This invention relates generally to power generation and, more specifically,
to programmable single-phase output voltage and frequency.

BACKGROUND OF THE INVENTION

In most power generation applications, synchronous motors are driven
to generate AC power. When so used, the frequency of AC power output is dependent
upon the process used to apply torque to the synchronous motor. Constant torque
at standardized values rarely exists in nature to precisely turn a synchronous
machine. The rotational rate or angular velocity varies greatly. Because angular
velocity is proportionate to the resulting frequency and voltage of AC power, where
the angular velocity of the torque source is variable, the frequency and voltage
of the resulting power is variable.

Randomly variable frequency AC power is not very useful. It is impossible
to synchronize such power with a power supply network in a commercially practical
manner. Such power cannot drive most applications designed for 60 Hz. line supply
voltages. Few applications can tolerate such frequency variability.

To overcome the frequency variability in power generation, the solutions
have been of two types, input and output solutions. Input solutions are mechanical
and govern the power transfer to the synchronous machine. Output solutions condition
electrical power garnered from the synchronous machine.

Traditionally, design constraints on power processing have required
selection of components based upon the peak power to flow through those components
rather than to design around the mean as in DC power systems. Periodically recurring
peaks in the voltage and current waveforms for each phase develop recurrent power
transfer well above the mean. Nonetheless, the power drawn from such a system as
an aggregate is constant. For instance, the total power drawn from a balanced three-phase
source by a balanced resistive load is constant. That is

where P_{t} = time value of power

V = peak line voltage

R = load resistance (per phase)

ω = source frequency

ϕ = 2π/3

This fact is exploited in the cycloconverter at discrete frequencies. Small fluctuations
in frequency are passed to the output. This need not be the case. The power transferred
from source to load is not a function of time. The transfer of power can be accomplished
without storing energy within the processor between input voltage cycles.

Because the power is independent of time and therefore phase, there
exists a means, by aggregating the outputs from the phases to subject several of
the components of the system to constant rather than varying power over time. Modulating
the waveforms as the inventive apparatus does, "knocks the corners off" of the
lower frequency waveforms as they come off the generator. In doing so, the inventive
apparatus removes the need to accommodate the peaks of instantaneous power. The
inventive system presents a system producing well-formed power output. Without
converting and inverting power to and from DC power, the inventive system comprises
smaller transformers and power storing filters than conventional means.

The traditional input solution to electrical power generation in applications
such as aircraft has been through the Constant Speed Drive (CSD) coupled to a generator
providing, for example, 115 VAC with three-phase power at a constant 400 Hz. In
more recent times this arrangement has combined the CSD and generator into an Integrated
Drive Unit, or IDU. With a constant frequency power output this has been a creditable
solution, albeit expensive to buy and to maintain.

More recently, the output solution has been the Variable Frequency
(VF) and cycloconverter systems. Cheaper of these two options, VF presents the
load with power such as 115 VAC, three-phase power but only has a distribution
capability at a frequency proportional to the engine speed. For a turbofan engine,
for instance, this is usually 2:1. However, because of the wide range of frequency
variation, power conditioning would be essential for almost all cases and, when
added to the procurement, the additional cost of attendant motor controllers becomes
prohibitively expensive.

In recent years, power has been generated with a Variable Speed Constant
Frequency (VSCF) cycloconverter. A cycloconverter is a power electronics device
designed to provide a variable voltage, constant frequency AC drive in a one stage
operation, suitable for supply to an AC motor. These devices work by generating
very high frequency three-phase power then selectively drawing voltages from the
peaks of the three phases in a manner to construct rough approximations of lower
frequency waveforms.

While a VSCF system unit (cycloconverter) does have the ability to
produce AC and DC simultaneously, it does not produce clean waveforms. The voltage
regulation is accomplished by a series of magnetic amplifiers, transformers, and
bridge rectifiers. The VSCF drive uses a simple drive system and lets the alternator
produce an electrical supply that is frequency wild, i.e. not well-controlled,
which is then shaped by a solid-state electrical unit. Nonetheless, the resulting
waveform includes several harmonics that impart an imaginary component to the power
and may interfere with the function of the load.

