This invention relates to power factor enhancement systems.

Control of power factor at the input of a power circuit, powered off
of an AC power line typically designated as an off line switcher (OLS), is critical
to both the integrity of the AC power line as well as the efficient operation and
transient response of the power circuit itself. In theory the power factor can
attain a unity value by forcing the input current waveform to conform exactly to
a sinusoidal waveform in phase with the fundamental of the sinusoidal voltage waveform
input. Many techniques have been advanced to achieve this current waveform control.
Some of the earlier techniques use passive networks with reactive components to
shape the input current waveform. As power factor and other operating requirements
of the power supply become more demanding the trend has been toward the use of
active power factor control networks to control the input current waveform.

Active power factor control networks typically sense input and output
signal parameters of the power circuit and utilize a rectifier followed by a boost,
buck, buck-boost, SEPIC or similar power trains connected between the AC line and
the power circuit to enhance the power factor. The boost power train includes a
power switch selectively switched or pulse width modulated in response to these
signal parameters to force the input current to conform to a desired or programmed
current waveform. In a particular illustrative arrangement disclosed in U. S. patent
4,412,277 a rectified input AC voltage waveform is multiplied with an error voltage
representing the deviation of the output voltage from a regulated value. The resulting
control signal is scaled to provide a programmed AC current waveform i_{p}.
This waveform is used to control the modulation of a pulse driving the power switch
of the boost power train to provide the desired input current waveform and hence
advance the power factor value more closely to a unity value.

An improved power factor control arrangement disclosed in U. S. patent
4,677,366 uses an instantaneous rms value of the input AC voltage as a control
variable to provide a suitable transient response to changes in the amplitude of
the waveform of the AC line voltage. This control arrangement includes a feed-forward
control, added to accommodate rapid changes occurring in the rms value of the input
AC voltage. This feed-forward arrangement scales the programmed current i_{p},
inversely by the square of the rms input voltage.

A problem with these existing arrangements is the effects of ripple
voltage due to rectification and other causes superimposed on the sensed voltage
waveforms. This ripple voltage in the sensed signals is a spurious signal which
is superimposed on the error voltages used to control the boost converter. This
prevents an accurate determination of the waveform of the programmed current i_{p},
and creates undesirable side effects in the operation of the control circuitry.
Present techniques advanced to deal with this ripple voltage slow the response
time of the power factor control circuitry.

Another problem with existing arrangements to enhance power factor
is the slow response time of output voltage regulation control loops to output
load transients. Existing arrangements to address this problem include output power
as a feed-forward variable in the feed-forward control loop controlling the power
switch of the power train. A key variable in the effect of the output power on
the control process is the energy stored in the output capacitor of the power train.

A controller designed to accurately accommodate changes in output
power has been implemented as a digital controller and is disclosed in the disclosure
entitled "A Digital Controller for a Unity Power Factor Controller" Mitwalli et
al, Workshop on Computers in Power Electronics, Berkley, CA, August, 1992. This
controller is based on modeling instantaneous power flows and is based on knowledge
of the value of the power train's output capacitance. It additionally requires
complex real time calculations to achieve satisfactory operation.

EP-A-0498553 describes a power converter that accepts an input voltage
*V*_{i}, absorbs an input current *i*_{i}, and delivers
output power *P*_{o}, which is equal to the input power
*P*_{in}. Based on this relationship, a control law is derived which
ignores any internal storage elements in the power converter. An exemplary derivation
is provided that results in an adaptive control process that regulates a constant
output voltage from a power converter which depends on a multiplicative gain parameter
K. Also provided are derivations disclosing how to regulate, alternatively, a constant
output current or an output voltage that is proportional to an input voltage. A
simple time-varying value, such as the instantaneous values of the input voltage
and current, are used as an input to the control process.

According to this invention there is provided a power factor enhancement
system as claimed in claim 1.

