Capacitive Transformer Having a Switch Responsive to Clock Signals

The present invention relates to the field of power conversion. More particularly, the present invention pertains to a high efficiency D.C. to D.C. voltage converter.

Two methods of power conversion are widely used. Linear regulation drops a supply current through a voltage equal to the desired difference between the supply voltage and the output voltage. Consequently, linear regulation wastes a great deal of power. For example, if a 5.0 volt input is to be dropped down to 3.3 volts the maximum power conversion efficiency that can be achieved is 66%. Efficiency is defined as the ratio of output power to input power. This means that 34% of the input power is wasted as heat dissipation.

Switching regulation is the other widely used method of power conversion. Switching regulators draw energy from an input supply in "packets" and temporarily store the energy in the magnetic field of a coil. The energy stored in the coil is then delivered to the load. The amount of power delivered to the load can be controlled by two methods. According to one method the amount of energy in each packet is held constant while the frequency at which the packets are drawn from the input supply is modulated. According to the second method the frequency at which the packets are drawn is held constant while the energy content of the packets is modulated in response to load requirements. Both methods generate noise pulses in the output because of the pulsed delivery of packets to the load. As a consequence, switching regulators typically use a low pass filter to remove the noise pulses from the output. That low pass filter may include a second coil to act as a choke. Switching voltage regulators cannot be fabricated as part of an integrated circuit, nor easily surface mounted within an integrated circuit package because they include coils.

The maximum theoretical power efficiency of a switching regulator is less than 100% because of the voltage drop across a coil-discharge diode within the regulator. The magnitude of the voltage drop can range between 0.5 volts to 0.7 volts. As a result, when stepping 5.0 volts down to 3.3 volts the maximum theoretical efficiency of a switching regulator is given by: ##EQU1## where V.sub.DROP is the voltage across the diode, and where .delta. is the duty cycle of coil energization, i.e., the fraction of the time the coil is connected to the supply. Under moderate Icad conditions, .delta. is approximately 0.5, which gives a theoretical maximum efficiently of approximately 90%. In practice, the maximum efficiency of switching regulators is less than 90% because of the non-ideality of switches and parasitic capacitance in the current path.

It is an object of the present invention to provide a power transformer with a maximum theoretical efficiency greater than that possible using linear regulators or switching regulators.

Another object of the present invention is to provide a voltage transformer with an actual efficiency of approximately 85% or greater.

Yet another object of the present invention is to provide a transformer that can step-up and step-down an input voltage.

A further object of the present invention is to provide a transformer that is relatively insensitive to large changes in load current, environmental noise and component precision.

A still further object of the present invention is to provide a transformer that reacts rapidly to load changes as a result of its open-loop structure.

Another object of the present invention is to provide a transformer that is cheaper to manufacture than a switching regulator.

A still further object of the present invention is to provide a voltage transformer that can be surface mounted inside integrated circuit packages in which a semiconductor device is mounted.

A capacitive transformer is described that includes two half capacitive transformers operating in antiphase. Each half capacitive transformer steps an input voltage down during one phase of a clock signal. Each half capacitive transformer does so using a number of capacitors and switching circuitry controlled by the clock signal. According to whether the clock signal is active or inactive, the switching circuitry places the half capacitive transformer in one of two configurations. In one configuration the capacitors are coupled in parallel between the input voltage and the output port, while in the other configuration, the capacitors are coupled in series between the output port and ground. The step-down ratio of the capacitive transformer is related to the number of capacitors included in a half capacitive transformer and not to the values of the capacitors themselves. The desired transitory response can be achieved by choosing a load capacitance in relation to the value of the capacitors within a half capacitive transformer.

An alternative embodiment of the half capacitive transformers is also described. In one configuration, the capacitors are coupled in series between the input voltage and the output port, while in the second configuration the capacitors are coupled in parallel between the output port and ground.

The capacitive transformer may be used to step up voltages by swapping the input and output ports.

Other objects, features, and advantages of the present invention will be apparent from the accompanying drawings and the detailed description that follows.

The present invention is illustrated by way of example and not limitation in the figures of the accompanying drawings in which references indicate similar elements and in which:

FIG. 1 is a block diagram of a capacitive transformer.

FIG. 2 is a schematic diagram of a half capacitive transformer.

FIG. 3A illustrates one configuration of a half capacitive transformer.

FIG. 3B illustrates the other configuration of a half capacitive transformer.

FIG. 4A illustrates the transitory response of a half capacitive transformer for C.sub.LOAD /C equal to 0.5.

FIG. 4B illustrates the transitory response of a half capacitive transformer for C.sub.LOAD /C equal to 1.0.

FIG. 4C illustrates the transitory response of a half capacitive transformer for C.sub.LOAD /C equal to 2.0.

FIG. 5A illustrates the transitory response of a capacitive transformer for C.sub.LOAD /C equal to 1.0.

FIG. 5B illustrates the transitory response of a capacitive transformer for C.sub.LOAD /C equal to 2.0.