Still another means of generating constant frequency power from a
variable source of torque is based upon converting and rectifying power to DC before
inverting the DC power to AC power such as 60 Hz. line voltage. This approach requires
a converter that can sink current of opposite polarity to the output voltage. However,
most converters cannot accommodate a non-unity power factor load. Additionally,
for a given power level, in a single phase rectifier, the current pulses that comprise
the ripple may be four to five times as large as the peak of the current waveform
for an equivalent unity power factor load. Such current requires much larger conductors
to minimize resistive losses. In polyphase rectifiers, the current peaks are not
as large since smaller peaks occur more frequently, but the power quality is not
as good because of large current transitions

Both the cycloconverter and the devices for converting and inverting
power use, transformer rectifier units that operate at line frequency chop power
at low frequencies creating low frequency fundamental sinusoids thus requiring
large and heavy transformers and large capacitors to store energy for smoothing
peaks and filling valleys in the waveform. Introduction of such elements often
adds reactive factors that affect the power factor. In such configurations the
reactance causes the current to either "lead" the voltage or to "lag" the voltage.
Like the input solution, power conditioning is necessary for power-factor correction.
Often this includes the use of synchronous motors spinning in "no load" states.
All of these solutions prove to be costly. The need for constant frequency power
has justified these solutions.

There exists, then, an unmet need for a power unit for converting
three-phase variable frequency AC power to constant frequency AC single-phase power
while maintaining a unity power factor. Internal control of the magnitude of the
voltage would further enhance the utility of such a power unit.

SUMMARY OF THE INVENTION

The inventive device exploits high frequency modulation of the signal
to balance the power throughout the input voltage cycle. By modulating both the
width and the amplitude of the power waveform, no power is lost and the power factor
remains substantially the same as that of the load when the power factor is calculated
at the input frequency.

The invention receives polyphase power of variable frequency and passes
power of a set frequency. Relationships between phases are exploited according
to known trigonometric identities. Each phase of the power is modulated distinctly
and is presented to a single-phase load where the output voltage represents difference
in potential pairs of terminals. The single-phase power delivered at the terminals
of the load is a very clean sinusoid. To describe the invention here, the circuitry
for handling a single-phase of the input power passed by the inventive device to
the load. For that purpose, an exemplary phase of the three-phase power supply
is set forth here.

The present invention comprises a system and method for a direct conversion
programmable power source controller. The invention controls three-phase power
by accepting a reference signal; generating three modulating signals, each for
one phase of the input power and having a frequency equal to the sum of the input
frequency and the desired output frequency; accepting the three-phase electrical
power from a wye connected power source; creating a high, chopping frequency for
sampling; phase-angle-modulating the source signal at the chopping frequency according
to each of the modulation signals. While in the high frequency state, the power
passes easily through small two-winding transformers used either for isolation
or for stepping up or stepping down the power. Once through the transformer, the
invention then re-reverses the reversed power and chops a portion of each half-cycle
in a manner that restores the pulse-width modulation of the power sinusoid by the
modulating sinusoid. Reversing the polarity of each phase of electrical power at
the phase-angle modulation frequency to reproduce the pulse-width modulated waveform
at the output.

Filters on the output integrate the power waves before feeding the
power to the load in delta. Between any two terminals of the delta connection,
the output voltage presents as a well-formed sinusoid.

According to further aspects of the invention, the invention converts
three-phase AC power to programmable frequency three-phase power without an intermediate
DC stage and without large power storage devices.

The invention presents the source a power factor at substantially
unity.

According to the invention, conversion to high frequency reduces the
size of isolation transformers. Further, the high-frequency conversion technique
reduces the size of the input and output filters and allows those filters to produce
near perfect sine waves. Also, inclusion of small high-frequency transformers allows
correction for low-line conditions. Wide separation of conversion and output frequencies
eliminates phase errors due to input and output filters. The presence of a phase
reference allows comparison with output waveforms for control loops. The control
loops eliminate line disturbances such as phase-voltage imbalance and short-term
spikes.

Users can also optionally program output voltage and frequencies without
component change by providing appropriate reference waveforms.

The invention also allows for bi-directional power flow thus to accommodate
non-unity power factor (reactive), or braking or regenerative loads.

Further, the control process does not require digital signal processing.
The inventive device may operate with a state machine with a look-up table for
the desired waveforms, and a standard multiplying digital-to-analog converter.

The present invention provides means of converting three-phase power
of variable frequency and amplitude to three-phase power of programmable and constant
frequency and amplitude without an intermediate DC conversion.

BRIEF DESCRIPTION OF THE DRAWINGS

The preferred and alternative embodiments of the present invention
are described in detail below with reference to the following drawings.

FIGURE 1a is a schematic diagram of a single phase of the inventive power controller;

FIGURE 1b is a block diagram placing the single phase as shown in FIGURE 1a
in context of a three-phase power supply;

FIGURE 2 is chart comparing the two constituent logic waveforms and the representative
output of the two waveforms through an XOR gate;

FIGURE 3 is a detail schematic of the gating circuit used in the primary and
secondary choppers;

FIGURE 4 is the tracing of a power wave pulse-width modulated by a second sine
wave; and

FIGURE 5 is a flowchart setting forth the method of presenting three-phase
programmable frequency and voltage to a load.

