Background

In the magnetic circuit system of electrodynamic (moving-coil) loudspeakers, the external magnet structure is widely adopted in various electroacoustic products including multimedia speakers, Bluetooth/smart speakers, car audio systems, TV receivers and telephone receivers thanks to its low cost and excellent stability. A complete external magnet magnetic circuit generally consists of three components: T-yoke (magnetic conductive pole / Bottom Plate + Center Pole), permanent magnet (ferrite or NdFeB ring magnet), and washer (magnetic conductive top plate / Top Plate). The T-yoke sits at the bottom, with its base flange and central magnetic pole forming the framework of the magnetic circuit. The annular permanent magnet is sleeved around the central pole and attached to the flange surface of the T-yoke. The washer is pressed on top of the magnet, forming an even, symmetrical magnetic gap typically only 0.3–0.8 mm wide between itself and the central pole of the T-yoke. The voice coil moves back and forth within this magnetic gap to convert electrical energy into acoustic energy.

All connections between the three magnetic circuit components rely entirely on adhesive bonding. A typical process involves dispensing a ring of AB adhesive onto the flange surface of the T-yoke, attaching the magnet and rotating it to spread the adhesive evenly. Adhesive is then applied to the mating surface between the washer and the magnet, followed by fitting the basin frame positioning gauge and rotating to control adhesive overflow. The continuity, uniformity, positional accuracy and overflow volume of the adhesive layer directly determine whether the magnetic gap remains uniform and whether failures such as adhesive discontinuity, core offset or adhesive overflow into the gap occur. These factors further affect the loudspeaker’s sensitivity, distortion and long-term reliability. It can be stated that adhesive dispensing on T-yokes constitutes the critical quality bottleneck for the magnetic circuit section of external magnet loudspeakers.

Objectives

Centered on Songqi Intelligent’s vision tracking dispensing machine, this solution delivers a fixture-free, operator-free, high-precision, flexibly convertible and traceable automated solution for adhesive dispensing on T-yokes in external magnet magnetic circuit production. Specific objectives are listed as follows:

• Replace manual labor for annular adhesive dispensing and positioning between T-yokes and magnets, as well as between magnets and washers. A single unit can steadily replace 1 to 2 skilled operators.
• Utilize visual recognition and dynamic tracking to cut the rework rate caused by adhesive overflow, missing dispensing and core offset from over 5% to below 0.5%, and raise the yield rate of magnetic circuits to above 99%.
• Eliminate reliance on precision fixtures. Product model changeover supports calibration within minutes, boosting line change speed by 80% compared with conventional methods and lifting overall equipment efficiency by more than 20%.
• Achieve seamless integration with existing production lines, and support connection to MES systems for process data traceability.

Gluing operations for the magnetic circuit of external magnet T-yokes still rely heavily on manual labor, and the industry is generally confronted with the following pain points:

Shortage of Skilled Workers Restricts Production Capacity

• Traditional manual operations for magnetic circuit assembly normally require 6 workers (T-yoke picking and gluing, magnet picking and gluing, T-yoke and magnet assembly, magnetic gauge installation, frame mounting, magnetic gauge removal), which are highly dependent on skilled technicians.
• In manufacturing hubs such as the Pearl River Delta and Yangtze River Delta, annual salaries for proficient dispensing operators have exceeded 100,000 RMB, while staff turnover remains high. Employee departures lead to unstable production line cycle times and sharply increased delivery risks.
• Manual working efficiency typically fluctuates by ±20%, making stable mass delivery difficult to achieve.

Uncontrolled Adhesive Overflow/Insufficient Gluing Contaminates Magnetic Gap and Causes Catastrophic Failures

• The magnetic gap width is merely 0.3–0.8 mm. If adhesive sagging, stringing or excessive overflow flows into the washer central hole or magnetic gap, the voice coil will be jammed and result in coil rubbing. This manifests as noise, squeaking and coil collision sounds during pure tone testing, ranking among the most common catastrophic failures of speakers.
• Industrial process standards impose stringent limits on adhesive overflow (typical requirement: the width × height of circumferential overflow ≤ 1.5 mm×1.5 mm with no unilateral offset; overflow along the long side is also restricted). Stable compliance is hard to achieve through manual visual inspection.

