REFERENCES

1. Rajasekaran, S.; Kulkarni, S.D.; Mallya, U.D.; Chaturvedi, R. C. 1998. Droplet detachment and plate fusion characteristics in pulsed current gas metal arc welding. Welding Journal 77 (6): 254-s to 269-s.

2. Rajasekaran, S. 1999. Weld bead characteristics in pulsed GMA welding of Al-Mg alloys. Welding Journal 78 (12): 397-s to 407-s.

3. Rajasekaran, S. 2000. Weld surface undulation characteristics in the pulsed GMA welding process. The Ninth International Conference on Computer Technology in Welding, September 28-30,1999, National Institute of Standards and Technology, United States Department of Commerce, Special Publication No. 949, May 2000, 349-360.

4. Rajasekaran, S.; Kulkarni, S.D.; Mallya, U.D.; Chaturvedi, R. C. 1995. Molten droplet detachment characteristics in steady and pulsed current GMA welding Al-Mg alloys. Proceedings of the Sixth International Conference on Aluminum Weldments (INALCO '95). April 3-5, 1995, Cleveland, Ohio, U.S.A. American Welding Society, Miami, Fla, 207-224.

5. Rajasekaran, S.; Kulkarni, S.D.; Mallya, U.D.; Chaturvedi, R. C. 1994. Droplet detachment and plate fusion characteristics in pulsed current gas metal arc welding. Proceedings of the Advanced Joining Technologies for New materials II, March 2-4, 1994, The Cocoa Beach Hilton, Florida, U.S.A, American Welding Society, Miami, Fla, 172-187.

DROPLET OSCILLATION AND WELD POOL IMAGING

USING COMPUTER-CONTROLLED COMPOSITE PULSE CURRENT

B. Zheng*

ABSTRACT

This paper presents a unique solution to real-time monitoring of both droplet detachment and weld pool during a pulsed gas metal arc welding process: a composite pulse current consisting of a square wave form followed by a sine wave form was designed. The instant for initiating a constant base current adaptively started at the instant droplet detachment was sensed. During the period of the base current, a flag signal was generated to trigger the imaging of a weld pool. The approach makes droplet detachment and image acquisition be proceeded without outside intervention. This provides the possibility for real time quantifiable monitoring and control for both droplet transfer and weld pool penetration.

INTRODUCTION

The results of the previous researches showed that metal transfer and weld pool penetration both have a large influence on the generation of the defects such as undercut, burn through, insufficient melting, spatter, gas pore, and even weld cracks during a gas metal arc welding (GMAW) process {Ref. 1-3). Hence, in-process monitoring and control for metal transfer and weld pool penetration are crucial in order to minimize the cost of post-weld inspection and repair. Currently, a droplet/arc oriented control strategy is often used even though it is inefficient in some applications. Since maintaining a consistent weld pool has a dramatically direct impact on weld quality, a weld pool oriented control strategy has been largely demanded. In a GMAW process with argon-rich shielding, the different metal transfer modes of shortcircuiting, globular, and spray can be observed in sequence as the welding current is increased (Ref. 2-5) when steel electrode wire is used. The critical welding currents at which the metal transfer mode changes are defined as transition currents, one of which is the spray transition current at which globular transfer becomes spray transfer (Ref. 2-3). The level of the spray transition current mainly depends on many factors such as wire diameter and composition of shielding gas. Quality welds can be achieved using the projected spray transfer mode of one droplet per pulse (ODPP). To ease the flexible selection of ODPP and simplify the adjustment of process parameters, pulsed current welding (GMAW-P) is a preferred process (Ref. 3 and 6-9). However, most of GMAW-P processes with ODPP mode use open loop control that regulates only the metal transfer mode and pay little attention to the weld pool (even though some researches have been pursued on sensing and closed-loop control of droplet transfer) (Ref. 4-21). The partial reason for this is the difficulty in measuring characteristic signals and control compatibility of droplet detachment with weld pool penetration.

Besides through-arc sensing of a weld pool (Ref. 22-28), machine vision systems have been used to sense the weld pool and control the full penetration during a PAW, or GTAW, or GMAW process (Ref. 29-45). These vision systems can be characterized into band-pass arc light filtering,

Edison Welding Institute, 1250 Arthur E. Adams Drive, Columbus, OH 43221

coaxial viewing, high-shutter-speed camera synchronizing with laser-assisted illumination, and infrared thermography. The instantaneous decrease of the welding current has also been used to weaken the intensity of the arc light to capture a clear weld pool image in GMAW-P (Ref. 44). However, the interference resulting from metal transfer while imaging a weld pool was not solved because of the inaccurate open-loop control that was used to determine the instant of the droplet detachment. Also, the camera had some limitations, such as the shutter speed and the luminance, which could not meet the needs of imaging and controlling a GMAW weld pool. The objective of this paper is to explore a unique solution to the compatibility of sensing and control of a droplet detachment and imaging a weld pool during a GMAW-P process, which will set a basis for real time control of a fully penetrated weld pool (Ref. 44-46).

