Evolution of power sources in arc welding - transition from passive to active role

TWI Bulletin, April 1987

By Chris Needham

Chris Needham, BSc(Eng), is Chief Control Engineer at Abington.

Arc welding was first recognised as an industrial process 100 years ago. Since then many variations have been developed, including gas shielded and constricted plasma arcs. Today there are at least two well established arc welding processes where arc behaviour depends inseparably on the power supply characteristics. These are short circuiting and pulse current MIG/MAG welding, to which may be added synergic control, which uses closed loop feedback between welding system and power supply.

The above and other processes could not be operated with conventional drooping characteristic power supplies (or even with more recently developed simple flat characteristic power supplies) without further modification or refinement. Particularly in the last 25 years the advent of power semi-conductor devices has permitted design of power supplies with flexible characteristics to suit the requirements of the welding arc. In spite of many developments that have already taken place, the author believes that there is further potential for the power supply to take a more-active role in arc welding.

In describing evolution from a passive to an active role, increasing interaction is as follows:

1. Nominally no interaction - power source only supplies voltage.
2. Moderate static response - output characteristic with preferred open circuit voltage and short circuit current;
3. Dynamic output characteristic - transient response down to 0.1 sec for intermittent short circuiting;
4. Strong static response - self-adjusting (MIG) system. Dynamic response down to 0.01 sec for regular short circuiting;
5. Molten pool control - by current pulse down to less than 1.0 sec, as in pulsed TIG welding;
6. Metal transfer control - by pulse current down to less than 0.01 sec, as in pulsed MIG welding;
7. Generalised dynamic interaction - as in synergic pulse current MIG operation.

History

The main types of power supply, including some recent developments, are illustrated in Fig.1-5.

Fig.1. DC power sources for MMA welding:

a) Storage accumulator and variable series resistor;

b) Self-excited shunt wound generator with regulating resistor;

c) Separately excited generator with bucking winding.

Fig.2. AC power sources for MMA welding:

a) Multi-operator variation of above;

b) Self-excited shunt wound generator with regulating resistor

c) Transformer with variable magnetic flux linkage.

Fig.3. Self-adjusting current sources for welding with constant electrode feed:

a) Low voltage transformer and minimal inductance for SA welding;

b) Experimental circuitry for MIG process evaluation;

c) Flat characteristic transformer-rectifier.

Fig.4. Semiconductor controlled current supplies for arc welding:

a) Transistor series linear regulator;

b) Switching on-off supply;

c) High frequency inverter.

Fig.5. Squarewave AC power sources;

a) Operating at line frequency (50/60Hz);

b) Low frequency inverter.

Briefly, power sources have evolved from standard storage accumulators, generators, and transformers with resistors and reactors for controlling current, to modern designs depending on power thyristors and transistors.

The first arcs were run from storage accumulators, with current control via series variable resistors (Fig.1). Accumulators were soon replaced by motor driven self-excited shunt wound generators, still with a regulating resistor. These in turn developed into the separately excited generator with a series bucking winding to provide the required drooping output characteristic, eliminating the regulating resistor and its power loss.

A similar development occurred with AC, where the power source was originally a simple transformer with a separate regulating inductor (Fig.2). Also multi-operator equipment evolved using a group of regulators each serving a separate arc on a common structure. Later, loosely coupled transformers with variable magnetic flux linkage between primary and secondary were developed to give the required drooping output characteristic.

With some of these simple AC and DC power supplies (Fig.1a, 2a) there is no interaction between the primary source of power, which is solely a voltage source, and arc. There is, however, some internal feedback in other types of AC and DC power sources, where output current directly reduces output voltage, giving the desired drooping characteristic (Fig.6).

Fig.6. Steeply drooping (constant current) power source characteristics with change in arc length L (where L₃ >L₂>L₁) for:

a) Simple DC resistive regulator;

b) Simple AC regulator.

