DESCRIPTION

Field of Invention

The object of the present invention is a three-phase inverter control method, particularly intended for applications in which the load is constituted by an electric arc. PRIOR STATE OF THE ART

It is well known that electronic devices called "in- verters", whether they are two-phase or three-phase, avail ^¬ able on the market for many years already, are increasingly being used today for many industrial applications.

Some of the best known of these inverter applications are :

- alternating current, variable-frequency-controlled motor control, for example motors for the actuation of fluid circulation pumps;

- fixed-frequency "UPS" control (UPS, Uninterruptible Power Supply) for the emergency power supply of machines or apparatuses, for example computers;

- control, in the use of photovoltaic panels, of transformation of the direct voltage into alternating voltage, that can be used in a home or enter the distribution network;

- "energy conversion", fixed-frequency system control, for example direct-current, high-voltage transmission sys ^¬ tems, such as HVDC (High Voltage Direct Current) systems.

All of these applications are aimed at converting an input current into an output alternating current. This con- cept also covers the "rectifier-inverter", which is fed in alternating current, transformed into direct and then further into alternating, used to vary the voltage and frequency of the alternating output current with respect to the input.

Inverters are made with electronic valves that perform the function of "electronic switch", characterised by two possible states, defined by the (voltage or current) level injected into a control terminal: closed switch or open switch. Inverters are divided into two main categories, de ^¬ fined by the following terms:

- voltage source inverter, characterised by the manipu ^¬ lation of the accumulated voltage on a capacitor bank;

- current source inverter, characterised by the manipulation of current fed into an inductor.

The increasing availability of new high-performance semiconductors (such as IGBT and IGCT) has accentuated the prevalence of the development of "voltage source inverters" on the other ones.

There is also an operational mode of the "voltage source inverters", referred to as "direct current control", wherein, while maintaining the typical "voltage source" structural feature, output current is however controlled in real time and forced to follow a waveform profile established by the control.

All these applications involve the creation of a sinusoidal waveform, with a frequency and amplitude, which may be fixed or variable. The most used method worldwide is sinusoidal PWM modulation (Pulse-width modulation) , which can be obtained with various implementation techniques, among which are mentioned in particular, as regards "voltage source inverters" in the proper sense:

a) pre-calculated PWM technique, wherein valve states, are pre-calculated with an algorithm that optimises certain performance aspects (minimum harmonic content, minimum current ripple, etc.).

b) technique for sub-oscillation, based on the inter- sections between the modulating sinusoidal wave and a bear ^¬ ing wave (generally having a triangular or "sawtooth" waveform) of higher frequency.

c) "space vector" technique, based on the simultaneous management of valve states through universally known mathematical algorithms, allowing to associate the position of a vector with each possible combination of valve states.

With regard to current-controlled "voltage source inverters", the following command techniques should be remembered:

d) "direct current control" technique, consisting in the comparison between the output current waveform and a "target" waveform. It is a technique with hysteresis control, in which the logical command is determined by comparing the instantaneous output current with a "target" waveform and limiting the frequency of switching operations through a convenient value assigned to the hysteresis (the transition from state "0" to state "1" takes place at a different level from the reverse transition) .

E) "current limit" technique, that can be added to a) , b) , c) techniques by setting the PWM when the instantaneous current exceeds the preset limits, which differs from the previous one for the fact that it is automatically activated when the current exceeds a predetermined level while, when this condition does not occur, impulse width modulation is established with the conventional methods, among which techniques a) , b) , c) are mentioned.

The present invention relates to the problem of how to use best an inverter for feeding a load constituted by an "electric arc". The type of load of an electric arc consists of a voltage source (Va) in series with a resistance (Ra) , all in series to an inductance (L_load) , as represented in the diagram of Fig. 1. It is well known that, in a circuit of this type, with alternating current supply, at the passage through zero of the current, the Ra value tends to assume high values, which cause a temporary increase of the voltage drop.

When the voltage at the Ra terminals reaches a sufficiently high value, arc re-ignition occurs but, if the load is powered by a modulated inverter with a PWM voltage, the interruption and the subsequent re-ignition could be repeated numerous times at each passage through zero of the current .

Therefore, in the use of these current feed control techniques at a load constituted precisely by an electric arc, particular difficulties are met, because the electric arc has a tendency to extinguish at each passage through zero of the current. This type of behaviour, which could also have a random component, can cause unstable operation, creating problems in terms of load power regularity.

