Note: Descriptions are shown in the official language in which they were submitted.
Patent
Atty. Docket No. 1459-0032US01
HIGH VOLTAGE, HIGH EFFICIENCY SINE WAVE GENERATOR
WITH PRE-SET FREQUENCY AND ADJUSTABLE AMPLITUDE
BACKGROUND
[0002] Class-D amplifiers are electronic amplifiers in which the
amplifying device
operates as an electronic switch, instead of as linear gain devices as in
other amplifiers. In
class-D amplifiers, the analog signal is amplified using a system of switches
that convert the
signal into a series of pulses by pulse width modulation, pulse density
modulation or another
method. The pulses that make up the amplified signals are then passed through
a low pass
filter that filters out the high frequency signal components to form an
amplified version of the
original signal. The filter is constructed from capacitive and inductive
elements so that the
energy losses associated with the filter are low. Class-D amplifiers are
designed to track the
input signal in real-time, and usually work in the frequency range of up to
tens of kHz,
making them suitable for amplifying audio signals.
[0003] The efficiency of class-D amplifiers is above 90%. This is
because the
amplified signal is binary and has segments in which the current passing
through the switch is
zero and the voltage finite, and segments in which the current through the
switch is finite and
the voltage is zero. In an electric circuit, heat loss is equal to the product
of current and
voltage. Therefore, in a system in which either the voltage or current are
equal to zero most
of the time, heat losses will be minimal. Class D amplifiers are therefore
highly efficient
amplifiers with minimal losses.
1
CA 2942319 2018-06-13
CA 02942319 2016-09-19
[0004] Using TTFields therapy to treat tumors is described in US patent
7,805,201,
and TTFields therapy requires the generation of a high voltage sinusoidal
signal. Previously,
generating this sinusoidal signal was implemented by generating a low
amplitude signal with
a function generator, then amplifying this signal using a linear amplifier and
subsequently
applying the signal to the electrodes that are positioned on the patient's
body. The use of
linear amplifier results in heat losses of close to 50%, reducing battery
life, and complicating
device design because a cooling system is required to dissipate the heat
generated.
[0005] Implementing a high-efficiency digital signal generation/amplifi
cation
architecture for TTFields would be beneficial. One possible approach for
building such a
system could be to replace the linear amplifier in the system with a Class-D
amplifier.
However, existing Class-D amplifier technology is not suitable for this task
for two reasons.
First, the signal distortion associated with Class-D amplifiers increases as
the signal
amplitude decreases, and TTFields requires sinewaves with very low levels of
distortion at all
signal levels. And second, TTFields therapy requires the generation of
sinewaves at
frequencies greater than 100 kHz. This requires very fast switches and complex
control that
are currently unavailable.
SUMMARY OF THE INVENTION
[0006] One aspect of the invention is directed to a first apparatus for
generating a
sinusoid at a pre-set frequency f. The first apparatus includes a DC power
source having a
voltage-control input that sets an output voltage of the DC power source, and
also includes a
transformer having a primary and a secondary, The first apparatus also has a
power switch
having a control input, and the power switch is configured to apply the output
of the DC
power source to the primary of the transformer in a first direction when a
first control signal
is applied to the control input, apply the output of the DC power source to
the primary of the
2
CA 02942319 2016-09-19
transformer in a second direction when a second control signal is applied to
the control input,
and remain off when neither the first control signal nor the second control
signal is applied to
the control input. The second direction is oppositc to the first direction.
100071 'File first apparatus also has a sequencer that (a) applies the
first control signal
to the control input for a duration of T/3, then (b) waits for a duration of
T/6, then (c) applies
the second control signal to the control input for a duration of T/3, and then
(d) waits for a
duration of T/6, then continuously repeat the sequence (a), (b), (c), and (d).
T is the
reciprocal of the pre-set frequency f. The first apparatus also has an output
filter connected to
the secondary of the transformer. The output filter passes the pre-set
frequency f and
attenuates frequencies above a cut-off frequency. The cut-off frequency is
between 2f and 4f,
and the output filter has a transfer function with a zero at 5f.
100081 Optionally, the first apparatus further includes a controller
programmed to
control an amplitude of the sinusoid at the pre-set frequency by adjusting a
third control
signal that is applied to the voltage-control input of the DC power source.
Optionally, it may
also include a sensor that detects an output current of the apparatus. The
controller is
programmed to increase the amplitude of the sinusoid when the detected output
current is
lower than a desired current, and to decrease the amplitude of the sinusoid
when the detected
output current is higher than the desired current Optionally, the sequencer
may be
configured to inhibit generation of both the first control signal and the
second control signal
when the output current detected by the sensor is indicative of an error
condition
100091 Optionally, the transfer function of the output filter in the first
apparatus may
have an additional zero at 7f. Optionally, this output filter may be a low
pass elliptic filter.
Optionally, the cut-off frequency of this output filter is at or near 3f.
3
CA 02942319 2016-09-19
=
[0010] Optionally, the frequency of the output filter of the first
apparatus is at or near
3f.