The first major change, leading to significantly greater involvement of the power supply in the welding process, came with the introduction of the self-adjusting arc system, in which the electrode wire is fed at a preset speed. Here there is no direct control of current by the power supply, which largely determines the operating arc length, but the operating current is set by wire feed rate. This system was first introduced in a variation of submerged-arc welding, using relatively low transformer voltages and minimum series inductance (Fig.3). However, this technique has found greatest use in MIG welding arcs. Initially, for experimental evaluation, the power supply comprised an accumulator with no regulating resistor (Fig.3) but later for industrial use this was replaced by a flat characteristic transformer-rectifier, again with no current regulation (Fig.3c).

Another major step forward, giving further power source flexibility, was the introduction of direct control of welding current via transistors, either as a series linear regulator or a switching (on/off) supply (Fig.4). The former is more versatile but requires a large number of transistors operating in parallel compared with the switching arrangement where the number of transistors is reduced to about one tenth. For the latter, mean output is varied by changing the mark/space ratio of the high frequency switching typically 3-15kHz. (Recently, commercial power sources have become available with switching frequencies of tens of kilohertz.) Another modern development which combines the advantages of both systems is the high frequency inverter (Fig.4). This is operated directly from the rectified mains supply, and probably represents the best combination of flexibility against cost at present.

Finally, for AC arcs (Fig.5) there are two developments of interest, both giving a more square waveform than the conventional sinusoidal wave. One system (Fig.5a) operates only at line frequency but provides relatively fast transitions from one polarity to the other. The output is AC which in appearance approximates to a Squarewave but with a sinusoidal top. The other system comprises a low frequency inverter (Fig.5b) giving a truly Squarewave output from a smooth DC supply. In practice commercial power sources have been designed to operate at line frequency, but with additional functions such as controlling the relative current and duration of the positive and negative periods.

Process characteristics

Operation with constant current supplies

Some of the static and dynamic interactions between power supply and welding arc are shown in Fig.6-9. The static operating point for all is given by the intersection of the volt-ampere output characteristic of the power supply with the corresponding volt-ampere characteristic of the load, i.e. the welding arc. Initially the function of the power supply was solely to provide current, as governed by a series regulator (Fig.1,2), nominally independent of arc length, L. The arc characteristic is nominally flat or slightly rising (Fig.6) over a range of current, while the power source characteristic is drooping, giving well defined intersection points according to the operating arc length. The latter is, in manual metal arc welding, controlled by the operator (and in submerged-arc welding by voltage controlled feedback to the wire feed system), while current is determined mainly by the power supply.

Fig.7. Typical nominally flat (constant voltage) power source characteristics.

Fig.8. Self-adjusting characteristics with change in wire feed rate W (Where W₂>W₁):

a) Linear throughout;

b) Less flat in operating zone.

Fig.9. Power source and arc interactions for dip transfer operation:

a) Voltage and current waveforms with respect to time;

b) Stylised cyclogram (plot of voltage against current).

With a simple resistive regulator the power source output characteristic is a straight line from open circuit voltage (OCV) to short circuit current (SCC) (Fig.6). However, with AC the characteristic is nominally a quadrant of an ellipse between open circuit and short circuit conditions (Fig.6). The reverse field generator and leakage reactance transformer (Fig.1c, 2c) also give similar, more steeply drooping characteristics such that, with change in arc length, there is a lesser change in operating current. Moreover, because of the limited short circuit current (not much greater than working current), the arc is normally ignited by touch and retract, as otherwise the short circuit current is insufficient to fuse the electrode.

Operation with constant voltage supplies

With a flatter output characteristic from the power supply (Fig.7) there is a much greater change in current with arc length or arc voltage, which reduces operating stability. In the limit (rising power source characteristic virtually coincident with the arc characteristic at a given length) it would be difficult, if not impossible, to control the operating current, since the slightest change in arc length leads to major changes in the nominal point of intersection. Nevertheless a stable method of working is obtained where, instead of endeavouring to maintain a constant arc length directly, the electrode feed is maintained constant, as in the self-adjusting arc system (e.g. semi-automatic MIG welding). With a constant electrode feed, the load characteristic is approximately vertical (Fig.8). Therefore the intersection with a flat characteristic or nominally constant voltage power supply is well defined. Thus, in contrast to the drooping characteristic power supply, current is determined principally by wire feed rate and the operating arc length by power source output level. As noted earlier this constant feed technique was first developed for flux shielded submerged-arc welding on AC, but later became well established with nominally constant potential DC power supplies for gas shielded MIG arcs (using relatively fine wire, and hence high feed rates compared with MMA welding).

