Improvements in variable-voltage & variable-frequency power converter.

0001The present invention relates to a variable-voltage and variable-frequency electric power converter for a load such as a motor, and a method of controlling the same.

0002A conventional power converter comprises a DC smoothing capacitor serving as a DC voltage source, a first inverter, a second inverter, an output transfor­mer, a load device, a load current detector, a com­parator, a current control/compensation circuit, a proportional amplifier, pulse-width modulation (PWM) controllers, etc.

0003The first and second inverters are PWM control inverters for converting a DC voltage into a variable-­voltage and variable-frequency AC voltage. The first inverter generates output voltage V_L1 through the output transformer. The second inverter directly generates output voltage V_L2. The load device is applied with a voltage of V_L = V_L1 + V_L2. Voltage V_L is adjusted to control load current I_L.

0004When output frequency f₀ is zero, since no voltage V_L1 can be generated from the output transformer, the second inverter controls load current I_L, and when fre­quency f₀ is increased to a certain level, the first inverter controls load current I_L in turn.

0005Figs. 17 and 18 show load terminal voltage V_L and input signals V₁* and V₂* of the PWM controllers of the first and second inverters as a function of output fre­quency f₀ in the conventional power converter.

0006When load current I_L is assumed to be constant, load terminal voltage V_L is expressed by output fre­quency f₀ as follows: V_L = V_C + I_L·R_L + jω₀L_LI_L      (1) V_C: counter electromotive force of load (for motor) R_L: resistance of load L_L: inductance of load ω₀: 2πf₀

0007A broken line in Fig. 17 indicates a voltage drop caused by resistance R_L of the load.

0008A broken line in Fig. 18 indicates input signal V₁* of the PWM controller of the first inverter. V₁* = 0 is established for f₀ ≦ f_min, and V₁* is proportional to f₀ in a region of f₀ > f_min. A solid line in Fig. 18 indi­cates input signal V₂* of the PWM controller of the second inverter. Signal V₂* is controlled to output voltage V_L given by equation (1) for f₀ < f_min, and V₂* = I_L*·R_L = constant in a region of f₀ > f_min.

0009In this manner, in the conventional power con­verter, the second inverter need only generate required voltage V_L up to minimum frequency f_min at which the output transformer can be operated, and can have a smaller capacity than the first inverter. Since the first inverter generates voltage V_L1 through output transformer, serial or parallel multiplexed operations of this inverter can be easily performed, and its capa­city can be easily increased. Therefore, the converter is convenient for one for driving a large-capacity AC motor.

0010According to the above conventional power con­verter, when output frequency f₀ becomes high and load current I_L is to be controlled by the first inverter, the following problems are posed.

0011In general, secondary voltage V_L1 of the output transformer is proportional to output voltage V₁ of the first inverter. However, when a DC component is included in load current I_L due to, e.g., drift of the current detector, the output transformer gradually suf­fers from DC polarized magnetization, and soon, its iron core is magnetically saturated to one side. As a result, secondary voltage V_L1 of the output transformer is decreased, and load current I_L is decreased to be lower than instruction value I_L*. Then, the inverter operates to increase voltage V₁ and further saturate the iron core. Therefore, primary current I₁ of the output transformer is abruptly increased, and may become an overcurrent to destroy elements of the first inverter. In order to prevent polarized magnetization of the out­put transformer, the sectional area of the iron core may be increased and an air gap is provided. In this case, the weight and dimensions of the output transformer are increased, resulting in an uneconomical system.

0012The present invention has been made in considera­tion of the above situation, and has as its object to provide a power converter which can correct polarized magnetization of an output transformer, and can accura­tely control a load current, and a method of controlling the same.

0013In order to achieve the above object, according to the present invention, there is provided a power con­verter which supplies an AC load with a sum of an output voltage from a first PWM inverter having an output transformer, and an output voltage from a second PWM inverter having no output transformer, wherein the first PWM inverter comprises a means for controlling a primary current or an exciting current of the output transfor­mer connected to the first PWM inverter, and a means for controlling a current supplied to the AC load.

0014The second PWM inverter controls a current to be supplied to the load regardless of output frequency f₀.

0015The first PWM inverter can be constituted by one or a plurality of PWM inverters. Each inverter controls a primary current or an exciting current of a corre­sponding output transformer. An instruction value of the primary current of the transformer is given by a sum of an instruction value of the load current control, and an instruction value of the exciting current of the transformer. When the instruction value of the exciting current is given to have an almost constant effective value, the output voltage of each output transformer is increased/decreased almost in proportion to the output frequency. A sum of the output voltages of the trans­formers and the output voltage of the second PWM inverter is applied to the AC load, and the output voltage of the second PWM inverter is adjusted to con­trol the load current. As a result, the first PWM inverter bears a voltage proportional to the output frequency, i.e., a voltage drop caused by a counter electromotive force (velocity electromotive force) and an inductance, and the second PWM inverter bears the remaining necessary voltage, i.e., a voltage drop caused by a resistance of the load and a transient voltage upon load current control. The capacity of the second PWM inverter can be considerably smaller than that of the first PWM inverter.

0016When the exciting current of each output trans­former is controlled to be almost constant, the output voltage of the transformer is increased/decreased in proportional to the output frequency, and an iron core is never saturated even if the frequency is lowered. Even when a DC component is included in an output voltage due to a variation of switching characteristics of self extinction elements constituting the first PWM inverter, and this causes DC polarized magnetization of the output transformers, since the exciting current of each transformer is controlled by the inverter, the DC voltage component can be automatically compensated for, and the transformers can always be operated in a state free from polarized magnetization. Therefore, the primary current of each transformer can be prevented from being increased too much, and elements can also be prevented from being destroyed by polarized magnetiza­tion of the transformers.