100111 Optionally, the output filter of the first apparatus is a
multi-stage low pass LC
filter, and a leakage inductance of the transformer provides at least half of
the inductance of a
first stage of the low pass LC filter. Optionally, the output filter of the
first apparatus is a
multi-stage low pass LC filter, and a leakage inductance of the transformer
provides all of the
inductance of a first stage of the low pass LC filter.
[0012] Optionally, the output filter of the first apparatus has an
output impedance of
70 ohms. Optionally, the output filter of the first apparatus has an output
impedance between
40 and 120 ohms. Optionally, the pre-set frequency f is 200 kHz. Optionally,
the pre-set
frequency f is 150 kHz, Optionally, the power switch of the first apparatus
has an H-bridge
configuration
100131 Another aspect of the invention is directed to a second
apparatus for
generating a sinusoid at a pre-set frequency f. The second apparatus includes
n DC power
sources, each of the n DC power sources having a voltage-control input that
sets an output
voltage of the respective power source, wherein n is a positive integer. The
second apparatus
also includes a transformer having a primary and a secondary, and a power
switch having a
control input. "File power switch is configured to either (a) apply the output
of a selected one
of the n DC power sources to the primary of the transformer in a selected
direction in
response to 2n states of a control signal that is applied to the control input
or (b) remain off in
response to an additional state of the control signal.
[0014] The second apparatus also includes a sequencer that
controls the generation an
oversampled version of a sine wave that is sampled N times per cycle using
evenly spaced
samples that include a sampling point at 0 , where N---2+4n, by setting the
output voltages of
4
CA 02942319 2016-09-19
the n DC power sources to levels that are present on the oversampled version
of the sine
wave, and then sequencing the control signal through the 2n states and the
additional state, so
that each of the n DC power sources is applied to the primary of the
transformer in each
direction at appropriate times in a sequence so as to generate the oversampled
version of the
sine wave.
100151 The second apparatus also includes an output filter connected to the
secondary
of the transformer. The output filter passes the pre-set frequency f and
attenuates frequencies
above a cut-off frequency, and the output filter has a transfer function with
a zero at a
frequency where a harmonic of the pre-set frequency f is expected to contain
power. The
controller controls an amplitude of the sinusoid by adjusting the output
voltages of the n DC
power sources via the voltage-control inputs, while maintaining a fixed ratio
between the
output voltages of each of the n DC power sources.
100161 Optionally, the transfer function of the output filter in the second
apparatus
has an additional zero a next frequency where a harmonic of the pre-set
frequency f is
expected to contain power. Optionally, the cut-off frequency of the output
filter in the second
apparatus is at or near 31 Optionally, the output filter in the second
apparatus is a multi-stage
low pass LC filter, and a leakage inductance of the transformer provides at
least half of the
inductance of a first stage of the low pass LC filter. Optionally, the output
filter in the second
apparatus is a multi-stage low pass LC filter, and a leakage inductance of the
transformer
provides all of the inductance of a first stage of the low pass LC filter.
100171 Optionally, the second apparatus further includes a second output
filter and a
switch. The second output filter passes a second pre-set frequency 12 and
attenuates
frequencies above a second cut-off frequency. The second output filter has a
transfer
function with a zero at a frequency where a harmonic of the second pre-set
frequency 12 is
CA 02942319 2016-09-19
expected to contain power. The switch selectively connects either the output
filter or the
second output filter to the secondary of the transformer.
100181 Optionally, the output filter of the second apparatus includes at
least one
component with a tunable reactance.
[0019] Another aspect of the invention is directed to a third apparatus for
generating a
sinusoid at a pre-set frequency f. The third apparatus includes n DC power
sources, each of
the n DC power sources having a voltage-control input that sets an output
voltage of the
respective power source, wherein n is a positive integer. The third apparatus
also includes a
power switch having output terminals and a control input. The power switch is
configured to
either (a) switch the output of a selected one of the n DC power sources
across the output
terminals in a selected direction in response to 2n states of a control signal
that is applied to
the control input or (b) remain off in response to an additional state of the
control signal.
100201 The third apparatus also includes a sequencer that controls the
generation an
oversampled version of a sine wave that is sampled N times per cycle using
evenly spaced
samples that include a sampling point at 00, where N=2+4n, by setting the
output voltages of
the n DC power sources to levels that are present on the oversampled version
of the sine
wave, and then sequencing the control signal through the 2n states and the
additional state, so
that each of' the n DC power sources is switched across the output terminals
in each direction
at appropriate times in a sequence so as to generate the oversampled version
of the sine wave.
[0021] The third apparatus also includes an output filter that filters a
signal received
from the output terminals of the power switch. The output filter passes the
pre-set frequency
f and attenuates frequencies above a cut-off frequency, and the output filter
has a transfer
function with a zero at a frequency where a harmonic of the pre-set frequency
f is expected to
contain power. The controller controls an amplitude of the sinusoid by
adjusting the output
6
CA 02942319 2016-09-19
voltages of the n DC power sources via the voltage-control inputs, while
maintaining a fixed
ratio between the output voltages of each of the n DC power sources.