DETAILED DESCRIPTION OF THE INVENTION

The inventive device exploits high frequency modulation of the signal
to balance the power throughout the input voltage cycle. By modulating both the
width and the amplitude of the power waveform, no power is lost.

The invention receives three-phase power of variable frequency and
passes power of a selectable frequency. Load current reflected to the source is
similarly modulated so that the source current frequency tracks changes in the
source voltage. Relationships between phases are exploited according to known trigonometric
identities. Each phase of the power is modulated distinctly until at the terminals
of the transformer where the voltage represents the difference in potential at
distinct terminals of the transformer. To describe the invention here, FIGURE 1a
portrays the circuitry for handling a single phase of the power passed by the inventive
device. For that purpose, an exemplary phase of the polyphase power supply is set
forth here.

Referring to FIGURES 1a and 1b. A generator 20 provides one of several
phases of power to the circuit 10 shown as is shown in FIGURE 1b. In most cases,
this will be a driven three-phase synchronous machine connected in wye. The generator
20 supplies a voltage to a pair of primary terminals of a transformer 55. A secondary
pair of terminals of the transformer 55 then feeds power to or receives it from
a load 80. The invention will receive current from a load as in the case of regenerative
braking of a load.

The inventive circuit includes inductive and capacitive filter elements
on the primary circuit, 42 and 46 respectively, and inductive and capacitive filter
elements on the secondary circuit, 72 and 76 respectively. The filters, 42 and
46, are used to remove ripple from the source currents and load voltages. Because
the filters need only remove very high frequency components of the signals, the
filter elements 42 and 46, 72 and 76 need not store significant power. Finally,
in electrical connection to the primary and secondary terminals of the transformer
55 there are a primary chopper 50 and a secondary chopper 60.

FIGURE 1a is a schematic diagram of a circuit corresponding to a single
phase of the inventive system. The system, in fact, comprises multiple circuits
of the configuration shown in FIGURE 1b. For purposes of illustration, this exposition
will present a three-phase system. Those skilled in the electrical arts will readily
see that the same inventive system is adaptable to any multiple-phase execution
by replicating the circuit shown in FIGURE 1a to correspond with each phase.

The primary chopper 50 performs one of the main roles, that of phase
modulating the power signal in the inventive power supply, by deriving a modulation
sinusoid in a feedback loop as set forth below. This modulation sinusoid as derived
from the source voltage and reference signals, has a signal logic voltage level
and a frequency representing the sum of the frequencies of sinusoids from the input
power source sinusoid and a reference sinusoid at the desired output frequency.
This reference sinusoid is, if desired, drawn from a power grid the power source
20 will supply.

The basic relationship between the elements of the circuit 10 exploits
the process of modulating the incoming signal with a local oscillator frequency.
The modulation is the multiplication of the power waveform by the reference sinusoid.
This modulation suppresses the two fundamental frequencies as well as produces
sum and difference frequencies. The modulation of the power signal relies upon
a trigonometric identity expressed as follows:
sin u x sin v = 1 / (2) [cos(u - v) - cos(u
+ v)]

Thus the arithmetic description of the modulation of the first phase
of three-phase power from a synchronous generator might look like this:
V_{A} = V_{peak}[sin(ω_{s}t)]×sin(ω_{m}t) V_{A} =V_{peak} / (2)(cos(t(ω_{s}
- ω_{m})-cos(t(ω_{s} + ω_{m})))
where V = peak phase voltage

α = ω_{s}t

β = ω_{m}t;

The ω_{s} and ω_{m}
are the source and the

modulating frequencies respectively

In a three-phase power system, each phase is offset from the other
phases by 2π/3 radians, or 120 degrees. The selection of the phase of the modulating
frequency yields algebraic opportunities to suppress the upper resulting sideband
from the modulation. Where the phase of the power is offset by 2π/3 radians
the modulating frequency is offset by -2π/3 radians or its phase equivalent
4π/3 radians. If the phases are designated A, B, and C, the equations describing
each of the remaining individual phases are:

The algebra now reveals the presence of a cosine sum term, cos(t(ω_{s}
- ω_{m})), present in each modulated phase. The presence of
this cosine sum term in each phase presents an opportunity to eliminate simplify
the sum of all of the phases. First, because the remaining terms represent the
sum of three phases of power, they cancel each other out according to the following
identity:

Thus, evaluating sum of the right-hand terms of the three phase voltage
equations 3, 4, and 5 yields:

Removing the right-hand terms leaves only the three cos(t(ω_{s}
- ω_{m})) terms such that:
V_{sum} = V_{peak} 3 / (2)(cos(t(ω_{s}
- ω_{m})))

Selecting a modulating frequency that represents the difference between
the input and the desired output frequencies means that the V_{sum}
term will result in supplied power at the desired frequency. The invention uses
feedback loops to generate both scaling of the amplitude of the power and the selection
of a proper modulating frequency to precisely match output to the needs of the
application.

