Category: PIR

How Fresnel Factory Built an AI-Driven Automated Production Line for Optical Sensors

Last Updated: 2026-04-25

Author: Ashton Myung Kim, CEO of Fresnel Factory Inc.

Quick Answer

Fresnel Factory’s automated production line is designed to deliver high-consistency optical components with full traceability through AI-based vision inspection, barcode-based tracking, and multi-stage quality control.

  • AI-based vision inspection covering 100% of parts
  • Barcode-based tracking from injection molding to shipment
  • Multi-stage QC: injection, assembly, and final inspection
  • Cycle time of approximately 3 seconds per part
  • Real-time detection of scratches, bubbles, misalignment, and missing components
  • Data-driven quality control for defect reduction and process stability

Why Automated Optical Production Matters in 2026

Optical components for sensors, especially PIR motion detectors, TMOS sensors, and infrared modules, are highly sensitive to surface defects, alignment errors, adhesive placement, and production variation.

A small scratch, bubble, or misalignment can affect optical performance, customer assembly yield, and final sensor reliability. Traditional manual inspection depends heavily on operator skill and fatigue, making it difficult to maintain consistent quality at high production volume.

Fresnel Factory developed its automated production line to reduce manual variation, inspect every part, and create traceable production records from molding to shipment.

How Does the Automated Production Line Work?

1. Injection Molding, Gate Cutting, and Barcode Traceability

The process starts immediately after injection molding. A robot picks the molded optical parts, performs gate cutting, places the parts into trays, and links each tray to a barcode.

Each tray contains 100 pieces, and each shipping box contains 3,000 lenses. By assigning barcodes at tray and box level, Fresnel Factory can trace production date, packing date, packing number, shipment history, and related inspection records.

If a customer reports a defect, Fresnel Factory can use the tray barcode or box barcode to review the production history and inspection images for that lot.

2. Multi-Stage AI Vision Inspection

The automated line is divided into three major quality control stages:

  • Injected Part QC: surface inspection, scratch detection, and dimensional consistency check
  • Assembly QC: protective film position, film angle, adhesive size, and adhesive placement
  • Finished Product QC: bubble detection, missing film, misalignment, and final dimensional confirmation

This is not sampling inspection. The system is designed for 100% inspection of the produced parts.

3. Protective Film and Adhesive Assembly

In the assembly section, robots pick the protective film and adhesive, measure their size and position, adjust alignment, and apply them to the optical part.

The system checks whether the protective film and double-sided adhesive are correctly positioned. It also measures the distance between the injected part edge and the adhesive film edge to confirm assembly consistency.

What Does the AI Vision System Inspect?

Fresnel Factory’s AI-assisted inspection system is designed to detect both appearance defects and assembly defects.

Inspection Area Detected Items Purpose
Injected optical part Scratch, contamination, dimensional variation Prevent defective molded parts from entering assembly
Protective film Missing film, position error, angle error Protect optical surface and maintain assembly accuracy
Double-sided adhesive Size variation, position error, edge offset Maintain stable bonding and customer assembly yield
Finished product Air bubbles, missing components, misalignment Prevent defective parts from being shipped

What Is the Production Speed and Efficiency?

The automated line can process approximately four parts every 12 seconds, which is about 3 seconds per part. This enables Fresnel Factory to combine high-throughput production with full inspection coverage.

Metric Value
Cycle time Approximately 12 seconds per 4 parts
Per-unit time Approximately 3 seconds per part
Inspection method 100% inspection
Traceability level Tray barcode and box barcode
Box quantity 3,000 lenses per box
Tray quantity 100 pieces per tray

Why Does AI-Based Inspection Improve Quality?

1. It Reduces Human Variation

Manual inspection quality can change depending on operator experience, concentration, and fatigue. AI-based vision inspection applies consistent inspection logic to every part.

2. It Enables Real-Time Filtering

Defective parts can be detected during the production flow instead of being found after shipment or during customer assembly.

3. It Creates a Data Feedback Loop

Inspection images and measurement data can be used for process analysis, Cpk review, yield monitoring, and continuous improvement.