0017This invention can be more fully understood from the following detailed description when taken in con­junction with the accompanying drawings, in which: <ul id="ul0001" list-style="none"><li>Fig. 1 is a circuit diagram showing an embodiment of a power converter according to the present invention;</li><li>Figs. 2A to 2D are timing charts for explaining a PWM control operation of the converter shown in Fig. 1;</li><li>Fig. 3 is a vector chart for explaining the opera­tion of the converter shown in Fig. 1;</li><li>Figs. 4 and 5 are graphs showing frequency-voltage characteristics for explaining the operation of the con­verter shown in Fig. 1;</li><li>Fig. 6 is a circuit diagram showing another embodi­ment of a power converter according to the present invention;</li><li>Fig. 7 is a circuit diagram showing still another embodiment of a power converter according to the present invention;</li><li>Fig. 8 is a circuit diagram of a controller for the converter shown in Fig. 7;</li><li>Figs. 9 and 11 are partial circuit diagrams of the converter shown in Fig. 7;</li><li>Figs. 10A - 10D and Figs. 12A - 12B are timing charts for explaining the operation of the converter shown in Fig. 7;</li><li>Fig. 13 is a circuit diagram showing another arrangement of the controller for the converter shown in Fig. 7;</li><li>Figs. 14A - 14N are timing charts for explaining the operation of the controller shown in Fig. 13;</li><li>Figs. 15A - 15C are views showing computer simulation results which demonstrate the effect of the control method of the present invention;</li><li>Figs. 16A - 16H are timing charts for explaining a control method of the present invention; and</li><li>Fig. 17 and 18 are graphs for explaining an opera­tion of the converter shown in Fig. 8.</li></ul>

0018Fig. 1 is a circuit diagram showing an embodiment of a power converter according to the present invention.

0019In Fig. 1, reference symbol C_d denotes a DC smoothing capacitor serving as a DC voltage source; INV-1, a first PWM inverter; INV-2, a second PWM inver­ter; and LOAD, a AC load.

0020First PWM inverter INV-1 is a so-called voltage type PWM inverter, and comprises self-extinction ele­ments (e.g., gate turn-off thyristors, and the like) S₁₁ to S₁₄, and diodes D₁₁ to D₁₄.

0021Second PWM inverter INV-2 is also a voltage type PWM inverter, and comprises self-extinction elements S21 to S24 and diodes D21 to D24.

0022AC load LOAD includes inductance L_L, resistance R_L, and counter electromotive force V_C.

0023As a controller, current detectors CT₁ and CT_L, calculation circuit CAL, comparators C₁ and C₂, current control/compensation circuits G0(S) and G_L(S), adder AD, and PWM controllers PWM₁ and PWM₂ are included.

0024The PWM control operation of first PWM inverter INV-1 will be briefly described below.

0025Figs. 2A to 2D are timing charts for explaining the PWM control operation of the converter shown in Fig. 1. In Fig. 2A, reference symbols X and Y denote PWM control carrier signals, and Y is an inverted value of X (in other words, having a 180° phase difference). V₁* deno­tes an input signal of PWM controller PWM₁ of first PWM inverter INV-1. Upon comparison between V₁* and carrier signals X and Y, gate signals g₁₁ and g₂₂ for elements S₁₁ to S₁₄ constituting first inverter INV-1 are generated. More specifically, When V₁* ≧ X, g₁₁ = "1", and S₁₁: ON, S₁₂: OFF When V₁* &lt; X, g₁₁ = "0", and S₁₁: OFF, S₁₂: ON When V₁* ≧ Y, g₁₂ = "1"′ and S₁₃: OFF, S₁₄: ON When V₁* &lt; Y, g₁₂ = "0" and S₁₃: ON, S₁₄: OFF

0026When the voltage of smoothing capacitor C_d is given by V_d, output voltage V₁ of first inverter INV-1 is given by: V₁ = +V_d when S₁₁ and S₁₄ are ON V₁ = -V_d when S₁₂ and S₁₃ are ON V₁ = 0 in other modes

0027More specifically, when the phases of carrier signal X for determining a firing signal of elements S₁₁ and S₁₂, and carrier signal Y for determining a firing signal of elements S₁₃ and S₁₄ are shifted by 180°, out­put voltage V₁* of first PWM inverter INV-1 has a wave­form controlled by a frequency twice a switching frequency of the elements. Average value <o ostyle="single">V</o>₁ (indicated by broken lines) of the waveform of V₁ is proportional to input signal V₁* of PWM₁.

0028The PWM control operation of second PWM inverter INV-2 is similarly executed.

0029A load current control operation will be described below.

0030Load current I_L is controlled by second PWM inverter INV-2.

0031First, current I_L supplied to load LOAD is detected by current detector CT_L, and the detected current is input to comparator C₂. Comparator C₂ compares load current instruction value I_L* with detected value I_L, and supplies its deviation ε_L = I_L* - I_L to the next current control/compensation circuit G_L(S). Circuit G_L(S) amplifies deviation ε_L, and supplies V₂* = G_L(S)·ε_L to PWM controller PWM₂ of the second inverter. Second inverter INV-2 generates voltage V_L2 proportional to input voltage V₂* of the PWM controller.

0032When I_L* > I_L, deviation ε_L becomes a positive value, and increases input signal V₂* of PWM₂. Thus, control is made so that output voltage V_L2 of second inverter INV-2 is increased and load current I_L is increased to establish I_L ≒ I_L*. Contrary to this, when I_L* < I_L, deviation ε_L becomes a negative value, and decreases output voltage V_L2 of second inverter INV-2 and load current I_L. In this case, control is balanced when I_L ≒ I_L*. When load current instruction value I_L* is sinusoidally changed, actual load current I_L is sinu­soidally controlled to follow the change. When output frequency f₀ is zero, load current control can be per­formed without any problem.

0033The control operation of first PWM inverter INV-1 will be described below.

0034First PWM inverter INV-1 generates voltage V_L1 almost proportional to output frequency f₀ through out­put transformer TR. Inverter INV-1 is basically not related to load current control. First PWM inverter INV-1 controls exciting current I₀₁ of output trans­former TR.