A power factor control system for an off line switching power supply
(OLS) is operative trough the generation of substantially ripple-free estimates
of control input parameters (a squared peak input voltage ε2m
,
output voltage and load power) and by the use of these substantially ripple-free
signals controls a boost, buck, buck-boost, SEPIC or other OLS type converter to
enhance power factor at the input to the OLS. The generation of these parameters
allows the derivation of a programmed current *i*_{p}, used to control
the waveform of the actual input current. The control procedure is based on a quarter
cycle avenged power basis that takes advantage of the energy stored in the output
capacitor of the convener which is significant compared with the amount of energy
that is drawn from the AC line during a quarter cycle of the AC voltage waveform.

The input power to the OLS is derived from the rms values of the input
voltage and current on a quarter cycle time scale. Given the output power and its
deviations in power due to load changes and error in the desired output voltage,
the programmed current i_{p}, is determined by deriving an input conductance
of the OLS and combining it with the input voltage.

__Brief Description of the Drawing__
In the Drawing:

- FIG. 1 is a schematic of a ripple-free ε2m
estimating
circuit forming part of the invention;
- FIG. 2 is a schematic of a power system with enhanced power factor embodying
the invention;
- FIG. 3 is a signal flow diagram for describing a control process of the power
system of FIG.2 with enhanced power factor;
- FIG. 4 is a schematic of a circuit for determining the absolute magnitude of
a voltage;
- FIG. 5 is a schematic of a digital implementation of a power factor controller;
- FIG. 6 is a schematic of another arrangement of a power system with enhanced
power factor embodying the invention;
- FIG. 7 is a schematic of another arrangement of a power system with enhanced
power factor embodying the invention; and
- FIG. 8 is a schematic of a power processing circuit suitable for use with the
power factor enhancement components and arrangements of FIGS. 1-7.

__Detailed Description__
An estimating circuit for determining specified input parameters to
the rectifier of a power factor enhanced power system, as shown in the FIG. 1,
takes the input AC voltage and extracts values representing the square of its peak,
and the in-phase time varying value of the fundamental. The input AC voltage
E_{m}sin(ωt)
is directly applied to the input terminal 101 of a controlled harmonic oscillator
110. The controlled harmonic oscillator 110 responds to the input AC voltage on
lead 101 and generates an AC sinusoidal voltage
ε_{m}sin(ωt)
on its output lead 102. The controlled harmonic oscillator 110 may comprise any
circuitry comprising integrators connected to generate both a sinusoidal fundamental
of the input AC voltage both in phase with it and a component displaced from it
by π/2.

The two derived signals displaced in phase by π/2 are coupled
respectively to the signal squaring circuits 111 and 112. The outputs of the squaring
circuits 111 and 112 are summed in the summing circuit 113 to produce the peak
square value ε2m
on lead 114. The form of the summed
output signal is a ripple-free magnitude governed in accord with the equation:
cos^{2} (&thetas;) + sin^{2} (&thetas;) = 1
where &thetas; is any value determined by the circuitry.

The outputs of the two squaring circuits 111 and 112 are also coupled
to an operational amplifier 115 which, by controlling the integrator gains, derives
a value for the frequency ω of the fundamental at lead 116.

A detailed description of an estimator suitable for application here
is disclosed in U.S. patent no. 5450029.

A power processing system, shown in FIG. 2 includes a power factor
control system to enhance power factor at the input of an off line switching circuit
shown schematically as a rectifier followed by a boost converter. The AC line
supplying the voltag
e E_{m} sin(ωt)
is connected to the input lead 201 which in turn applies this voltage to a full-wave
rectifier 205 and to an estimator 210, such as is shown in FIG. 1.

The output of the rectifier 205, as schematically shown in the FIG.
2, is coupled to a switching type converter 230, such as a boost,buck or buck-boost
type DC/DC converter, whose power switch is pulse width modulated to generate a
DC voltage having a time-varying current related to the AC voltage waveform input
at lead 201. The power switch is controlled by the controller 235 which responds
to the programmed current parameters supplied by the estimator 210. Controller
235 may be embodied as an IC circuit which responds to the programmed current i_{p}
to produce a pulse width modulated drive signal for driving the converter's power
switch to achieve intended power factor and regulation results. IC circuits to
respond to an input signal such as i_{p} are available commercially.

