Rotating Arc Seam Tracking Signal Processing for Wire-Feeding TIG Welding

Literature Overview

The paper by Jia Jianping, Ai Shixiong, Wang Xin, and Ye Yanhui, published in Hot Working Technology in 2019 (Vol. 48, No. 19, pp. 124-127), presents a seam tracking system based on rotating arc TIG welding with wire feeding. This work was supported by the National High-Tech Research and Development Program of China (863 Program, Grant No. 2013AA041003). The research was conducted at the Jiangxi Provincial Key Laboratory of Robotics and Welding Automation, Nanchang University.

Core Technical Findings

The researchers designed a rotating arc TIG welding seam tracking system that employs a rotating arc sensor to scan the weld seam. By collecting and processing welding voltage variations, the system achieves real-time seam tracking. The key contributions include:

Development of a rotating arc sensor system specifically adapted for TIG welding applications.
Analysis of actual welding voltage signals to characterize the signal structure and noise characteristics.
Proposal of a combined digital filter and wavelet filter method for signal processing.
Establishment of a model relating rotating arc height to voltage output.
Demonstration that the designed filtering algorithm is practical, reliable, and effectively identifies arc height variations during rotating arc wire-feeding TIG welding.

Signal Processing Architecture

Processing Stage	Method	Purpose
Signal acquisition	Voltage sampling from welding circuit	Capture raw arc voltage signal
Pre-filtering	Digital filter (e.g., low-pass, notch)	Remove high-frequency noise and power supply ripple
Feature extraction	Wavelet transform	Isolate seam-related features from noise
Signal modeling	Arc height-voltage relationship	Convert processed signal to positional information
Tracking output	Error signal generation	Provide feedback to torch positioning system

Technical Interpretation

Seam tracking is a critical technology for automated welding, particularly for long weld seams where groove alignment errors accumulate over distance. The rotating arc sensor principle exploits the fact that when an arc is rotated (either by rotating the torch or by using a rotating electrode), the arc length varies as the arc scans across the groove geometry. Over a groove, the arc encounters different surface geometries at different rotational angles, producing characteristic voltage variations that encode positional information about the torch relative to the seam.

Rotating Arc Sensor Operating Principle

The rotating arc sensor operates on the following physical principles:

Arc length variation: As the arc rotates, the distance between the electrode tip and the workpiece surface varies depending on the groove geometry. Over a V-groove, the arc is shorter at the groove bottom and longer at the groove edges.
Voltage-arc length relationship: The arc voltage is approximately proportional to the arc length (V = V₀ + k·L, where V₀ is the arc drop at the electrodes and k is the voltage gradient per unit length). Therefore, voltage variations directly reflect arc length changes.
Signal symmetry: For a centered torch position, the voltage signal exhibits symmetry about the rotation axis. Asymmetry indicates torch misalignment, and the direction and magnitude of asymmetry provide the error signal for correction.
Wire feeding interaction: In wire-feeding TIG welding, the deposited wire creates a molten pool and weld bead that affects the arc scanning pattern. The rotating arc must account for the presence of the wire and the evolving weld geometry.

Signal Processing Challenges

The raw voltage signal from rotating arc TIG welding contains multiple components that must be separated:

Signal Component	Frequency Range	Source	Treatment
Power supply ripple	50/60 Hz and harmonics	AC power supply	Notch filter
Arc noise	1-100 kHz	Arc instability	Low-pass filter
Seam signal	0.1-10 Hz	Torch rotation	Wavelet extraction
Process disturbances	Variable	Spatter, wire contact	Adaptive filtering

The combination of digital filtering (for removing known frequency components) and wavelet transform (for extracting multi-scale features) is a robust approach that addresses both stationary and non-stationary noise components.

Engineering Practice Implications

This technology has significant relevance for automated pipeline welding, particularly for:

Long pipe butt welds: In pipeline construction, butt welds can extend over several meters in circumference. Manual seam tracking is impossible, and optical sensors may be affected by arc radiation. Rotating arc sensors are inherently immune to arc light interference.
Underwater welding: In underwater pipeline repair, optical sensors are impractical. Rotating arc sensors can operate effectively in submerged conditions.
Pipe fitting fabrication: For automated welding of elbows, tees, and reducers, where the seam geometry varies continuously, adaptive seam tracking is essential.

Integration with Welding Control Systems

For practical implementation, the seam tracking system must be integrated with the overall welding control architecture:

Control Loop	Function	Response Time
Seam tracking	Torch position correction	10-100 ms
Arc height control	Arc length maintenance	5-50 ms
Current/voltage regulation	Power source control	1-10 ms
Travel speed control	Weld bead geometry	100-1000 ms

The seam tracking loop operates at a slower timescale than arc control but faster than travel speed control. The separation of these control loops is essential to avoid instability.

Key Questions and Reflections

Several aspects of this research warrant further consideration:

Signal robustness in wire-feeding mode: The presence of a continuously fed wire introduces additional signal disturbances, including wire contact noise and molten pool oscillation. The effectiveness of the proposed filtering algorithm in the presence of these disturbances should be validated under production conditions.
Groove geometry adaptability: The rotating arc signal characteristics vary significantly with groove geometry (V-groove, X-groove, square butt). The system's ability to adapt to different groove profiles without recalibration is critical for multi-application deployment.
Multi-pass tracking: In multi-pass welding, the seam tracking must account for the geometry of previously deposited weld beads. The rotating arc signal characteristics change with each pass, requiring adaptive signal processing.
Integration with robotic systems: The tracking system must interface with the robotic welding cell's control system. Communication latency and synchronization between the tracking system and the robot controller are critical for stable operation.
Validation methodology: The study reports on signal processing effectiveness but does not provide quantitative data on tracking accuracy (e.g., tracking error in millimeters) under various misalignment conditions. Such data would be essential for engineering design.

Study Insights and Implications

This research addresses a fundamental challenge in automated welding: reliable seam tracking under harsh arc conditions. The rotating arc sensor concept is attractive because it uses the welding arc itself as the sensing medium, eliminating the need for external sensors that are susceptible to arc radiation, spatter, and smoke. The combination of digital and wavelet filtering demonstrates a practical approach to extracting useful signal from noisy measurements. For pipeline and fitting fabrication engineers, this technology offers a path to fully automated, high-quality welding with reduced reliance on skilled operators and improved consistency. The key to successful implementation lies in thorough validation under production conditions, including characterization of tracking accuracy, robustness to disturbances, and compatibility with different welding configurations.