machines
Sandustry Filter and Advanced Filter Routing Guide
Sorting failures are dangerous because they may pass most material correctly while a small contaminant reaches an expensive stage. This guide treats Filters as tested interfaces rather than magic cleanup blocks and avoids assuming that demo-era direction or material support remains current.
Early Access · logistics verification · Checked 2026-08-22
Confirm the current filter type
Check the official Wiki for the current building name and documented material class. Do not assume a solid Filter handles liquids or gases, or that an Advanced Filter is required for every case, without current evidence. Record the checked date beside the blueprint. If terminology differs from a video, use the current game and official reference as the boundary.
Build an isolated orientation test
Place the Filter where all sides are visible and connect temporary destinations. Feed a tiny clean sample of the selected material, then a tiny clean sample that should be rejected. Label the observed input, selected output, and rejection output. Rotate or rebuild only after clearing the test so old pixels cannot make the new orientation look successful.
Test mixtures after clean samples
A clean pass proves basic orientation but not behavior under a mixed pile. Combine only a small measured sample and watch for bridging, burial, or one material blocking another. Keep both destinations open. If behavior changes with fill level, use a thinner feed or a buffer that spreads material before sorting rather than increasing upstream pressure.
Give rejection a real destination
The rejected stream is part of the design, not waste that can be ignored. Route it to quarantine or a named return buffer with a full-line marker. Never return it directly to the same mixed inlet, which creates an invisible loop and repeated work. If the rejection output fills, the selected output may appear faulty even when the filter decision remains correct.
Prevent backflow
Check whether a full destination can push or settle material back across the Filter boundary. Use elevation, one-way transport behavior that you have tested, and buffer space to keep outputs distinct. Simulate a full selected destination and a full rejection destination separately. The safe response should contain material without allowing either stream to contaminate the inlet.
Handle fluids in containment
When testing a documented liquid or gas path, isolate it from the main reservoir and valuable storage. Provide an emergency basin or vent path appropriate to the currently verified system. A small leak can spread farther than a solid sorting mistake. Confirm pump, pipe, and vent behavior separately before combining them with an Advanced Filter.
Commission at increasing feed rates
Start below normal supply and observe both outputs. Increase in steps while watching the inlet depth and destination levels. If sorting fails only when piled material reaches a certain height, record that physical limit and redesign delivery. Do not describe the highest brief rate as filter throughput; use a sustained conservative rate that preserves separation.
Recheck after patches
Retest orientation, accepted class, full-output behavior, and mixed-feed separation after material or logistics updates. Keep the old blueprint labeled with its build rather than silently changing arrows. Community posts about wall mode or direction are useful clues, but publish a workaround only after it is reproducible and does not depend on an undocumented bug.
Maintain a filter verification card
For every production Filter, keep a short card listing building type, selected material, input side, selected output, rejection output, allowed feed depth, overflow destination, current build, and last mixed-sample test. Add the full-state result for both outputs. Attach the card to the blueprint notebook rather than relying on color arrows in a screenshot. When a contaminant appears, compare the live block with the card before rotating or replacing it; a destination backup or buried interface can imitate wrong selection. After a patch, retest one clean selected sample, one clean rejection, and one small mixture. Update the card only after all three behave consistently, and quarantine anything produced during uncertain trials.
Sources and evidence notes
- Sandustry on Steam ↗
Tier 1 · Early Access retail store listing. Official identity, release state, features, platforms, and requirements
- Official Sandustry Wiki ↗
Tier 1 · Current official community wiki. Named materials, buildings, and documented game systems
- Sandustry Official FAQ ↗
Tier 1 · Pinned Early Access FAQ and support discussion. Current support topics, features, and reporting guidance