However, even with the self-adjusting arc there may be instability because of the flat power source characteristic. If there are transient changes in arc length caused by disturbances in the molten pool, these are accentuated by corresponding changes in operating current. To reduce this effect, particularly at high current, the power source characteristic is made less flat in the operating zone (Fig.8). At high current the load characteristic of the process is also less vertical, as in the meso-spray regime.

Short circuit arc operation

Flat characteristic power supplies represent the first major step in the evolution of the power supply from being merely a passive source of current for the arc. The next development is exemplified by the short circuit MIG arc where, with a flat source, it was necessary to introduce inductance (Fig.3) to control rate of rise and fall of current.

Here, in addition to the long term (static) effect of electrode feed rate on average current, there is also a short term dynamic interaction whereby the current fluctuates about the mean during successive short circuit and arcing stages (Fig.9). The operating ranges of arc and short circuit voltages and currents in general do not intersect the power source output characteristic but lie on either side of it. Also the occurrence of the short circuits is substantially random, even though an average frequency is sometimes quoted for given working conditions. Nevertheless, the long term equilibrium current is closely related to the ratio of average arc time to average short circuit time.

Pulsed current TIG

A further step in evolution of a more active role is application of current pulsing to TIG and MIG welding (Fig.10-12). Modulation of the power source output to enforce current changes in the arc produces desired modifications of the welding process which are generally associated with the higher current. The actual pulse current waveform, often considered to be square, is subject to limitations in rate of rise and ripple (Fig.10). In TIG welding, pulse current waveshape was distorted by the earlier transductor controlled power supplies, but waveforms with modern inverter equipment are more rectangular. The current pulse is of relatively long duration, upwards of 0.1 sec and, for thicker materials with larger molten pools, in excess of 1.0 sec. The current pulse produces rapid penetration which is prevented from becoming excessive by reverting to a low or background current which allows the pool to solidify partially or completely between pulses. This technique is used in TIG and plasma welding, and in principle can also be extended to MIG welding (see discussion).

Fig.10. Pulsed current TIG welding waveforms:

a) Idealised squarewave;

b) Distortion typical of transductor welding set.

Fig.11. Pulsed current MIG welding waveforms from:

a) Sinusoidal line supply;

b) Linear transistor regulator;

c) Secondary switched or inverter welding set.

Fig.12. Synergic operation for 50% increase in wire feed rate:

a) Constant added pulse;

b) Constant total pulse.

Pulsed current MIG

On a shorter time scale, typically milliseconds, current pulses are also used to control metal transfer in MIG welding (Fig.11). Initially, pulses were derived from the sinusoidal line supply but later pulses of nominally square waveform were produced by a linear transistor regulator (Fig.4). The latter equipment, though versatile, uses many transistors in parallel. Commercial equipment today uses secondary switched or inverter power supplies (Fig.4) which produce reasonable pulse waveforms albeit with a degree of ripple at the switching frequency (Fig.11c). This pulse current technique is well established for operating MIG arcs at average currents below that at which spray transfer normally occurs. However, pulse controlled transfer is also beneficial in the spray range, particularly where otherwise the wire tip tapers with associated streaming transfer and directional instabilities.

Synergic MIG arcs

In the last few years synergic operation has become commercially available, where the pulse parameters (for controlled metal transfer in MIG welding) are varied according to the average operating condition. Thus for higher wire feed speeds, corresponding to higher average currents, pulse repeat frequency is increased substantially in the same proportion, together with adjustment of the background current (or pulse duration, etc) to maintain equilibrium between average current and wire feed rate.

There are two types of synergic control (Fig.12) in which the background current is altered to suit the change in average operating conditions. In one the pulse is treated as an addition to the background and increased in frequency proportional to wire feed rate, together with proportional increase in background current. This has the advantage of a wide operating range, but the disadvantage that the total pulse amplitude is increasing and hence changing the transfer characteristics in detail. In the alternative approach total pulse amplitude is maintained constant and increased proportionally in frequency. The background current, however, has to increase more rapidly to maintain the same coulomb content (current x time). Although this system has a more limited range in operating feed rates, the advantage is that transfer conditions are unchanged throughout the range.

