Fluid Dispensing Inspection AI

Bring real-time intelligence to every dispensed dot, bead, and coating on the line with Roboflow vision AI for fluid dispensing inspection.

Dots, beads, and micro-dispensing:

• Confirm every dot and bead is present, in position, and within size tolerance, with segmentation that measures dot diameter and bead width as a volume proxy on each cycle.
• Detect satellites, stringing, tailing, and skips on adhesive and silicone dispenses, including dots at and below 100 microns with a high-resolution camera on the dispense head.
• Flag drift in dot size or placement across a run so process engineers can catch a clogged needle or pressure fault before it produces a full tray of rejects.

Solder paste, underfill, and coatings:

• Inspect solder paste deposits for presence, bridging, insufficient volume, and smearing as a lightweight complement to SPI on lines that do not have one on every product.
• Verify underfill and potting fill coverage, voids, and overflow onto keep-out areas.
• Check conformal coating for coverage, thin spots, and masking-area contamination under UV or white light.

Embedded inspection and traceability:

• Embed inspection in dispensing equipment through the Roboflow API or edge runtime so OEMs can ship a system that inspects what it dispenses without a separate vision vendor.
• Log every dispense with image, measurement, and pass/fail against the board or device serial for IPC and ISO 13485 traceability.
• Retrain in-house when a new fluid, substrate, or product variant launches so the same inspection model covers the next program.

Bring intelligence to every dispense cycle today. Stop skipped dots and bridged pads from becoming rework, field failures, and held lots.

‍

Frequently asked questions

What is dispensing inspection with Vision AI?

Dispensing inspection with Vision AI uses a camera on the dispense head, an inline station, or an assembly cell and a trained computer vision model to verify that each dispensed adhesive dot, bead, solder paste deposit, underfill, potting fill, or conformal coating is present, correctly positioned, within size and shape tolerance, and free of satellites, stringing, skips, and bridging. Dot diameter and bead width are measured on each cycle as a proxy for dispensed volume, and results are logged against the board or device serial. The inspection can be documented under IPC-A-610 for electronics assemblies and ISO 9001, with ISO 13485 and FDA 21 CFR Part 820 for medical devices.

Can Vision AI inspect micro-dispensed dots at 100 microns?

Yes. The approach uses a high-resolution camera close to the dispense head, telecentric or macro optics where the working distance allows, and a segmentation model that measures each dot rather than only detecting it, with image tiling in the Workflow so small features are not lost when the image is resized.

Does fluid dispensing inspection support IPC-A-610 and ISO 13485?

Yes. IPC-A-610 (Acceptability of Electronic Assemblies) defines acceptance criteria for solder, adhesive, and coating conditions such as bridging, insufficient paste, and coating coverage, and ISO 13485 (the quality management standard for medical device manufacturing) expects validated inspection with documented acceptance criteria and traceable records, alongside FDA 21 CFR Part 820 for devices sold in the US. Roboflow serves as the inspection engine: it runs the model, records every image, measurement, and decision, and supports IQ/OQ/PQ documentation. Your quality and process engineering teams keep ownership of the acceptance criteria, class definitions, and reaction plans.

Can it integrate with our dispensing equipment, SMT line, and MES?

Yes. Results are available over REST, MQTT, and OPC UA and can be written directly to a database, so pass/fail and measurement data flow into MES and quality systems such as Aegis FactoryLogix, Siemens Opcenter, SAP, Ignition, and MasterControl. For equipment OEMs, the Roboflow API and edge runtime embed inspection in the dispensing platform itself, so a fail can pause the head, trigger a re-dispense, or divert the board through a PLC or machine controller signal. Every decision is stored with its image, timestamp, and model version, and for medical assembly an FDA 21 CFR Part 11 audit trail.

‍