4. It Supports Cost Optimization

When the process is stable and traceable, redundant manual inspection can be reduced, rework can be minimized, and the total cost of quality can be improved.

How Does Barcode Traceability Help Customer Quality Communication?

Barcode traceability is important because reported quality issues do not always originate from the same supplier, production date, or process condition.

When a customer or contract manufacturer reports a defect using a tray barcode or box barcode, Fresnel Factory can check:

  • Injection date
  • Production lot
  • Packing date
  • Shipment record
  • Inspection image history
  • Whether the part belongs to Fresnel Factory’s production lot

This helps separate actual production defects from reporting errors, supplier mix-up, or customer-side classification issues.

How Is This Different from Traditional Optical Component Manufacturing?

Item Traditional Manufacturing Fresnel Factory Automated Line
Inspection method Manual or sampling inspection AI-assisted 100% inspection
Traceability Lot-level or limited tracking Tray and box barcode tracking
Defect detection Operator-dependent Vision system and AI-based detection
Data collection Limited Inspection images and measurement records
Root-cause analysis Slow and subjective Data-based and traceable
Production consistency Depends on operator and shift System-controlled process

What Optical Sensor Applications Can Use This Production Model?

Fresnel Factory’s automated production model is suitable for optical components that require stable mass production, tight assembly control, and traceable quality records.

  • PIR Fresnel lenses for motion detection
  • TMOS sensor optical covers
  • IR sensor windows
  • Optical parts with protective film
  • Optical parts with double-sided adhesive
  • Custom sensor optics requiring mass production quality control

What This Means for Engineers and Buyers

For hardware engineers, the main value is process consistency. A stable optical component helps reduce variation in final sensor performance.

For sourcing and quality teams, the main value is traceability. When a defect is reported, the barcode system allows production records and inspection images to be reviewed quickly.

For program managers, the automated line supports scalable production while keeping quality communication data-based and objective.

Key Takeaway

Fresnel Factory’s AI-driven automated production line combines injection molding, robotic assembly, AI-based inspection, and barcode traceability into a single production flow.

This system is designed to reduce manual variation, improve defect detection, support high-volume optical component production, and provide reliable quality records for customers developing PIR, TMOS, and infrared sensor products.

FAQ

Q1. Is AI inspection better than manual inspection?

Yes. AI inspection provides consistent criteria and reduces operator-to-operator variation, especially for small surface or assembly defects.

Q2. Does 100% inspection slow down production?

No. The automated line processes approximately four parts every 12 seconds, or about 3 seconds per part, while maintaining 100% inspection coverage.

Q3. Can defects still escape detection?

A small possibility always exists in any manufacturing process, but multi-stage AI vision inspection and barcode traceability significantly reduce the risk and make root-cause analysis faster.

Q4. How is traceability handled?

Traceability is managed through tray barcodes and box barcodes. Each tray contains 100 pieces, and each box contains 3,000 lenses.

Q5. Can this system reduce cost?

Yes. By improving yield, reducing rework, and minimizing redundant manual inspection, the automated line can support total cost optimization.

Q6. What products can be produced on this automated line?

The system can be applied to PIR Fresnel lenses, TMOS optical parts, infrared sensor windows, and optical components requiring adhesive or protective film assembly.

Next Step

If you are developing PIR motion sensors, TMOS-based sensing modules, or infrared optical assemblies, Fresnel Factory can support optical design, prototyping, performance evaluation, and scalable production.

Contact Fresnel Factory for custom optical component development or mass production support.

Author

Ashton Myung Kim
CEO, Fresnel Factory Inc.
IEC and ISO Sensor Standard Expert
LinkedIn: Ashton Myung Kim

Related Articles

  • AI-Based Optical Inspection Systems in Sensor Manufacturing
  • How to Reduce Defect Rates in Injection-Molded Optical Parts
  • PIR Fresnel Lens Design Guide for Smart Sensors


Preparing the Device Under Test (DUT) for IEC 63180 Compliance

Preparing the Device Under Test (DUT) for IEC 63180 Compliance

Platform: Fresnel Factory Scaled Performance Test (SPT) machine

Overview

IEC 63180 defines test conditions and procedures for evaluating motion detectors based on passive infrared (PIR) sensing.
Fresnel Factory’s SPT machine is a compliant evaluation system that can assess performance using a detector’s
simple on/off event signal; it also supports recording the signal amplitude when needed. For IEC 63180 testing specifically,
the amplitude capture is not required, so your DUT only needs to provide a clean detection (on/off) signal and a common ground.

