Project
PRJ-0287
Industry
Automotive & Mobility
Application
Inline inspection & traceability
Commissioned
Jun 20, 2025
Status
Commissioned
Customer-reported defects reduced from 210 to 38 PPM, with an image and result stored against every serial number.
Project
PRJ-0287
Industry
Automotive & Mobility
Application
Inline inspection & traceability
Commissioned
Jun 20, 2025
Status
Commissioned
Customer-reported defects reduced from 210 to 38 PPM, with an image and result stored against every serial number.
FIG. 1
Inline inspection & traceability
PRJ-0287
01
Commissioning results
Measured outcome
01
Commissioning results
Measured outcome
Customer-reported defects
210
→
38
PPM
6 months after SAT vs. prior 6 months · OEM-reported
Customer-reported defects
210
→
38
PPM
6 months after SAT vs. prior 6 months · OEM-reported
Inspection time per part
780
ms
4 cameras in parallel · inside 6.5 s station takt
Inspection time per part
780
ms
4 cameras in parallel · inside 6.5 s station takt
Data matrix first-read rate
99.96
%
22 production days · ≈183,000 parts
Data matrix first-read rate
99.96
%
22 production days · ≈183,000 parts
False-reject rate
0.38
%
Same sample · rejects re-inspected manually
False-reject rate
0.38
%
Same sample · rejects re-inspected manually
02
Engineering record
Challenge, approach and architecture
02
Engineering record
Challenge, approach and architecture
Challenge
Context. A brake-caliper assembly line supplying two OEM platforms, running at a 6.5-second takt over two shifts.
Existing process. Final inspection was visual: one inspector per shift checked bleed-screw presence, dust-boot seating and pad orientation. Part serial numbers were laser-marked but not read, so a complaint could not be traced to a shift, station or torque record.
Constraints.
No increase in takt; the inspection had to fit inside the existing final station cycle.
OEM change approval required for any change to the control plan.
Installation in two weekend stops.
Challenge
Context. A brake-caliper assembly line supplying two OEM platforms, running at a 6.5-second takt over two shifts.
Existing process. Final inspection was visual: one inspector per shift checked bleed-screw presence, dust-boot seating and pad orientation. Part serial numbers were laser-marked but not read, so a complaint could not be traced to a shift, station or torque record.
Constraints.
No increase in takt; the inspection had to fit inside the existing final station cycle.
OEM change approval required for any change to the control plan.
Installation in two weekend stops.
Engineering approach
We built a defect library with the client's quality team — 11 defect types, each with good and bad reference parts — and ran a feasibility study on 400 production parts before choosing optics.
Scope. Four 5 MP cameras in an enclosed station, with ring and dome lighting and a telecentric lens on the dimensional check; fixed-mount data matrix readers at line entry and exit; integration with the line PLC for reject handling; a traceability service writing results, images and process data to the plant database by serial number.
Engineering approach
We built a defect library with the client's quality team — 11 defect types, each with good and bad reference parts — and ran a feasibility study on 400 production parts before choosing optics.
Scope. Four 5 MP cameras in an enclosed station, with ring and dome lighting and a telecentric lens on the dimensional check; fixed-mount data matrix readers at line entry and exit; integration with the line PLC for reject handling; a traceability service writing results, images and process data to the plant database by serial number.
System architecture
Vision: 4 × 5 MP cameras, 430 mm working distance, 60 × 45 mm field of view per view.
Code reading: fixed-mount readers at entry and exit, verification grade logged per part.
Control: existing S7-1500 line PLC extended with a vision interface and a locked reject chute.
Data: results and images stored by serial number; OPC UA link to the MES for part genealogy.
System architecture
Vision: 4 × 5 MP cameras, 430 mm working distance, 60 × 45 mm field of view per view.
Code reading: fixed-mount readers at entry and exit, verification grade logged per part.
Control: existing S7-1500 line PLC extended with a vision interface and a locked reject chute.
Data: results and images stored by serial number; OPC UA link to the MES for part genealogy.
Deployment & commissioning
Acceptance criteria were agreed before FAT: detection of every defect type in a 220-part challenge set, ≤ 0.5 % false rejects and ≥ 99.9 % first-read rate. FAT was run on client parts in our integration hall; on-site installation used two weekend stops.
The client updated their control plan and obtained OEM approval before the station was used for release.
Deployment & commissioning
Acceptance criteria were agreed before FAT: detection of every defect type in a 220-part challenge set, ≤ 0.5 % false rejects and ≥ 99.9 % first-read rate. FAT was run on client parts in our integration hall; on-site installation used two weekend stops.
The client updated their control plan and obtained OEM approval before the station was used for release.
- INPart arrivalStation index
- READCode readSerial at entry
- CAMInspect4 views, 780 ms
- DECDecisionOK / reject chute
- DBRecordImages + results by serial
03
Specification & results
Specification, acceptance and results
03
Specification & results
Specification, acceptance and results
Cameras
Parallel acquisition
4 × 5 MP
Cameras
Parallel acquisition
4 × 5 MP
Working distance
430 mm
Working distance
430 mm
Field of view
Per camera
60 × 45 mm
Field of view
Per camera
60 × 45 mm
Station takt
Unchanged
6.5 s
Station takt
Unchanged
6.5 s
Defect library
Agreed with client quality team
11 types
Defect library
Agreed with client quality team
11 types
Parameter
Specified
Measured
Result
Inspection time
≤ 900 ms
780 ms
Pass
Inspection time
Specified
≤ 900 ms
Measured
780 ms
Result
Pass
Challenge-set detection
100 %
220 / 220
Pass
Challenge-set detection
Specified
100 %
Measured
220 / 220
Result
Pass
Data matrix first-read rate
≥ 99.9 %
99.96 %
Pass
Data matrix first-read rate
Specified
≥ 99.9 %
Measured
99.96 %
Result
Pass
False-reject rate
≤ 0.5 %
0.38 %
Pass
False-reject rate
Specified
≤ 0.5 %
Measured
0.38 %
Result
Pass
Results
Measure | Before | After |
|---|---|---|
Final inspection | Visual, 1 inspector per shift | Automated, all parts |
Traceability | Serial marked, not read | Serial, images, results, torque |
Customer-reported defects | 210 PPM | 38 PPM (−82 %) |
Complaint investigation | Days, by shift | Minutes, by serial number |
Results
Measure | Before | After |
|---|---|---|
Final inspection | Visual, 1 inspector per shift | Automated, all parts |
Traceability | Serial marked, not read | Serial, images, results, torque |
Customer-reported defects | 210 PPM | 38 PPM (−82 %) |
Complaint investigation | Days, by shift | Minutes, by serial number |
04
Drawings
Layout and detail drawings
Original technical drawings prepared for this record. Illustrative, not to scale.
04
Drawings
Layout and detail drawings
Original technical drawings prepared for this record. Illustrative, not to scale.
FIG. 1 · Inspection and traceability stations
FIG. 2 · Camera station elevation
When a complaint comes in, we can show the image of that exact part. That changed the discussion with our customer.
Quality Manager
Client · tier-1 supplier (anonymized)
PRJ-0287
05
Related
Solutions, platforms and resources
05
Related
Solutions, platforms and resources
A
Related solutions
B
Technology stack
C
Related resource
Have a similar application?
Discuss an inspection requirement
An engineer replies within one business day.
Have a similar application?
Discuss an inspection requirement
An engineer replies within one business day.