min / 1.2mm max | 0.3mm min / 1.0mm max | > 1.5mm without substrate test |
| Draw ratio (height ÷ width) | ≤ 0.45 for coated SBS | ≤ 0.35 for coated SBS | Not calculated at all |
| Emboss-to-foil register gap | 0.8mm min | 0.8mm min | 0.3–0.5mm in final artwork |
| Substrate caliper tolerance | ±0.03mm accepted | ±0.02mm preferred | Not specified in brief |
Translating CAD Geometry to Die-Ready Files — Where the Process Forks #
If your emboss design lives as a vector file, the translation to a die-prep file depends on the relief profile you need. Flat-top relief elements — a geometric pattern, a border, a logo mark — are straightforward: the vector path defines the cavity perimeter and the toolmaker interpolates shoulder angle at 70–80° from baseline, which is standard. For sculptural or continuous-tone relief (a portrait, a landscape, a gradiated texture), a vector file is insufficient. You need a greyscale heightfield, typically a 16-bit TIFF at 300–600 dpi, where pixel luminance maps to Z-height. Our tooling partners work in ArtCAM or JewelCAM for this conversion. If you send us a flat vector and expect a sculptural result, the die shop will make interpretive decisions you may not agree with.
For designs combining emboss with registered print or foil, we require that all elements share a single coordinate origin in the mechanical file. Separate Illustrator layers with different origins — common in agency workflows — are one of the most consistent sources of misregistration in our incoming briefs. When we receive artwork this way, it triggers a mandatory DFM-02 re-alignment review before we issue a die quote, which adds 3–5 working days.
If thermal simulation inputs are needed for high-volume die design (some clients request FEA validation for tooling expected to run above 500,000 impressions), the parameters we pass to the die supplier are: substrate elastic modulus at working temperature (typically 2.0–4.5 GPa for coated SBS at 20–25°C), Poisson’s ratio (0.3–0.35 for cellulosic board), and target impression dwell time (80–120ms on our flatbed press). These values feed a simplified plane-stress compression model. The output is a predicted die deflection value under peak load; above 0.015mm deflection, the die geometry is revised before cutting.
Some clients ask whether digital embossing (Scodix, MGI JETvarnish) removes the need for any of this analysis. For low-relief texture and gloss-matte effects up to 0.25mm, digital processes do bypass die engineering entirely. Above 0.25mm or for any structural emboss on a rigid board panel, traditional die embossing is still the correct process. The two are not interchangeable above that threshold.
Specification Notes for Brand Partners #
When you brief us on an embossed packaging project, the information that most directly affects quote accuracy and sample success is: substrate grade and target caliper, a list of all finishing processes sharing the same panel (foil, spot UV, laminate type), and a high-resolution export of the emboss artwork with all elements on a single layer sharing one coordinate origin.
The brief gap that consistently causes sample iterations is missing substrate caliper data. Brands often specify “350gsm SBS” without confirming the caliper from their nominated board supplier. If we tool a die for 0.47mm caliper and the production board measures 0.44mm, the emboss depth perception changes noticeably, and the register to foil drifts. Confirm your board caliper with your paper merchant before the die is cut, or let us specify the board from our qualified supplier list.
Our standard sampling timeline for a new emboss die is 15–20 working days from approved die-ready file to first press sample. Complex sculptural relief or multi-process registration designs extend that to 25 working days. What extends timelines most reliably is artwork that arrives without a confirmed substrate spec.
What file format should I send for a sculptural emboss design?
A 16-bit greyscale TIFF at 300 dpi minimum, where luminance maps to Z-height. A vector PDF defines perimeter only — it cannot describe a gradiated surface. If you only have a vector, our team can convert it to a heightfield interpretation, but you need to approve that interpretation before tooling begins.
Can I combine a 1.0mm emboss with a registered foil stripe on the same panel?
Yes, but the foil edge needs to sit at least 0.8mm from the emboss shoulder line. Closer than that, the combined tolerance of the two processes (±0.25mm each) makes consistent registration across a production run unreliable. If your design places them closer, we’ll flag it in the DFM-02 review before any tooling is quoted.