The estimator 210 supplies the parameter
ε_{m} sin(ωt)
on lead 251 and the parameter ε2m
on lead 232.
Additional parameters are fed back in response to signals sensed at the circuit
output, mainly the output voltage on output lead 241, sensed by lead 229 and the
output current I_{load} sensed by the current sensor 231 on lead 241.

The sensed output current I_{load} is multiplied with the
output voltage V_{out}
sensed on lead 233 in the multiplier 225 and the resulting
product representing the output power of the DC-to-DC converter 230 is applied
to the summing circuit 224. The output voltage sensed on lead 229 is applied to
the gain control impedances 227 and 228 and the inverting input of the operational
amplifier 226. A reference voltage is applied to its non-inverting input of the
operational amplifier 226. Its output on lead 234 is representative of an error
in the output power, δ P_{out}, of the converter, This value δ
P_{out} is applied to the summing circuit 224.

The signal on lead 242 (i.e. the output of summing circuit 224) is
applied to the numerator input of a dividing circuit 222. The peak square voltage
value ε2m
on lead 232 is applied to the denominator
input of the dividing circuit 222.

The output of the dividing circuit 222 is combined with the
ε_{m} sin(ωt)
output, on lead 251, of the estimator 210 in the multiplier 221. The output of the
multiplier is scaled by scaling circuit 223 to form its absolute magnitude and
applied to the controller 235. The output of the rectifier 205 is directly connected
to the controller 235 via lead 215.

The operation of the circuit in FIG. 2 can be understood by discussing
the theoretical basis underlying its operation. The underlying principle is the
equating of the average input power to the rectifier 205 to the average output
power from the rectifier 205, in combination with an accounting for the imperfect
efficiency η of the power conversion process (i.e., efficiency is less than
100%). The efficiency η is a ratio of the average output power to the average
input power.
η &peseta; P_{in,avg} = P_{out}
η &peseta; E_{rms}&peseta; I_{rms} = P_{out}
Satisfying the equation (2) requires that the power factor be unity at the input
to the rectifier. The time interval for the averaging process for both Equation
(1) and Equation (2) is any positive integer multiple of a quarter cycle of the
input AC sinusoidal waveform.

A control law for the programmed current i_{p}, corresponding
to Equation (2) may be formulated. This control law, which is based on a quarter-cycle
averaged basis, is:
i_{p} = 2(P_{out})ε_{m}sin(ωt) / (η&peseta;ε^{2}_{m})
This uses the following relationship for sinusoidal waveforms:
ε ^{2}_{m} = 2&peseta;ε^{2}_{rms}
.

For practical converters of the boost or similar type for high power
factor applications, η is about 0.93 to 0.98, and is relatively constant under
substantial variations in load power.

Thus Equation 3 allows the development of a control strategy to compute
the programmed current i_{p}, based on knowledge of the rms value of the
input voltage and the output power, or a filtered value of the output power.

A controlled conductance value G for the rectifier is defined such
that the instantaneous value of the programmed current i_{p} is given by:
i_{p} = G&peseta;e_{in}
where the value G is:
G = 2&peseta;P_{out} / (η&peseta;ε^{2}_{m})
and where e_{in} is the instantaneous value of the input voltage, or, preferably,
the instantaneous value of the fundamental harmonic
ε_{m} sin (ωt)
of the input voltage. For a sinusoidal input voltage and for constant load power,
G is constant and i_{p} is sinusoidal. In addition, the determination of
G is "well conditioned", i.e., it does not require division of small uncertain
numbers by other small uncertain numbers. In practice, Equation (3) or, equivalently,
Equation (5) is not sufficient for a complete control process because it is almost
always required that the output voltage from the power supply be regulated to a
predetermined voltage such as 400 volts. Accordingly, Equation (3) is modified
to include a "power increment" δP_{out} to reflect the error between
the desired (or reference) output voltage V_{ref} and the actual output
voltage, V_{out}.
i_{p} = 2(P_{out} + δP_{out})&peseta;ε_{m}
sin(ωt) / (η&peseta;ε^{2}_{m})
The purpose of Equation (7) is to use the output power P_{out} to control
the principal portion of the controlled conductance G, and to use δP_{out}
as a small increment to regulate the output voltage. Thus, the OLS can have the
capability to respond quickly to large step changes in output power, with the feedback
term δP_{out} only accounting for modeling errors.