A system for generating an electric arc is known for example from DE4416353. This document discloses an arc welding apparatus with a direct current power supply unit; the control unit acts on the inverse rectifier so that sequences of alternate positive and negative impulses are produced with a frequency well above the operating frequency of the transformer; impulses are rectangular and amplitude modulated at a frequency lower than impulse generation frequency, so that the transformer input windings are supplied with alternating voltage at the modulated frequency .

Brief description of the invention

However, the problem underlying the invention is to propose a control method of an inverter, in particular a three-phase inverter, which ensures the re-ignition and, thus, the electric arc continuity even when, due to an ex- cessively high voltage value necessary to re-ignite it, in relation to the voltage present on the DC bus supplying the inverter, this goal cannot be achieved with a traditional- type control. The harmonic content of the output current is not the main focus here, being a secondary importance objective and, as such, negligible.

The main purpose, that is, the arc re-ignition, is reached through a control method which provides two modes of operation (three-phase and single-phase) having the features mentioned in claims 1 and 3 as regards the three- phase mode, 1 and 13 as regards the single-phase mode. The dependent claims describe preferred features of the same invention .

The invention therefore consists in the identification of the optimal choice of the control strategy in relation to the particular type of load control consisting in an electric arc and, in the case of operation as a voltage generator, of the features that must have the waveform of the modulating wave (which can be sinusoidal or non- sinusoidal) allowing to achieve the specific objectives of this application, also allowing to optimise the use of semiconductors and other electrical and electronic components used by the system.

In the single-phase operation, the invention further consists in the identification of a technique which allows to equally distribute the load current on the three phases.

In the mode "voltage generator", although the generation of the PWM which realizes the waveform which will be described may be obtained with each of the known techniques mentioned above in paragraphs a) , b) , c) , d) , e) (or with others) , in this description only the "sub-oscillation" technique (paragraph b) and the "current limit" (paragraph d) are considered, but the invention also comprises all the other techniques with which it is possible to obtain, as a result, the same type of command. The invention also in ^¬ cludes the use of "current control" techniques (paragraphs d) and e) ) , describing the characteristics that such tech- niques must possess to meet the requirements posed by this application. This technique may be applied to all the different known configurations of inverters among which, only by way of example, we point out: Three-phase inverter, three-phase inverter with IGBT / IGCT in series, NPC In- verter, H-bridges.

The sub-oscillation is a technique that consists in generating the logic level of the state of a semiconductor command (on / off) on the basis of the comparison between a high frequency bearing wave and a modulating wave which re- produces the target waveform, which, theoretically, could be obtained with a bearing wave of infinite frequency. The Figure 2 represents the case of a triangular bearing wave and a sinusoidal modulating wave. However, it is clear that other type of waveforms may be used, e.g. with a sawtooth bearing wave and the square modulating wave.

Brief description of the drawings

Further features and advantages of the invention will anyhow be more evident from the following detailed description of a preferred embodiment, given by mere way of non- limiting example and illustrated in the accompanying drawings, wherein:

Fig. 1 and 2 have already been described and represent the known technique regarding the theoretical circuit to represent a load constituted by an "electric arc" and, re- spectively, to generate the P M waveforms associated to a triangular bearing wave and to a sinusoidal modulating wave ;

Fig. 3 shows the trend of the modulating wave, accord- ing to a first embodiment of the invention, obtained by modifying a basic square modulating wave;

Fig. 4 shows a second possible embodiment of the modulating wave, obtained by modifying a sinusoidal wave;

Fig. 5 shows a third possible embodiment of the modulating wave, obtained by modifying a sinusoidal wave with the addition of a third harmonic;

Fig. 6 schematically shows a circuit which allows to obtain the waveform of Fig. 4;

Fig. 7 schematically shows a variant of the circuit of Fig. 6;

Fig. 8 schematically shows another variant of Fig. 6;

Fig. 9 shows the voltages to the ground of an output phase and the star point in the case of a square modulation wave of the traditional type;

Fig. 10 shows the voltages to the ground of an output phase and the star point in the case of a square modulation wave modified by implementing the present invention;

Fig. 11 schematically shows the control system in which the invention is inserted;

Fig. 12 schematically shows a three-phase circuit of square wave current references.