Uneven Adhesive Volume and Mixing Ratio Cause Variations in Bonding Strength and Curing Quality

• Magnetic circuit adhesives are mostly two-component AB adhesives (epoxy resin / modified acrylate systems) that require thorough mixing at a strict 1:1 ratio. Manual dispensing produces inconsistent glue volume and unbalanced amounts of Component A and Component B, which easily trigger incomplete curing, insufficient bonding strength and major long-term reliability risks.
• Adhesive stringing may span into the voice coil gap and burn out the voice coil after power-on; adhesive sagging contaminates the effective area of the diaphragm and alters vibration mass.

Centering Deviation Is Hard to Eliminate

• Concentric assembly is required among T-yokes, magnets and washers. Manual magnetic gauge fitting and rotary lamination tend to create centering deviation. Such deviation leads to unilateral magnetic gaps, increased non-linearity of the BL factor, higher distortion, and even collision between the curled voice coil tail and the T-yoke.

Slow Product Changeover and High Fixture Costs Hinder Flexible Production

• Conventional equipment requires custom precision fixtures for each product, costing several thousand to over ten thousand RMB per set. Product changeover demands re-teaching and programming, which takes hours to several days.
• Against the backdrop where multi-variety, small-batch and fast-delivery orders have become commonplace, manufacturers dare not accept urgent orders and are reluctant to take small orders, severely undermining competitiveness in order acquisition.

Lack of Traceable Quality Data

• Manual gluing fails to record process data including adhesive volume, dispensing position and speed. When batch coil rubbing or abnormal noise occurs, root cause identification becomes difficult, leaving quality improvement without data support.

The long-term reliance on seasoned technicians for adhesive dispensing of external magnet T-yokes stems from multiple stringent requirements inherent to the process:

Narrow Annular Adhesive Path with High Difficulty in Trajectory Planning

• The dispensing surfaces are the narrow annular edge of the T-yoke flange and the bonding ring between the magnet and washer. The adhesive path is thin and must form a closed loop without breaks or gaps. The adhesive volume should sufficiently cover the bonding surfaces while preventing overflow that contaminates the central hole.

Extreme Magnetic Gap Tolerance Requiring Zero Margin of Error for Adhesive

• The magnetic gap of 0.3–0.8 mm demands continuous adhesive application to avoid loss of adhesion, and prohibits adhesive overflow that causes coil jamming. Extremely high standards are imposed on the positioning accuracy of the dispensing nozzle, adhesive output control, and adhesive profile (no stringing, no sagging, full bead formation).

Unavoidable Incoming Material Tolerances

• T-yokes and magnets have dimensional and geometric tolerances. Random variations exist in the placement position and angle of incoming materials on the conveyor belt. Conventional positioning relying on precision fixtures fails to provide adaptive compensation, which requires equipment capability for visual recognition and dynamic compensation.

Metering, Mixing and Dispensing Control for Two-Part AB Adhesive

• AB adhesive requires precise proportional metering, homogeneous mixing, on-demand dispensing with no stringing or dripping. Immediate suction back to cut off the adhesive flow upon completion of dispensing is mandatory; otherwise, residual adhesive will cure and block the valve body.

Process Coordination Between Rotary Bonding and Magnetic Gauge Insertion

• After dispensing, magnets are fitted and rotated 360° for uniform adhesive distribution. Magnetic gauges are then inserted for positioning prior to curing (7–10 minutes at ambient temperature, or 5–15 minutes in an oven at 65–75 °C). The process cycle must match the production line speed; otherwise, it will become a bottleneck limiting throughput.

Strict Specifications for Adhesive Profile

• The adhesive must exhibit no stringing, no sagging and full bead formation: stray adhesive strings bridging gaps may lead to coil burnout, sagging adhesive contaminates diaphragms, and insufficient adhesive results in weak bonding with risks of lead or component detachment.

Vision System ("Eyes")

• Hardware: 5-megapixel color Hikvision industrial camera + integrated customized Model 360 flicker-free LED light source, capturing workpiece images at high speed within a field of view of 320×256 mm.
• Capabilities: Automatically identifies the precise position, angle and posture of T-iron/magnetic steel on the assembly line with a positioning accuracy of ±0.15 mm. Products can be placed arbitrarily; the vision system performs automatic positioning and deviation correction, eliminating the need for physical fixtures entirely.