Specific design is required for the signal processing elements that
provide the inputs for the determination of the controlled conductance G. First,
the estimate of the peak-squared voltage or equivalently, the rms-squared value
of the AC input voltage should be substantially ripple free and should respond
quickly to changes in amplitude of the waveform of the AC input voltage. The preferred
method as well as other alternatives for determining the peak-squared value of
the AC input voltage are described in U.S. patent no. 5450029.

Second, the estimate of output power should also be substantially
ripple free during steady-state operation, and it too should respond quickly to
changes in the load. The preferred approach for loads that draw constant power
in the steady state is to multiply the rectifier output voltage by the output current.
Commercially available analog ICs such as the MC1495 or AD532 can be used. An alternative
is to rely on the pre-regulated, known value of the rectifier output voltage (i.e.
the reference voltage applied to the operational amplifier 226 in FIG. 2) and to
scale the measured OLS output current to estimate the output power, (i.e the current
sensing circuit 231 in FIG. 2) with filtering supplied, as necessary, to reduce
ripple. Ripple contributes to unwanted periodic variations in the derived value
G that can cause distortion of the programmed current i_{p}.

Third, the feedback term δP_{out} should also be substantially
ripple free. Filtering, to eliminate the ripple, is required. The filtering selected
must be consistent with stability requirements of the feedback loop in which it
is embedded. This filter may be embodied in circuitry using the same observer technique
described in our co-pending application referenced hereinabove. Such a design is
also used in the estimator 210 which estimates the rms value of the input AC waveform.
This low pass filter using observer techniques estimates a dc (i.e. ripple free)
component of the output voltage or output current. A two state variable model of
the 100 or 120 Hz ripple can be used to estimate the ripple, which is then subtracted
from the output voltage or output current

FIG. 3 shows the signal flow control, required for the controller
235 and supplied as shown by the output portion of the estimator of FIG. 1, for
a control process described by Equation (7). The output voltage is sensed at output
lead 241 and summed in summer 305 with a reference voltage applied to lead 306.
The summed output is applied to the low pass filter 301. Filter 301 is a low-pass
filter for sensing output voltage error and includes the required gain and phase
shaping networks necessary to assure system feedback stability. The detailed required
characteristics of the LPF 301 are readily apparent to persons skilled in the art

The output of filter 301 is representative of error in output voltage
at lead 241. The required output power change (i.e. required by the voltage change)
denoted δ P_{out} is applied to a summer 224. The estimated steady
state power is derived by multiplying the output voltage and the output current
in the multiplier 225. The estimated steady state power is summed with the δ
P_{out} by the summer 224. If gain and phase control of the output of multiplier
225 is desired an optional low pass filter 243 may be inserted between the multiplier
225 and the summer 224, as shown in the FIG. 2.

The output of summer 224 is applied to the numerator input "n" of
the divider 222. The peak squared voltage ε2m
output
of the estimator 220 is applied to the denominator input "d" of the divider 222.
The resultant of the division is the required OLS conductance "G" and is supplied
on the lead 219 and designated as the value "G".

In the FIG. 3 the value of
ε_{m} sin(ωt)
on lead 214 is applied to the multiplier 221 whose other input is the value "G".
The multiplier combines the two inputs to derive the programmed current value i_{p}.