Each of the arithmetic processes set forth above can be effected in
the combination of the primary chopper 50, the transformer 55, and the secondary
chopper without the use of digital logic at power levels. "Chopping" occurs at
a frequency significantly higher than the frequency of either the power supply
or the desired output. According to the invention, the multiplication is a byproduct
of the chopping process in conjunction with pulse-width modulation. Pulse-width
modulation characterizes the output of the secondary chopper 60, that is the multiplication
modulation. The width of each pulse represents the value derived from the sinusoidal
modulation function, the height of the pulse, from the power sinusoid. Thus, when
the resulting waveform is integrated over time, the resulting waveform is a smooth
product-of-sines.

The pulse-width modulation is implemented by means of two distinct
steps. First, the primary chopper 50 achieves phase-angle modulation. Then, once
so modulated, the circuit 10 passes near-square-waves through the transformer 55
at chopping frequency that is very high relative to the power and reference sinusoids.
The synchronous secondary chopper 60 inverts the second of each square wave pair
to create a pulse-width-modulated waveform at the same chopping frequency.

The ability to create a phase-angle modulated square wave arises from
parameters implicit in the nature of a sine wave itself. The magnitude of the sine
function is bounded by the value one or unity, i.e.:
|sin x|≤1, for all x
Without resorting to digital logic and without amplification, this relationship
allows the modulation of the power waveform solely with switching devices. Rather
than to mathematically multiply the instantaneous values of the power sinusoid
with the instantaneous values of the reference sinusoid at each given moment,
t, to produce a smooth function, the invention admits power during a portion
of a sampling period corresponding to the value of the reference sinusoid.

Before considering the modulation of a power waveform by means of
the primary chopper 50, consider the chopping of a simple constant DC voltage.
"Chopping," as used, means the opening and closing of a switching device at a given
frequency. A full chopping cycle comprises one period when the switch is in an
open or non-conducting state and an immediately adjacent period when the switch
is in a closed or conducting state. In the context of chopping, the duty cycle
is the proportion the period of the conductive state bears to the whole cycle.
Chopping is often used to supply a lesser DC voltage to a load from a higher DC
voltage. For each segment, the power is chopped according to the following equation:
V_{out} = V_{in} ×
D, where D is the duty cycle.

To modulate the power wave according to a second sine wave, one can
vary the duty cycle according to the magnitude of the sine. Thus, for each period
of one sampling interval, Δt, the switch will remain closed (or conductive)
for a smaller period in proportion to the value of sin(ω_{m}nΔt).
For each integer n, and for Δt = 1 / (ω_{samp}),
V_{out}(nΔt) = V_{in} ×
sin(ω_{m}nΔt)
Outside of the bounds of the first lobe of the sine wave, the value of the sine
function goes negative. Equation 11 would normally be very problematic when seeking
to modulate the power sinusoid. A duty cycle cannot become negative in length where
the sine term becomes negative. It is for this reason that an intermediate step
makes sense.

This average output is the key to the invention's two-step modulation.
The primary chopper modulates the power signal by reversing the polarity at a high
frequency. The onset of the reversal occurs after a fixed clock signal according
to the term .5 x (1 + M sin(ω_{m}nΔt)). Thus
for a phase angle Φ, the phase delay of the chopped power wave is:
Φ(nΔt) = 0.5(1 + M sin(ω_{m}nΔt))2π / (Δt)

This primary chopping allows a balanced high-frequency power wave
to pass through the necessary transformer 55. After the power passes through the
transformer 55, the secondary chopper re-reverses the polarity of the power wave
according to the fixed frequency clock signal, folding the power signal onto itself
according to the constant term in the expression.5×(1+M sin(ω_{m}nΔt)).
Because the bridge operation inverts the polarity of the incoming voltage, the
average of the bridge output is zero at D = 0.5. As D varies from
zero to 1, the output of the bridge varies from 100% negative polarity to 100%
positive polarity; or from -1 to 1 times the source voltage. The synchronized and
periodic reversal of the polarity by the secondary chopper suppresses the power
wave causing the averaged output of the bridge to be:
V_{out}(nΔt) = 0 + M sin(ω_{m}nΔt).
The M term is a coefficient between zero and one, chosen for purposes of scaling
the modulation, increasing or decreasing the magnitude of the sine curve corresponding
to regulating the voltage output in the equation (10).

