Programmable PWM Output Voltage Independent of Supply

The present application claims priority from U.S. provisional application number 60/009,985, filed Jan. 2, 1996.

The present invention relates to a pulse width modulator circuit for providing a square wave whose duty cycle compensates for variations in supply voltage. The application for this invention arises when it is necessary to provide a programmable voltage across a load such as a motor or lamp. A power device, such as a power MOSFET, is connected to the load, either on the supply or ground side. The output of the pulse width modulator is applied to the gate, or base, of the power device, driving it to conduct and pull the output up to positive supply, when the pulse width modulator output is high. When the pulse width modulator output is low, the flow of current through the power device is prevented, thus allowing the output to float to ground. By adjusting the pulse width in response to a programming signal, a specified average output voltage can be maintained.

In many applications, the supply voltage is not constant, owing to variations in load or other causes. Furthermore, the input voltage which is intended to control the output voltage has no correlation to the supply voltage and thus cannot compensate for supply voltage variations. Unless variations in supply voltage are compensated, the variance in voltage across the load will be directly proportional to the variance in supply voltage. This is undesirable in applications where, for example, dependence on supply voltage results in a noticeable variation in lamp intensity or motor speed. In accordance with this invention, the duty cycle can be accurately altered to compensate for variations in supply voltage to yield a constant average output voltage.

In accordance with one formulation of the present invention, a pulse width modulator circuit is provided for supplying a train of square pulses whose width is proportional to an applied control voltage and inversely proportional to a power supply voltage. The pulse width modulator circuit has a comparator with two inputs. One comparator input is connected to an oscillator which provides a periodic voltage ramp whose peak output voltage is proportional to the power supply voltage. The second comparator input is connected to the output of a summing means which receives a control voltage and a signal proportional to the power supply voltage and supplies a linear combination of the control signal and the power supply voltage to the second comparator input.

In accordance with an embodiment of the present invention, the pulse width of the output of the pulse width modulation is controlled so that the average supply voltage across the load and power device will not vary in response to variations in the supply voltage. The pulse width modulator circuit described herein advantageously provides the capability to drive a power device with a square wave whose duty cycle accurately compensates for variations in supply voltage. The circuit of the present invention further advantageously provides a novel voltage-compensating pulse width modulator circuit which is of relatively simple and inexpensive construction. Other objects and advantages of the invention are in part apparent and in part pointed out hereinafter.

FIG. 1 is a schematic diagram of a pulse width modulator constructed in accordance with the present invention.

FIG. 2 is an illustration of the voltage level at the output of an oscillator as a function of time showing the effect of compensated reference voltage V.sub.dc on duty cycle in the present invention.

Referring now to FIG. 1, the pulse width modulator (PWM) is indicated generally by reference character 10. In the preferred embodiment, pulse width modulator 10 is implemented as a portion or entirety of an integrated circuit (not shown). As generally practiced in the art, pulse-width modulator circuits supply a square wave to the gate or base 12, of a power device M1. The duty cycle of the square wave is dependent upon a control voltage V.sub.control, supplied by the application. Like conventional pulse width modulators, PWM 10 consists, fundamentally, of an oscillator 16 and a comparator 20. Oscillator 16 provides an oscillating voltage output to a first input port 22 of comparator 20. The temporal shape of the oscillating voltage output of oscillator 16 is that of a periodic sawtooth or triangle wave. A second input port 24 of comparator 20 is maintained at a compensated reference voltage V.sub.dc such that output 14 of comparator 20 is driven high during part of the period of oscillator 16, and during the remainder of the period of oscillator 16, output 14 of comparator 20 is driven low. Thus, output 14 of comparator 20 consists of pulses whose width, or duty cycle, is governed by adjustment of compensated reference voltage V.sub.dc.

While operation of the invention requires that the waveform of oscillator 16 consist of linear ramps, either a triangle (symmetric) or sawtooth (asymmetric) waveform may be employed. Power device M1 may be a MOSFET, as shown by way of example, or any other power device known in the art. The drain or collector of power device M1 is connected to voltage supply V.sub.cc, if the application requires a high-side driver, or the source or emitter of power device M1 is connected to ground, if the application requires a low-side driver. While the invention is described in terms of a high-side driver embodiment, as depicted in FIG. 1, it is to be understood that the low-side driver embodiment is also within the scope of the invention.

In high-side applications, when power device M1 is in the conducting ("on") state, power device M1 sources current I.sub.LOAD, and pulls output voltage V.sub.OUT up to positive supply voltage V.sub.cc. When power device M1 is off, no current flows and output V.sub.OUT floats to ground. The duty cycle of PWM 10 is the fraction of time output 14 of PWM 10 is high and drives power device M1 to conduct. During the balance of the time, output 14 of PWM 10 is at ground and power device M1 is off. Therefore, the average of output voltage V.sub.OUT changes with the duty cycle of PWM 10. In particular, the average of output voltage V.sub.OUT can be expressed as:

The invention is directed toward systems where control voltage V.sub.control is supplied in order to control average output voltage Vout.sub.avg according to:

where x is a gain that expresses the constant of proportionality between output voltage V.sub.out and control voltage V.sub.control and is set by the application's needs. In particular, the peak values of control voltage V.sub.control and output voltage V.sub.out are related as:

since output voltage V.sub.out can be as large as the nominal value of supply voltage V.sub.cc. For example, if the nominal supply voltage is 14 V and the Maximum of V.sub.control is 5 V then x=2.8.