As a further practical consideration, it is necessary that the absolute
magnitude of the programmed current be supplied to the controller if the input
voltage to the boost or similar type switching regulator is already rectified by
a diode bridge. This function is provided by the scaling circuit 223 shown in the
FIG. 2. This circuitry to derive the absolute magnitude value can be implemented
by an operational amplifier and comparator as shown in FIG. 4. This circuitry includes
the operational amplifier 401 having its inverting input 402 connected to receive
the i_{p}
output of the multiplier 215 shown in the FIG. 3. This value of
i_{p}, is also applied to the inverting input port of an operational amplifier
411. Its non-inverting input is connected to an FET device 410 whose control electrode
413 is controlled by the output of the operational amplifier 401. The FET device
410, whose conductance is switched with the sign of the amplitude of i_{p}
on lead 402, connects the non-inverting input of amplifier 411 to ground. Thus
the conductance of FET 410 is switched depending on whether operational amplifier
401 is saturated positive or negative, in determining whether operational amplifier
411 is inverting or non-inverting. Resistors 421 and 422 are equal, suitable values
being about 10K ohms. The resistance of resistor 423 should be about 500 times
larger than the on-resistance of FET 410. The output of lead 420 represents the
absolute magnitude of i_{p}.

Several approaches can be used to implement a controller based on
Equation 7. One method uses analog multipliers and dividers to perform the nonlinear
operations such as squaring, multiplication and division, and operational amplifiers
for summing and subtracting. One implementation of this method has been described
above. Another approach, such as shown in the FIG. 5 implements the controller
with digital technology by using a microcontroller 510 such as an 80C51-type which
can include multiplexed analog-to-digital conversion on the same integrated circuit
to compute the slowly varying quantities such as the controlled conductance G.
Quickly varying quantities such as i_{p} can be formed using a multiplying
digital-to-analog converter (DAC) 520 such as the DAC1022. Thus, the multiplying
DAC 520 conveniently multiplies a quickly varying analog signal (i.e.,
ε_{m} &peseta; sin(ωt)
) by a slowly varying digital signal, i.e., the digital representation of the controlled
conductance G, to form the quickly varying output, i_{p} on lead 521.

The inputs
ε_{m}sin(ωt)
,
ε_{m}cos(ωt)
, V_{0} and I_{0} are applied to the sample and hold circuit 501.
The sampled values are applied either directly or via operational amplifiers to
form the absolute value of the signals which are applied to the microprocessor
510. The processed output of the microprocessor 510 is applied to the digital to
analog converter 520 and from thence to the circuitry for deriving the absolute
magnitude i_{p}.

A power factor enhancement system shown in the FIG. 6 is similar to
that of the FIG. 2 system. In the system of FIG. 6 the input terminal 201 is directly
connected to the multiplier 221 via lead 202. In this arrangement the waveform
of the input AC voltage is assumed to be essentially free from distortion.

Another variant of the power factor enhancement system is shown in
the FIG. 7 in which the rectified sine wave output of the rectifier 205 is applied
to the multiplier 221.

The power factor control system of the invention allows the utilization
of many various configurations of the OLS. For example an OLS in which the boost
convener and rectification are merged into one unitary circuit combining both
functions is shown in FIG. 8. Such a circuit arrangement allows a minimization of
circuit components incurring a loss as compared with a combination of a separate
rectifier with a separate power converter for wave shaping as a power factor enhancing
circuit. The circuit of FIG. 8 accepts an AC line voltage at the input terminals
801 and 802. This AC voltage is applied to the junction nodes 811 and 812 of a
diode bridge circuit 805 which includes the diodes 806, 807, 808 and 809. A bidirectional
power switch 803 selectively connects the two junction nodes 811 and 812. The output
nodes 817 and 818 of the bridge are connected to the output terminals 821 and 822.
A charge storage capacitor 813 shunts the output terminals 821 and 822.

The power switch 803 is driven by the output of the controller to
pulse width modulate the rectified current to control the waveform of this signal
and the voltage at the output terminals 821 and 822.

This arrangement may be readily substituted for the rectifier converter
arrangement of the FIG. 2 with the control drive applied directly to the bidirectional
power switch 803.