TTML Test System

Automated LED Lamp Testing

Raspberry PiPythonOpenCVNode-RED

The Mission

Automate visual pass/fail decisions for an LED lamp and automatically measure voltages and currents to reduce test time from minutes to seconds.

Overview

TTML was a self-built automated test system developed in an EMS environment to eliminate manual judgement from repetitive LED lamp quality checks. I designed both the physical test rig and the full software stack, combining hardware control, computer vision and workflow automation. The system guided operators through a structured sequence while handling validation, timing and traceability automatically.

// FIG_01: Operator UI with real-time test results
Result // 01

Reduced test time from minutes to seconds through full automation

Result // 02

Eliminated subjective judgement by validating LED count, colour and illumination via computer vision

Result // 03

Delivered repeatable, traceable test results linked to RFID serial numbers

Result // 04

Successfully deployed into a live manufacturing cell

01 // Discovery & Design

// Discovery

Mapped the full manual test workflow and identified repetitive validation steps suitable for automation

// Key Decisions

Developed on a Raspberry Pi, using the Pi-Camera for vision and Node-RED for workflow orchestration

// Trade-offs

The system was not networked, which means that data was stored locally only, so analysing data involved a manual process

🛠️ One thing we changed after launch

Added clearer instructions on the screen for the operators, so they knew what test was being done at all times and the outcome, as they found it quite interesting!

02 // Technical Execution

// Tech Stack

Raspberry-Pi based system, with NodeRed as orchestrator of Python and Bash scripts and OpenCV for images

// Implementation

Raspberry Pi with weekly (manual) back-ups of test results

// System Design

RFID tags attached to the PCB were used as a unique identifier for each lamp

💡 One thing I would do different now

Refine the UI design and network the tester for automated back-ups and analysis

03 // Ownership

// Team

I designed and built the hardware and software and Production Engineering designed and built the mechanical test jig

// Timeline

From idea to conception in a few months; the hardest and most rewarding part was automating the LED count, the illumination level and the colour accuracy

// Ownership

Owned the full system before transitioning it to Production Engineering

❤️ The thing I enjoyed the most

Designing a complete hardware-software system independently and seeing it operate reliably on the manufacturing floor

04 // Evidence

// FIG_01: Operator UI with real-time test results
// FIG_02: Electro-mechanical jig with Raspberry Pi
// FIG_03: Mapping the manufacturing cell workflow to understand assembly flow and where automated testing should sit
Previous
Next