One of the virtues of the inventive primary chopper 50 is presentation
of phase-angle-modulated square waves to the primary terminals of the transformer
55. Rather than presenting a pulse-width-modulated square wave of the power at
the sampling frequency, the invention seeks only to accomplish half of the process
of multiplying the sinusoids before placing the power through a transformer 55
(isolation, "step up," or "step down" as the determined by the application). Rather
than to modulate the width of the pulses according to the sine, the voltage sent
through the transformer as represented in Equation 13 is a phase modulated square
wave.

The need for using a nearly true square-wave over a pulse-width-modulated
wave is that the pulse-width-modulated waveform contains very low frequency components,
much lower than those of the chopper square wave. To achieve this primary chopping
of the waveforms, a frequency significantly higher than either the power or the
modulating sine waves is selected. This is the frequency that will be passed through
the transformer and is suitably any frequency selected to minimize hysteresis and
eddy current losses while remaining high enough to yield good resolution of the
power wave. Frequency might be chosen to optimize the characteristics of a chosen
transformer. So long as it is suitably high to yield sufficiently formed sinusoids
when passed through the final filters, the frequency is not critical to the operation
of the invention.

A square wave that is phase modulated contains all of the information
necessary to effect the product of sines modulation, but the fundamental frequency
will be very close to the frequency of the phase modulated square wave. This fact
yields an opportunity to select a frequency much higher than that of the input
power wave and the modulation sine wave. Doing so will allow the use of a much
smaller transformer in the invention as transformers designed for high frequency
are generally much smaller than those necessary for lower frequency so long as
the average voltage sent through the transformer is zero. The higher the lowest
frequency, the shorter the period before the average returns to zero.

Transformers also require symmetrical waves for efficiency. Where
the average voltage across the primary terminals over a period exceeds zero, the
net resultant current will push the transformer towards core saturation. A symmetrical
(the voltage peaks displace equally from zero, i.e. the magnitude of the negative
and positive portions of the waveform is the same) will fulfill the requirement
that the average is zero over two sampling periods. The higher frequency square
wave allows the transformer to pass power without a long-term (greater than a single
period at the higher frequency) effect on the transformer flux. Thus it is advantageous
to pass symmetric square waves through the transformer and only then to further
chop the phase modulated square waves to re-create the "product of sines" pulse-width-modulated
waves discussed above.

In FIGURE 2, the logic signals for effecting Equation 12 are shown.
Curve 110 is a clock curve used to set the sampling frequency, ω_{samp},
for purposes of admitting a portion of the power waveform. The curve 110 has a
period of Δt, reflecting a sampling period. Δt is used
here to prevent confusion with the period of either the power source or modulating
sinusoids and to reflect the analogy to Riemann Sums. The next portion of the algorithm
necessary to effect Equation 12 is the curve relating to the 0.5(1+M sin(ω_{m}nΔt))factor.
To effect that portion of the equation, the triggering of gates in the primary
chopper 50, discussed below in FIGURE 3, will delay the onset of a square-wave
120 with a phase angle delay of Φ(nΔt)=0.5(1+M sin(ω_{m}nΔt))2πΔt
radians where ω_{m} is the frequency of the modulating sinusoid
130 referred to above. For each period, Δt, commencing at either the
leading or the falling edge of the clock pulse 122, the pulse will remain logically
high for a proportion of the Δt period equal to 0.5(1+M sin(ω_{m}nΔt))
and then immediately fall to negative logic levels 124. Sending the clock curve
110 and the pulse-width-modulated square-wave pulse through an XOR gate
results in the curve shown as curve 140. The result is a digital gate capable of
directing switching at a rate to effect the 0.5(1+M sin(ω_{m}nΔt))
phase modulation.

Curve 140 is a phase-angle-modulated square-wave with glitches 142.
In digital logic, a glitch is a sudden break in function or continuity, usually
caused by switching error, of a transient nature. Here, during the period when
curves 110 and 120 simultaneously transition from low to high or high to low, a
glitch occurs. Because the logic is keyed to the clock curve 110, this glitch 142
is very transient but, it is, nonetheless, noted here. The glitch has no functional
significance in this application because the inductive properties of the transformer
remove the affects of the glitch 142 from practical consideration. Nonetheless,
the glitch 142 serves as a benchmark for discussion of later processing.

