Transistor Inverter Motor Drive Having Voltage Boost at Low Speeds

This invention relates to static electric inverters and in particular to transistor bridge inverter drives of the pulse width modulation (PWM) type for driving an electrical motor.

A PWM tranistor bridge inverter drive is disclosed in my copending U.S. application Ser. No. 558,497 filed Mar. 14, 1975, entitled Transistor Bridge Inverter Motor Drive Having Reduced Harmonics and having the same assignee as this invention, which controls frequency and magnitude of stator voltage applied to an induction motor so as to maintain constant flux in the motor over a wide speed range and to substantially reduce harmonics in the current supplied to the motor. The inverter drive disclosed in my copending application controls the voltage-to-frequency (volt/hertz) ratio so as to maintain constant flux in the motor and can be adjusted so that motor output torque is constant over the entire speed range or can alternatively be adjusted so that the motor provides substantially constant rated torque below motor base speed and rated horsepower above base speed. The inverter drive of my copending application integrates the output voltage of the bridge inverter with respect to time over a fixed fraction of a cycle and compares the integral to a fixed level to effect constant flux in the motor and also has adjustable means to selectively vary the volt/hertz ratio and thus regulate the slope of the motor terminal voltage versus frequency characteristic.

The inverter drive disclosed in my copending application maintains the volt/hertz ratio constant and thus should theoretically maintain the magnetic flux (whose magnitude is proportional to the rate of change of voltage) in the motor constant over the entire speed range. However, the magnetic flux in a motor decreases at low motor speeds even though the volt/hertz ratio remains constant. The effective resistance drop in a motor is substantially constant in magnitude but is so much lower than the leakage reactance drop at speeds above motor base speed that it can be considered negligible. However, at low motor speeds the resistance voltage drop across the copper approaches the magnitude of the reactance voltage drop across the magnetizing inductance of the stator winding, thereby decreasing the magnitude of magnetizing current (and the magnetic flux density which is aproximately proportional to ampere turns, neglecting saturation) at low frequencies.

It is an object of the invention to provide an improved transistor bridge inverter drive for an electric motor which provides constant volts per hertz and can be adjusted to boost stator voltage at low motor speeds.

It is a further object of this invention to provide such an improved transistor bridge inverter drive for an induction motor which is adjustable to selectively vary the slope of the low motor speed portion of the motor terminal voltage versus frequency characteristic and thus provide a voltage boost and is also selectively adjustable to vary the predetermined motor speed below which voltage boost occurs.

Another object is to provide an improved transistor bridge inverter device for an electric motor which increases the number of constant volt/hertz PWM pulses per cycle of fundamental voltage at low motor speeds to thereby provide a voltage boost.

These and other objects and advantages of the invention will be more readily apparent from the following detailed description when considered in conjunction with the accompanying drawings wherein:

FIG. 1 is a schematic block diagram of the inverter motor drive embodying the invention;

FIG. 2a illustrates the output characteristic of the inverter drive shown in FIG. 1 when adjusted to drive the motor with constant rated torque below motor base speed and rated horsepower above base speed, and FIG. 2b illustrates different terminal voltage versus frequency charcteristics to which the inverter drive shown in FIG. 1 can be adjusted as well as different slopes to which the low motor speed portion of the characteristics can be selectively adjusted to provide different amounts of voltage boost;

FIG. 3 illustrates pulses which exist at various points within the inverter motor drive of FIG. 1;

FIG. 4 illustrates typical reduced-harmonic output voltage PWM pulse patterns provided to the motor by the inverter drive of FIG. 1;

FIG. 5 shows the circuit diagram of the resettable integrator and of the first and second comparators of the inverter drive of FIG. 1;

FIG. 6 shows the circuit diagram of the voltage boost circuit including the second voltage controlled oscillator of the inverter drive of FIG. 1;

FIG. 7 shows the circuit diagram of the frequency switch of the inverter drive of FIG. 1 and pulses which appear at various points within the frequency switch; and

FIG. 8 shows the circuit diagram of the volt-second switch and the volt-second latch of the inverter drive of FIG. 1.

