What do you mean by voltage?

TWI Bulletin, September 1985

by J C Needham

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

A companion article in the August issue of the Research Bulletin examined how arc welding currents should be correctly represented in numerical form for procedural and QA purposes. This article deals with the other major variable in arc welding voltage - which is largely ameasure of arc length, but is also indicative of arc behaviour.

While current is an independent variable governed essentially by the power supply ^[1] (and hence is a measure of what the power source is providing), voltage is a dependent variable governed by the arc itself, and is a measure of the arc's response to the effect of the current through it.(The main exception to the last statement is found in MIG welding with true flat characteristic power supplies, where the voltage is essentially governed by the power source whilst the current is a measure of the arc's response withconstant wire feed speed).

Moreover, while current can be correctly measured anywhere in the welding circuit, voltage measurements are only true of that particular part of the circuit. Thus measurements physically (and electrically) close to the arc are takento be that of the arc itself, while measurements taken from inside the power supply unit are representative of power source output and are only loosely related to true arc voltages.

Current always appears as a smooth continuum (as in oscilloscope traces) with, at most, abrupt changes in slope (rate of change of current) ( Fig 1a). On the other hand voltage can change very abruptly, i.e. appear as a discontinuity, or develop such high frequency oscillation that it appears as a blurred trace on an oscilloscope. The most common discontinuity is that arising from short circuiting in dip transfer welding, asshown in Fig.1b. Here in effect the (true) arc voltage ( i.e. the voltage while the arc is burning), is interrupted by the occurrence of the short circuit and its rupture in parallel.

The instantaneous voltage peaks can be significant (rather than the overall average level), for instance when they follow the moments of short circuit rupture or when they occur as reignition voltages on alternating current, which peaks may be several times greater than the average arc voltage. There is also the voltage required to initiate an arc from cold ( i.e. without contact between electrode and work) which is upwards of 1kV.

Where to measure arc voltage

Strictly arc voltage should be measured by making potential connections directly across the arc itself, but this is impracticable. Therefore measurements are made as close to the arc as reasonable while at the same time taking care to eliminate extra potential drops caused by the passage of welding current. ^[2] Additional voltage drops caused by current in cables, connections, loose ground returns and so forth must be avoided. Since, in this context, the voltage drop along the stickout can be considered part of the true arc system ( Fig.2) the simplest arrangement is to make potential connections to the electrode wire and to the surface of the workpiece being welded ( Fig.3). In the case of MIG or submerged arc welding ( Fig.3a) one potential connection can be made to the wire reel since there is no current flow in the wire (except that along the stickout which, as indicated above, is correctly included in the voltage measurement).

The potential connection to the work should be made such that the current return path is not included in the voltage measurement. The simplest arrangement is to make contact with the upper part of the work on a side which is away from the ground return, and to ensure that the current path is down through the thickness of the workpiece.

The voltage drops across mechanical junctions carrying full welding current can be significant and readily amount to 0.05V per connection. The example shown in Fig.4 was obtained for a good quality fresh clean connector. However in practice, with normal deterioration under shop conditions, the voltage drop including the effect of frayed flexible cable connections can amount to some three times greater voltage drop per junction. Also cable resistance increases appreciably with temperature so that a 'hot spot' in a poor connector can heat the associated cables for a considerable distance and add further to the extraneous voltage drops.

Steady dc operation

Where the voltage is inherently smooth and steady, as in a TIG arc at constant length and with a steady current, ( Fig.5a) there is no difficulty in specifying the voltage limits such as 9.9 ± 0.1 V.

One potential connection should be made direct to the tungsten rod, and the electrode allowed to reach equilibrium temperature before measurement, see Fig.3c. However, the majority of welding arcs employ a consumable electrode where the arc length (although consistent in the long-term) is changing significantly in the short-term because of the metal transfer that is occurring. Thus, as shown in Fig.5b, the arc voltage for a steady MIG arc in the spray region is consistent, but with a superposed ripple which is indicative of the instants of metal transfer. Although the ripple itself could be quoted as a peak-to-peak value, it is in fact better ignored, as the instantaneous peaks, depending on the resolution of the recording system, can be quite high, of the order of 3-5V. It is therefore advisable to include a time constant of at least 1msec in the voltage measuring system to smooth out the high frequency spikes in the ripple and to determine the longer term mean value.

Similarly in submerged arc welding there is a cyclic variation in arc voltage, ( Fig.5c) under otherwise consistent operating conditions which is caused largely by the welling up and collapse of the bubble cavity and the associated metal transfer from the relatively thick electrode wire. This fluctuation occurs typically at a rate in the region of 10Hz and gives rise to a nominally saw-tooth like fluctuation in the instantaneous arc voltage. Here again the long-term mean can be quoted together with, where relevant, the peak-to-peak fluctuation as an index of the cavity behaviour.