Motion Control ("Hands")

• Controller: Self-developed YMC-08-01 motion control card. Vision coordinates and motion trajectories are directly linked at the bottom layer to avoid latency and precision loss caused by intermediate-layer protocol conversion.
• Platform: All-steel KK modules (X-axis KK8620 / Y-axis KK6020 / Z-axis KK6010) equipped with AC servo motors (400 W for X/Y axes, 100 W with brake for Z-axis), featuring high rigidity and fast response.
• Rotary Axis: Songqi patented 360° rotary R-axis. Combined with self-developed algorithms, it enables circular glue path tracking and posture adjustment at any angle, perfectly suited for circumferential gluing of T-iron flange rings and magnetic steel rings.
• Precision: Motion accuracy of XYZ axes: ±0.02 mm; rotational accuracy of R-axis: ±0.15° (360° full rotation).

Glue Supply System

• Dual-barrel glue supply with low-level alarm: Enables non-stop glue feeding during production. Optional screw pump glue supply boosts efficiency by 20%–30% compared with conventional glue feeding methods, while reducing glue waste and lowering wear and maintenance costs of dispensing valves.
• Precise metering and anti-dripping design: Equipped with an instant automatic suck-back system that cuts off the glue path immediately after dispensing, preventing residual glue dripping and valve blockage. Compatible with various two-part AB glues such as epoxy resin and modified acrylate (1:1 ratio and adjustable ratios).
• Glue Compatibility: Supports two-component epoxy/acrylate structural adhesives commonly used in magnetic circuits, meeting requirements for non-stringing, non-sagging and full glue bead profile.

Self-developed Software & Data Closed Loop

• Fully self-developed vision software: Custom-tailored for the equipment structure with deep coordination between "hand (dispensing valve) and eye (camera)". No commercial licensing fees for third-party software such as Halcon, nor fees for version upgrades or secondary development.
• Template-based programming: Product changeover only requires loading or creating new vision templates, with calibration completed within minutes. Offline programming and storage of multiple sets of process parameters are supported.
• MES Integration & Traceability: Real-time recording of parameters including glue volume, speed, position and temperature. Supports in-depth connection with MES/ERP systems to provide complete data support for quality tracing and process optimization.

Integration with Existing Production Lines

• Embedded model: Mounts directly above customers’ existing conveyor belts and utilizes on-site belt conveying systems for material transfer without conveyor replacement or modification. The deployment workflow is simplified to "Workstation Confirmation → Equipment Positioning → Signal Connection → Parameter Calibration". Commissioning from unpacking to mass production normally takes only several hours.
• Cabinet model: Built-in integrated conveyor belt. It can be flexibly deployed as an independent workstation at any section of the production line without interfering with upstream and downstream processes, and capable of closed-loop independent operation.

Comprehensive Economic Benefits: Equipment efficiency exceeds competing products on the market by over 20%, customer repurchase rate stands at 83.2%+. The payback period in the Pearl River Delta region is generally 6–12 months. Fluctuation of production cycle is narrowed from ±20% to within ±3%, supporting stable 24-hour non-stop operation.

Main Technical Specifications of Equipment (Take GP-26-740 Embedded Model as an Example)

RX Series (Cabinet Type with Fixed Point Function) provides longer travel under equivalent accuracy (e.g. RX-26-940 with travel of 760×660 mm and belt width of 500 mm) and optional integrated conveyor belt. Effective dispensing travel and machine dimensions depend on model configuration.

Recommended Dispensing Process Parameters

Q1: What is the fundamental difference between vision follow-up gluing and traditional manual/T-yoke gluing?

A: Traditional gluing relies on precision fixtures and visual judgment and experience of skilled workers, featuring slow production line changeover, significant output fluctuation and poor traceability. Songqi vision follow-up gluing adopts 500W industrial cameras to identify workpiece poses in real time. The self-developed software automatically conducts deviation correction and drives the dispensing head to follow dynamically. It requires no fixtures, no manual operation and no line stoppage, stabilizes accuracy at ±0.02 mm, and fundamentally eliminates failures such as core offset and glue overflow into gaps.

Q2: Which T-yoke/speaker specifications are applicable to this solution?