With the resulting phase-angle-modulated curve 140, the invention
is capable of appropriately chopping the incoming power curve. Because power line
voltage levels would burn up most digital logic chips, an appropriate gating apparatus
is necessary. FIGURE 3 portrays a schematic diagram of one an exemplary gating
device 52 that is a part of the primary chopper 50 where the switching devices
are shown as actuated switches. Although any suitable switching device will serve,
the gating device 52 comprises gangs of switching devices 501, 502, 503, and 504;
that is the gating device 52 is a simple bridge. A logic signal fed to diagonal
pairs of switching devices (e.g. 501 and 503 or 502 and 504) opens and closes the
switching devices to appropriately pass the power curve to the transformer 55.

As indicated in the discussion of FIGURE 2, curve 140 is the output
of an XOR
gate (not shown). Most commercial XOR gates also provide
the complementary logic curve. In the complement, logical highs in the original
curve correspond to logical lows in the complement and vice versa. The output of
the XOR gate is fed to the gate terminals 15 and 17 of one diagonal pair
of switching devices 501 and 503, respectively. The complementary output is fed
to the gate terminals 16 and 18 of the remaining diagonal pair of switching devices
502 and 504, respectively. By this logical scheme, a voltage potential corresponding
to the amplitude of the power curve is fed to the primary side of the transformer
55, with rapidly reversing polarity according to the logical transitions of the
output of the XOR gate. The waveform at the primary terminals at the transformer
55 is a square wave at the clock frequency and power levels; it is offset from
the clock wave by a time period corresponding to the magnitude of the modulating
sine according to Equation 13. The wave is high frequency and symmetrical and thus
allows the use of a smaller transformer.

After the primary chopper 50 sends this output wave through the transformer
55, the secondary chopper 60 demodulates the waveform so as to remove the half
amplitude power wave that remains:
V_{out}(nΔt) = V_{in}
× [0.5(1 + M sin(ω_{m}n2πΔt))]
- 0.5V_{in} = MV_{in} sin(ω_{m}n2πΔt)
The secondary chopper 60 uses one half-cycle of the clock pulse 110 to invert that
portion of the wave that either precedes or follows the clock transition 144. By
this means, the secondary chopper inverts the curve between every other glitch
portrayed in FIGURE 2. The same gating device 52 shown in FIGURE 3 is suitably
used in the secondary chopper. The resulting pulse-width-modulated curve 120 now
determines the periods of conductivity, and the clock transition 110 causes the
reversal of the polarity of the output. Because the demodulation is synchronized
to the modulation, the resulting waveform is the product of sines of Equation (3)
shown in FIGURE 4 as curve 160. By conducting that output to the load in delta
and filtering the same with a low pass LC filter, 72 and 76, the invention produces
the power described in Equation 8: a smooth single-phase power sinusoid at the
desired frequency.

Another inventive aspect of the present invention is the use of a
feedback loop to produce the modulating waveform as the difference of frequencies,
ω_{m}, used for modulation. As indicated in Equation 3 above,
the modulating frequency, ω_{m}, will have a value of the
difference between the desired output or reference frequency, ω_{ref},
and the source frequency, ω_{s}. As frequencies are not readily
added, the trigonometric identity set forth in Equation 2 is, again, employed to
generate the modulating frequency, ω_{m}.
V_{sum} = V_{peak} 3 / (2)(cos(t(ω_{s}- ω_{m})))=V_{peak}3 / (2)(cos(t(ω_{ref})))

The interrelation of the power and control circuits provides a feedback
loop that assures the regularity of the output waveform by assuring that the difference
between the input frequency and the sum of the input and modulating frequencies
remains the same. To maintain this difference in the feedback loop, the invention
draws from the input power curves from each phase of the power from the source
20. For each phase, the inventive device uses as input a reference waveform representing
the desired output frequency. This might be a waveform drawn from the power grid,
from a local oscillator or from a look-up table.

Using this source frequency, ω_{s}, from above
and ω_{ref}, the invention generates a product of the sines
according to the same trigonometric identity:
sin u×cos v=1 / (2)[sin(u+v)+sin(u-v)] Γ_{A} = sin ω_{s}t×cosω_{ref}t
= 1 / (2) [sin(ω_{s} + ω_{ref})t
+ sin(ω_{s} - ω_{ref})t]

The mathematics is easily performed at signal levels using an analog
four-quadrant multiplier, a multiplying digital-to-analog converter, or state machine
recalling a sine form from a memory look-up table, for example. The feedback loop
employs one small but critical difference between feedback and the forward algorithms
is the direction of rotation, or phase sequence, of the reference sine multipliers.
For the feedback path, the direction is the reverse of the modulator set shown
in the forward path. That is why the sum frequency, rather than the difference
frequency, is preserved. The difference between phases results in a simple sine
wave at a frequency equal to the sum of the input and modulation frequencies. For
instance, as between the B and C phases:

The result is a sinusoid at a frequency representing the sum of the
input and the reference frequencies, ω_{s} + ω_{ref}
= ω_{m}, as the modulator for the input power A phase. The
other two possible differences will yield similar results for the B and C phases
of the input. Where the source frequency departs from a given value, the difference
of frequencies term in Equations 3, 4, and 5 will remain constant; the sum of frequencies
terms cancels in the resulting Equations 7, 8, and 9 and therefore has no effect
upon the resulting waveform. The inventive device reliably provides constant frequency
power.