Figure 5d shows an even greater disturbance in the arc voltage caused by metal transfer short-circuiting the arc, as in manual metal arc welding with rutile and, to a lesser extent, with basic type electrode coverings: Accurate measurement is not practicable with this gross degree of variation unless a longer term average, based on say at least two and preferably five seconds, is used. A further complication arises from the heating up and consumption of the electrode itself in the course of operation. Thus the apparent voltage measured at the core wire, ( Fig.2), is that of the arc plus the potential drop along the core wire. The latter, as shown in Fig.6, is initially low at about 1.5V with the wire cold and increases to a maximum of around 2V as the wire is heated up before finally falling as the total length is reduced to the stub end. (Radial lines drawn from the point P have a slope corresponding to the resistivity of the core wire which is continuously increasing under the operating conditions concerned). The additional voltage drop typically ranges from about 1-2V in the course of burning an electrode at the normal recommended current. For accurate definition of the operating arc voltage, i.e. to better than about 1V, it is necessary to introduce a correction to allow for this variation in potential drop in the core wire.

Pulsed DC operation

Pulsed current is normally applied only to the gas shielded processes TIG and MIG, where the pulsed current may be three or more times greater in amplitude than the background current condition. The arc voltage does not vary so greatly since, from the static arc characteristic, ( Fig.7a), the operating voltage at a given arc length is approximately constant over a wide range of current. In detail, however, the instantaneous voltage during a pulse depends partly on the sharpness of the current pulse. For rates of change of current of less than 10A/msec the static arc characteristic, ( Fig.7a), is valid. On the other hand for high rates of change of current (greater than 100A/msec) the instantaneous voltage changes as if the arc were for that instant behaving as a constant resistance, ( Fig.7b). This phenomenon is illustrated by the dashed lines in Fig.7a where, on suddenly increasing the current the voltage momentarily tends towards 2' (and conversely on suddenly decreasing the current the voltage momentarily tends to 1') before recovering to the corresponding static states 2 and 1 respectively.

Thus with a TIG arc, the voltage during a current pulse may in the longer term be substantially equal to that during the background period (or slightly greater or less according to the specific points on the static characteristic concerned). Nevertheless, with high rates of change of current (as is possible with transistor type power supplies), the leading edge of the voltage for a current pulse may show an overshoot and undershoot corresponding with the rise and fall in current, ( Fig.7b).

In MIG welding, where current pulses are used to control the metal transfer, there is a further disturbance in the instantaneous arc voltage because of the transfer itself, which may occur during the current pulse as in Fig.8a or in the subsequent background period, Fig.8b. For all practical purposes these spikes in the instantaneous voltage can be ignored and the voltage averaged over the period concerned. Because the principal work done is during the current pulse, it is better to determine the average voltage from the pulse-on period alone and to ignore that during the background period. This is particularly true when using low background currents where the arc may wander and cause anomalous voltage fluctuations which have no significance in practice. However a clear distinction should be made between whether the arc voltages quoted are those averaged over the complete pulse and background period, that is an overall average, or are the mean voltages during the current high period alone.

Alternating current operation

With AC the arc voltage is further complicated by the fact that following current zero the arc by definition has to be re-established for each half cycle in succession. The voltage required to do this can vary from virtually negligible (and generally less than the arc voltage itself) to a distinctive peak which may be several times the arc voltage in magnitude ( Fig.9). With rutile electrodes the reignition voltage ( Fig.9a), is indistinguishable from the arc voltage which is correctly given by the mean (rectified) value. ^[3] However with bare nonthermionic electrodes, as would be the case for MIG wires on AC, ( Fig.9b), the voltage required to re-establish the arc for either polarity, is in excess of 150V. ^[4] For AC TIG, the reignition characteristics are a combination of the above two examples, where, with electrode-negative, the reignition is indistinguishable but with electrode-positive a significant excess voltage occurs, ( Fig.9c). Also with other grades of manual metal arc covering, particularly the cellulosic and basic electrode types, there is a significant reignition stage where the instantaneous voltage is some 2-3 times the arc burning voltage and may last for a significant fraction of each half cycle ( Fig.9d). ^[3]

These reignition voltage peaks add significantly to the apparent arc voltage, but are not necessarily consistent in themselves. Therefore the overall voltage appears to fluctuate when in fact the true arc voltage can be more consistent. The effect of the voltage peak is shown in Fig.10 where, considering one half cycle only, if the overvoltage is of sufficiently short duration the error is relatively small (~2%, Fig.10a) but if it lasts say 1msec the overall mean is increased by some 15% over the true arc voltage level ( Fig.10b). If a true RMS indicating meter is used then the increases over that for the arc voltage alone are further exaggerated; for example some 5% for the conditions shown in Fig.10a where the short duration spike lasts only 1% of the half cycle and over 23% in the case of Fig.10b.

To improve the resolution and accuracy of measurement of arc voltage on AC, it is desirable to eliminate the voltage spikes arising from cyclic reignition. This feature is currently designed into some experimental instruments developed by The Welding Institute. Alternative methods of eliminating the effect of the reignition stage are being considered.