A transistor inverter drive in accordance with the invention provides variable frequency and variable voltage to an electric motor and pulse width modulates a predermined number of harmonic-reduced, constant volt/hertz pulses per cycle of motor frequency to selectively vary motor terminal voltage so that constant torque can be effected over a wide speed range and increases the number of constant volt/hertz PWM pulses per motor frequency cycle at low motor speeds to provide a boost in motor terminal-voltage and thereby compensate for decrease in magnetic flux which would otherwise occur at low frequencies.

The preferred embodiment of bridge inverter drive has first and second oscillators which derive trains of first and second clock pulses that vary at different rates in proportion to an analog speed signal; a frequency switch which selects the higher frequency train of pulses; a three-phase generator which derives three-phase reference waves displaced 120.degree. each of whose period includes a predetermined number of first clock pulses; a volt/hertz integrator which derives ramp signals whose leading edges are of the output pulses from the frequency switch and which vary in magnitude as the time integral of the inverter output voltage; a comparator for deriving a control pulse when each ramp signal reaches a reference voltage; and means for controlling the fundamental output voltages of individual phases of the inverter bridge in accordance with respective reference waves and including pulse width modulation chopping means for controlling conduction of transistors of the bridge inverter by variable width PWM pulses whose leading edges are at the output pulses from the frequency switch and whose trailing edges are at the control pulses, thereby increasing the number of constant volt/hertz PWM pulses in each fundamental voltage period at lower motor speeds and providing a voltage boost in the motor terminal voltage versus frequency characteristic.

The inverter motor drive of my aforementioned copending application Ser. No. 558,497 can control the stator voltage to frequency (volt/hertz) ratio so as to maintain substantially constant flux in the motor and thus provide substantially constant torque over a wide speed range as shown, for example, by the linear, constant torque, motor terminal voltage versus frequency characteristic EF in FIG. 2b. It can also be adjusted to change the volt/second ratio and thereby vary the slope of the terminal voltage versus frequency characteristic to that shown as GKHF wherein approximately constant torque is obtained from a low motor speed (such as point K) to motor base speed (as shown in FIG. 2a) and rated horsepower is obtained above base speed. The present invention permits the slope of the low frequency portion of each voltage versus frequency characteristic to be selectively varied, for example, the slope of the low speed portion GK of characteristic GKHF may be selectively adjusted to G'K or to G"K shown in dotted lines to provide a voltage boost which compensates for decrease in magnetizing current and in motor torque which would otherwise occur at low motor frequencies. The present invention also permits selection of the "breakpoint" frequency, such as that at point K, below which voltage boost is provided.

The schematic block diagram of the present invention shown in FIG. 1 is generally similar to that of the inverter motor drive of my aforementioned application with the principal difference being that a second voltage controlled oscillator VCO2 and a frequency switch FS are added to provide a voltage boost (such as shown by curve portion G'K) at low motor speeds. In the transistor bridge inverter drive disclosed in my copending application, power from a three-phase A.C. source may be rectified in a full wave rectifier BR to provide 300 volts unidirectional power to a bridge inverter INV having a pair of power transistors in each phase leg connected in series between the 300 volt positive supply bus + and the zero voltage reference bus -. Phase one leg of the inverter bridge may have transistors 1A and 1B connected in series with the node therebetween constituting phase one terminal .phi..sub.1 and also have feedback diodes 1AD and 1BD respectively connected in inverse polarity relation with 1A and 1B. Phase two may include transistors 2A and 2B connected in series between the + and - buses, and phase three of the inverter may similarly include transistors 3A and 3B connected in series between the + and - buses. Bridge inverter INV may supply voltage of variable frequency and magnitude to a three-phase induction motor M having star connected stator windings W1, W2, W3 connected respectively to the inverter phase output terminals .phi..sub.1, .phi..sub.2, .phi..sub.3.