0035Voltage V_L applied to load LOAD is expressed by: V_L = V_L1 + V_L2 = V_C + R_L·I_L + I_L·p·I_L      (2) p: differential operator

0036When load current I_L is sinusoidally controlled as follow: I_L = I_Lm·sinω₀t      (3) I_Lm: current peak value ω₀ = 2πf₀ equation (2) is rewritten as: V_L = V_C + R_L·I_L + jω₀L_L·I_L      (4) and, coincides with equation (1).

0037First PWM converter INV-1 generates a component proportional to output frequency f₀ in equation (4), and given by: V_L1 = V_C + jω₀L_LI_L      (5) where V_C is the counter electromotive force of an AC motor, and is proportional to a rotational speed of the motor, i.e., output frequency f₀ of the inverter.

0038In order to cause first PWM inverter INV-1 to generate voltage V_L1 given by equation (5), an instruc­tion value given by the following equation is input to PWM controller PWM₁ through adder AD: V_L1* = V_C* + jω₀·L_L·I_L*      (6) When the ratio of the numbers of turns of the primary winding to secondary winding of output transformer TR is 1 : 1, output voltage V₁ of first PWM inverter INV-1 is directly output from output transformer TR as V_L1 in a normal operation mode, and becomes a voltage complemen­tary to a sum of counter electromotive force V_C of the motor and voltage drop jω₀L_LI_L by inductance L_L. As a result, second PWM inverter INV-2 need only generate output voltage V_L2 corresponding to voltage drop I_L·R_L caused by resistance R_L, and load current I_L can also be easily controlled.

0039However, when output transformer TR suffers from DC polarized magnetization, an iron core may be saturated, and an exciting current of transformer TR may become too high. In order to prevent this, the first PWM inverter controls exciting current I₀₁. The control operation of the exciting current will be described below.

0040Primary current I₁ and secondary current (load current) I_L are respectively detected by current detec­tors CT₁ and CT_L, and a difference between the two currents is calculated to obtain exciting current I₀₁ of the transformer. That is, I₀₁ = I₁ - I_L      (7) In this case, the ratio of the numbers of turns of the primary and secondary windings of transformer TR is assumed to be 1.

0041On the other hand, exciting current instruction value I₀₁* is obtained by the following calculation by calculation circuit CAL using the above-mentioned output voltage instruction value V_L1*.<maths id="math0001" num=""><img file="EP0390184A2_D0001.tif" /></maths> where M is the exciting inductance of the transformer.

0042In a synchronous motor load, when load current I_L is sinusoidally controlled to be in phase with counter electromotive force V_C, V_C* and I_L* are respectively given by: V_C* = V_Cm*·sinω₀t      (9) I_L* = I_LM*·sinω₀t      (10) V_Cm*: voltage peak instruction value I_Lm*: current peak instruction value

0043Thus, exciting current instruction value I01* given by equation (8) is expressed by:<maths id="math0002" num=""><img file="EP0390184A2_D0002.tif" /></maths>

0044Fig. 3 is a vector chart of these values. More specifically, vector V_L1* of the secondary voltage of output transformer TR is equal to a sum of counter electromotive force vector V_C* and vector jω₀L_LI_L* of a voltage drop caused by inductance L_L of the load, and, hence, exciting current I₀₁* is necessary for generating secondary voltage V_L1*.

0045Comparator C₁ compares exciting current instruc­tion value I₀₁* with detection value I, , and deviation ε₀₁ = I₀₁* - I₀₁ therebetween is amplified by the next current control/compensation circuit G₀(S). The output signal of G₀(S) is added to output voltage instruction value V_L1* by adder AD, and V₁* = V_L1* + G₀(S)·ε₀₁ is input to PWM controller PWM₁ of first PWM inverter INV-1.

0046When I₀₁* > I₀₁, deviation ε₀₁ becomes a positive value, and is amplified by G₀(S), thus increasing input signal V₁* of PWM₁. Therefore, output voltage V₁ of the first inverter is increased, and exciting voltage of the transformer is increased. In this case, although secondary voltage V_L1 of output transformer TR is also slightly increased, output voltage V_L2 of second inverter INV-2 which performs load current control is decreased by a value corresponding to an increase in V_L1, and voltage V_L = V_L1 + V_L2 to be applied to load LOAD is left unchanged. Therefore, control is balanced to yield I₀₁ ≒ I₀₁*.

0047In contrast to this, when I₀₁* &lt; I₀₁, deviation ε01 becomes a negative value, and decreases output voltage V₁ of first inverter INV-1 and exciting current I₀₁ of transformer TR. Thus, control is balanced when I₀₁ ≒ I₀₁*. In this case, load current I_L is also controlled by second inverter INV-2 to satisfy I_L ≒ I_L*.

0048When exciting current instruction value I₀₁* is sinusoidally changed, as expressed by equation (11), actual exciting current I₀₁ is sinusoidally controlled to follow it.

0049In equation (11), since peak value V_Cm* of the counter electromotive force is proportional to output frequency ω₀, when peak value I_Lm* of the load current is constant, the peak value of exciting current I₀₁* becomes constant regardless of output frequency f₀.

0050In the vector chart of Fig. 3, when load current instruction value I_L* is increased, inductance drop jω₀L_LI_L* of the load is increased accordingly, and secondary voltage V_L1* of output transformer TR is also increased. As a result, the magnitude of exciting current vector I₀₁* is also increased.

0051Fig. 4 shows the relationship between output frequency f₀ and the magnitude of load terminal voltage V_L. First PWM inverter INV-1 bears component V_C + jω₀L_LI_L proportional to output frequency f₀, and second PWM inverter INV-2 bears voltage drop I_L·R_L by the resistance.

0052Fig. 5 shows the relationship between secondary voltage V_L1 of output transformer TR of first PWM inverter INV-1 and output frequency f₀. When exciting current I₀₁ of the transformer is assumed to be constant, the magnitude of secondary voltage V_L1 is pro­portional to output frequency f₀. When excitation current I₀₁ is decreased like I′₀₁, the magnitude of V_L1 is also decreased as indicated by a broken line.