In the particular case of AC TIG, there is considerable asymmetry in the electrode voltage for the positive and negative arc polarity. From the point of view of automatic arc length control, it is preferable to base the AVC system on the average electrode-negative voltage or possibly better still to sample the instantaneous voltage at current peak on the negative half cycle.

Short circuit MIG operation

For this gas shielded welding process the situation is similar to that shown in Fig.5d except that the short circuits occur 10 times more frequently. In fact the average welding voltage normally quoted is that of the arc and short circuit stages together ( Fig.11).

Since the short circuit voltage is nearly zero, this means that the true arc voltage has been reduced by an amount proportional to the relative short circuit time. However, as in the case of current, ^[5] the overall mean or average voltage is significant and is indicative of the operating conditions for the purposes of weld procedure specification. As before the voltage has to be averaged over a significant period, at least 0.5sec and preferably over 1 or 2sec to eliminate the effect on the average of the gross fluctuation in voltage because of individual short circuits. The average voltage, like the average current, is then consistent for a given welding condition.

Special instrumentation is required to extract the arc burning voltage alone. Although this is not essential, it is desirable in so far as it would help identify whether or not the correct combination of wire and gas composition is being utilised. This refinement is desirable since the average welding voltage (including the effects of short circuiting) can be apparently correct, although in fact it could represent a wrong combination of relative short circuit time and true arc voltage.

RMS or mean reading instruments?

Although an RMS instrument gives the same reading as the average or mean value for a steady level, RMS instruments should be used with considerable care as there is a greater tendency to misleading results. The effect of the spike voltages discussed above are more severe with an RMS instrument than with a true mean reading instrument. Moreover with short circuit MIG welding, the mean reading is itself significant, but an RMS reading is not because of the irregular waveform concerned. As long as like is being compared with like, there is no difficulty, but when translating information from one laboratory to another, where different types of instrument might be employed, it is essential to specify what kind of instrument was used.

When using AC there is a particular problem with one type of RMS meter which is based on registering the true (mean) rectified voltage but which is scaled to read an RMS value, presuming that the waveform is sinusoidal. Now with alternating current arcs the voltage waveform even with a sinewave current is approximately more square than sine. Therefore such an instrument reads apparently about 10% high compared with the true RMS value. Therefore it is necessary on AC to ensure that the instruments read either mean (rectified) directly or are true RMS reading. As the effects of spikes are more exaggerated with the RMS reading instruments the latter are to be avoided.

Therefore it is strongly suggested that all measurements are based on mean reading instruments to avoid the possibility of error and confusion, and to ensure consistency between DC and AC data.

Peak voltage measurements

Apart from the average arc voltage, there is interest in specific peak conditions such as the reignition voltage with alternating current. These peak voltages can themselves be averaged and their standard deviations found in the same manner as for short circuit current peaks in dip transfer welding. ^[6] Data has been obtained in the past on typical reignition voltages and their distribution in magnitude for various grades of electrode and for the effects of different types of so-called square wave AC supply compared with the normal sine wave. ^[7] The lower the average value of the reignition peak and the shorter its duration the more stable the arc. Hence measurement of this reignition-stage serves as an index of stability or reproducibility of the arc condition (or of the flux used) in either manual metal arc or submerged arc welding.

Peak voltages also occur at the instant of short circuit rupture in dip transfer welding, and again such voltage could be used as an index of the stability of the operation or maybe as an index of the gas composition. The latter aspect is based on the fact that the arc has to be reignited following each short circuit, and the voltage peak would be a measure of the work that has to be done to establish the arc. As far as the author is aware this aspect has not yet been investigated and could prove a fruitful field in defining operating stability.

Finally there is the voltage required to break down the arc gap for contactless ignition of the arc in gas shielded welding, both TIG and MIG. The voltages involved here are large, typically in excess of 1kV, and for TIG with a finite gap of some 2mm the average breakdown voltage is around 3 or 4kV in argon ( Fig.12a). Again a single numerical value is imprecise as there is considerable variation in breakdown potential from instant to instant. In some early work ^[8] on the necessary voltage to strike a clean tungsten electrode in argon, there was at least 1kV standard deviation. This means that in practice although the gap might break down at voltages as low as 2kV there were situations where voltages in excess of 5kV would be required to ensure 100% satisfactory operation, as indicated in Fig.12b.

Summary

As in the case of current, the overall recommendation is to use mean (not RMS) reading instruments (but rectified for AC) for deriving the average arc voltage. Special instruments are required to derive the true arc burning voltage (in, short-circuit arc welding) and to estimate stability as by the reignition voltage stage on AC. When using AC the average true arc voltage is increased by the voltage in the reignition stage, and again special instrumentation is needed to eliminate this effect.

Finally arc voltage measurements should not be taken from the power supply, but from as close to the arc as is practicable. For most purposes one potential contact should be made to the electrode wire (or even to the wire reel) for MIG and submerged arc welding, and to the electrode (core wire) in TIG and MMA welding. The other potential contact should be made to the near side work surface, but remote from the welding current return path.

References