A speed potentiometer SPEED POT which is set by the operator to provide desired motor speed derives an analog speed signal voltage which is converted by a voltage controlled oscillator VCO1 to a variable frequency train of clock pulses f1 shown in FIG. 3a whose frequency is proportional to the magnitude of the analog speed signal. The f1 clock brain determines the frequency of the variable width PWM power pulses conducted by transistors 1A through 3B and applied to the motor windings W1, W2, and W3 in the invertor motor drive of my aforementioned application and also the PWM pulse period is established by the interval between f1 clock pulses.

The f1 pulse train is converted by a two phase, divide-by-six frequency divider DIV into two timing pulse trains f1' /6 (FIG. 3a) and f1"/6 (FIG. 3b) that are 180.degree. phase displaced. The f1'/6 pulse train is the clock frequency for a three-phase generator GEN which derives a set of three-phase reference square waves A, B, C (See FIG. 3d) which are displaced 120.degree. and establish the fundamental frequency of the line-to-reference bus inverter output voltages such as V.sub.1.sub.- 0 and V.sub.2.sub.- 0. Each full cycle of fundamental voltage from inverter INV contains 36 pulse periods w shown in FIG. 4 each which is equal to the interval between successive f1 clock pulses in the drive disclosed in my aforementioned application, whereas in the present invention each motor frequency fundamental voltage period may include more than 36 PWM pulse periods and thus provide a voltage boost at low motor speeds.

A volt-second feedback control for providing constant flux in motor M includes a resettable integrator RI shown in FIG. 5 which senses the motor stator voltage at terminals .phi..sub.1, .phi..sub.2, .phi..sub.3 and integrates such voltage over a fixed fraction of a cycle. The motor terminal voltage signals at .phi..sub.1, .phi..sub.2, .phi..sub.3 are coupled to an OR circuit of diodes D22, D23, D24 whose output is applied to a voltage divider comprising resistors R60 and R61 and a volt-second adjustment potentiometer R59. A resistor R62 coupled to the junction of R60, R61 and a capacitor C14 form an integrator which derives ramp signals v.sub.c14 shown in FIGS. 3d and 5. The ramp pulses v.sub.c14 increase in magnitude as a function of the time integral of the inverter output voltage and are coupled through a follower operational amplifier OP3 to a first level comparator LC1 wherein they are applied to the noninverting input of an operational amplifier OP4 which has a first reference voltage applied to its inverting input. When ramp signal v.sub.c14 becomes equal to the first reference voltage, comparator LC1 generates a first control pulse HVT (shown in FIGS. 3g and 5) which triggers a volt-second latch VSL. When volt-second latch VSL is triggered, it initiates a longer pulse VT (shown in FIGS. 3i and 5) which begins at pulse HVT and is terminated by the succeeding f1 or f2 clock pulse. Pulse VT is fed back to integrator RI to reset it and thereby terminate ramp pulse v.sub.c14 and hold it reset until the succeeding f1 or f2 clock pulse. Pulse VT is also applied to a volt-second switch VSS where it establishes the a width of the PWM chopping pulses (see FIG. 4) which accomplish substantially constant volt-second ratio of stator voltage applied to motor M. Power is applied to motor M beginning at the leading edge of each a width PWM pulse (whose leading edge is synchronized to an f1 or f 2 clock pulse) and continues until the succeeding first control pulse HVT which terminates the PWM pulse. VOLT-SEC potentiometer R59 permits regulation of the slope of the ramp signals v.sub.c14 and thus adjusts the magnetic flux level in the motor M and the slope of the motor terminal voltage versus frequency characteristic output from bridge inverter INV, for example, the slope of the constant torque characteristic EF shown in FIG. 2B wherein the torque is approximately constant from a low speed up to 120 Hz. VOLT-SEC potentiometer R59 also permits adjustment of the bridge inverter drive of my copending application to provide the characteristic GKHF wherein torque is approximately constant along the portion KH over a speed range from a low speed up to 60 Hz and the motor M provides constant horsepower above base speed 1.0.

