Block Surface Inspection After Washing: Three Stations, One Pass

Overview

A block that leaves the washer looking clean can still fail for two different reasons that appear identical in an image: a casting defect that was always there, and a chip that the washing stage failed to remove. Separating those two is the point of inspecting immediately after the wash rather than at the end of the line.

The inspection also serves the washer. If chips are found on a consistent area, the finding points at a nozzle position, a filtration stage or a fixture contact, and it points at it within the same shift rather than three weeks later when the part fails at assembly.

Big Bird Industrial (formerly Shimada Big Bird Industrial) has been building cleaning and deburring systems for automotive powertrain components since 2001, and the station arrangement described here follows production lines for diesel blocks, using a robot-carried high-resolution camera with broad diffuse illumination across three stations.

1. Why The Position After The Washer Matters

The same defect carries different information depending on where it is found, and the position immediately after washing is the one that separates cause from consequence.

Attribution Without Guessing

At the end of the line, a chip on a machined face could have arrived from the washer, from the assembly operation, from a fixture, or from the environment. Immediately after the washer there is only one candidate, which turns an investigation into a check.

The Washer Gets A Result

A washing stage is normally judged by a sampled extraction, which reports a mass figure hours after the part was produced. An inspection station reports per part and per location, which is a different kind of feedback: it says which face, not just how much.

2. Three Stations And What Each Sees

A block cannot be inspected in one view. Its relevant surfaces face in several directions, and the surfaces that are hardest to see are often the ones where chips collect.

Underside, Perimeter And Review

The arrangement used on diesel blocks covers the bottom face, the perimeter faces, and a review position where ambiguous findings are examined by a person. The three are not redundant: the bottom face collects chips by gravity, the perimeter carries the machined faces that mate with other components, and the review position exists because not every ambiguous finding should be decided by a model.

Robot-Carried Camera Versus Fixed Cameras

A robot carrying a camera can reach several surfaces with one sensor, at the cost of motion time and of path accuracy. Fixed cameras covering a station are faster but need one per view. On a block, the robot arrangement is common because the number of views is large and the part is presented in one orientation.

3. A Chip And A Defect Look Alike

The hardest part of surface inspection after washing is that the two things it must distinguish are visually similar.

Why The Distinction Is Difficult

A metal chip sitting on a machined face has an edge, a shadow and a contrast against the background, which is exactly what a scratch or a dent presents. A model trained only on defect examples will classify chips as defects, and a station that does that produces false calls that erode trust within a week.

How The Distinction Is Made

The reliable separators are physical rather than statistical. A chip sits above the surface and casts a shadow whose direction moves with the light, while a dent is a depression whose shading does not. A chip can be displaced by an air puff; a defect cannot. Using a second exposure with a different light direction, or an air puff before re-imaging, converts an ambiguous classification into a physical test.

4. The Human Review Station

A review position is not a failure of automation. It is what allows the automated stations to be set at a sensitivity that catches everything without generating a queue of decisions nobody can resolve.

What Review Is For

Review handles the cases where the image does not separate the classes: a mark that might be a scratch or might be dried coolant, a spot that might be porosity or might be a chip. A person with a light and a magnification looks once and decides, which takes seconds and settles the question.

Keeping Review Honest

The risk is that review becomes a place where everything is passed under time pressure. Recording what the reviewer decided, and sampling those decisions against a laboratory check, keeps the station's overall escape rate measurable rather than assumed.

A review position also needs the right tool for the question being asked. A magnifier and a low-angle lamp settle a scratch; an air nozzle settles whether a mark is loose material; a borescope settles a bore that the camera cannot see into. Providing all three at the review position is cheaper than sending the part back to the washer to be re-examined.

5. Lighting A Wet Part

Inspecting immediately after washing means inspecting a part that may still be damp, and water changes how a surface looks. A film of water fills shallow scratches and hides them; droplets create highlights that read as features.

The practical answer is either to inspect after drying or to accept a wet part and design the illumination for it, using broad diffuse light that does not produce a specular image of the droplets. Where the criterion includes dryness, inspecting after drying is the only option, because a wet part cannot be judged for water.

6. Feeding The Result Back To The Washer

An inspection station produces data that is only useful if it reaches the people who can act on it. The useful form is a map: which surface, which location, how often, and whether the finding is a chip or a defect.

A cluster of chip findings on one face points at a nozzle or a filtration problem. A cluster of scratch findings points at handling or fixturing. Both are actionable within a shift, and both are invisible in a report that says only how many parts failed.

7. Cycle Time And Where The Station Sits

Inspection after washing competes with the line for the part, and the station has to release it quickly enough that the washer is not waiting.

Three Stations In Sequence

With a robot carrying the camera, the three views run one after another on the same part, so the station time is the sum of the moves plus the image processing. On a block that is a substantial fraction of a minute, which is why the station is often placed where a small buffer exists rather than directly in the washer exit path.

Sampling Versus Full Inspection

Inspecting every part is only possible when the station time fits inside the takt. Where it does not, the usual compromise is to inspect every part on the surfaces that carry the highest risk and to audit the rest on a sampling plan. That compromise should be written into the specification, because a station bought for full inspection and operated on sampling has a different justification.

The position also determines what the station can prove. Immediately after the washer it proves the state of the part as washed; a hundred metres down the line it proves the state of the part after transfer, handling and any waiting, which is a different claim and a weaker one. Moving a station downstream is easy on a layout drawing and expensive in what it can no longer explain. The same logic applies to placing it after a buffer or after cooling, where condensation and handling have already changed the surface.

Frequently Asked Questions

Should The Station Be Before Or After Drying?

After drying if the dryness is part of the acceptance criterion, because water on the surface prevents judging both the surface and the dryness. Before drying only if the process has no dryness requirement.

How Small A Defect Can Be Detected?

On machined block surfaces, defect contours at the half-millimetre scale are detectable with a high-resolution camera and diffuse light. Smaller features need a narrower field of view and more stations.

Does The Station Replace The Cleanliness Test?

No. Surface inspection finds what is visible on a face; the cleanliness test measures what is extracted from a region, including material inside passages. The two answer different questions and both are needed.

How Is The Station Validated?

With a set of parts carrying known defects and known chip contamination, both presented repeatedly. A station tested only on clean parts has demonstrated nothing about its detection rate.

Conclusion

Inspecting immediately after the washer turns a late failure into a same-shift signal, and it separates a chip from a casting defect at the only point where the distinction is unambiguous. Three stations cover the surfaces; the review position handles what the image cannot settle.

The station is worth building only if its output reaches the washer as a location map rather than as a pass rate. A count of failures tells nobody which nozzle to check.