100221 Optionally, the transfer function of the output filter in the third
apparatus has
an additional zero at a next frequency where a harmonic of the pre-set
frequency f is expected
to contain power. Optionally, the cut-off frequency of the output filter in
the third apparatus
is at or near 31
[0023] Optionally, the third apparatus further includes a second output
filter and a
switch. The second output filter passes a second pre-set frequency f2 and
attenuates
frequencies above a second cut-off frequency. The second output filter has a
transfer
function with a zero at a frequency where a harmonic of the second pre-set
frequency f2 is
expected to contain power. The switch selectively connects either the output
filter or the
second output filter to the secondary of the transformer.
[0024] Optionally, the output filter of the third apparatus includes at
least one
component with a tunable reactance.
[0025] Another aspect of the invention is directed to a first method for
generating a
sinusoid at a pre-set frequency f The first method includes setting n DC power
sources to
respective output voltages, where n is a positive integer. The first method
also includes
generating an oversampled version of a sine wave that is sampled N times per
cycle using
evenly spaced samples that include a sampling point at DO, where N=2+4n, by
setting the
output voltages of the n DC power sources to levels that are present on the
oversampled
version of the sine wave, and then switching the outputs of the n DC power
sources to an
output in a controlled sequence such that each of the n DC power sources is
switched to the
output in each direction at appropriate times in the sequence so as to
generate the
oversampled version of the sine wave.
7
CA 02942319 2016-09-19
100261 The first method also includes filtering the oversampled version of
the sine
wave to pass the pre-set frequency f and attenuate frequencies above a cut-off
frequency.
The filtering implements a transfer function with a zero at a frequency where
a harmonic of
the pre-set frequency f is expected to contain power. The amplitude of the
sinusoid is
controlled by adjusting the output voltages of then DC power sources,
[00271 Optionally, n in the first method equals 1. Optionally the cut-off
frequency is
at or near 3f.
[0028] Alternatively, n in the first method is greater than 1, and the
amplitude of the
sinusoid is controlled by adjusting the output voltages of the n DC power
sources, while
maintaining a fixed ratio between the output voltages of each of then DC power
sources.
100291 Optionally, the transfer function in the first method has an
additional zero at a
next frequency where a harmonic of the pre-set frequency f is expected to
contain power.
BRIEF DESCRIPTION OF THE DRAWINGS
100301 FIG. 1 is a block diagram of a first embodiment of a sinusoid
generator that
generates a sinusoid at a pre-set frequency f, with a controllable amplitude.
100311 FIG. 2 depicts a block diagram of one preferred approach for
implementing
the power switcher and a suitable architecture for implementing the output
filter.
[0032] FIG. 3 depicts a sine wave and an oversampled version of that sine
wave that
is sampled 6 times per cycle.
[0033] FIG. 4 is a schematic diagram of an embodiment of the output filter.
100341 FIG. 5 is a block diagram of a second embodiment of a sinusoid
generator that
generates a sinusoid at a pre-set frequency f, with a controllable amplitude
8
Patent
Atty. Docket No. 1459-0032US01
[0035] FIG. 6 is a block diagram of one preferred approach for
implementing the
power switcher in the FIG. 5 embodiment.
[0036] FIG. 7 depicts a sine wave and an oversampled version of that
sine wave that
is sampled 10 times per cycle.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0037] The embodiments described herein are useful in connection with
generating
TTFields, as described in US patent 7,805,201. The embodiments described
herein can be
used to generate the high voltage sinusoidal signals, as required for
TTFields, using a high
efficiency architecture. This enables the reduction in size and weight of the
device that
generates the high voltage sinusoids and of the associated batteries.
[0038] Note that when generating a high-voltage signal for TTFields
delivery, the
exact shape of the signal is known at every moment (pure sine wave at a known
frequency)
and it is only the amplitude of the output signal that changes over time,
based on external
inputs (e.g., control based on the skin temperature of the patient).
Furthermore, the rate at
which the output signal has to change is very slow, as the decision to change
the signal level
is made a long time before the devices changes the level, on the timescale of
one or more
seconds. This relatively long timescale stands in contrast to the rapid
response-time provided
by conventional class-D amplifiers in which the exact shape of the
input/output signals is
unknown and changes are made in real-time on a millisecond or microsecond time
scale.
[0039] The embodiments described herein generate high voltage sinusoidal
signals by
generating a specific pulse train that, when filtered using a specific low
pass filter, will result
in a low distortion sine wave of the desired amplitude and frequency.
9
CA 2942319 2018-06-13
CA 02942319 2016-09-19
100401 FIG. 1 is a block diagram of a first embodiment of a sinusoid
generator that
generates a sinusoid at a pre-set frequency f, with a controllable amplitude.
Ultimately, the
amplitude of the output sinusoid will be proportional to the output of the DC
power source
50, which is preferably a controlled DC-DC converter.
100411 In the illustrated embodiment, the DC to DC converter 50 is
configured to
multiply an analog voltage-control input signal by 10, so when a 1 V voltage-
control signal is
applied the output will be 10 V, and when a 5 V voltage-control signal is
applied the output
will be 50 V. with proportional control there between. The output of the DC-DC
converter
50 can therefore take any value between 0 and 50 -V, depending on the voltage
(e.g., 0-5 V)
that is applied to the analog voltage-control input. A controller 40 controls
the output voltage
of the DC-DC converter 50 by writing a control word to a digital-to-analog
converter (DAC)
42. The DAC 42 then generates an analog voltage that is proportional to the
control word,
and this analog voltage is applied to the voltage-control input of the DC-DC
converter 50.
