Short answer: a solar-fed factory needs enough stored energy to carry the complete machine load after sunset. If the grid still has charge, diagnose the material route and output before adding more generation.

Use a four-stage night test

  1. Record the daytime state. Note active machines, generation, storage and demand while the line is known to work.
  2. Watch sunset, not only the stopped machine. If production falls as solar generation disappears, check whether stored charge begins supplying the same demand.
  3. Read the grid at the failure moment. Empty storage with unmet demand supports a power diagnosis. Remaining charge means the next check belongs at the machine.
  4. Run one small batch. Confirm inputs reach the machine and the output has somewhere to go while power is available.
Success condition

At night, stored power continues to cover the active factory load and one test batch moves from input through output without pausing.

Do not mix these failure types

What you observeDiagnosis path
All powered machines stop together at sunsetGeneration and stored charge first.
One machine stops while others continueThat machine’s ingredients, connection and output.
“Could not pack a capsule” with materials and powerConfirm 1.0.4; this bug is listed as fixed.
Power returns at sunrise without rebuildingStrong evidence that nighttime storage or load is the bottleneck.

Exact power tables are not settled

Two current guides publish conflicting power figures for the same processing machines. Without an official specification or an editor-run measurement, copying either table would create false precision. Use the demand shown by your current build, then size generation and storage against that observed load. This guide therefore does not prescribe a fixed number of panels or batteries.

Change one variable at a time

Adding panels, storage, machines and belts together can make a broken line appear fixed without revealing why. Start with one known daytime load. If stored charge reaches zero before sunrise, add capacity or reduce the active load, then repeat the same test.

Evidence limit: the power workflow is synthesized from third-party automation guides; the official patch notes reviewed do not document the night-power system. Official 1.0.4 support on this page is limited to the separate capsule-packing bug fix. This is not a measured all-night blueprint.

Sources