FIG. 5C illustrates the transitory response of a capacitive transformer for C.sub.LOAD /C equal to 3.0.

FIG. 6 illustrates schematically the general form of a half capacitive transformer having a step-down ratio greater than or equal to 1/2 and less than one.

FIG. 7A illustrates one of the configurations of the half capacitive transformer of FIG. 6.

FIG. 7B illustrates the other configuration of the half capacitive transformer of FIG. 6.

FIG. 8 illustrates schematically the general form of a half capacitive transformer having a step-down ratio less than or equal to 1/2.

FIG. 9A illustrates one of the configurations of the half capacitive transformer of FIG. 8.

FIG. 9B illustrates the other configuration of the half capacitive transformer of FIG. 8.

FIG. 10 illustrates schematically operation of a half capacitive transformer in step-up mode.

FIG. 11 illustrates schematically an embodiment of a half capacitive transformer.

FIG. 12 is a timing diagram of the phasing signals coupled to the half capacitive transformer of FIG. 11.

FIG. 13 is a block diagram of a capacitive transformer using the half capacitive transformer of FIG. 11.

FIG. 14A illustrates a simulated change in load current of the capacitive transformer of FIG. 13.

FIG. 14B illustrates the change in output voltage of the capacitive transformer of FIG. 13 in response to the change in load current.

FIG. 1 illustrates in block diagram form capacitive transformer (CXFR) 30 of the present invention. Capacitive transformer 30 very efficiently transforms direct current (D.C.) input voltage V.sub.in 32 into D.C. output voltage V.sub.out 34. Capacitive transformer 30 may be fabricated as part of an integrated circuit (IC) to permit circuitry designed to operate from a 3.3 volt supply to operate directly from 5 volts. In other words, in one embodiment capacitive transformer 30 steps a 5 volt input at port 32 down to 3.3 volts at port 35. Capacitive transformer 30 operates bilaterally and may also be used to step-up D.C. voltages. That is to say, a voltage input at port 35 will be stepped up at port 32.

Capacitive transformer 30 includes two half capacitive transformers (HCXFR) 36. Half capacitive transformers 36 operate in complementary fashion because they are identical circuits controlled by complementary clock phase signals PHl1 40 and PHl2 42. Clock phase signals 40 and 42 are complementary and nonoverlapping binary signals. Stated another way, phase signals 40 and 42 each have two phases, one active and the other inactive, per clock period. Only one clock phase signal 40 or 42 is active at a time.

FIG. 2 illustrates schematically an embodiment of a half capacitive transformer 36. HCXFR 36 includes two capacitors 50 and 52 and switching circuitry 54. Switching circuitry 54 includes a single-pole-double-throw (SPDT) switch coupled to each plate of capacitors 50 and 52. The position of each switch is controlled by clock input 37. Thus, switching circuitry 54 places capacitors 50 and 52 in two different configurations according to whether the phase signal applied to clock input 37 is active or inactive.

HCXFR 36 steps V.sub.in 32 down to generate V.sub.out 34. HCFXR 36 of FIG. 2 provides an accurate stepping ratio of 2/3 even when inaccurate capacitors 50 and 52 are used. The relationship of V.sub.out 34 to V.sub.in 32 can be determined by applying the rule of charge conservation to the redistribution of charge within HCXFR 36 during three consecutive phases of clock input 37. Let us call the first of the three consecutive phases phase 1 of clock cycle n-1 and denote it in the following equations as: n-1, PHl1. During this phase HXCFR 36 is in the configuration illustrated in FIG. 3A. In FIGS. 3A and 3B and subsequent figures the dots near capacitors 50 and 52 serve only to differentiate between the two capacitor plates. Similarly, the second of the three consecutive phases is called phase 2 of clock cycle n-1 and is denoted as: n-1, PHl2. During n-1, PHl2 HCXFR 36 is in the configuration illustrated in FIG. 3B. Finally, the third of the consecutive phases is called phase 1 of clock cycle n and is denoted as: n, PHl1. HCXFR 36 returns to the configuration of FIG. 3A during the third consecutive phase.

The charge stored on capacitors 50, 52 and 39 during the first of the three consecutive phases are:

During the second of the consecutive phases, the total charge within HCXFR 36 remains the same; however, some charge from capacitors 50 and 52 is redistributed to capacitor 39. Denote this change in charge as .DELTA.Q(n-1 , PHl2). Thus, the charge on capacitors 50, 52 and 39 during the second phase can be expressed as:

The new output voltage, V.sub.OUT (n-1, PHl2), is given by ##EQU2## V.sub.OUT (n-1, PHl2) is also given by ##EQU3##

By substituting Eq. 4, Eq. 5 and Eq. 6 into Eq. 7 and Eq. 8, we arrive at two equations in two unknowns: V.sub.OUT (n-1, PHl2) and .DELTA.Q(n-1, PHl2). Once we have solved for .DELTA.Q(n-1, PHl2) we can solve for the charges on the three capacitors using Eq. 4, Eq. 5 and Eq. 6.

