ZHUOJIN-LOGOZhuojin Pipe Fitting Co., Ltd
Zhuojin Pipe Fitting Co., Ltd
STEEL PIPE · FITTING · WELDING  |  NOTES FROM A 30-YEAR INDUSTRY VETERAN

Arc Tracking Sensor for Gas Metal Arc Welding

Introduction and Literature Context

This paper by Ai Yongyi and Yao Li from Chongqing University, published in the Journal of Welding in 1991, presents an arc tracking sensor for gas metal arc welding (GMAW) that utilizes a low-speed arc rotation mechanism to improve the surface appearance of fillet welds. The authors developed a mechanical device that rotates the welding torch in an elliptical path, and they established a theoretical relationship between the welding current variation and the deviation of the arc rotation center from the groove center. By extracting the signal in the frequency domain using band-pass filtering and cross-correlation noise reduction, they constructed an analog circuit-based arc sensing and tracking system. The system was tested in CO2 welding and achieved a tracking error of ±0.5 mm. This is a classic example of arc sensing technology, which has been widely used in automated welding systems for decades, and it remains relevant today as a cost-effective alternative to more sophisticated laser or vision-based seam tracking systems.

Key Technical Findings and Interpretation

The core innovation of this work is the use of a low-speed elliptical arc rotation to generate a measurable current signal that is proportional to the deviation between the arc rotation center and the groove center. When the torch rotates in an elliptical path, the arc length varies periodically, which causes the welding current to vary correspondingly. If the rotation center is aligned with the groove center, the current variation is symmetric and the signal has no component at the rotation frequency. If the rotation center is offset, the current variation becomes asymmetric, and the signal at the rotation frequency carries information about the magnitude and direction of the deviation. The authors used band-pass filtering to isolate the signal at the rotation frequency and cross-correlation to reduce noise, which is a robust approach for extracting a weak signal from a noisy welding environment. The theoretical model accounts for the dynamic changes in arc length and wire extension during rotation, which is important for accurately relating the current signal to the torch position. The tracking error of ±0.5 mm is impressive for a CO2 welding application, and it demonstrates the practical viability of this approach.

Practical Implications for Automated Welding

From a practical standpoint, this work addresses a fundamental challenge in automated welding: how to maintain the welding torch accurately aligned with the joint throughout the weld. In many applications, particularly in the fabrication of large structures such as ship hulls, pressure vessels, and steel bridges, the joint position may vary due to fit-up tolerances, thermal distortion, and fixturing errors. An arc tracking sensor provides a means of correcting for these variations in real time, without the need for expensive vision systems or complex mechanical fixtures. The low-speed arc rotation mechanism described in this paper is relatively simple and robust, making it suitable for harsh industrial environments. The use of frequency-domain signal processing is particularly effective for rejecting noise from the welding arc, which is a major source of interference in arc sensing. The tracking error of ±0.5 mm is sufficient for most fillet weld applications, where the acceptable tolerance is typically in the range of 1 to 2 mm. The system is also applicable to other welding processes, such as pulsed MIG or flux-cored arc welding, provided that the signal processing is appropriately tuned.

Process Window and Signal Processing Considerations

The performance of the arc tracking sensor depends on several factors, including the rotation speed, the rotation radius, the welding current level, and the signal processing parameters. The rotation speed must be low enough to avoid disturbing the molten pool but high enough to generate a measurable signal at the rotation frequency. The rotation radius determines the amplitude of the arc length variation and hence the amplitude of the current signal; a larger radius produces a stronger signal but may also cause the arc to wander outside the groove, which can lead to lack of fusion. The welding current level affects the sensitivity of the current signal to arc length changes; higher currents generally produce larger current variations for a given arc length change, but they also increase the risk of burn-through on thin sections. The band-pass filter must be centered on the rotation frequency with a bandwidth that is narrow enough to reject noise but wide enough to accommodate variations in the rotation speed. The cross-correlation technique is effective for extracting the signal from noise, but it introduces a time delay that must be accounted for in the control system. In practice, the welding engineer must tune these parameters for each specific application, using the theoretical model as a guide and verifying the performance through welding trials.

Integration with Engineering Practice and Quality Control

In automated welding systems, the arc tracking sensor is typically used in conjunction with a welding robot or a mechanized welding carriage. The sensor provides real-time feedback on the torch position, which is used to correct the torch path and maintain the arc in the groove center. This is particularly important for long welds, where the cumulative effect of thermal distortion and fit-up errors can cause significant deviation. The tracking error of ±0.5 mm achieved in this study is well within the acceptable range for most fillet weld applications, and it ensures consistent weld quality and bead appearance. From a quality control perspective, the use of an arc tracking sensor reduces the need for manual inspection and rework, as the weld is consistently placed in the correct position. It also reduces the risk of defects such as lack of fusion, undercut, and excessive reinforcement, which are often caused by torch misalignment. In my experience, the implementation of arc tracking sensors in automated welding systems has resulted in significant improvements in productivity and quality, particularly in the fabrication of large structures where manual welding would be impractical or inconsistent.

Key Questions and Reflections

Several questions arise from this study that are worth considering. The authors tested the system in CO2 welding, which is a relatively forgiving process; the performance in other processes such as pulsed MIG or flux-cored arc welding may differ. The tracking error of ±0.5 mm was measured in a laboratory setting with controlled conditions; in production, the error may be larger due to variations in fit-up, welding position, and environmental factors. The low-speed arc rotation mechanism adds complexity to the welding torch, which may increase maintenance requirements and reduce accessibility in confined spaces. The signal processing is based on analog circuits, which are reliable but may be less flexible than digital signal processing; modern systems would likely use digital signal processors or field-programmable gate arrays to implement the filtering and cross-correlation algorithms. The theoretical model assumes a specific relationship between arc length and current, which may not hold exactly in all conditions; the model should be validated for each application. Finally, the system does not provide information about the joint geometry or the weld pool shape, which limits its ability to correct for other types of misalignment.

Study Insights and Implications

This study provides a valuable contribution to the field of arc sensing technology, and it demonstrates a practical approach to seam tracking that is both effective and cost-efficient. The use of low-speed elliptical arc rotation to generate a measurable signal is an elegant solution to the problem of detecting torch misalignment, and the frequency-domain signal processing is robust against the noise that is inherent in welding. The tracking error of ±0.5 mm is a significant achievement, and it validates the theoretical model and the practical implementation. For the welding industry, this work highlights the importance of arc sensing as a means of improving weld quality and productivity in automated welding systems. The principles described in this paper are still relevant today, and they form the basis for many modern arc tracking systems. The study also underscores the broader principle that simple, robust solutions are often more effective in industrial environments than complex, high-tech alternatives.

Reference Value and Outlook

This paper is a valuable reference for welding engineers and researchers working on automated welding systems, particularly those interested in arc sensing and seam tracking. The experimental methodology is sound, and the results are clearly presented with supporting diagrams and data. The paper's contribution lies in its systematic development of a practical arc tracking system, from theoretical modeling through signal processing to experimental validation. The tracking error of ±0.5 mm is a benchmark that demonstrates the viability of this approach, and it provides a useful reference for evaluating the performance of other seam tracking systems. Looking forward, the principles described in this paper can be extended to more advanced applications, such as adaptive welding where the arc tracking signal is used to adjust not only the torch position but also the welding parameters. The combination of arc sensing with vision or laser-based sensing could provide a more comprehensive solution for seam tracking in complex joint geometries. The findings also have implications for the development of welding robots, where arc tracking can be used to compensate for variations in the workpiece position and orientation, thereby reducing the need for expensive fixtures and improving the flexibility of the welding cell.