Tech note: detection rate vs cycle time — one station or two? Let the data decide
On the same metal-part samples, two stations with nine images caught 6/9 scratches in about 3 s; one station with four images caught 4/9 in about 1.8 s. Both caught 100% of contamination and black spots. Which should you choose?
One of the most common questions is: “Can one station inspect everything?” The answer depends on how much detection rate you are willing to trade for cycle time. Here is a measured comparison on metal parts.
Two module designs
- Option A (two stations): three tiers of square bar lights, then a large ring light — 9 + 2 images
- Option B (one station): RGB ring light plus a white square bar light — 4 images
Results
- Scratches: A detected 6/9, B detected 4/9
- Contamination: 5/5 for both
- Black spots: 5/5 for both
- Imaging plus inspection time: A about 2.85–3.35 s, B about 1.6–2.1 s
Timing: about 100 ms per image and 50 ms per light switch, so 0.6 s for four images and 1.35 s for nine, plus computation time.
Why some defects were missed
In a larger follow-up test, the one-station option detected about 40% of faint bright spots. These are tiny surface glints that only appear when lit from one exact angle — a lighting setup too specific to be practical. For exposed copper, combined RGB light brought out the colour better than a white dome light, which sometimes made the copper bright too and caused false calls.
How to choose
If your spec focuses on contamination, black spots and exposed copper, one station is enough and nearly twice as fast. If shallow scratches and faint glints drive customer complaints, an extra station — or extra lighting for specific zones — pays off. We hand you this table during the assessment so you can decide with data.