The ramp signals v.sub.c14 from integrator RI are also applied to the noninverting input of an operation amplifier OP5 of second level comparator LC2 to obtain reduction in fifth and seventh harmonics in the current supplied to motor M. When the integrated voltage pulses v.sub.c14 reach the level of a second reference voltage which is applied to the inverting input of operational amplifier OP5, second comparator LC2 initiates a second control pulse LVT (shown in FIGS. 3h and 5) which is coupled to the volt-second switch VSS and establishes the b width of PWM pulses (see FIG. 4) in the inverter output voltages such as V.sub.1.sub.- 0 and V.sub.2.sub.- 0. A HARMONIC adjustment potentiometer R77 permits setting of the desired percent reduction in fifth and seventh harmonics by varying the magnitude of the second reference voltage to thereby regulate the width of the b width PWM pulses relative to that of the a width PWM pulses (which controls the steps in the twelve-step inverter output voltages).

The f1"/6 timing pulse train from frequency divider DIV is coupled to a sixty degree interval generator IG which also receives the three-phase references waves A, B, C from the three-phase generator GEN and derives square wave pulses H1 (see FIG. 3e) H2, H3 of 60.degree. duration which determines the steps in the corresponding twelve-step, phase to ground inverter output voltages V.sub.1.sub.- 0, V.sub.2.sub.- 0, V.sub.3.sub.- 0 from inverter INV. The H1, H2, H3 step-controlling pulses correspond respectively to the A, B, C reference waves which establish the fundamental output voltage frequency from phases .phi..sub.1, .phi..sub.2 and .phi..sub.3 of bridge inverter INV. Step-controlling pulses such as H1 span 60.degree. at the zero crossing transitions between (A and A) of the corresponding phase reference wave.

The H1, H2, H3 signals are coupled to volt-second switch VSS which also receives the LVT and HVT pulses and selects the a or b width of PWM pulses in accordance with signals H1, H2, H3. The output from volt-second switch VSS comprises three trains of variable width, duty-cycle regulating pulses VT1 (see FIG. 3k), VT2, VT3 which are associated respectively with phases .phi..sub.1, .phi..sub.2, .phi..sub.3 of inverter bridge INV. Volt-second switch VSS provides the narrow b width of pulses in the VT1 pulse train (regulated by second control pulse LVT) when H1 is present and provides the wider a width pulses of the VT1 pulse train (regulated by first control pulse HVT) when H1 is logic 0.

The three-phase reference waves A, B, C from GEN and the trains of reduced-harmonic, variable-width duty cycle regulating pulses VT1, VT2, VT3 from volt-second switch VSS are combined in a modulation logic current ML to form three-phase pulse patterns 1a (see FIG. 3m), 2a, 3a which respectively control chopping of current by transistors 1A, 2A and 3A of bridge inverter INV and also form three-pulse patterns 1b, 2b, 3b which are the complements of 1a, 2a, 3a and control chopping of motor current by transistors 1B, 2B, 3B respectively. During the 180.degree. reference wave positive half cycle A, the pulse pattern 1a for phase one of the inverter follows the complement VT1 of the corresponding duty cycle regulating pulses, and during the 180.degree. negative half cycle A the pulse pattern 1a follows the corresponding duty cycle regulating pulse pattern VT1. The portion of modulation logic ML for phase 1 of inverter INV also derives 60.degree. width pulses 1A* (see FIG. 3l) and 1B* and superimposes them on the VT1 train of duty cycle regulating pulses so that the pulse train 1a which controls chopping by transistor 1A follows VT1 during the positive half cycle A of the corresponding reference wave with a 60.degree. maximum duty cycle pulse at the middle thereof (see FIG. 3m) and follows VT1 during the negative half cycle A with pulses deleted for 60.degree. during the midportion thereof.