0053A case will be considered below wherein a DC com­ponent is superposed on load current I_L due to, e.g., drift of current detector CT_L in the converter shown in Fig. 1, and output transformer TR suffers from DC polarized magnetization. Due to the DC polarized magne­tization, the iron core of transformer TR is saturated to one side. For this reason, mutual inductance M of transformer TR is decreased, and causes exciting current I₀₁ to increase. However, in the converter of the present invention, since first PWM inverter INV-1 controls exciting current I₀₁ of transformer TR to coincide with instruction value I₀₁*, output voltage V₁ of first inverter INV-1 is automatically decreased, and an increase in exciting current I₀₁ is prevented. As a result, primary current I₁ of transformer TR, i.e., the output current of inverter INV-1 is no longer increased. As a result, elements can be prevented from being destroyed by an overcurrent caused by polarized magnetization as a conventional problem.

0054Fig. 6 is a circuit diagram showing another embodi­ment of a power converter according to the present invention. A difference from the converter shown in Fig. 1 will be described below.

0055More specifically, two stages of inverters INV-11 and INV-12 having output transformers are stacked and multiplexed, and direct-connection inverter (having no output transformer) INV-2a bridge-connects three phases. Neutral point N of a load is connected to an inter­mediate tap of DC voltage V_d.

0056In this manner, the number of multiplexed stages of inverters having output transformers can be increased to increase an output capacity of the power converter. Since neutral point N of the load is connected to the intermediate tap of DC voltage V_d, an electrical power can be supplied to a three-phase, four-line type load.

0057In the converter shown in Fig. 7, load current I_L can be controlled by direct-connection inverter INV-2, and exciting currents of corresponding output trans­formers TR_U1 and TR_U2 can be respectively controlled by inverters INV-11 and INV-12 with the output transfor­mers. Thus, the same effect as in the converter shown in Fig. 1 can be provided.

0058According to the power converter of the present invention, even when DC polarized magnetization occurs in an output transformer, the primary current (output current of the first PWM inverter) of the transformer is prevented from being increased, and elements can be pre­vented from being destroyed by an overcurrent.

0059Fig. 7 is a circuit diagram showing still another embodiment of a power converter according to the present invention, and Fig. 8 is a circuit diagram showing an embodiment of a controller of the converter shown in Fig. 7.

0060In Fig. 7, reference symbol V_d denotes a DC voltage source; INV-1 to INV-4, multiplexed PWM inverters; INV-5, a direct-connection PWM inverter; TR₁ to TR₄, output transformers; and LOAD, an AC load (U phase). Fig. 7 illustrates a circuit corresponding to only one output phase of the converter. However, circuits corresponding to V and W phases have the same arrange­ments.

0061In Fig. 8, reference numerals A₁ to A₆ denote adders; C₁ to C₅, comparators; G₁ to G₄ and G_L, current control/compensation circuits; K_N, an operational ampli­fier; PWM₁ to PWM₅, PWM controllers; and TRG, a carrier (triangular wave) generator. Note that CT₁ to CT₅ in Fig. 7 designate current detectors.

0062Fig. 9 shows a detailed arrangement of PWM inverter INV-1 shown in Fig. 7. Inverter INV-1 comprises self-­extinction elements S₁₁ to S₁₄ and free-wheeling diodes D₁₁ to D₁₄.

0063Figs. 10A to 10D are timing charts for explaining the PWM control operation of the inverters shown in Fig. 9.

0064Reference numerals X₁ and Y₁ denote PWM control carrier signals. Y₁ has an inverted value (or a signal having a phase difference of 180°) of X₁. Control input signals e₁ and X₁ are compared to generate gate signal g₁ for elements S₁₁ and S₁₂ of inverter INV-1. When e₁ ≧ X₁, g₁ = 1, and S₁₁: ON, S₁₂: OFF When e₁ &lt; X₁, g₁ = 0, and S₁₁: OFF, S₁₂: ON Control input signals e₁ and Y₁ are compared to generate gate signal g₁′ for elements S₁₃ and S₁₄. When e₁ ≧ Y₁, g₁′ = 1, and S₁₃: OFF, S₁₄: ON When e₁ &lt; Y₁, g₁′ = 0, and S₁₃: ON, S₁₄: OFF

0065When the ratio of the numbers of turns of primary to secondary windings is assumed to be 1, output voltage V₁ of inverter INV-1 is given by: V₁ = +V_d when S₁₁ and S₁₄ are ON V₁ = -V_d when S₁₂ and S₁₃ are ON V₁ = 0 in other modes

0066Thus, a waveform shown in Fig. 10D is obtained. Average value <o ostyle="single">V</o>1 (indicated by a broken curve) is pro­portional to control input signal e₁.

0067In this manner, output voltage V₁ of inverter INV-1 is controlled by a frequency twice the PWM control carrier frequency.

0068The three remaining PWM inverters INV-2 to INV-4 are similarly controlled. In this case, carrier signals X₂ to X₄ have a phase difference of 45° (electrical angle) from each other (Y₂ to Y₄ are inverted values of X₂ to X₄, respectively). As a result, a sum V₁ + V₂ + V₃ + V₄ of voltages generated through output transfor­mers TR₁ to TR₄ becomes a multiplexed voltage, and has a voltage waveform controlled by 8·f_C with respect to PWM control carrier frequency f_C. A lower-order harmonic component upon PWM control is canceled, and only a higher-order harmonic component appears in output voltage V₁ + V₂ + V₃ + V₄. This higher-order harmonic component can be easily canceled by a filter, e.g., a reactor.

0069Direct-connection inverter INV-5 has an arrangement as shown in Fig. 11. Fig. 11 shows a circuit corresponding to a U phase of a three-phase Graetz con­nection, which comprises self-extinction elements S₅₁ and S₅₂ and free-wheeling diodes D₅₁ and D₅₂. A neutral line of a load is connected to an intermediate line of DC voltage V_d.