Fig. 13 shows the mechanism by which the invention operates to determine the correct operation of the converter

Fig. 14 shows the use of the invention by means of a sinusoidal circuit of current references

Fig. 15 schematically shows the technique to obtain an even distribution of the load currents in case of single- phase mode operation;

Fig. 16 schematically shows a circuit which allows to obtain the waveforms of Fig. 15;

Detailed Description of Preferred Embodiments

The solution proposed by the present invention consists in adding to the waveform of the modulating wave a first impulse-type forcing wave characterized by a width, a shape and a duration that, with a delay "TO" with respect to the passing through zero of the modulating wave, increases the absolute value for a certain time; in three-phase mode of operation, the invention also provides for the addition of a second impulse-type forcing with opposite sign, having equal or lower width than the first, synchronous to one of the first forcing waves applied to the other phases and thus delayed with respect to the first forcing wave of an angle which can be of 120° or 240°.

In . the case of the modulating circuit of Fig. 3, in the form of a square wave, the first impulse-type forcing wave may be constituted, in its simplest form, by a rectangular impulse with a width "A" and a duration "Tl".

Particularly, and by way of example, taking as unit of measure the width of the bearing wave, the width "A" of the first impulse-type forcing wave, which value may be determined using one of the methods described later, added to the basic modulating wave (i.e. to the square wave) will bring the modulating wave at a value which can vary from 50% to 100% of such width. The second forcing wave of opposite sign corresponds to the first forcing wave of the previous phase, in the sequence U, V, W.

Therefore, also the delay TO and the duration Tl, which may be determined automatically with one of the methods described below, or determined by two constant parameters, will be, by way of example, both comprised between 0.5 and 3.5 ms . In any case - as highlighted in Figure 3 - in order to ensure the safety and quick re-ignition of the electric arc just after the passage through the zero of the current - it is essential that the sum of the two parameters "TO" and "Tl" is longer than the phase displacement of the cur- rent with respect to the voltage generated by the control.

The same object may be obtained, of course, also with forcing waves having a shape other than the rectangular impulse, applied to base modulating waves having a non- rectangular shape. For example, Fig. 4 represents the case of a circuit of sinusoidal base modulating waves, to which a first impulse-type forcing wave having a non-rectangular shape is coupled, aimed at bringing the value of the modulating wave to a level (either positive or negative) equal to "Vmax", and a forcing wave, with an opposite sign, corresponding to the first forcing wave of the previous phase, in the sequence U, V, W.

The figure 5 represents a variant of Figure 4, when the circuit of the sinusoidal base modulating wave is coupled to a third harmonic, which allows to obtain a higher output voltage, at an equal voltage. Also in this case, a first impulse-type forcing wave having a non-rectangular shape is applied, aimed at bringing the value of the modulating wave at a level (positive or negative) equal to "Vmax" and a forcing wave, with an opposite sign, corresponding to the first forcing of the previous phase, in the sequence U, V, W.

In the initial phase of the arc furnace operation, when the material is still in the solid state, it may be more advantageous to operate with a single-phase operation, cyclically alternating the pairs (U, V) , (V, W) , ( , U) . In this operating mode, which will be subsequently better described in detail, it is used only one forcing wave, given that the second forcing wave, with opposite sign, should have a phase displacement of 180°, equal to that of the current .

In FIG. 6 a circuit is schematically shown which carries out a possible technique that allows to obtain the modulating wave of phase "U" having the characteristics il ^¬ lustrated in FIG. 4, starting from a base modulating wave having a sinusoidal waveform. The base modulating (1), af ^¬ ter having being squared in (2), enters into the two phase displacer (3) and (4), which apply the pre-set phase displacements TO and Tl . The two signals enter the "Or exclusive" device (5) (XOR) , whose output (IMP impulses) assumes the state "1" when the two input signals are different. The (6) "half-wave separator" device provides two signals, the first of which assumes the state "1" during the positive half-wave, while the other one assumes the state "1" during the negative semi-wave.

The two half-wave signals and the output (IMP pulses) of the device (XOR) enter in the "AND" gates (7) and (8) and are multiplied by the difference between the parameter value "Vmax" and the value absolute of the base modulating wave; the signal thus obtained is added to the positive half-wave and subtracted from the negative half-wave, thus obtaining the first forcing pulse of the phase "U" . Said pulse is added to the base modulating wave, while from the latter is subtracted instead the first forcing pulse generated in the same way on the phase "W" . The wave thus obtained is the modulating wave to be used for the generation of the P M.