Figure 12 shows the effect of a 50% increase in operating conditions. It should be noted that the second system reaches a limiting condition at double the initial feed rate shown, as background current is then equal to pulse current, giving continuous DC. This limit is not reached in the first system until initial feed rate has quadrupled. It is suggested for the widest possible operating range with these types of synergic control that the system operates according to the constant total pulse, up to some level of background current, and that thereafter the system continues whereby background current increases with a constant excess pulse current superposed.

Discussion and future developments

Current waveform

One question which has not yet been clearly resolved is 'What are the advantages and disadvantages of different current waveforms, particularly in pulse MIG and TIG welding?' The basic waveforms shown in Fig.13 are smooth, as produced by a linear regulator, such as a transistor power source, or more heavily rippled, as produced by an inverter operating at relatively high frequency, above 10kHz. It is believed that in practice the differences are not significant, but this point needs to be clearly established, particularly concerning limitations of one or other type of waveform. Another alternative is where pulse current waveform is composed entirely of short duration impulses of constant amplitude and duration, as illustrated (Fig.13c). Here the average is governed by the mark: space ratio of pulses operating at frequency above 20kHz. Although the average effect of the current is similar to that for more conventional waveforms there could well be detailed modification of the operating arc characteristics. For example, high frequency pulses are claimed to give a stiffer arc in TIG welding, but this information needs to be more clearly established in quantitative terms. The advantage or disadvantage of high frequency modulated pulses in MIG welding is also not known.

Fig.13. Types of pulsed current:

a) From linear regulator;

b) Inverter equipment output;

c) Constant high frequency pulse with variable interval.

Inductance

Use of series inductance (or reactance) is important to successful operation of the short circuit MIG process. Although necessary, inductance is in practice a compromise between that which satisfies the short circuit condition and that which is required by the arcing stage separately. Thus with a given inductance the current rises relatively sharply during short circuit and also falls sharply so that, for most of the time between short circuits, arc current is low (Fig.14).

Fig.14. Dip transfer current waveforms with:

a) Constant inductance;

b) Varied electronic inductance.

With transistor controlled power supplies of good response it is possible to regulate output current to simulate any degree of inductance without using an inductor. Thus, current can rise rapidly, equivalent to low inductance, to a set level but following rupture of the short circuit fall much more slowly, equivalent to high inductance (Fig.14). It is also noted that in such a synthetic inductance system the maximum short circuit current required is reduced, giving less spatter than that found even for the best compromise setting with a pure inductor.

With modern inverter power supplies of high switching frequency it is feasible to reproduce controlled current waveforms simulating the effects of inductance but without excessive current ripple. Advantages of using these over conventional waveforms in short circuit arc welding need to be established.

Arc: short circuit duration ratio

This ratio, termed M, has been shown to be the principal parameter determining equilibrium conditions in short circuit MIG welding. In fact the average current is substantially constant, even with relatively wide fluctuations in apparent short circuit frequency, so long as the M ratio is constant (Fig.15). Conversely variation in the M ratio, even at constant short circuit frequency, gives rise to significant fluctuations in mean current. It is also found that high values of M correspond to a hot (or more fluid) molten pool and low values of M to a relatively cold pool leading to lack of fusion defects. It is suggested that power supplies operating under controlled M ratio would give more consistent welding in practice and avoid lack of fusion defects in vertical welding and similar difficult applications.

Fig.15. Dip transfer welding control by M ratio.

Wire feed systems

Another aspect to be resolved, particularly with respect to MIG welding, is the detailed nature of the wire feed; should it be smooth and steady or discontinuous (incrementally pulsed)? With conventional drives the wire feed is steady, at the rolls at least, giving a linear displacement with time (Fig.16). However, unless the feed rolls are close to the welding head there is a degree of freedom in the conduit such that the wire feed at the arc tends to be discontinuous, occurring in irregular discrete jumps.

Fig.16. Types of wire feed:

a) Linear displacement with time;

b) With small finite modulation;

c) With motion in discrete steps.