0070Figs. 12A and 12B are timing charts for explaining the PWM control operation of inverter INV-5. Carrier signal X₅ and control input signal e₅ are compared to generate gate signal g₅ for elements S₅₁ and S₅₂. When e₅ ≧ X₅, g₅ = 1, and S₅₁: ON, S₅₂: OFF When e₅ &lt; X₅, g₁ = -1, and S₅₁: OFF, S₅₂: ON

0071Output voltage V₅ of inverter INV-5 is given by: V₅ = +(V_d/2) when S₅₁ is ON V₅ = -(V_d/2) when S₅₂ is ON

0072Average value <o ostyle="single">V</o>5 is proportional to control input signal e₅. In this manner, output voltage V₅ of direct-­connection inverter INV-5 is controlled by the PWM control carrier frequency.

0073The control operation of the converter of the pre­sent invention will be described below with reference to Figs. 7 to 12.

0074Load current I_U is controlled by direct-connection inverter INV-5.

0075More specifically, load current I_U is detected by current detector CT₅, and the detected current is input to comparator C₅. Comparator C₅ compares load current instruction value I_U* with detected value I_U to obtain deviation ε_U = I_U* - I_U. Deviation ε_U is amplified by the next current control/compensation circuit G_L, and the amplified deviation is supplied, as input signal e₅, to PWM controller PWM₅ of direct-connection inverter INV-5 through adder A6. Direct-connection inverter INV-5 generates voltage <o ostyle="single">V</o>₅ proportional to input signal e₅, as described above.

0076When I_U* > I_U, deviation ε_U has a positive value, and increases output voltage V₅ of direct-connection inverter INV-5, thus increasing load current I_U so to establish I_U = I_U*. In contrast to this, when I_U* < I_U, deviation ε_U has a negative value, and decreases output voltage V₅, thus decreasing load current I_U. Thus, the load current is balanced when I_U = I_U*. When current instruction value I_U* is sinusoidally changed, actual current I_U is sinusoidally controlled to follow the change.

0077Direct-connection inverter INV-5 basically genera­tes voltage R_L·I_U corresponding to a voltage drop caused by resistance R_L of the load. Thus, added A₆ receives a signal obtained by multiplying load resistance R_L with current instruction value I_U*. Input signal e₅ to PWM₅ is given by: e5 = R_L·I_U* + ε_U·G_L      (12A)

0078More specifically, direct-connection inverter INV-5 always generates a voltage corresponding to a voltage drop caused by the load resistance, and additionally generates a voltage according to current deviation ε_U to control the load current. Since a voltage drop caused by the resistance dies not depend on output frequency f₀, when f₀ = 0, multiplexed inverters INV-1 to INV-4 having output transformers cannot generate a voltage, and a voltage must be supplied from the direct-­connection inverter.

0079Multiplexed PWM inverters INV-1 to INV-4 generate voltages proportional to output frequency f₀. More specifically, for a motor load, these inver­ters supply a voltage corresponding to a sum of a voltage of counter electromotive force V_CU and voltage drop jωL_L·I_U caused by inductance L_L of the load.

0080In Fig. 8, compensation signal V_CU* + jωL_L·I_U* obtained by counter electromotive force compensation value V_CU* and load current instruction value I_U* is input to operational amplifier K_N and adder A₂ to supply input signal e₁ to PWM controller PWM₁ of inverter INV-1. Similarly, input signals e2 to e4 are input to controllers PWM₂ to PWM₄. For the sake of simplicity, if other signals are assumed to be zero, e₁ = e₂ = e₃ = e₄ = (V_CU* + jωL_L·I_U*)/n      (12B) = 2π·f₀, n = 4 where <u style="single">n</u> is the number of stages of inverters consti­tuting a multiplexed inverter, and n = 4 in this case.

0081As a result, output voltage V₁ + V₂ + V₃ + V₄ of the multiplexed inverter is proportional to the sum of input signals e₁ to e₄, and a voltage proportional to V_CU* + jωL_LI_U* is generated. Counter electromotive force V_CU* is proportional to output frequency f₀, and second term jωL_LI_U* is also proportional to output fre­quency f₀ second term jωL_LI_U*. Therefore, output voltage V₁ + V₂ + V₃ + V₄ of the multiplexed inverter is proportional to output frequency f₀, and effective values of exciting currents IOU₁ to IOU₄ of output transformers TR₁ to TR₄ become almost constant.

0082In practice, however, a DC bias or the like may be slightly applied to the output transformers due to drift of the controller or unbalanced switching charac­teristics of elements. When a DC bias voltage is applied to a transformer, the transformer gradually undergoes polarized magnetization, and finally, its iron core is saturated, so that an excessive exciting current flows through the transformer. As a result, not only the transformer is burnt, but also elements constituting the inverter are destroyed by an overcurrent.

0083In the converter of the present invention, four inverters INV-1 to INV-4 constituting the multiplexed PWM inverter respectively control primary currents I_U1 to I_U4 of their output transformers. The control opera­tion of primary current I_U1 of inverter INV-1 will be described below.

0084Current detector CT₁ detects primary current I_U1 of transformer TR₁, and the detected current is input to comparator C₁. Comparator C₁ compares primary current instruction value I_U1* with current detection value I_U1 to obtain deviation ε₁ = I_U1* - I_U1. Deviation ε₁ is amplified by the next current control/compensation cir­cuit G₁, and is then input to adder A₂ to supply input signal e₁ of PWM controller PWM₁. Thus, e₁ given by equation (12B) is rewritten as: e₁ = (V_CU* + jωL_L·I_U*)/4 + ε₁·G₁      (13)

0085Primary current instruction value I_U1* is given by the sum of load current instruction value I_U* and instruction value I_OU1* of exciting current I_OU1 of transformer TR₁. If the four transformers have the same capacity and rated value, I_OU1* = I_OU2* = I_OU3* = I_OU4*. Therefore, I_U1* = I_U* + I_OU* = I_U2* = I_U3* = I_U4*      (14)

0086When I_U1* > I_U1, deviation ε₁ has a positive value, and increases input signal e₁ given by equation (13), thus increasing primary current I_U1 to establish I_U1 = I_U1*. In this case, load current I_U is controlled by direct-connection inverter INV-5 to I_U ≒ I_U*. As a result, exciting current I_OU1 of transformer TR₁ is controlled to coincide with instruction value I_OU* described above.