The inverter motor drive of the present invention schematically illustrated in FIG. 1 provides a voltage boost at low motor speeds to compensate for decrease in magnetizing current and in motor torque that would occur at low frequencies in the absence of such voltage boost. The analog speed signal voltage set on the SPEED POT potentiometer is applied to a voltage boost circuit VB shown in FIG. 6 which includes a second voltage controlled oscillator VCO2 that is similar in function to oscillator VCO1 and generates a train of clock pulses f2 whose frequency varies as a different function of the speed signal than clock pulse train f1.

The general purpose of the voltage boost circuit VB is to apply extra PWM pulses to motor M at low speeds by substituting f2 for f1. The inverter drive of my copending application may apply a maximum of 36 PWM pulses per fundamental voltage cycle, whereas the inverter drive of the present invention may apply more than 36 pulses per fundamental cycle to motor M and each PWM pulse still has the same volt-second level. A frequency switch FS receives both the f1 and f2 clock pulse trains and chooses the higher frequency train as the PWM pulse rate.

The analog speed signal from the SPEED POT potentiometer is coupled through a slope adjustment potentiometer R19 to the base of transistor Q5 of relaxation oscillator VCO2 which converts the speed signal to a pulse train f2. Transistor Q5 converts the speed signal voltage to a constant charging current flowing into a timing capacitor C2 through breakpoint adjustment potentiometer R20 and collector resistance R23. Breakpoint adjustment potentiometer R20 establishes the predetermined motor speed below which a voltage boost is applied to motor M (such as illustrated at point K on characteristic GKHF in FIG. 2), and such predetermined motor speed is established by the frequency at which the voltage versus frequency characteristics of oscillators VCO1 and VCO2 cross. Variations in magnitude of the speed signal voltage changes the charging rate of timing capacitor C2. The voltage across timing capacitor C2 is coupled to the emitter of a transistor Q6 whose base is referenced to a voltage divider R24, R25, R26 so that Q6 functions as a differential switch. When the voltage across timing capacitor C2 reaches a predetermined magnitude which forward biases Q6, it conducts and develops a voltage drop across resistance R27 which forward biases transistor Q7 and turns it on. Conduction by Q7 lowers the reference voltage at the base of Q6 so that Q6 and Q7 are latched into conduction by positive feedback. Capacitor C2 is discharged through conducting transistor Q7 to a level which is insufficient to forward bias Q6 and Q7, whereupon oscillator VCO2 returns to its quiescent state. Conduction by Q7 pulls its collector to ground and generates a negative f2 clock pulse through diode D13 on the f2 lead to the frequency switch.

The frequency switch FS regulates whether clock pulses f1 or f2 determine the rate of PWM pulses applied to motor M. Frequency switch FS selects the higher frequency train f1 or f2, and since oscillators VCO1 and VCO2 vary in frequency with speed signal voltage at a predetermined but different rate, a crossover point, or breakpoint (such as point K), exists at which the frequencies of f1 and f2 are switched.

The range of possible boost of motor terminal voltage which can be set by slope adjustment potentiometer R19 is illustrated in FIG. 2. When no voltage boost is provided, the motor terminal voltage versus frequency characteristic such as GKHF passes through the origin. When maximum voltage boost is selected on R19, the low voltage portion assumes zero slope such as G"K wherein the motor terminal voltage is approximately constant up to the breakpoint frequency. One end of slope adjustment potentiometer R19 is coupled to the junction of two resistances R21 and R22 (see FIG. 6) connected in series to form a voltage divider across a voltage source +18V, and potentiometer R19 can be adjusted so that the base of transistor Q5 receives the unattenuated speed signal (when the slider of R19 is to the extreme left position) or an approximately fixed voltage from divider R21, R22 (when the slider is at the extreme right position). The low frequency portion (such as GK) of terminal voltage versus frequency characteristic (such as GKHF) shown in FIG. 2b has the highest slope so that the characteristic passes through the origin and no voltage boost is provided when oscillator VCO2 receives the unattenuated speed signal. The amount of voltage boost is progressively increased as the slider of R19 is moved to the right, and the low frequency portion of the characteristic may assume positions such as G'K and G"K shown in dotted lines. Maximum boost is provided and the characteristics assumes a horizontal, or zero slope when the slider R19 is at its extreme right position so the base of Q19 receives a fixed voltage from voltage divider R21, R22.