To aid the passage of wire through a conduit system, a small finite modulation in wire feed could be introduced (Fig.16). This may also aid control of metal transfer. For example, a system in which the wire feed consists of a series of controlled impulses (with discrete movement around 2mm per step) (Fig.16) gives consistent short circuit performance in CO₂ welding.

Again these aspects of the wire feed as part of the MIG welding system need to be quantified to establish their relative advantage.

MIG molten pool control

Analogous to low frequency current modulation for control of molten pool characteristics and penetration in TIG welding, it is also feasible to modulate average current in MIG welding. The practical difficulty is in maintaining suitable metal transfer over a relatively wide range of average currents and corresponding wire feed rates. This problem is resolved by pulse current control of metal transfer, but operating pulse parameters need to be altered according to operating feed rate to maintain equilibrium.

The above requirement is easily resolved with synergic control, where pulse repeat frequency corresponds to wire feed rate and remaining pulse parameters are adjusted to maintain burnoff equilibrium. The arrangement (Fig.17) of a constant added pulse of variable repeat frequency (superposed on background current which is also proportional to feed rate) is one method of accomplishing simultaneous control of metal transfer with modulation of average current, to control the molten pool. This modulation is at low frequency (as in pulse current TIG) at about 1-2Hz.

Fig.17. Synergic operation with modulated wire feed rate.

There is evidence of the advantage of this combined technique for control of wetting in narrow gap welding, and also for weld pool control as in making the first pass in a V joint preparation. There have also been several examples of combining spray and short circuit arc operation together with weaving or gun oscillation for better control in positional welding. These various systems, including the synergic approach whereby metal transfer is maintained by discrete current pulses, could be usefully explored to determine their relative advantage and disadvantage with respect to the more critical welding applications.

Conclusions and summary

Several steps can be identified in the increasing role taken by the power supply with respect to the welding operation. Although arc welding itself is nominally a century old, these significant changes have occurred in only the last third. These are principally related to the development of power semiconductors, which have led to increased flexibility in power source design and operation. Further interactions between power source and welding system are envisaged, and these need to be quantified to establish clearly which is preferred in given applications.

Steps in evolution of the power supply towards a more active role can be summarised as:

1. The change from steeply drooping to flat output characteristics to give self-adjustment with a constant wire feed rate (where the power source controls arc length, and feed rate controls current);
2. Use of inductance to control dynamic rise and fall of current, particularly in the short circuit MIG process, while maintaining self-adjustment with a flat source;
3. Introduction of low frequency (around 1Hz) current modulation for control of the molten pool and its penetration, as in pulse current TIG welding;
4. Use of medium frequency (around 100Hz) current pulsing for control of metal transfer from the consumable electrode wire in MIG welding in inert or substantially inert gas shields;
5. Synergic or interactive operation, whereby the working conditions at the arc are used to modify the operating parameters for control of metal transfer over a range of average currents and wire feed rates.

Further developments, which are in the experimental stage at present, could lead to new industrial equipment for gas shielded arc welding, as follows:

1. Use of a constant, high frequency (>10kHz) impulse current in place of a steady current in MIG welding, with control of operating current by mark: space ratio in a manner analogous to that used with some TIG, weldingequipment;
2. Feedback to the power source for short circuit MIG welding based on the ratio of arc duration to short circuit duration for control of average heat input;
3. Use of low frequency (around 1Hz) modulation of wire feed rate (preferably with synergic control of the arc) to give molten pool control in MIG welding similar to that obtained with low frequency pulsed current in TIG welding.
4. Combination of incremental wire feed with incremental current pulsing for control of the arc, both in short circuit and open arc transfer modes.

Acknowledgements

The author wishes to acknowledge the many contributions from members of Commission XII of the International Institute of Welding on new arc techniques, particularly in MIG welding. These contributions are too numerous to be referenced in this short review article, but they nevertheless clearly indicate the potential for further evolution of the role of the power supply in arc welding, particularly for the gas shielded processes. This overview also includes work carried out by The Welding Institute, UK.

This article was originally presented at a symposium held in Nagoya, Japan, by Commission XII of the International Institute of Welding in July 1986.