Quick answer
Clear the test lane, send one product, and watch until it reaches the Maw. Fix the first place it slows, catches, escapes, or bounces back. Do not add another machine or more product until one clean pass works.
The five-layer check
| Layer | What failure looks like | First correction |
|---|---|---|
| Source | Nothing leaves the machine | Test unlock, power, and output alone |
| Movement | Product stops at a seam or turn | Straighten the first hard transition |
| Containment | Product escapes the lane | Add side control before adding speed |
| Release | Stockpile compresses at the gate | Widen and simplify the release |
| Maw | Product bounces at the final edge | Reduce the last drop and align the mouth |
Recovery sequence
Stop feeding the failure
Pause the upstream source or isolate the lane. More product hides the original catch point.
Clear loose objects
Use the v1.0.5 Clear Items control for a full reset when appropriate. Community clips also show the Broom as a useful manual recovery tool, but treat it as cleanup—not a substitute for stable geometry.
Run exactly one product
Keep the camera on the object from source to Maw. The first bad transition is the work order.
Change one variable
Move one Panel, soften one corner, or change one Bumper/Fan position. A multi-part rebuild makes the successful change impossible to identify.
Repeat, then scale
Require several clean single-item passes before restoring continuous production.
Tool-specific cautions
- Broom: useful for community-observed loose-object recovery; it does not prove the line is fixed.
- Bumper: can redirect a product toward the hole, but an energetic final impact may trade one clog for a bounce-out.
- Fans: use them to deliver through a proven channel, not to break fragile products at random.
- Hydraulic Press: isolate its input and exit when testing; a dramatic machine can hide a basic transport bottleneck.
Community examples
These Twitch clips show three common failure and recovery patterns: a clogged line, Broom cleanup, and a Bumper test. Use them to recognize a failure pattern, not to predict exact throughput.