What the SPT Machine Needs from the DUT

  • Detection Output (on/off): A single digital output that changes state when motion is detected.
    Configure the SPT machine to trigger on the edge (rising or falling) that matches your DUT’s output polarity.
  • Common Ground: Share ground between the DUT and the SPT machine to ensure a stable reference.

Learn about DigiPyro-based PIR modules

LowPower DigiPyro Application Note (PDF)

Electrical Interface Requirements

The SPT machine accepts logic-high inputs in the range of 3 V to 12 V, so typical 3.3 V or 5 V
PIR module outputs can be connected directly. To maximize signal integrity:

  • Output Level: 3.3 V or 5 V logic is recommended (both are compatible).
  • Drive/Impedance: The SPT input is high impedance (≈10 MΩ), so a standard MCU pin or buffer op-amp can drive it.
  • Damping (optional): A small series resistor (≈100–330 Ω) at the DUT output can reduce ringing on long or noisy runs.

Cabling and Layout Guidance

  • Length: Keep the detection line short—preferably ≤ 1 m—to minimize edge delay and noise pickup.
  • Cable Type: Use twisted pair (signal+GND) and shielding where practical.
  • Environment: During tests, minimize nearby EMI sources (e.g., Wi-Fi radios, high-power LEDs) and thermal disturbances that can influence PIR behavior.

Minimal Wiring: Example

Basic Connections between DUT and SPT Machine
Line DUT Pin SPT Input Signal Type Typical Level
Power VDD (3.3 V or 5 V) — — —
Ground GND GND (shared) Reference 0 V
Detection Motion/Interrupt Output Trigger Input On/Off (edge-triggered) 3–5 V logic

Note: The SPT machine can record signal amplitude if you provide an additional measurement connection. However, IEC 63180 testing does not require amplitude data—only the on/off detection signal is needed.

Pre-Test Checklist

  1. Confirm the SPT input range (3–12 V) matches your DUT output level (3.3 V or 5 V).
  2. Share ground between DUT and SPT; verify a solid, low-noise return path.
  3. Set the SPT trigger edge to match your DUT’s output polarity (rising or falling).
  4. Use short, shielded cabling and tidy routing; add a small series resistor if edges ring.
  5. Allow PIR warm-up time, then verify a clean state transition when motion occurs.

Summary

For IEC 63180, the SPT machine only needs a reliable on/off detection signal from the DUT to evaluate motion-detection performance.
While the platform can also capture amplitude, that feature is optional for this standard. Prepare a clean digital output, keep wiring short and quiet,
and you’ll get repeatable, standards-aligned results with minimal setup.

PIR Sensor Lens Development, Verification, and Mass Production FAQ – Outdoor Camera Development

1. Should PIR lens surfaces be matte or glossy?

The surface finish affects not only the look but also how well the lens works. Glossy surfaces transmit more infrared light, giving the sensor a stronger signal. Matte finishes cut down glare and blend better with product design, which is why many consumer devices use them. In practice, engineers usually polish the actual lens area to be glossy while leaving the non-lens areas either matte or glossy, depending on design goals.

2. How are the lens and housing joined?

For high-volume production, double-sided adhesive tape is the simplest and cheapest option. It holds the lens firmly enough for indoor products. But outdoor cameras face rain and humidity, so additional sealing is common. Depending on the design, engineers add O-rings, cut grooves in the housing, or use waterproof tape and silicone gaskets. In factories overseas, especially in Vietnam, suppliers sometimes deliver lenses with waterproof tape already applied, so assembly workers only need to press the lens in place.