With the feedback process in place, the invention is self-regulating
such that fluctuations in the input frequency do not affect the output frequency.
So, too, as earlier identified, the amplitude of the output sine is governed by
the value of the modulating constant, M, in Equation 14. M can be used to control
a simple feedback loop.

It is worthwhile to touch on the resulting unity or near unity power
factor. Recalling that at the stage of the inventive device where the power is
routed through the transformer 55, the square-waves are at a frequency much higher
than that of either the source or the output power. After secondary chopping at
the chopper 60, the fundamental frequency does drop but still remains much higher
than the source and input power frequencies. The inductive filter element 72 (FIGURE
1a) and the capacitive filter element 76 (FIGURE 1a) can energize and de-energize
at frequencies that are two orders of magnitude higher than that of the output
power. It will be appreciated, though, that the filter elements 72 and 76 may energize
and de-energize at frequencies that are either greater than or less than two orders
of magnitude higher than that of the output power, as desired. Thus such leading
and lagging effects that they might introduce are de minimus
at the output.

FIGURE 5 sets forth a method 205 of producing programmable three-phase
voltage and frequency according to the invention. Electrical power is received
from a driven synchronous machine at block 210. To construct the feedback leg,
at block 215, each phase of the source power is sampled. A reference sinusoid is
drawn from either the power grid or a generated oscillation at block 220.

Having the two distinct logic-level voltage sine waves, the device
then multiplies the sine values to create a product of sines for each phase according
to Equations 15, 16, and 17 at block 225. At block 230, subtracting each phase
from the next adjacent phase to result in three distinct sinusoids, each at a frequency
representing the sum of the source frequency and the desired output frequency,
according to Equation 18, results in three modulation sinusoids, each for one source
phase.

At block 235, the invention employs one of two modes of the feedback
process. Because the magnitude of the output voltage is proportionate to the coefficient
used to scale the modulation sinusoid, the method monitors the output voltage and
adjusts the coefficient between zero and one to achieve a constant or a programmable
output voltage. The scaling is a dynamic and constant process thus overcoming fluctuations
in the source voltage. As properly scaled, the modulating sinusoid then dictates
the phase-angle modulation of the source power as it travels through the transformer.

The invention phase-angle creates and then continuously shifts a significantly
higher frequency square-wave at block 235. The shifting is according to the corresponding
modulation sinusoid for the phase of the source power to be modulated. At block
240, the polarity of the corresponding source power phase is then periodically
reversed according to this phase-angle shifted square wave. The resulting power
wave is sent through the primary terminals of a transformer 55 (FIGURE 1a) at block
245. At block 250, on the secondary side, the resulting voltage is chopped and
re-reversed according to the original square-wave to produce a pulse-width-modulated
product of the source power phase and the "sum of the frequencies" modulating sine
wave 160. Filtering the sine to integrate the resulting wave results in a smooth
wave at 255. The power is supplied to the load in delta connection at block 260.

While the preferred embodiment of the invention has been illustrated
and described, as noted above, many changes can be made without departing from
the spirit and scope of the invention. For example, four, six, or eight phase power
is equally susceptible to frequency and voltage programming according to the same
modulation scheme. Furthermore, many simplifications to the presented embodiment
are possible, including operation without a transformer, and the algorithm shown
is applicable, or adaptable. Accordingly, the scope of the invention is not limited
by the disclosure of the preferred embodiment. Instead, the invention should be
determined entirely by reference to the claims that follow.

Anspruch[en]

A direct conversion programmable power controller receiving three-phase input
power at an input power frequency and having a polarity, the controller comprising:

a primary chopper for each phase of electrical power, each primary chopper
having an input, electrically connected in wye connection to each phase of power,
having an output;

a two-winding transformer for each phase of power, each transformer having
primary and secondary terminals and connected electrically by its primary terminals
to the output of the primary chopper; and

a secondary chopper for each phase of power, each secondary chopper having
an input connected electrically to the secondary terminals of the transformer and
connected with the remain phases in series producing single-phase output power.