100421 The output of the DC-DC converter 50 is routed to the power
switcher 60.
The power switcher 60 has a control input, and depending on the state of the
control input, it
will route the output of the DC-DC converter 50 to the primary of the
transformer 70 in either
direction. More specifically, when a first control signal is applied to the
control input, the
power switcher 60 will apply the output of the DC-DC converter 50 to the
primary of the
transformer 70 in a first direction. When a second control signal is applied
to the control
input, the power switcher 60 will apply the output of the DC-DC converter 50
to the primary
of the transformer 70 in a second direction that is opposite to the first
direction. When neither
the first control signal nor the second control signal is applied to the
control input, the power
switcher 60 will remain off, in which case power from the DC-DC converter 50
is not routed
to the primary of the transformer 70.
CA 02942319 2016-09-19
100431 FIG. 2 includes a block diagram of one preferred approach for
implementing
the power switcher 60 using a set of four electronically controlled switches
61-64 connected
to the primary of the transformer 70 in an H-bridge configuration. These
switches 61-64
open and close in response to signals that are applied to a control input 68.
A wide variety of
technologies may be used for implementing these switches, as will be
appreciated by persons
skilled in the relevant arts. For example, the switches 61-64 may be
implemented using
MOSFET transistors (e.g., BSC109NIONS3 manufactured by Infineon) along with
appropriate logic to switch them on and off in response to a control signal.
In order to apply
the output of the DC-DC converter 50 to the primary of the transformer 70 in
the first
direction, only switches 63 and 62 should be closed. In order to apply the
output of the DC-
DC converter 50 to the primary of the transformer 70 in the opposite
direction, only switches
61 and 64 should be closed. When all four of these switches 61-64 are off, no
power is
routed into the primary of the transformer 70.
100441 Transformer 70 is preferably a step-up transformer with a step-up
ratio
between 1:4 and 1:9. In some preferred embodiments, transformer 70 is a step-
up transformer
with a step-up ratio of 1:6. For example, when a transformer with a 1:6 step-
up ratio is used
in combination with a DC-DC converter 50 that can output up to 24 V. the
resulting voltage
at the secondary of the transformer 70 can go as high as 300 V.
[0045] Returning to FIG. 1, the sequencer 55 applies control signals to the
control
input of the power switcher 60 in a time-choreographed sequence in order to
construct an
oversampled version of a sine wave that is sampled six times per cycle using
evenly spaced
samples. More specifically, FIG. 3 depicts a sine wave 110 and an oversampled
version of
that sine wave 112 that is sampled at 0 , 60', 120', 180', 240 , 300 , and 360
. Looking at
this oversampled version 112, it becomes apparent that it contains only three
voltage levels: a
11
CA 02942319 2016-09-19
positive voltage IV between 60 and 180 , a negative voltage -V between 240
and 3600, and
zero volts between 0 and 60 and also between 180 and 240'. Note that the
zero Volt level
exists because we have chosen the sampling times such that one of the sampling
points
occurs at 0 and another one of the sampling points occurs at 180 , where the
sine function
equals zero. This choice advantageously reduces the number of voltage levels
that must be
generated to construct the oversampled version 112 of the sine wave. It also
advantageously
reduces the number of switching events, which minimizes losses that are
incurred during the
switching process.
10046] As a result, an oversampled version of a sinusoid at a pre-set
frequency f can
be constructed at the output of the transformer 70 by continuously repeating
the following
four steps: (a) applying the first control signal to the control input 68 for
a duration of T/3,
which corresponds to the 60-180 segment of waveform 112 in FIG. 3; then (b)
waiting for a
duration of T/6, which corresponds to the 180-240 segment of waveform n 112;
then (c)
applying the second control signal to the control input 68 for a duration of
T/3, which
corresponds to the 240-360 segment of waveform 112; and then (d) waiting for
a duration of
T/6, which corresponds to the 0-60 segment of waveform 112. Note that T is
the reciprocal
of the pre-set frequency f.
[0047] The sequencer 55 is responsible for generating these control signals
in this
sequence. The sequencer 55 may be implemented using a wide variety of
approaches that
will be apparent to persons skilled in the relevant arts including but not
limited to state
machines, counters, and microcontrollers.
100481 The output of the secondary of the transformer 70 is routed to an
output filter
80 that has a cut-off frequency between 2f and 4f. The output filter 80 passes
the pre-set
frequency f and attenuates frequencies above the cutoff frequency.
12
CA 02942319 2016-09-19
100491 Note that when the oversampled version of the sine wave (112 in FIG.
3) is
converted to the frequency domain, all of the even harmonics will be zero as a
result of the
fact that waveform 112 being symmetric. In addition, because sampling is
performed 6 times
per period, the third harmonic of waveform 112 will also be zero.