The technique described involves the application of two pre-set phase shifts, but, in reality, it may be convenient to determine these parameters automatically. The figure 7 shows the schema with which these functions are realized. In this case, the input to the phase shifter (3) is pre-set to a maximum phase shift equal to half-wave (90°), while a comparator (9) is triggered at the logic value "0" when the absolute value of the output current exceeds the value "I_min, at which the electric arc is considered definitely triggered, thus determining the instant (T0+T1) in which the first forcing impulse must cease. The time (TO) is instead determined by inputting to "NOR exclusive" device (10) the waveforms coming out from the squarer (2) of the modulating base wave and from the squarer (12) of the output current, thereby obtaining a signal which assumes the logic state "0" when the two inputs are different, that is, from the passage through zero of the modulating wave to the passage through zero of the current. After the passage through zero of the modulating wave, the output of the "AND" logic block (11) will assume the logic value "1" only when the "0" state of the "exclusive NOR" device (10) (time TO) will be ceased and will return to the "0" state, stopping the application of the "IMP" impulses to the modulat- ing wave, after the comparator (9) has reported that the absolute value of the output current has exceeded the threshold set to "I_min" (time Tl) .

In all the examined representations, the impulse width is defined by the desired level of forcing (Vmax) to be reached when the " IMP" signal assumes a logic level "1". This level can be pre-set with a parameter, or determined automatically. Figure 8 is a scheme allowing to automatically obtain this value. It exploits the property that, the higher the level of forcing, the more rapid the arc re- ignition will be and therefore the shorter the impulse duration which, however, must not be less than a pre-defined parameter. The " IMP" signal is sent to the integrator (13) limited between "min" and "max" by the limiter (14) and, if impulse duration is less than the "imp_min" parameter value, the "Vmax" value is lowered.

Example of operation with saturated modulating wave One of the main objects of the invention is to obtain the triggering and maintaining the electric arc with the maximum arc voltage as possible, at an equal voltage on the DCbus. The maximum voltage that may be theoretically obtained from a three-phase inverter, at an equal voltage on the DC bus, is realised with three phases controlled in full square wave, phase-shifted by 120°. This type of power supply is not suitable, however, to ensure the continuity of the electric arc, because the potential to ground of each single phase assumes the waveform of Figure 9, due to potential fluctuation of the neutral point which, at each instant, is equal to the average of the three square waves, phase-shifted by 120°; as shown in Figure 9, this potential decreases by one third the maximum voltage that the "U" phase can take to the ground in the first 60° from the leading edge of the square control wave to the same "U" phase.

Figure 10 represent the result that is obtained using the proposed technique in this invention. The application of the first forcing impulse, which starts after the time TO", has no effect, since the modulating wave is already saturated at the level "1", which represents the entire voltage present on the DCbus and the same also happens on the other phases (V and W) . However, the application of the second impulse, phase-shifted by 120°, is effected; this impulse being negative, it temporarily reduces for a time Tl the amplitude of the modulating wave; this reduction does not cause the electric arc extinction, supported by load inductance, but causes a lowering of the potential of the neutral point in the critical time for the next V phase. On the U phase, this effect is well visible in that, because of the negative forcing that is applied to the W phase after time TO, the potential of the neutral point is reset for the time Tl, with the result that the potential of the output phase rises up to the present voltage value on the DC bus.

Example of adjustment system with implementation of the current limit

Figure 11 represents the complete scheme of the regula- tion that incorporates the invention constituting the object of this patent. The process regulator (which may be a voltage, or current, or power regulator) sets the modulation index "km" that defines the "base modulating wave" (1) entering the "modulating generator" (15) , which changes the modulating wave as described above. The modified modulating wave (16) is compared to the triangular bearing wave (17) by the comparator (18), while the regulator represented in Figure 11 also includes a logic circuit that realises the "current limit" function.

The output current is compared by comparators (19) and

(20) to the thresholds (16) representing the maximum positive and negative peak value that the output current can assume. Said comparators assume the logic state "0" when the current exceeds the limits set by the thresholds in ab- solute value. The two AND logic gates (21) and (22) remove the commands to the power semiconductor when the outputs of the comparators (19) and (20) assume the logic state "0".

Example of operation in direct current control

The objectives that are achieved with the "voltage source" operation mode described previously can also be made with a "current direct control", generating three current profiles phase-shifted by 120°, whose main feature must be to have, instant by instant, a sum equal to zero.

Figure 12 represents a first example of three-phase square wave; each waveform has three states ("positive", "negative" and "zero"), each of which has a total duration of 120°, that is, a third of the period; the "zero" state is also divided into two intervals of equal duration, equal to a sixth of period each (60°) .