The controller of Claim 1 wherein the primary chopper periodically reverses
the polarity of each phase of power from the power source at a frequency significantly
higher than the input power frequency.

The controller of Claim 2 wherein the primary chopper periodically reverses
the polarity of the phase of power from the power source wherein an onset of the
polarity reversal within a period varies according to a sinusoid.

The controller of Claim 3 wherein the sinusoid has a frequency equal to difference
between the frequency of the power input frequency and a desired frequency of the
output.

The controller of Claim 4 wherein a feedback loop generates the sinusoid according
to the power input frequency and a reference sinusoid.

The controller of Claim 5 wherein the feedback loop samples the input power
to determine the power input frequency.

The controller of Claim 5 wherein the feedback loop samples a power grid to
generate the reference sinusoid.

The controller of Claim 5 wherein the feedback loop generates the reference
sinusoid by means of a state machine.

The controller of Claim 5 wherein the feedback loop generates the reference
sinusoid by means of a local oscillator.

The controller of Claim 5 wherein the feedback loop generates the reference
sinusoid from values stored in a look-up table.

The controller of Claim 5 wherein the feedback loop generates the reference
sinusoid by means of a GPS receiver.

The controller of Claim 5 wherein the secondary chopper periodically reverses
the polarity of the phase of power from the transformer at the frequency significantly
higher than the input power frequency.

The controller of Claim 12 wherein the secondary chopper conducts the phase
of the power according to a duty-cycle within each period and where the secondary
chopper periodically reverses the polarity of the phase of power.

The controller of Claim 13 wherein the duty-cycle varies according to the sinusoid.

The controller of Claim 13 wherein feedback loop samples the power at the load
and varies a coefficient between zero and one according to the amplitude of the
sampled power.

The controller of Claim 15 wherein the duty-cycle varies according to the product
of the coefficient and the sinusoid.

The controller of Claim 13 wherein the secondary chopper modulates the input
power applied to the load according to the sinusoid.

The controller of Claim 1 wherein the transformer for each phase of power comprises
a multi-phase two winding core transformer.

The controller of Claim 1 wherein the transformer for each phase of power comprises
a "step-up" transformer.

The controller of Claim 1 wherein the transformer for each phase of power comprises
an "isolation" transformer.

The controller of Claim 1 wherein the transformer for each phase of power comprises
a "step-down" transformer.

The controller of Claim 1, further comprising an input low-pass power conditioning
filter including:

an inductive element electrically connected in series between the three-phase
power source and the input of the primary chopper; and

a capacitive element electrically connected in to the input of the primary
chopper in parallel with the three-phase power source.

The controller of Claim 1, further comprising an output low-pass power conditioning
filter including:

an inductive element electrically connected in series between the load and

the output of the secondary chopper; and

a capacitive element electrically connected in to the output of the secondary
chopper in parallel with the load.

A method for controlling three-phase electrical power, the power having a polarity
and power frequency, the method comprising:

generating a first electrical signal, having a first frequency greater than
the power frequency of the three-phase electrical power;

generating three second signals each of the second signals having a second
frequency and being offset from the other second signals by phase-angles of 2π/3
radians;

providing the three-phase electrical power from a wye connected power source,
each phase of electrical power being offset from the other phases of electrical
power by phase-angles of 2π/3 radians;

phase-angle-modulating the first signal according to each second signal to
produce a third signal at the first frequency to correspond with each phase of the
three-phase power;

reversing the polarity of each phase of electrical power at the first frequency
according to the third signal that corresponds with the phase of electrical power;

providing a two-winding transformer having primary and secondary terminals
for each phase of electrical power;

applying the phase of electrical power and the reversed phase of electrical
power across each set of the primary terminals;

chopping each phase of electrical power induced across each set of secondary
terminals to pulse-width-modulate each phase of electrical power according to an
instantaneous value of the second signal; and

applying the pulse-width-modulated electrical power from the secondary terminals
to a load wherein the individual phases are fed in series to the load.

The method of Claim 24, wherein providing three-phase electrical power includes
filtering the power to remove ripple.

The method of Claim 24, wherein applying the pulse-width-modulated electrical
power includes filtering to integrate the output.

The method of Claim 24, wherein generating a first electrical signal includes
generating a square-wave.

The method of Claim 24, wherein generating a first electrical signal includes
generating a clipped-sine wave.

The method of Claim 24, wherein generating a second signal includes generating
a second signal at a frequency representing the sum of the frequency of the power
frequency and a frequency of a desired output.

The method of Claim 29, wherein generating a second signal further includes
scaling the second signal to regulate the three-phase electrical power applied
to the load.