A: It is suitable for external magnet round speakers (rotating outer diameter of frames ≤ φ200 mm) and external magnet square speakers (e.g., 200×60 mm frames), covering common 2.5–8 inch products. The outer diameter of T-yokes ranges approximately from φ40 to φ95 mm, the outer diameter of magnetic steels ranges from φ40 to φ100 mm, and the thickness of magnetic steels is 6–10 mm. Larger sizes can be supported according to machine models.

Q3: What types of glue are supported? How is the mixing ratio of AB glue controlled?

A: It is compatible with two-component magnetic circuit AB glues including epoxy resin and modified acrylate, with a standard mixing ratio of 1:1. The glue supply system adopts dual barrels and precision metering (screw pump optional for 20%~30% efficiency improvement). Instant suction back prevents glue dripping to ensure uniform mixing and consistent glue volume.

Q4: How to prevent glue from overflowing into magnetic gaps or the central hole of washers and causing voice coil rubbing?

A: The following measures can mitigate or resolve this issue:

1. The vision system precisely locates the flange of T-yokes and inner diameter of magnetic steels to keep the glue path outside the central hole;
2. Closed-loop control of glue output and glue height based on process parameters. Annular glue overflow is strictly controlled within ≤1.5×1.5 mm without offset;
3. Optimized glue morphology (no stringing and no sagging);
4. Full-process data recording to facilitate abnormality tracing and fine-tuning of processes.

Q5: How long does production/product line changeover take?

A: Repeated mechanical teaching and customized fixtures are not required. Operators can call up corresponding vision templates or create new calibrations in the software, and the equipment completes line change preparation within minutes. The line change speed is improved by approximately 80% compared with traditional methods.

Q6: What is the output and cycle time of a single machine?

A: Vision recognition only takes 0.1–0.3 seconds. The dispensing head operates synchronously with the assembly line belt without line stoppage. The equipment delivers over 20% higher efficiency than competing products on the market, can steadily replace 1–2 skilled workers, and supports 24-hour continuous operation.

Q7: How to connect to existing production lines? Will renovation lead to prolonged production shutdown?

A: Embedded models (GP Series) can be directly mounted across existing conveyors and utilize on-site belt transportation without conveyor replacement or modification. The deployment process follows the sequence: "Workstation Confirmation → Equipment Positioning → Signal Connection → Parameter Calibration". Commissioning from unpacking to mass production usually takes only several hours, realizing "replacement at original position and seamless upgrade". Cabinet-type models (RX Series) are also available as independent workstations.

Q8: Is MES/data traceability supported?

A: Yes. The self-developed software can achieve in-depth connection with MES/ERP systems. It monitors and records parameters such as glue volume, moving speed, position and temperature in real time, providing complete data support for quality tracing and process optimization.

Q9: Are there hidden costs for the equipment software? How is after-sales service guaranteed?

A: Songqi adopts fully self-developed vision software and self-developed YMC-08-01 control cards with no annual licensing fees. After-sales services include dedicated on-site service (1-on-1 / 1-on-2), 7×24 multi-channel technical support (15-minute response mechanism), lifelong maintenance and regular follow-up visits. Sufficient spare parts are kept in stock to enable direct troubleshooting without complicated coordination among multiple parties.

Q10: Can sample testing be carried out for verification before purchase?

A: Yes. Songqi provides free process scheme design and free sample testing. After customers send samples, we conduct dispensing tests on actual equipment and issue a complete test report for reference before customers make decisions.

• Company: Dongguan Sukiauto Intelligent Technology Co., Ltd. (SUKIAUTO)
• Address: Unit 4, 9th Floor, Building 2, (Dongguan Daotong Intelligent Valley), approximately 160 meters northeast of the square of Dongcheng Foreign Economic Industrial Park, Huijing Road, Dongcheng Subdistrict, Dongguan City, Guangdong Province
• Tel: 13761860627
• Email: service@sukiauto.cn
• Services: Free sample testing · One-on-one dedicated after-sales service · 7×24-hour response · Lifelong maintenance

The equipment parameters, efficiency and yield data listed in this solution are based on internal tests and sales statistics of Sukiauto Intelligent. Industry process standards are derived from publicly available production specifications, patents and literature of the electroacoustic industry. The final model selection and process parameters shall be subject to actual sample verification and the process plan confirmed by both parties.