Does the laminate type affect the emboss result?
Significantly. Soft-touch laminate applied before embossing cushions the surface and reduces peak relief perception by roughly 15–20% compared to bare board or gloss laminate. If your target is a sharp, high-definition relief, emboss after laminating with gloss or matte, not soft-touch. If your target is a subtle tactile texture, soft-touch pre-lamination gives a more diffuse, premium feel that works well for cosmetics packaging.
What’s the maximum impression run before a steel die needs inspection?
Our internal tooling maintenance schedule (logged under the MM-09 die lifecycle register) calls for dimensional inspection every 80,000 impressions for steel dies running on coated SBS. For designs with inter-element walls below 0.5mm, we inspect at 50,000. Brass dies on short-run work get a visual check every 10,000 impressions. These are inspection intervals, not replacement intervals — most dies run well past those points, but problems caught at inspection are correctable. Problems caught at the end of a production run are not.
Planning a packaging project? Contact our team to request a complimentary specification review and sample quote.
Did the 0.9mm relief height on that run get flagged by DFM-02 or did it pass — because at 0.35 draw ratio on coated SBS that’s already pushing the production minimum, and I’m wondering if the 120-tonne impression setting was dialled in for a lighter emboss spec originally.
The 0.3mm inter-element spacing failure tracks exactly with what we’ve seen, but the cold-flow timeline depends heavily on whether the coated SBS is clay-coated or cast-coated — we ran a near-identical 48-element repeat pattern on cast-coated 350gsm last spring and didn’t see base merging until sheet 1,100, almost four times later than described here. Cast coat’s harder surface layer seems to redistribute compression differently under sustained impression force, which doesn’t change the outcome on a 5,000-sheet run but does affect where in the run you’d catch it on a quality pull.
We had something almost identical happen on a 350gsm GC1 folding carton for a fragrance house — tight repeating diamond pattern, 52 elements per panel, designer had them at 0.25mm walls between cavities and we flagged it twice before the brand overruled us citing the 3D render. Die cut at 1.1mm relief, by impression 400 the land areas were already flowing together and the finished texture looked like someone had pressed a thumb into warm wax. Scrapped 4,200 sheets, non-recoverable tooling. The 3D mockup doesn’t compress at 120 tonnes.
The scrapped run scenario is exactly why we started absorbing a mandatory DFM review into our tooling quotes — the male die alone on a job that size runs £1,800–£2,400 with us, and that’s before you account for the 5,000 sheets of 350gsm SBS. We’ve found that catching the inter-element wall issue at vector review stage costs maybe 2 hours of engineering time versus a full retool plus material write-off that can hit 3–4x the original tooling budget.
Switching to a matched-pair aluminium die for sampling runs on repeat-geometry patterns cut our short-run tooling cost from around £1,600 to £850 per set — the tradeoff is you’re locked to a softer relief ceiling, roughly 0.7mm max before definition starts degrading, but for a 2,000-sheet validation run it’s usually the right call before committing to steel.
On the tolerance stackup point — did the substrate caliper measurement used in the DFM-02 review reflect the board under ambient warehouse conditions, or was it taken post-conditioning to something like TAPPI T411 equilibrated stock, because we’ve had 350gsm SBS come in 40–50 microns thinner after 48 hours in our humidity-controlled intake area and that alone has shifted our impression settings meaningfully.
The merged-element failure mode the article describes took us three sampling cycles to diagnose on a 48-element botanical pattern we ran for a skincare client in late 2022 — each cycle was 4 weeks out and back from our die maker in Düsseldorf, so by the time we had a clean proof the brand had already missed their Q4 launch window.
We caught a nearly identical tolerance stackup failure on a 32-element repeat for a spirits client in 2021, but our trigger was fibre delamination rather than cold-flow merge — 350gsm GC2 uncoated, and the relief height was only 0.85mm, well inside what the spec sheet suggested was safe. The delamination started appearing around sheet 180 and was consistent across 60% of the panel face by sheet 400. Uncoated board under that impression load behaves nothing like the caliper reading implies once the fibre orientation runs diagonal to the emboss geometry.