Quality mechanics: the cascade, modules and recycling
Quality is the system that lets crafted items come out at higher tiers with better stats. The mechanic that makes it tractable is a fixed cascade ratio: when an item upgrades, it does not simply go up one level. It jumps one level most of the time, two levels some of the time, three levels rarely, and straight to legendary in a tiny fraction of crafts. Understanding that cascade is the difference between guessing how many modules you need and calculating it.
Figures computed from game version 2.1.12, updated 2026-08-23.
The five playable quality levels
The quality data defines six entries including the hidden "unknown"
level used as a default before quality is researched. The five playable
levels are normal, uncommon, rare, epic and legendary. The game's
internal level field runs 0, 1, 2, 3 and then jumps to
5 for legendary — it skips 4 entirely. That gap is a display number,
not an array position. When the cascade computes outcomes it uses five
consecutive indices (0 = normal … 4 = legendary); using the game's
level field directly as an index would create a quality tier that does
not exist and silently absorb probability into it.
| Quality | Game level | Array index | Next quality |
|---|---|---|---|
| normal | 0 | 0 | uncommon |
| uncommon | 1 | 1 | rare |
| rare | 2 | 2 | epic |
| epic | 3 | 3 | legendary |
| legendary | 5 | 4 | none (top tier) |
The next_probability field is 1.0 for normal through epic.
It describes the cascade chain being connected — normal's next tier is
uncommon, uncommon's next is rare, and so on. It is not
the recycler's quality chance. The recycler's quality chance comes from
the quality modules fitted in it, exactly like an assembler. The
next_probability field was misread as "recycler always has
q = 100 %" in an earlier draft; that is incorrect.
The cascade ratio is fixed at 1/10
The single most important fact about the quality system is that the distribution of upgrade distances is a fixed geometric cascade, independent of the quality probability q. When a craft produces an upgrade, the probabilities are:
| Outcome | Probability |
|---|---|
| Stays at current quality | 1 − q |
| Goes up 1 level | 0.9 q |
| Goes up 2 levels | 0.09 q |
| Goes up 3 levels | 0.009 q |
| Jumps to legendary | 0.001 q |
The coefficients 0.9, 0.09, 0.009 and 0.001 form a geometric sequence with ratio 1/10. That ratio is hardcoded in the game's quality prototype and does not change with modules, beacons or machine tier. What q controls is only the total probability that any upgrade happens at all, which is q itself: 0.9q + 0.09q + 0.009q + 0.001q = q.
A common mistake is to assume that an item with q = 10 % has a 10 % chance to go up exactly one level. The one-level probability is 0.9 × 0.10 = 0.09, or 9 %, not 10 %. The remaining 1 % is split across the higher jumps. The error hides when q = 10 % because 9 % is close to 10 %, but it becomes obvious at larger q values. At q = 24.8 %, the one-level probability is 22.32 %, not 24.8 %.
Quality modules provide q
The quality probability q for a crafting machine comes from quality modules installed in it. Each module adds its effect value to the machine's total q. The modules also impose a 5 % speed penalty each.
| Module | Quality bonus (q per module) | Speed penalty |
|---|---|---|
| quality module | 1.0% | -5% |
| quality module 2 | 2.0% | -5% |
| quality module 3 | 2.5% | -5% |
The total q is the sum of all installed quality modules. For example, an assembling machine 3 with four quality module 3s has q = 4 × 0.025 = 0.10, or 10 %. An electromagnetic plant with five module slots can reach q = 12.5 %, and a cryogenic plant with eight slots can reach q = 20 %. The speed penalty is also additive: four modules at −5 % each reduce crafting speed by 20 %, so an assembling machine 3 normally at speed 1.25 runs at effective speed 1.0 when fully loaded with quality modules.