Adjustment of the breakpoint (such as point K) in the terminal voltage versus frequency characteristic (below which voltage boost occurs) is accomplished by breakpoint adjustment potentiometer R20 which regulates the charging current to timing capacitor C2 to thereby adjust the frequency of oscillator VCO2 (for a given magnitude of speed signal from SPEED POT). The adjustments of the slope and breakpoint potentiometers are interdependent. The slope setting potentiometer R19 should be adjusted first by comparing the change in frequency of clock pulses f2 with the change of frequency of pulses f1, and the breakpoint can then be adjusted at R20 by making f2 equal to f1 at the breakpoint frequency. Characteristic LMN illustrates the condition where volt-second potentiometer R59 is set to provide the highest slope of motor-terminal voltage versus frequency characteristic and slope adjusting potentiometer R19 is set to provide maximum voltage boost at low motor speeds.

Frequency switch FS shown in FIG. 7 determines whether the frequency of clock pulses f1 or f2 is higher and includes first and second JK flip-flops FF1 and FF2 which are respectively clocked by pulses f1 and f2 and reset by the other pulse train. Both FF1 and FF2 have logic 1 and logic 0 voltages applied to their J and K inputs respectively. Consequently, the Q output of FF1 goes to logic 1 on the trailing edge of each f1 pulse and goes to logic 0 on the leading edge of the succeeding f2 pulse. Similarly, the Q output of FF2 goes to logic 1 on the trailing edge of each f2 pulse and goes to logic 0 on the leading edge of the succeeding f1 pulse. The Q outputs of FF1 and FF2 are respectively applied to the J and K inputs of a JK flip-flop FF3. The f1 and f2 pulses are inputs to a NAND gate 13b whose output is coupled to the clock input of FF3 so that FF3 can be clocked by either an f1 or f2 pulse. The Q outputs of FF1 and FF2 are inputs to a NAND gate 13a whose output is also coupled to the clock input of FF3 so that the output of 13a goes to logic 0 and prevents clocking of FF1 when the outputs of FF1 and FF2 are both logic 1, which occurs during transistion between states.

FIG. 7 illustrates successive situations wherein the f2 pulse rate begins at a higher frequency than f1, then decreases to a lower frequency than f1, and then returns to a higher frequency than f1. Inasmuch as the Q output of FF2 goes to logic 1 on the trailing edge of each f2 pulse and is only returned to logic 0 at the succeeding f1 pulse, the situation whereon the frequency of f2 is greater than f1 results in simultaneous logic 1 signals on the clock and K inputs of FF3 so that its Q output is logic 0 and its Q output is logic 1. The Q output of FF3 is an input to an NAND gate 16a which also receives the f1 clock pulses as an input, and the Q output of FF3 is an input to a NAND gate 16b which also receives the f2 clock pulses as an input. The output of gates 16a and 16b are commonded and connected to a CP lead to the voltage second switch VSS and also to resettable integrator RI. Consequently, when the f2 frequency is higher than the f1 frequency so logic 1 exists on the Q lead from FF3, the output of gate 16b goes to logic 0 on each f2 pulse so that f2 pulses appear on the CP lead to the VSS switch and to integrator RI. Similarly, when f1 clock pulse frequency is higher than f2 so that logic 1 exists on the Q lead from FF3, the output of gate 16a goes to logic 0 on each f1 pulse and the train of f1 pulses appear on the CP lead.

The clock pulses (f1 or f2) on the CP lead to resettable integrator RI are coupled through gate NOT 5 and diode D30 to the base of transistor Q13 and forward bias it into conduction to thereby short circuit and discharge capacitor C14 (if Q13 has not been turned on by a VT pulse).