0087In contrast to this, when I_U1* &lt; I_U1, deviation ε₁ has a negative value, and decreases input signal e₁, i.e., the output voltage of inverter INV-1 to satisfy I_U1 ≒ I_U1*.

0088In this manner, primary current I_U1 of transformer TR₁ is controlled to coincide with instruction value I_U1*, and exciting current I_OU1 of transformer TR₁ consequently coincides with instruction value I_OU*. Even when a DC bias is applied to transformer TR₁ due to, e.g., unbalanced characteristics of elements, exciting current I_OU1 ≒ I_OU* is established, and the control system is operated to finally cancel the DC bias.

0089Other inverters INV-2a to INV-4 are similarly controlled.

0090As described above, the multiplexed PWM inverter must basically generate a sum voltage of counter electromotive force V_CU and voltage drop jωL_LI_U caused by inductance L_L of the load. Therefore, exciting currents I_OU1 to I_OU4 of transformers TR₁ to TR₄ must be adjusted to corresponding values. When a mutual induc­tance upon a rated output of each transformer is repre­sented by M, instruction value I_OU* of the exciting current input to adder A₁ is given by:<maths id="math0003" num=""><img file="EP0390184A2_D0003.tif" /></maths>

0091In this manner, the multiplexed PWM inverter generates a voltage corresponding to counter electromo­tive force V_CU and voltage drop jωL_LI_U caused by the inductance of the load, and the direct-connection inverter generates a voltage corresponding to a voltage drop caused by the resistance of the load and a tran­sient voltage necessary for load current control, thereby controlling load current I_U to coincide with its instruction value I_U*. In this case, even when a DC bias voltage is applied to the output transformer due to unbalanced characteristics of elements, since primary current control (in other words, exciting current control) of the transformer is executed, no DC polarized magnetization occurs. Thus, no overcurrent caused by polarized magnetization flows through the transformer, and elements constituting the inverter can be prevented from being destroyed.

0092However, the power converter shown in Fig. 7 poses the following problems.

0093That is, output voltage V₁ + V₂ + V₃ + V₄ generated by the multiplexed PWM inverter can be a sinusoidal voltage free from distortion. However, since the direct-connection inverter cannot execute multiplexed PWM control, the waveform distortion of its output voltage V₅ is increased. In particular, in a region of low output frequencies f₀, a current to be supplied to AC load LOAD includes many pulsation components, and causes torque pulsation for a motor load.

0094If the carrier frequency of direct-connection inverter INV-5 can be increased up to about 8 times those of multiplexed inverters INV-1 to INV-4, the above-mentioned voltage distortion can be decreased, and a balanced power converter as a whole can be realized. However, the switching frequency of a GTO (gate turn-off thyristor) as a typical self-extinction element is at most about 500 Hz, and it is impossible to increase only the carrier frequency of direct-connection inverter INV-5.

0095As the power converter has higher-voltage, larger-­capacity specifications, elements constituting an inverter must be serially connected. In order to reduce a switching loss and snubber circuit loss, a carrier frequency is preferably reduced as much as possible during an operation.

0096Thus, the present invention also provides a power converter which can always supply a sinusoidal current free from distortion to an AC load regardless of an out­put frequency level without increasing a carrier fre­quency of a direct-connection inverter, and a method of controlling the same.

0097Fig. 13 is a circuit diagram showing another embodiment of a controller of the converter according to the present invention.

0098In Fig. 13, reference numerals SW₁ and SW₂ denote switch circuits; C₁ to C₆, comparators; A₁ to A₈, adders; SH, a Schmitt circuit; G₁ to G₄ and G_L, current control/compensation circuits; K_N1, K_N2, and K_N, opera­tional amplifiers; PWM₁ to PWM₅, PWM controllers; and TRG₁ and TRG₂, carrier generators.

0099A difference between the embodiment shown in Fig. 13 and the controller shown in Fig. 8 is as follows.

0100In the controller shown in Fig. 8, four PWM inverters INV-1 to INV-4 constituting the multiplexed inverter control primary currents I_U1 to I_U4 of their output transformers. In this embodiment, these inver­ters control exciting currents I_OU1 to I_OU4 of the transformers. If the four transformers are assumed to have the same capacity and rated value, instruction value IOU* of its exciting current is given by equation (15). Exciting currents I_OU1 to I_OU4 of the transformers can be obtained by calculations from the detection values of primary currents I_U1 to I_U4 of the transformers, and the detection value of load current I_U as follows:<maths id="math0004" num=""><img file="EP0390184A2_D0004.tif" /></maths>

0101Comparator C₁ compares exciting current detection value I_OU1 and its instruction value I_OU*, and control/compensation circuit G₁ amplifies deviation ε₁ = IOU* - I_OU1. Inverter INV-1 is controlled under PWM control to satisfy I_OU1 ≒ I_OU*. Exciting currents I_OU2 to I_OU4 of other transformers are also controlled to coincide with instruction value I_OU*. As a result, polarized magnetization of the output transformer can be eliminated, and element destruction can be prevented as in the controller shown in Fig. 8.

0102The next difference is as follows. That is, when output frequency f₀ of the inverter is lower than mini­mum frequency f_min (e.g., f_min = 1 Hz is selected as a minimum frequency at which the output transformer can be operated), a fundamental wave (output frequency com­ponent) voltage generated from the multiplexed PWM inverter is controlled to be zero.