100501 Many filter designs have inherent instabilities at their cutoff
frequencies. But
because the third harmonic component of the oversampled waveform 112 is zero,
the lowest
harmonic that will have any significant power will be the fifth harmonic. If
the output filter
80 is designed so that its cutoff frequency coincides with the third harmonic,
the oversampled
waveform 112 will not be affected by the instabilities in the vicinity of the
cutoff frequency,
because the waveform contains no power at 3f It is therefore most preferable
to design the
output filter 80 with its cutoff frequency at 3f, in which case (a) the
fundamental component
will be far enough below the cutoff frequency so as not to activate the
instabilities and (b) the
fifth harmonic will be far enough above the cutoff frequency so as not to
activate the
instabilities.
100511 To further reduce the higher order harmonics, the output filter 80
is preferably
designed so that the transfer function of the output filter has a zero located
at the fifth
harmonic. This may be accomplished, for example, by selecting the components
within the
output filter 80 to implement an elliptic low pass filter or a Chebyshev-2 low
pass filter.
Ordinarily, elliptic filters and Chebyshev-2 filters are not suitable for
filtering square waves
into sine waves because they have significant ripple in the stop band As a
result, if an
incoming signal happens to contain a frequency component that coincides with a
crest within
that ripple, that component would not be filtered out from the incoming
signal. The FIG. 1
embodiment avoids this situation by generating the oversampled waveform 112 at
a pre-set
frequency, which means that the frequency of the fifth harmonic will be known
in advance.
13
CA 02942319 2016-09-19
By selecting the components within the output filter 80 so that its transfer
function has a zero
at the fifth harmonic, we ensure that the fifth harmonic will never coincide
with a crest within
the ripple in the stop band.
10052] To reduce the higher harmonics even further, the output filter 80
may be
designed so that its transfer function has an additional zero located at the
seventh harmonic.
Here again, because the frequency of the seventh harmonic will be known in
advance, the
components within the output filter 80 can be selected so that its transfer
function has a zero
at the seventh harmonic.
[0053] Designing the output filter 80 with zeros at the fifth and seventh
harmonics
reduces the attenuation at other frequencies located between the harmonics,
which would
ordinarily be very undesirable. However, because the frequency of the
oversampled
waveform 112 is pre-set in advance and because it only contains signals
centered around the
odd harmonics (starting with the fifth harmonic), this design will actually
decrease the overall
distortion of the output signal in the FIG. I embodiment.
[0054] When the output filter 80 is designed with zeros at the fifth and
seventh
harmonics, the initial harmonic that will contain any significant power will
be the ninth
harmonic. But because the power in the ninth harmonic of the oversampled
waveform 112 (in
FIG. 3) is relatively low to begin with, and because the ninth harmonic is 6f
above the cutoff
frequency, the power in the ninth harmonic (and all higher harmonics) at the
output 100 of
the output filter 80 will be low enough to produce an excellent sine wave.
[0055] FIG. 2 depicts a suitable architecture for implementing the output
filter 80
with the cutoff frequency and the zeros at the locations indicated above.
Preferably, the
output filter 80 is a multi-stage low pass LC filter. In this case, the first
stage of the output
filter 80 comprises inductor 82 and capacitor 83, and the subsequent stages
are represented by
14
CA 02942319 2016-09-19
block 85. In some embodiments, the filter 80 is a fourth order LC low pass
filter. In some
embodiments, the filter 80 is a dual M-type element low pass filter.
[0056] When the electrical characteristics of transformer 70 are modelled,
the leakage
inductance of the transformer appears in series with the secondary of
transformer 70. As a
result, this leakage inductance must be accounted for when calculating the
inductance of the
first inductor 82 in the first stage of the output filter 80. In some
embodiments, a transformer
70 with a leakage inductance that is large enough to supply all of the
inductance that is
needed for the first inductor 82 is selected. In this case, the first inductor
82 can be
eliminated entirely from the output filter 80 and replaced with a wire. For
example, if the
calculated desired value for the first inductor in the output filter is 60 p.H
and the leakage
inductance of the transformer 70 is 60 RH, the first inductor 82 of the output
filter can be
eliminated entirely.
[0057] In alternative embodiments, the leakage inductance of the
transformer 70
accounts for at least half of the inductance of the first stage of the low
pass LC filter. In these
embodiments, we start with the calculated value for the first inductor 82 and
reduce that value
by the leakage inductance of the transformer 70, For example, if the
calculated value for the
first inductor in the first stage of the output filter is 100 [tH and the
leakage inductance of the
transformer 70 is 60 [LH, a 40 [tH inductor should be used as the first
inductor 82 of the
output filter (because 100 p.H - 60 itI-1 = 40 PH).
10058] FIG. 4 is a schematic diagram of an embodiment of the output filter
80 in
which the inductance of the transformer 70 provides all of the inductance that
is needed to
serve as the first inductor for the first stage of the output filter. The
transformer in FIG. 4 is a
Zolotov TRM085, which has the following characteristics: a turn ratio of 6:25;
an inductance
of 0 25 mH in the primary (at 200 kHz); an inductance of 4.5 in the
secondary (at 200
CA 02942319 2016-09-19
kHz); and a leakage inductance between 32 and 36 [LH (at 200 kHz). The
capacitors C33,
C35, C36, C42, C43, and C44 are all 3300 pF capacitors. C40 is a 4.7 nF
capacitor. C41 is a
470 pF capacitor. The inductor L5-L8 are all 4 H inductors. The values of
these
components were selected to position the zeros of the filter at the fifth
harmonic and the
seventh harmonic when the operating frequency is 200 kHz.