The portion of volt-second switch VSS for phase one of the inverter INV is shown in FIG. 8 and described in my aforementioned application and includes volt-second latch VSL comprising cross-connected NAND gates 27a and 27b. Each first control pulse HVT from first level comparator LC1 is an input to gate 27a and sets the latch VSL to generate the positive-going leading edge of the VT pulse at the output of 27a which is fed back to resettable integrator RI to discharge capacitor C14 and terminate the ramp signal v.sub.c14, thus terminating each a width PWM pulse (when H1 is absent) in the 1a train which controls chopping by transistor 1A. The subsequent f1 or fpulse on the CP lead is an input to gate 27b and resets latch VSL to terminate the VT pulse. When the VT pulse ceases, transistor Q13 in no longer forward biased and turns off, thus permitting capacitor C14 of the integrator RI to begin to charge and generate the succeeding ramp pulse v.sub.c14.

The period w of the PWM pulses is the interval between clock pulses f1 or f2 on lead CP. Clock pulses f1 appear on lead CP when the frequency of f1 is greater than that of f2, and f2 clock pulses appear on lead CP when the frequency of f2 is greater than that of f1.

When square wave H1 from interval generator IG is absent, the duty cycle regulating pulses of the VT1 pulse train are identical to the VT pulses which begin at each HVT pulse and terminate at the subsequent clock (f1 or f2) pulse on lead CP. Since transistor 1A is controlled by pattern 1a which follows the complement VT1 during the positive half cycle of the A reference wave, the PWM pulses conducted by 1A are of a width (w-VT), where w is the pulse period. Pulse period w is variable and depends upon whether f1 or f2 pulses appear on the CP lead to volt-second latch VSL to integrator RI. The reference waves A, B, C (see FIG. 3d) which determine the fundamental component of the inverter phase output voltages are controlled by clock pulses f1, but the number of PWM pulses in each cycle of fundamental voltage is dependent upon whether the frequency of f1 or f2 pulses is higher. The output frequencies of oscillators VCO1 and VCO2 both vary in proportion to the magnitude of the speed signal from SPEED POT potentiometer but vary at different rates so that the frequencies of f1 and f2 pulses cross at the breakpoint, and the frequency of clock pulses f2 is higher than that of clock pulses f1 at frequencies below the breakpoint. Further, the difference in frequency between f1 and f2 clock pulses increases as motor speed diminishes. Since the number of PWM pulse periods w during each cycle of fundamental motor voltage is a function of the higher frequency pulses f2 below the breakpoint while the fundamental motor voltage period remains constant, it will be appriciated that average voltage applied to motor M is boosted at low frequencies. As described hereinbefore, the low motor speed portion of each motor voltage versus frequency characteristic (such as GKHF) may be adjusted by slope setting potentiometer R19 so that it passes through the origin and has the highest slope (such as GK) when no voltage boost is provided, or may be selectively adjusted by R19 so that it assumes different slopes, such as shown at G'K or G"K, wherein different amounts of voltage boost are provided. Further, breakpoint adjusting potentiometer R20 establishes the motor speed (such as that corresponding to point K) below which voltage boost occurs because the f2 clock pulse frequency is higher than the f1 clock pulse frequency.

Volt-second switch VSS also receives the H1 and LVT pulses as inputs, and when H1 pulse is present the output pulse on the VT1 lead begins at the LVT pulse and terminates at the succeeding clock pulse (f1 or f2) on the CP lead. The duty cycle regulating pulses of the VT1 train are thus of a width (w-b) as long as H1 is present. Inasmuch as transistor 1A is controlled by pulse train 1a which follows the complement VT1 during the positive half cycle of the A reference wave, the pulses of the 1a train from modulation logic ML are of b width during the first 30.degree. and during the last 30.degree. of the A wave positive half cycle since H1 spans the transitions between the A and A half cycles, thereby resulting in the twelve step, reduced-harmonic, average line-to-line voltages, such as V.sub.1.sub.-2 shown in FIG. 4, being applied to the motor windings.

While only a single embodiment of my invention has been illustrated and described, many modifications and variations thereof will be readily apparent to those skilled in the art, and consequently it should be understood that I do not intend to be limited to the particular embodiment shown and described.