EXPERIMENT PROCEDURE

A PC-controlled, inverter welding power source and a wire feeder are used in this experiment. A high frame rate digital camera with adjustable frame rate and a frame grabber is used for monitoring the droplet with the back illuminating light from a He-Ne laser. Current and voltage sensors are integrated with a data acquisition system. A positioning system driven by stepper motors is also interfaced with the system. A CCD camera with high shutter speed is attached to the front of the torch to synchronously image the weld pool.

To explore a droplet detachment method associated with the imaging of a weld pool, a controlled signal generator was designed, the current wave form of which is shown in Fig. 1. The period of pulse consists of a square wave form and a sine wave form with a pre-selected frequency. The peak current for a square pulse is experimentally set at around 160 A that is below the spray transition current, the function of which is to initiate the droplet growth and minimize the energy input to a workpiece. The function of a sine wave pulse following the square pulse is to activate an externally forced oscillation of a droplet before it detaches from the tip of an electrode wire. The sine wave pulse with its peak level (260 A) being higher than transition current (220-240 A, in this case) adaptively terminates at the instant of the droplet detachment monitored

by the arc voltage sensor. The imaging of a weld pool is initiated during the base current (40 A) period after a time delay that is experimentally set. The frequency of the sine wave pulse can be varied from 0 to 500 Hz. The total cycle frequency is less than 30 Hz due to the limitation of the

tsine

Pulse peak current = Ips1
Sine wave peak current =
= Ips2

Base current = b

Transition current = la

Pulse period = tps1 + tps2

Ips2

y

Base period = tb

la

Ips1

tps1

Time

tb

Vs-Synchronous pulse

Time VT-Camera trigger pulse

Time

Fig. 1 Composite Pulse Current (Ic), Synchronous Pulse (Vs),

and Trigger Pulse (VT)

recording frequency by the high shutter speed camera. A droplet cannot detach during the period of a square pulse because of the low electromagnetic pinch-off force resulted from the acting current. The sine wave current pulse not only promotes the continuous growth of a droplet as it oscillates but also assists in droplet detachment and further decreases the energy input to a workpiece. At the end of the base period for maintaining an arc, a new cycle repeats the same process.

Bead-on-plate welds were made in a flat position on 6.0-mm thick mild steel plates. The electrode wire is 1.2-mm diameter steel (AWS standard ER 70S) and the shielding gas is argonrich (85%Ar+15%CO2).

RESULTS AND DISCUSSION

According to the static force balance theory (Ref. 10-12), the forces promoting droplet detachment in a flat welding position are the electromagnetic force resulting from the welding current, gravity and the plasma gas drag force. The force resisting the droplet detachment is mainly the surface tension. The other resistant forces are anode spot pressure and metallic vapor force acting on the anode spot. When the forces promoting droplet detachment are equal to or greater than the resisting forces, the droplet will detach from the tip of an electrode wire. The electromagnetic force will be a promoting force for droplet detachment when the lower half of the droplet is covered with the spread arc anode spots and the welding current is above the transition current. Otherwise, it will become a resisting force for droplet detachment. The magnitude of the electromagnetic force depends on the level of welding current (Ref. 11-13). The control strategies for droplet detachment may be categorized into two types: electrical and mechanical (Ref. 6-19).

Since electrically controlling the electromagnetic force inside the droplet and arc is reliable and fast, most investigations on droplet detachment control focus on this approach. In an ODPP mode, the current during the pulse period of a cycle forms a droplet with a certain size, while the current during the base period of a cycle prevents the arc from extinguishing. The melting rate of the wire (the wire feed speed) is directly proportional to the pulse frequency: the higher the frequency, the higher both the average welding current and the melting rate. The pulse period and level also have the positive effect on the melting rate when the pulse frequency is constant. Therefore, the mass and heat transferred to the workpiece may be flexibly regulated by the adjustment of average welding

current.

Both the static force balance theory and the

pinch instability theory reveal that the ODPP approach, characterized by approximately the same