100591 An alternative design for implementing an output filter 80 with an
operating
frequency of 150 kHz can be realized by starting with the schematic of FIG. 4
and (a) adding
an additional 4.7 nF capacitor in parallel with C40; and (b) swapping in 5600
pF capacitors in
place of the 3300 pF capacitors C33, C35, C36, C42, C43, and C44. These
components were
selected to position the zeros of the filter at the firth harmonic and the
seventh harmonic
when the operating frequency is 150 kHz.
100601 The output impedance of the output filter 80 is preferably as close
as possible
to 70 ohms. In alternative embodiments, the output impedance of the output
filter 80 is
between 40 and 120 ohms. Using an output impedance in this range is
appropriate because
the current and voltage of the output signal 100 can change depending on the
load that is
presented (i e , the patient and the transducer arrays in the context of
TTFields treatments).
But because the output impedance is between 40 and 120 ohms, even if there is
a short circuit
on the exit, the current will not search to dangerous values. In addition, if
the impedance of
the load suddenly increases (e.g., if an electrode becomes partially
disconnected from a
patient), then the drop in current will be a lot less significant. This is
very useful as a safety
feature in the context of TTFields treatment.
100611 The controller 40 controls the amplitude of the output signal 100
by adjusting
the control signal that is applied to the voltage control input of the DC-DC
converter 50. In
the illustrated embodiment, this is accomplished by having the controller 40
write a control
16
CA 02942319 2016-09-19
word to the DAC 42. The DAC 42 responds by outputting an analogue voltage,
which serves
as the control signal that is applied to the voltage-control input of the DC-
DC converter 50
Assume, for example, that the output of the DAC 42 starts at 1 V, that the DC-
DC converter
is outputting 10 VDC, and that the transformer 70 has a step-up ratio of 1:6.
Under these
conditions, the pulses at the output of the secondary of the transformer 70
will be 60 V.
When the controller 40 writes a new control word to the DAC 42 that causes the
output of the
DAC 42 to increase to 2 V. The DC-DC converter 50 will respond to the new
signal that is
being applied to its voltage-control input by increasing its output voltage to
20 V DC, which
(after passing through the step up transformer 70) will cause the pulses at
the output of the
secondary of the transformer 70 to increase to 120 V.
[00621 Preferably, the voltage and/or current of the output signal 100 are
monitored
by a voltage sense circuit 92 and/or a current sense circuit 94. The output of
these circuits 92,
94 is preferably fed back to the sequencer 55, and the sequencer 55 is
preferably configured
so that when and error condition is detected at the output 100 (e.g.,
overvoltage, overcurrent,
severe voltage drop, etc.), the sequencer will shut down the power switcher 60
by inhibiting
the generation of both the first control signal and the second control signal
that are applied to
the control input 68 of the power switcher 60. Optionally, shut down of the
power switcher
60 may also be triggered by an over-temperature condition at the load by
including
appropriate temperature sensors and routing a signal back from those
temperature sensors to
the sequencer 55.
100631 Optionally, the functionality of the controller 40 and the sequencer
55 may
both be implemented in a single microcontroller that is programmed to perform
the tasks of
both the controller 40 and the sequencer 55.
17
CA 02942319 2016-09-19
[0064] In some embodiments, the output of the current sense circuit 94 and
or the
voltage sense circuit 92 is fed back to the controller 40. In these
embodiments, the controller
can adjust the voltage at the output of the DC-DC converter 50 by writing
appropriate control
words to the DAC 42 in order to adjust the current or voltage of the output
signal 100 to a
desired level. For example, when the controller 40 is set to adjust the
current to a particular
level and the output of the current sense circuit 94 indicate that the current
is too low, the
controller can increase the voltage at the output of the DAC 42, which will
cause an increase
in amplitude at the output signal 100. Similarly, if the output of the current
sense circuit 94
indicate that the current is too high, the controller can decrease the voltage
at the output of the
DAC 42, which will cause a corresponding decrease in amplitude at the output
signal 100.
[0065] In alternative embodiments, the transformer 70 (shown in FIGS. 1
and 2) can
be omitted, in which case the two conductors at the output of the power
switcher 60 are
hooked up directly to the two conductors at the input of the output filter 80.
But these
alternative embodiments are less preferred, especially in situations when
isolation is desirable
and in situations where a high voltage output is desirable. In addition, these
alternative
embodiments cannot rely on the leakage inductance of the transformer to
provide some or all
of the inductance needed for the first stage of the filter.
[0066] Note that the design of the FIG embodiment stands in sharp contrast
with
the design of a conventional class D amplifier, because conventional class D
amplifiers are
designed to handle random input signals that can include frequency components
anywhere
within the operating range of the amplifier. The design of the FIG. 1
embodiment relies on
advance knowledge of the incoming signal, and the intentional construction of
both the signal
and the output filter 80 so that the most significant harmonics are either
inherently zero (e.g.,
the even harmonics and the third harmonic) or zeroed out by the output filter
80 (e.g., the
18
CA 02942319 2016-09-19
fifth and seventh harmonics). This helps provide a very clean high voltage
output signal at
the desired frequency, with very high efficiency.