Machines that accept quality modules
Not every machine can host quality modules. The table below lists machines whose prototype allows the quality effect. Note that assembling machine 1 has no module slots and cannot use them, while assembling machine 2 has two slots. The recycler also accepts quality modules — they set its quality chance, just as they do for an assembler.
| Machine | Module slots | Crafting speed | Max q (all QM3) |
|---|---|---|---|
| assembling machine 2 | 2 | 0.75 | 5.0% |
| electric furnace | 2 | 2 | 5.0% |
| assembling machine 3 | 4 | 1.25 | 10.0% |
| chemical plant | 3 | 1 | 7.5% |
| centrifuge | 2 | 1 | 5.0% |
| recycler | 4 | 0.5 | 10.0% |
| biochamber | 4 | 2 | 10.0% |
| crusher | 2 | 1 | 5.0% |
| foundry | 4 | 4 | 10.0% |
| electromagnetic plant | 5 | 2 | 12.5% |
| cryogenic plant | 8 | 2 | 20.0% |
Recycling and the quality loop
Quality modules alone produce upgraded items at a low rate. The system
becomes powerful when non-legendary outputs are recycled. A recycler
takes an item and returns a fraction of its ingredients. In the
prototype data, recycling recipes return 25% of each
ingredient (the extra_count_fraction field is 0.25).
The recycler's quality chance is not hard-coded to
100 %. It comes from the quality modules installed in the recycler,
using the same cascade as any crafting machine. If you fit four quality
module 3s, the recycler runs at q = 10 %; if you fit none, q = 0 and
recycled items do not upgrade at all. This corrects a misunderstanding
in an earlier version of this page, which treated the data field
next_probability = 1 as the recycler's quality chance. That
field only records which tier follows which in the cascade chain; it
says nothing about the odds of an upgrade occurring.
The practical loop is: craft items with quality modules in the assembler, keep legendary outputs, send everything else to a recycler that also carries quality modules, and use the returned ingredients to craft again. Each recycling pass returns only 25% of the material, so there is a material cost to the loop, but the quality upgrades from both machines compound.
The recycler itself runs at crafting speed 0.5, has 4 module slots, and draws 180kW. Because its quality chance is module-driven, the choice between quality modules and speed modules in a recycler is a real trade-off: quality modules increase the upgrade chance on each recycle pass, while speed modules process more items per second.
Worked example: the verified anchor
The cost model on this page is a reference implementation that has been checked against an independent derivation to nine significant figures. The anchor point is q = 24.8% in both the assembler and the recycler, with the standard 25% recycling return. At that setting, the expected material cost of one legendary item — starting from normal inputs — is 154.7× the base recipe cost.
| Outcome from normal craft (q = 24.8%) | Probability |
|---|---|
| Stays at normal | 75.200% |
| Up 1 level (uncommon) | 22.320% |
| Up 2 levels (rare) | 2.232% |
| Up 3 levels (epic) | 0.223% |
| Jumps to legendary | 0.025% |
Direct legendary crafts from normal happen at 0.0248%, roughly one in four thousand. That is too rare to rely on directly. The recycling loop is what makes legendary production practical — but only when the recycler also carries quality modules.
The model solves a system of linear equations for each quality tier. Starting from one unit of material at tier k, a craft produces a distribution of outputs; non-target outputs are recycled, returning 0.25 material at each output's tier, which re-enters the craft step. The loop converges to an expected total material cost. Starting from progressively higher tiers:
| Target quality | Expected material per target item |
|---|---|
| uncommon | 2.91× |
| rare | 11.91× |
| epic | 43.12× |
| legendary | 154.7× |
The dramatic jump from uncommon (2.91×) to legendary (154.7×) reflects how many recycling cycles an item must survive to climb four tiers, losing 75 % material at each pass. The epic-to-legendary step alone costs 154.7× from normal inputs — but that number includes every prior tier climb. If you can feed epic ingredients into the final assembler, the remaining cost to legendary is much smaller because only one tier is left. This is why a strategy of pre-upgrading ingredients through a few recycling passes before final assembly can reduce total material consumption.