The particularity of this type of control consists in the fact that, at each instant, the (target) current entering a phase always comes from one other phase; although, due to the inductance of the load, the actual waveform of the current cannot follow with accuracy the reference "target" one; the control signal that is generated is, however, the optimal one, and meets the requirements that constitute the object of this invention.

Figure 13 is, relatively to the two "U" and "V" phases, the trend of the reference signals, of output currents and of control signals. It is noted that, at the "U > V" transition, the "U" phase current, supported by the inductance of the load, is greater than the reference, while the current of the "V" phase is lower than its reference; in these conditions, as shown in Figure 12, the current of the "W" phase is definitely negative and the command of the "U" phase is certainly brought to the low level, while the command of the "V" phase is definitely at high level, thus providing the output terminal the maximum possible value of voltage to ground, thus creating the optimal conditions for the arc ignition of the "V" phase.

Figure 14 is a second example of the three references, of sinusoidal shape, which is also suitable to ensure the correct operation. It observes that, at the passage through zero of the current of the "U" phase, the current of the "V" phase is slightly less negative than its reference, thus the command of the "V" phase is negative, while the current of phase "W" follows the reference and therefore its command level is intermediate. Instead, because of the difficulty of re-igniting the arc, the "U" phase current is lower than its reference, thus the command of the "U" phase is forced to the maximum positive value, thus favouring the re-ignition of the arch.

Example of single-phase operation

In certain operating conditions, in particular when the material to be melted is still in the solid state, the trigger and the maintenance of the electric arc is more problematic. To properly handle this operating mode, it may be convenient to provide a single-phase feed to the electrodes, feeding only two electrodes, for example, only the (U, V) pair. With this type of power supply, in fact, the electrical circuit is closed only if the electric arc is formed between both the phase "U" and the ground and be ^¬ tween the phase "V" and the ground, but the voltage that is applied to these two arcs, which are connected in series, is directly controlled by the inverter output voltage, while, in the three-phase operation, the voltage to ground, which is applied to each individual electrode, depends on the potential of the star point that, if one of the elec ^¬ trodes is temporarily isolated from earth, can be very unbalanced. Moreover, in single-phase operation, the two phases are fed with a phase shift of 180° instead of 120°.

The single-phase power supply, however, consumes only two electrodes at a time and, to avoid this inconvenience, it is necessary to provide a control system that distributes the load on the three electrodes, while always main- taining a single-phase mode enabled. Figure 15 shows the technique proposed in this invention, applied to the particular case in which the alternation between the pairs of active electrodes is achieved every n = 5 cycles.

The pairs of active electrodes are highlighted by three windows (+U / -W) , (+V / -U) , (+W / -V), each of which has a duration of "n" basic cycles of the square wave (in the example, with a supposed 50 Hz frequency) . In the first window, the base square wave is brought to the "U" phase and its denied wave to the "W" phase; in the second window, the first is brought to the "V" phase and the second to the "U" phase. It is noted that, in this way, in the second window, the "U" phase continues to remain active but, at the transition between the first and the second window, the command status is extended by a half-cycle of the base sguare wave; in the third window, instead, the "U" phase is kept turned off, or is modulated to 50%, that is, the amplitude of the modulating wave is brought to zero. The same sequence is also applied to the other phases.

Figure 16 represents the logical scheme with which the sequence relative to the phase command "U" is implemented. Signals of the modulating base "Q" and the window (+U / - ) , "UW" are sent to the AND logic gate (23), while the re- spective denied "U N" and "QN" signals are sent to the AND logic gate (24) ; the outputs of the two logic gates are added together by the "OR" device (25) , the output of which provides the enabling of the modulating wave of the "U" phase when the logical switch (26) , controlled by the de- nied signal of the window (+W / -V) , is closed; otherwise, the modulating wave is positioned on the "50%" level, thus ensuring the idle state of the phase.

Even the single-phase modulating wave generator is inserted in the control system described in Figure 11 but, in this case, there is only the first forcing impulse given that, with the phase shift of 180° typical of the single- phase operation, the second impulse coincides with the first impulse the other controlled phase. Furthermore, unlike what occurs in the three-phase operation, the modi- fied modulating wave (16) could also be a square wave, giving to the comparators (19) and (20) the task of generating the PW that, in this case, sets a current output having a square waveform, impossible to obtain with a three-phase inverter