The requisite duty cycle for providing the desired average output voltage is derived by combining (1) and (2), yielding:

It is evident from Equation (4) that, in order to provide an output voltage V.sub.out independent of supply voltage V.sub.cc, it is necessary that the duty cycle be varied in inverse relation to the variations in supply voltage V.sub.cc.

The invention is specifically intended for applications which employ an oscillator 16 whose peak and valley voltages are both fixed fractions of supply voltage V.sub.cc. The transfer function described in equation (2) is easily implemented if the oscillator valley goes all the way to the ground supply rail, or, similarly, if the oscillator peak goes all the way to supply voltage V.sub.cc. For example, if the oscillator ramped from ground to the supply voltage, and the comparator output were high when V.sub.dc is higher than the oscillator ramp voltage, then the duty cycle would simply be the fraction of V.sub.cc constituted by V.sub.dc, and

would obtain, as required, for an appropriately chosen gain factor x. Therefore the system compensates itself.

It is not practical, however, to design an oscillator in which at least one peak is a supply rail. That is because integrated oscillators charge and discharge a capacitor with current sources and allow comparators to determine the maximum and minimum voltages. Operating a trip point too close to a rail makes the triangle wave non-linear and therefore a poor oscillator for PWM applications. In order to avoid this, it is common practice to make the trip points for the oscillator be a percentage of the supply voltage. That gives the current sources enough head room to operate accurately. This will add an offset error term dependent upon Vcc to the Vout.sub.avg equation defined in equation (2). The invention cancels out this error term, in a manner now described.

Referring now to FIG. 2, the ramp voltage V.sub.triag during half of the triangle wave is depicted as being compared to the compensated reference voltage, V.sub.dc required to produce the proper duty cycle for supply voltage compensation, while the horizontal axis represents time. In particular, T/2 is the duration of the positive-going ramp 50, while z is the duration during which V.sub.dc exceeds V.sub.triag. Thus,

The variables k.sub.p and k.sub.v are percentages of V.sub.cc that represent the peak 52 and valley 54 of oscillator ramp voltage V.sub.triag respectively. While the following analysis assumes that the output is high when V.sub.dc is larger than V.sub.triag, it is to be understood that implementation with respect to the opposite polarity is also within the scope of the invention. The variable y represents the oscillator ramp voltage V.sub.triag, with respect to valley 54 of oscillator ramp voltage V.sub.triag, at which output 14 switches between high and low, i.e.,

Using similar right triangles, the duty cycle is derived as:

and, by combining (8) with (9):

Equating the expression (10) for the duty cycle with that of (4), and solving for Vdc yields the requisite transfer function:

An implementation of this transfer function results in an average PWM voltage that is independent of supply voltage. Expressed in general terms, voltage V.sub.dc applied to second port 24 of comparator 20 is a linear combination of V.sub.control and V.sub.cc.

A preferred embodiment of the invention for achieving the linear combination described by equation (11) is described with further reference to FIG. 1. Compensated voltage V.sub.dc is compared by comparator 20 to the triangle wave output of oscillator 16 to produce a PWM signal at output port 14 to drive the external power device M1. The peak 52 and valley 54 voltages (shown in FIG. 2) are set up by the resistor divider R1-R4, such that

In this preferred embodiment, a first operational amplifier OP1, in conjunction with a transistor Q1 converts a voltage at input port 28 proportional to Vcc into a current I1 that is applied to a summing node 26 of a second operational amplifier OP2. Control voltage V.sub.control is applied to noninverting input port 30 of OP2 such that compensated reference voltage V.sub.dc appearing at second input 24 of comparator 20 equals:

Since operational amplifier OP2 functions to maintain noninverting input port 30 and summing node 26 at an equal potential, a current 12 flows through resistor R7 which is the sum of current I1, proportional to V.sub.cc and sourced by transistor Q1, and current I3 through resistor R6 which is proportional to voltage V.sub.control at input port 30 and, thus, equivalently, at summing node 26. Consequently, compensated reference voltage V.sub.dc is the linear combination of V.sub.control and Vcc as given by Equation 14.

By equating (14) with (11), resistors R1 through R7 can be chosen to accommodate the parameters of a given application. This means of deriving Vdc provides isolation of Vcc from second operational amplifier OP2 to prevent undesirable interaction of circuit elements. Equation (14) shows that the accuracy of the system is only dependent on the accuracy of the resistor matching (R7 to R6 and R5, and R1 through R4 to each other) and the input offset voltages of operational amplifiers OP1 and OP2 and of comparator 20.

While FIG. 1 shows the preferred means for implementing the transfer equation described by (11), Some variations include:

(1) Providing a voltage proportional to V.sub.cc at non-inverting input port 28 of first operational amplifier OP1 by connecting non-inverting input port 28 of first operational amplifier OP1 to a resistor string between Vcc and ground which is separate from resistor string R1-R4;

(2) Using a current source referenced through a resistor to Vcc to develop current I.sub.1 ;

(3) Having summing node 26 of the inverting terminal of OP2 connect through a resistor to a voltage divider off of supply voltage Vcc.

Variations (2) and (3) have less accuracy than the preferred embodiment shown in FIG. 1, but they are simpler implementations.

The described embodiments of the inventions are intended to be merely exemplary and numerous variations and modifications will be apparent to those skilled in the art. All such variations and modifications are intended to be within the scope of the present invention as defined in the appended claims.