3. What is the impact of adding ribs in mold design?

Ribs around the lens edge make the structure stronger and improve how well the parts stay bonded. The trade-off is that the mold becomes more complex and expensive. Without ribs, the mold is simpler, but the finished product might be weaker or harder to seal against water. Adding ribs also often requires small hook holes, which further raises mold cost. So the decision usually comes down to the waterproof rating required and the target price of the product.

4. How are IP ratings reflected in the design?

If the device is for indoor use, engineers often choose the cheapest option: tape bonding only. Outdoor devices like CCTV units need at least a basic IP rating for dust and water. That means design choices such as grooves, silicone seals, or O-rings must be considered right from the mold design stage. Skipping this step early almost always causes redesign later.

5. What tools are used for PIR lens design?

Designers start with CAD software such as SolidWorks for the mechanical layout. At Fresnel Factory, engineers also run their own calculation tools to predict how the lens will shape the infrared field. After that, ray-tracing simulations check the detection zones, the field of view, and energy distribution. Using both mechanical and optical tools helps catch mistakes early, which saves time and money once molds are made.

6. What standards apply to PIR-based motion detectors?

Most companies follow IEC 63180. It defines three key tests: Radial (movement toward the sensor), Boundary (maximum distance and angle), and Tangential (side-to-side motion). These tests are easier to repeat with automated machines than with people walking back and forth. At Fresnel Factory, the Scaled Performance Machine was built to meet IEC 63180, so results are consistent across projects.

7. How much does performance testing cost?

A full set of the three IEC tests costs around USD 2,000. That covers radial, boundary, and tangential measurements. If a customer asks for extra scenarios—for example, testing the sensor at a higher mounting height—those are priced separately.

8. How long does it take to get a test report?

From the time Fresnel Factory receives the sample, testing and reporting usually take about three weeks. That includes running the measurements, reviewing the data, and preparing the report. For urgent projects, the three-week timeline can feel long, so it’s wise to build that time into the project schedule from the beginning.

9. How is the contract typically structured?

Most development contracts require a 50% down payment to begin work. The remaining 50% is due once the customer validates the sample. Mold costs are bundled into the development fee. Although the customer holds the rights to the mold, Fresnel Factory stores and maintains the physical mold and uses it for production. This arrangement is standard in the industry, since it protects quality and avoids problems with mold transfers.

10. Why are 3D modeling files required?

Without a 3D file, designers cannot check how the lens and housing fit together. Customers usually send STEP or IGES files, which show the exterior geometry. From that, engineers create lens patterns, plan mold details, and design seals for waterproofing. Even a basic exterior file is enough to start. Later, once molds are made, prototypes confirm that the design matches expectations. Sharing these files early helps prevent design errors and saves weeks of rework.

11. What materials are used for PIR lenses?

Most PIR lenses are made from Poly FIR200 or its compounded forms like SBK150 and HGW335. Poly FIR200 has good infrared transmission at 8–14 µm and flows well in injection molding. SBK150 is formulated for outdoor use and maintains performance even after years of UV exposure. Accelerated weathering tests show SBK150 still transmits more than 93% of its original level after five years, while cheaper alternatives may drop below 50%.

12. How are lens zones designed?

PIR lenses split the detection area into zones. For example, one design might use three zones covering 35°, 15°, and 7°. This allows the sensor to detect people moving at different heights and angles. If the zones are poorly planned, blind spots appear where motion is missed. That’s why zone design is usually checked both in simulation and in field tests before finalizing.

13. What is the service life of a lens mold?

A well-maintained PIR lens mold can last around four years and produce roughly 150,000 units per year. Actual lifetime depends on the plastic used, production conditions, and how often the mold is polished. Because the lens surface has fine patterns, wear shows up as lower optical performance, not just cosmetic flaws. Regular inspection and light re-polishing extend mold life.

14. How are final product shipment tests carried out?

Before shipment, customers may request a full IEC test or rely on the factory’s own quality checks. Final inspections usually measure detection distance, angle coverage, noise resistance, and waterproofing. In large production runs, even a 0.5% increase in defect rate can mean serious financial loss. For that reason, suppliers and customers should agree on the test scope and acceptance criteria early, before the first shipment leaves the factory.