100671 The FIG. 1 embodiment uses a single DC-DC converter 50, and
implements
six equally-spaced sampling points per cycle. In alternative embodiments, the
number of
sampling points may be increased to N=2+4n, where n is a positive integer.
When n=1, we
have the situation described above in connection with FIG. 1. When n=2, we
have the
situation described below in connection with FIG. 5, which uses two DC-DC
converters.
Other embodiments may be implemented for n > 2 following the same framework
using
additional DC-DC converters and even more samples (following the rule that
N=2+4n).
100681 FIG. 5 is a block diagram of a second embodiment of a sinusoid
generator that
generates a sinusoid at a pre-set frequency IT, with controllable amplitude,
in which n=2. As a
result, there are two DC-DC converters 50, SOB and (following the formula
N=214n) 10
samples per cycle are used. Note that in the FIG. 5-6 embodiment, components
with similar
reference numbers operate in a manner similar to the description above in
connection with the
FIG. 1-2 embodiment.
[0069] FIG. 7 depicts a sine wave 120 and an oversampled version of that
sine wave
122 that is sampled 10 times per cycle (i.e., at 0 , 36 , 72 , ... 324 , and
360 ). Looking at
this oversampled version 112, it becomes apparent that it contains only five
voltage levels: a
low positive voltage V1, a higher positive voltage +V2, a low negative voltage
-V1, a higher
negative voltage -V2, and zero volts (between 0 and 36 and also between 180
and 216').
Here again, the zero Volt level exists because we have chosen the sampling
times such that
one of the sampling points occurs at 0 and another one of the sampling points
occurs at
1800, where the sine function equals zero. This choice advantageously reduces
the number of
19
CA 02942319 2016-09-19
voltage levels that must be generated to construct the oversampled version 122
of the sine
wave to two levels (i.e., V1 and V2).
10070:1 As a result, a sequencer 55B can be used to control the generation
of an
oversampled version of a sine wave that is sampled N times per cycle using
evenly spaced
samples that include a sampling point at 00, where N=2+4n, by setting the
output voltages of
the DC power sources to levels that are present on the oversampled version of
the sine wave,
and then sequencing the control signal through the 2n states and the
additional off state, so
that each of the DC power sources is applied to the primary of the transformer
in each
direction at appropriate times in a sequence so as to generate the oversampled
version of the
sine wave.
100711 When n=2 (as it is in the FIG. 5-6 embodiment), an oversampled
version of a
sinusoid at a pre-set frequency f can be constructed at the output of the
transformer 70 by
continuously repeating the following eight steps: applying Vito the primary of
the
transformer 70 in the first direction between 36 and 72'; applying V2 in the
first direction
between 72' and 144'; applying VI in the first direction between 144 and
180'; remaining
off between 180 and 216'; applying V1 in the second direction between 216'
and 252';
applying V2 in the second direction between 252 and 324'; applying V1 in the
second
direction between 324 and 360'; and remaining off between 00 and 36 . Note
that in order
for the resulting waveform to properly track an oversampled version of a
sinusoid (122 in
FIG. 7), the ratio between V1 and V2 must remain constant. More specifically,
the ratio
V2/VI must equal sin(72 ) / sin(36 ), which comes to 1.618.
10072] The sequencer 55B is responsible for generating control signals that
cause the
power switcher 60B to apply these voltages to the transformer 70 in the
sequence identified
CA 02942319 2016-09-19
above. The sequencer 55B is similar to the sequencer 55 in the FIG 1
embodiment, except
that it sequences through 10 states per cycle instead of six states per cycle.
100731 Referring now to FIG. 6, the power switch 60B has a control input
68, and the
power switch is configured to either (a) apply the output of a selected one of
the DC power
sources to the primary of the transformer 70 in a selected direction in
response to 2n states of
a control signal that is applied to the control input 68 or (b) remain off in
response to an
additional state of the control signal.
10074] FIG. 6 is a block diagram of one preferred approach for implementing
the
power switcher 60B. This power switcher is similar to the power switcher 60 of
the FIG. 1
embodiment, except that it contains additional switches 65-66 for switching
the output of the
second DC-DC converter across the transformer 70 in either direction. More
specifically, this
power switcher 60B uses a set of six electronically controlled switches 61-66
connected to
the primary of the transformer 70 as depicted in FIG. 6. These switches 61-66
(which are
similar to the corresponding switches in the FIG. 1-2 embodiment) open and
close in
response to signals that are applied to a control input 68. In order to apply
the output of the
first DC-DC converter 50 to the primary of the transformer 70 in the first
direction, only
switches 63 and 62 should be closed. In order to apply the output of the first
DC-DC
converter 50 to the primary of the transformer 70 in the opposite direction,
only switches 61
and 64 should be closed. In order to apply the output of the second DC-DC
converter SOB to
the primary of the transformer 70 in the first direction, only switches 65 and
62 should be
closed. In order to apply the output of the second DC-DC converter SOB to the
primary of
the transformer 70 in the opposite direction, only switches 61 and 66 should
be closed. When
all six of these switches 61-66 are off, no power is routed into the primary
of the transformer
70.