0103More specifically, comparator C₆ compares output frequency f₀ and setup value f_min of the minimum fre­quency, and the Schmitt circuit switches the contacts of switch circuits SW₁ and SW₂. When f₀ < f_min, circuits SW₁ and SW₂ are connected to contacts <u style="single">a</u>. Thus, exciting current instruction value I_OU* of the transformer becomes zero, and a voltage of the fundamental wave generated from the multiplexed PWM inverter also becomes zero. When f₀ > f_min, switch circuits SW₁ and SW₂ are switched to contacts <u style="single">b</u>, and the multiplexed PWM inverter generates counter electromotive force V_CU of the load and voltage drop jωL_LI_U caused by the inductance. Meanwhile, when f₀ < f_min, direct-connection PWM inverter INV-5 bears a total of counter electromotive force V_CU of the load, voltage drop jωL_LI_U caused by the inductance, and voltage drop R_L·I_U caused by the resistance, and controls load current I_U. In this manner, in a region of f₀ < f_min, output transformers TR₁ to TR₄ are controlled not to flow wasteful exciting currents.

0104In the embodiment shown in Fig. 13, compensation control is executed so that a harmonic voltage generated by direct-connection inverter INV-5 is canceled by multiplexed inverters INV-1 to INV-4. More specifi­cally, pulsation component ΔV₅ of voltage V₅ generated from inverter INV-5 is predicted from PWM control input signal e₅ of direct-connection inverter INV-5 and output signal g₅ of PWM controller PWM₅, and multiplexed PWM inverters INV-1 to INV-4 are caused to generate its inverted value -ΔV₅.

0105As has been described with reference to Figs. 12A and 12B for explaining the PWM control operation of the direct-connection inverter, carrier signal X₅ and control input signal e₅ are compared to generate gate signal g₅. Output voltage V₅ is determined as follows. When e₅ ≧ X₅, g₅ = 1, and S₅₁: ON, S₅₂: OFF When e₅ &lt; X5, g₅ = -1, and S₅₁: OFF, S₅₂: ON Output voltage V₅ is separated into fundamental wave component <o ostyle="single">V</o>₅ and harmonic component ΔV₅. More specifi­cally, ΔV₅ = V₅ - <o ostyle="single">V</o>₅. This can be converted to the level of PWM control input signal e₅ as follows.

0106Peak values of carrier signals X₁ to X₅ are repre­sented by E_max, and gate signal g₅ is multiplied with E_max by operational amplifier K_H. Output g₅·E_max of amplifier K_H is a voltage proportional to output voltage V₅, and input signal e₅ proportional to V₅ is subtracted from the output: Δe₅ = g₅·E_max - e₅∝ΔV₅      (17) The output from adder A₇ corresponds to an inverted value of a value given by equation (17), and is multiplied with (1/2n) by operational amplifier K_N2 to obtain compensation control signal H_OU*. H_OU* = -(g₅·E_max - e₅)/(2n)      (18) where <u style="single">n</u> is the number of series-connected multiplexed inverters, and n = 4 in this case. Compensation signal H_OU* is input to controllers PWM₁ to PWM₄ through adder A₅ and adders A₁ to A₄. In the following description, other signals are assumed to be zero for the sake of simplicity.

0107Figs. 14A to 14N are timing charts for explaining the operation of the multiplexed PWM inverter when e₁ = e₂ = e₃ = e₄ = H_OU*.

0108Reference numerals X₁ to X₄ and Y₁ to Y₄ denote PWM control carrier signals. Signals Y₁ to Y₄ are inverted signals of signals X₁ to X₄. Signals X₁, X₂, X₃, and X₄ have a phase difference of 45° from each other.

0109The principal circuit arrangement of inverter INV-1 is as shown in Fig. 9 described above.

0110Input control signal H_OU* is compared with carrier (triangular wave) X₁ to generate gate signal g₁, thereby firing-control elements S₁₁ and S₁₂ of inverter INV-1. That is, When H_OU* ≧ X₁, g₁ = "1", and S₁₁: ON, S₁₂: OFF When H_OU* &lt; X₁, g₁ = "0", and S₁₁: OFF, S₁₂: ON H_OU* and Y₁ are compared with each other to generate gate signal g₁′, thereby firing-control elements S₁₃ and S₁₄. That is, When H_OU* ≧ Y₁, g₁′ = "1", and S₁₃: OFF, S₁₄: ON When H_OU* &lt; Y₁, g₁′ = "0", and S₁₃: ON, S₁₄: OFF As a result, output voltage V₁ of inverter INV-1 becomes V₁ = +V_d when elements S₁₁ and S₁₄ are ON; V₁ = -V_d when S₁₂ and S₁₃ are ON; and V₁ = 0 in other modes. Thus, output signal V₁ has a waveform, as shown in Fig. 14D.

0111Similarly, input control signal H_OU* is compared with triangular waves X₂ and Y₂ to generate gate signals g₂ and g₂′ for inverter INV-2. H_OU* is compared with X₃ and Y₃ to generate gate signals g₃ and g₃′ for inverter INV-3. H_OU* is compared with X₄ and Y₄ to generate gate signals g₄ and g₄′ for inverter INV-4.

0112As a result, output voltages V₂, V₃, and V₄ of inverters INV-2 to INV-4 are as shown in Figs. 14G, 14J, and 14M.

0113A sum of output voltages V₁ to V₄ of inverters INV-1 to INV-4 has a waveform as shown in Fig. 14N, and is proportional to input signal H_OU*. Sum voltage V₁ + V₂ + V₃ + V₄ = -ΔV₅ cancels harmonic component ΔV₅ of voltage V₅ generated by direct-connection inverter INV-5.