How module configuration changes the cost
The calculator below lets you set the quality chance independently for the assembler and the recycler. The table shows several common configurations with the same q in both machines. At q = 0 (no quality modules anywhere), legendary is unreachable and the model returns no finite cost.
| Configuration | q | Material / legendary (from normal) | Material / legendary (from epic) |
|---|---|---|---|
| 1× quality module 1 | 1.0% | 48,282× | 5211× |
| 4× quality module 1 | 4.0% | 6882× | 911.1× |
| 4× quality module 2 | 8.0% | 1922× | 316.0× |
| 4× quality module 3 | 10.0% | 1212× | 217.9× |
| 5× quality module 3 (EM plant) | 12.5% | 747.3× | 148.2× |
| 8× quality module 3 (cryogenic plant) | 20.0% | 255.9× | 63.8× |
Beacons and quality
Beacons can transmit module effects to surrounding machines, but in the Space Age expansion, beacons do not transmit quality effects. A beacon accepts speed, efficiency and consumption modules but not quality modules. This means quality probability must come from modules inside the crafting machine itself, which caps q at the machine's module slot count times the best quality module's effect. A beacon loaded with speed modules is still useful alongside quality modules: the speed modules compensate for the quality modules' speed penalty, letting the machine craft faster while retaining its quality chance. The beacon's distribution effectivity means each beacon transmits a fraction of its module effects, so the exact speed compensation depends on beacon count and placement.
Why quality module bonuses are small
A quality module 3 adds only 2.5 % quality chance per module, which looks modest compared to a speed module 3 at +50 % speed. The numbers are small because the cascade amplifies them: a single q value produces legendary outcomes through both direct legendary jumps and multi-level upgrades that reach legendary after recycling. If quality modules granted large bonuses, legendary items would be trivial to produce and the recycling loop would have no purpose. The small per-module values are what make the craft-recycle cycle necessary rather than optional. The speed penalty is also deliberate: quality modules trade throughput for upgrade chance, and a machine running four of them crafts 20 % slower.
The practical implication is that quality production should be planned as a separate block from normal production. A factory that makes legendary items is not simply a factory with better modules; it is a factory with a recycler loop, a return path for ingredients, and enough excess input capacity to absorb the 75% material loss at each recycling pass. At the reference anchor of q = 24.8% in both machines, the steady-state multiplier is 154.7× from normal inputs. The initial ramp-up, when no higher-quality ingredients exist yet, consumes more material because every item starts at normal and must pass through the cascade multiple times before any legendary output appears.
Common mistakes
The most frequent error is treating q as the probability of going up exactly one level. The cascade splits q across four jump distances, and the one-level probability is always 0.9q, not q. This error is invisible when q = 10 % because 9 % is close to 10 %, but it causes significant miscalculation at higher q values.
The second mistake — and the one that corrupted an earlier version of
this page's calculator — is treating the recycler as if it always had
q = 100 %. That misreads next_probability = 1 in the
quality data as a quality chance, when it only describes how the
cascade tiers are linked. The recycler's q is set by its own modules.
A recycler with no quality modules does not upgrade items at all; the
returned ingredients stay at the same tier.
The third mistake is using the game's display level field (0, 1, 2, 3, 5) as an array index. The gap at 4 is intentional in the display numbering, but the calculation engine uses five consecutive indices 0 through 4. Treating level 5 as array position 5 creates a sixth slot that absorbs probability into a nonexistent tier, producing cost estimates that are wrong by orders of magnitude.
The fourth mistake is ignoring the material loss from recycling. Each pass returns only 25% of ingredients, so the loop is not free. The expected cost model above accounts for this loss, and the calculator lets you adjust the yield to see how sensitive the result is.
Applies when…
- The cascade ratio of 1/10 is fixed by the game prototype and does not change with modules, beacons, machine tier or game settings.
- Quality probability q comes from quality modules inside the crafting machine and the recycler. Beacons do not transmit quality effects.
- The recycler's quality chance is module-driven, not hard-coded to 100 %. A recycler without quality modules returns ingredients at the same tier.
- The material multiplier assumes a steady-state craft-recycle loop where all non-legendary outputs are recycled and all ingredients are recrafted.
- The recycling yield of 25% is read from the generated recycling recipe data. Mods that change recycling yield will change every multiplier on this page.
- Array indices run 0 through 4 (normal, uncommon, rare, epic, legendary). The game's display level skips 4, but the calculation does not.
Related
- Beacons and modules — how beacons transmit module effects and why they do not carry quality
- How production ratios work — the arithmetic behind machine counts that quality builds still depend on
- Science pack ratios — quality research is gated behind the science pack chain