21
CA 02942319 2016-09-19
100751 Returning to FIG. 5, an output filter 80B is connected to the
secondary of the
transformer 70, and the output filter passes the pre-set frequency f and
attenuates frequencies
above a cut-off frequency. The output filter 80B is similar to the output
filter 80 in the FIG.
1-2 embodiment, except the location of the zeros in the transfer function of
the output filter
80B must be adjusted to account for the different frequency content of the
oversampled
waveform 122 (shown in FIG. 7). More specifically, the output filter 80B
should have a
transfer function with a zero at a frequency where a harmonic of the pre-set
frequency f is
expected to contain power.
100761 For example, because the waveform 122 has 10 samples per cycle, the
initial
harmonic that we would expect to appear will be the ninth harmonic.
Accordingly, a transfer
function with a zero at the ninth harmonic would be useful when this waveform
122 is being
used. The cut off frequency of the filter should also be adjusted accordingly,
based on the set
of harmonics that are expected to appear (which can be calculated in advance
by taking the
Fourier transform of the waveform that is being used).
100771 Optionally, the transfer function of the output filter 80B can also
be designed
to have a zero at the next frequency where a harmonic of the pre-set frequency
f is expected
to contain power. In the case of the waveform 122, this would be the eleventh
harmonic.
100781 The controller 40B controls an amplitude of the sinusoid at the
output 100B of
the output filter 80B by adjusting the output voltages of the DC power sources
50, 50B via
their voltage-control inputs, while maintaining a fixed ratio between the
output voltages of
each of the DC power sources. In the illustrated embodiment, this is
accomplished by writing
appropriate control words to DAC 42 and DAC 42B, taking care to maintain the
required
ratio of sin(72 ) / sin(36 ) as described above. In alternative embodiments,
the second DAC
22
CA 02942319 2016-09-19
42B can be eliminated, and replaced by a I.618x hardware multiplier that is
inserted between
the output of DAC 42 and the voltage control input to the second DC-DC
converter 50B.
100791 In alternative embodiments, the transformer 70 can be omitted from
the FIG. 5
embodiment, in which case the two conductors at the output of the power
switcher 60B are
hooked up directly to the two conductors at the input of the output filter
80B. But these
embodiments are less preferred for the same reasons discussed above in
connection with FIG.
1.
100801 Note that the system descried above is suitable for generating high
voltage
signals of any shape, as long as the pulse train that will result in these
signals can be
determined before use either through calculations or experiments, and the
filters are designed
accordingly.
100811 When the output signal generated by the system is applied to
electrodes to
generate TTFields (as described in patent 7,805,201) changes in the load
associated with the
body of the patient and the transducer arrays can change the output signal due
to interactions
with the output filter. This means that any changes to this load (e.g.,
lifting of a disk off a
patient's body, short circuiting etc.) immediately influence the output
signal, which is
constantly monitored Hence, it is possible for the device to respond very
quickly to these
changes (e.g, by shutting down the power switcher 60 in response to the
detection of a short
circuit or overload condition).
100821 Notably, in the embodiments described above, the exact shape of the
output
signal is known in advance at every moment because we are generating a sine
wave at a
known frequency. It is only the amplitude of the output signal that changes
over time based
on the controller responding to external inputs (e.g., current measurements or
temperature
measurements). Furthermore, the rate at which the output signal has to change
is very slow
23
CA 02942319 2016-09-19
(i.e., on the order of seconds or tens of seconds). The embodiments described
above can
advantageously be used to generate very clean narrow band limited signals in
the frequency
range of 100-500 kHz, with very low losses and very low sensitivity to the
external load to
which the signal generator is connected
100831 In alternative embodiments, the system can be used to generate a
sinusoid at
any desired frequency within a pre-set range by building the filter using a
component with a
tunable reactance (e.g. a tunable capacitance or a tunable inductance). In
these embodiments,
the reactance of the tunable components is set to imbue the filter with the
desired transfer
function characteristics. Then, an appropriate oversampled sinusoid is
generated and fed into
the filter as discussed above in connection with FIGS. 1 and 5.
100841 In other alternative embodiments, the system can be used to
generate a finite
number of pre-defined signals at a plurality of different pre-set frequencies.
These
embodiments can be implemented by saving the characteristics of the pulse
trains for each of
the pre-defined signals in a look up table, and providing a bank of filters
that can be
selectively switched in to the signal path so as to provide the filtering
characteristics
necessary to generate the desired one of the pre-defined signals. When using
the system to
generate one of the pre-defined signals, the characteristics of the required
pulse train are
retrieved from memory and the appropriate filter (i.e., the one that matches
this pulse train) is
switched in to the signal path.
100851 In other alternative embodiments, composite signals that contain a
small
number of discrete frequencies (e.g., between two and five frequencies) can be
generated by
generating an oversampled version of the composite signal, and passing the
oversampled
version of the composite signal through an appropriate filter.
24
CA 02942319 2016-09-19
100861 While the present invention has been disclosed with reference to
certain
embodiments, numerous modifications, alterations, and changes to the described
embodiments are possible without departing from the sphere and scope of the
present
invention, as defined in the appended claims Accordingly, it is intended that
the present
invention not be limited to the described embodiments, but that it has the
full scope defined
by the language of the following claims, and equivalents thereof.