0114When compensation control signal H_OU* expressed by equation (18) is input as PWM control input signal e₁ = e₂ = e₃ = e₄ = H_OU* of inverters INV-1 to INV-4, average values of output voltages V₁ to V₄ of the inverters are proportional to corresponding input signals e₁ to e₄, and their sum voltage V₁ + V₂ + V₃ + V₄ is proportional to 4·H_OU*.<maths id="math0005" num=""><img file="EP0390184A2_D0005.tif" /></maths> When equations (17) and (18) are substituted in equation (19) for n = 4, equation (19) is rewritten as:<maths id="math0006" num=""><img file="EP0390184A2_D0006.tif" /></maths>

0115In this manner, harmonic voltage V₅ generated by direct-connection inverter INV-5 can be compensation-­controlled to be canceled by multiplexed inverters INV-1 to INV-4. As a result, the carrier frequency of the direct-connection inverter can be lowered, and a switching loss and a snubber circuit loss can be reduced.

0116Figs. 15A to 15C show computer simulation results of the converter of the present invention, and illustrate three-phase output currents I_U, I_V, and I_W, U-phase output voltage V_U, counter electromotive force V_CU, output voltage V₁ of one multiplexed PWM inverter, and output voltage V₄ of the direct-connection inverter when an output frequency is 30 Hz. Note that the load comprises an AC motor, the multiplexed inverter has a three-stage structure, the PWM control carrier frequency of the multiplexed inverter is 350 Hz, and the PWM control carrier frequency of the direct-connection inverter is 120 Hz. Although the carrier frequency of the direct-connection inverter is lowered, load currents I_U, I_V, and I_W of its component are free from pulsation, and are controlled to be sinusoidal currents free from distortion.

0117In this manner, load current I_U is controlled according to its instruction value I_U*. In this case, one problem is posed. More specifically, when PWM control carrier frequency f_CA of the multiplexed inverter is set to be the same as PWM control carrier frequency f_CB of the direct-connection inverter, voltages generated by inverters constituting the multiplexed inverter are deviated to the positive or negative side when a harmonic voltage component generated by the direct-connection inverter is to be compensated for.

0118Figs. 16A to 16H are timing charts when compen­sation voltage H_OU* as a PWM control input signal is perfectly synchronous with carrier signals X₁ to X₄ and Y₁ to Y₄ of the multiplexed inverter. Since compen­sation voltage H_OU* is synchronous with carrier signal X5 of the direct-connection inverter, the above-­mentioned state occurs when carrier signals X₁ to X₄ or Y₁ to Y₄ have the same frequency and PWM control input signal e₅ of direct-connection inverter INV-5 is constant.

0119H_OU* and X₁ are compared with each other to generate gate signal g₁, and H_OU* and Y₁ are compared with each other to generate gate signal g₁′, as described above. As a result, output voltage V₁ of inverter INV-1 becomes as shown in Fig. 16D. Similarly, output voltages V₂ to V₄ of inverters INV-2 to INV-4 become as shown in Figs. 16E to 16G, and output voltage V₁ + V₂ + V₃ + V₄ of the multiplexed inverter has a waveform shown in Fig. 16H. This voltage is propor­tional to compensation voltage H_OU*, and corresponds to an inverted value of harmonic component V₅ of output voltage V₅ of direct-connection inverter INV-5.

0120In this case, in output voltages V₁ to V₄ of the inverters constituting the multiplexed inverter, integrated values of positive voltage components do not coincide with those of negative voltage components. For example, in output voltage V₁, negative voltage com­ponents are larger than positive voltage components. The same applies to V₂. Contrary to this, in V₃ and V₄′ positive voltage components are larger than negative voltage components. In the multiplexed inverter as a whole, V₁ + V₂ + V₃ + V₄ is balanced between the positive and negative sides.

0121When carrier signals X₁ to X₄ (Y₁ to Y₄) are synchronous with compensation voltage H_OU*, this state permanently continues. For example, this is equivalent to a state that a negative DC bias voltage is applied to output transformer TR1, and DC polarized magnetization occurs in the iron core. The same applies to other out­put transformers TR₂ to TR₄.

0122DC polarized magnetization is finally corrected since inverters INV-1 to INV-4 control primary or exciting currents of their transformers, as described above. However, it is not preferable to create a cause of DC polarized magnetization, and this may disturb current control. When the DC polarized magnetization is corrected, a compensation control operation of a har­monic voltage component of the direct-connection inverter is disturbed, and proper compensation may become impossible.

0123In the controller shown in Fig. 13, therefore, carrier generators TRG₁ and TRG₂ respectively generate carrier signals X₁ to X₄ (Y₁ to Y₄) to be supplied to PWM controllers PWM₁ to PWM4 of the multiplexed inverter, and carrier signal X5 to be supplied to PWM controller PWM5 of the direct-connection inverter so as to have different frequencies f_CA and f_CB.

0124In this manner, for example, voltage V₁ generated by one multiplexed inverter INV-1 may instantaneously have a large positive or negative voltage level. However, in terms of an average value, voltage V₁ is proportional to input signal H_OU* (assuming that e₁ = H_OU*), and will never cause DC polarized magnetization of output transformer TR₁. The same applies to other output transformers TR₂ to TR₄.

0125As a result, primary current control (or excitation current control) of the output transformers by the corresponding inverters will not be disturbed, and a current free from waveform distortion can be supplied. A harmonic voltage component generated by the direct-­connection inverter can be accurately compensated for by the multiplexed inverter.

0126The inverter output corresponding to one phase (U phase) has been described above. However, the same applies to V and W phases.

0127Compensation signal H_OU* for canceling harmonic voltage component ΔV₅ generated by direct-connection inverter INV-5 is directly obtained from gate signal g₅ and PWM control input signal e₅, thereby canceling all the harmonic components. Alternatively, signal g₅ may be frequency-analyzed to extract and compensate for a specific harmonic component therefrom.

0128As described above, according to the power con­verter of the present invention, DC polarized magnetiza­tion of the output transformers of the multiplexed PWM inverter can be prevented, and elements constituting the inverter can be prevented from being destroyed by an overcurrent. At the same time, the load current can be accurately controlled by the direct-connection inverter, and a sinusoidal current free from a waveform distortion can be supplied to an AC load.