Fall voltages are important characteristics that researchers and engineers seek to understand for applications including power source design, and welding output prediction (e.g. heat input and bead width). Currently, there is limited understanding of how the voltage set at the machine is isolated to distinct fall voltages that make up the total voltage loss (VTot) in the Gas Metal Arc Welding (GMAW) system. This has led to a costly trial and error approach to achieve desired settings for a successful weld. In a typical welding system, the anode and cathode fall voltages (Van and Vcath) are the expected main contributors for heat input. However, all fall voltages including the contact tip (VCT), electrode extension (VEE), arc column (VAC), and welding leads (VWL) must be accounted for to understand heat input and the required voltage set at the machine.
Previous work has posed important questions around the definition of arc length (LAC), which is required to determine VAC. Here, a definition for point of arc attachment has been considered as the point where the electrode meets the boundary between the shielding gas and metal vapor dominant regions. Understanding the relationship between voltage and LAC is paramount to controlling the GMAW process. The self-regulating arc in GMAW allows for the quantification of effects to LAC from changing variables such as VTot, welding current (Iw), and contact tip to work piece distance (LCTWD).
In this study, a Phantom v210 high speed camera was synchronized with an external data acquisition (NI USB 6351) system to collect synchronized video, current, and voltage data for a series of welds using 1.2 mm ER4043 electrode on 6061 base material. The high-speed videography was analyzed to obtain time weighted LAC measurements. These measurements were taken manually using Phantom PCC software, which served as a baseline for comparing LAC measurements using a Convolutional Neural Network algorithm (UNET).
The arc column potential (V’AC) was found to be 0.62 ± 0.05 V mm-1 from an empirical relationship (VM =0.62LAC+18.5) considering arc length (LAC) and a measured voltage (VM) from two weld sets at different wire feed speeds (UC). The sum of the anode and cathode (VF) had an average of 18.3 ± 0.4 V and was approximately independent of current. Fall voltage at the electrode extension, contact tip, and welding leads were found to be minor contributors with equal to or less than 1% contribution to the overall voltage loss. Also, the arc length was found to be independent of LCTWD and had a consistent decrease of 0.07 mm A-1 as current was increased.
This work provides a practical approach to determine expected arc lengths and voltage contributions for each fall voltage constituent as results are compared to corresponding values using desired settings from the Lincoln Procedure Handbook. Ongoing work using the UNET model is to form part of the final presentation and will reveal results for experiments with different material grades and consumable diameters at varying Iw ranges to verify the impact of Iw on V’AC and LAC.