During the final of the three consecutive phases the charge on capacitors 50 and 52 is the sum of the charges from the two previous phases plus a small charge that flows into them from V.sub.in 32. Designate that change in charge as .DELTA.Q(n, PHI1). Thus, the charge on capacitors 50, 52 and 39 during the last clock phase is given by: ##EQU4## The new output voltage V.sub.OUT (n, PHl1) is given by ##EQU5## V.sub.OUT (n,PHl1) is also given by ##EQU6## By substituting Eq. 9, Eq. 10 and Eq. 11 into Eq. 12 and Eq. 13, we arrive at two equations in two unknowns: V.sub.OUT (n,PHl1) and .DELTA.Q(n,PHl1).

Solving these two equations for V.sub.OUT (n,PHl1) as a function of V.sub.OUT (n-1, PHl), and calling V.sub.OUT (n, PHl1) V.sub.OUT (n) and calling V.sub.OUT (n-1,PHl1) V.sub.OUT (n-1), the following equation is obtained:

where .alpha. and .beta. are given by ##EQU7## Solving Equation 14 recursively, we get ##EQU8## .beta. is less than 1 for 0<C.sub.LOAD <.infin., i.e., for any practical value of C.sub.LOAD. Therefore, as n increases towards .apprxeq., .beta..sup.n tends towards zero, and we get ##EQU9##

Thus, as stated previously, according to equation 18 the step-down ratio of HCXFR 36 is independent of the values of capacitors 50 and 52. The only requirement is that capacitor 39 have a non-zero value. Therefore, accurate step-down ratios can be obtained using inaccurate capacitors.

The transient response of V.sub.out, as a function of n, can be graphed by computing .beta. for assumed values of capacitors C.sub.1 50, C.sub.2 52 and then solving equation 17. The simplest case is when C.sub.1 and C.sub.2 are equal to the same value, which we will designate C. Defining x=C.sub.LOAD /C then .alpha. and .beta. are given by ##EQU10## An examination of the behavior of V.sub.OUT during each clock phase shows that

where V.sub.OUT (n,PHl.sub.1) is the value of V.sub.OUT at the end of PHl.sub.1 in clock n. Furthermore,

where .alpha.', .alpha.", .beta.", and .beta." are given by: ##EQU11## The relationships of a and .beta. to .alpha.', .alpha.", .beta.', and .beta." are given by: ##EQU12## which correspond exactly to Equations 19 and 20.

From equations 24 and 26, V.sub.OUT 34 has an oscillatory transient characteristic for C.sub.LOAD less than C and an exponential-decay transient characteristic for C.sub.LOAD greater than C. Furthermore, V.sub.OUT 34 has a null in its transient characteristic, i.e., it responds instantaneously, when C.sub.LOAD =C.

The transient behavior of V.sub.OUT in the time domain can be obtained by substituting t/T.sub.p for n in equation 17, where t is the elapsed time and T.sub.p is the time period of the phase signal applied to clock input 37. Equation 17 then becomes:

where .tau.=-T.sub.p /In.beta..

FIGS. 4A, 4B and 4C illustrate simulation results which confirm equation 29 under three different conditions. FIG. 4A illustrates the results of simulation for C.sub.LOAD equal to 0.5 C. The simulation confirms the oscillatory transient characteristic expected. FIG. 4B illustrates the result of a simulation for C.sub.LOAD equal to C. Again, the simulation confirms the expected null in the transient characteristic. Finally, FIG. 4C illustrates exponential delay in the transient characteristic for C.sub.LOAD greater than C.

A single HCXFR 36 within CXFR 30 draws current from Vin 32 during only half the clock period. Power is, therefore, drawn from the HCXFR 36 only during half the clock period so that its output voltage fluctuates between each half of the clock period. CXFR 30 does not exhibit the same characteristics because the pair of HCXFRs 36 are connected in parallel and operate in antiphase. Therefore, the CXFR 30 draws essentially a D.C. current from V.sub.in 32 throughout the clock period. So, power is drawn from CXFR 30 during the entire clock period, minimizing voltage fluctuations in V.sub.out 34.

The relationship of V.sub.out 34 to Vin 32 within CXFR 30 is described by equation 17, which was derived during the discussion of the behavior of a HCXFR. However, .alpha. and .beta. for the full CXFR are given by: ##EQU13##

Furthermore, the definition of n when equation 17 is applied to CXFR 30 differs from that when equation 17 is applied to a HCXFR 36. Within a half capacitive transformer 36 n refers to the number of elapsed clock periods; however, within capacitive transformer 30 n refers to the number of elapsed phase periods. Because CXFR 30 employs two HCXFRs operating in antiphase, CXFR 30 changes configurations twice as many times per clock period as a HCXFR. Thus, when equation 17 is applied to CXFR 30 .tau. equals-T.sub.p /2ln.beta. rather than -T.sub.p /ln.beta. as with a HCXFR.

The step-down ratio of CXFR 30 is also given by equation 18, which describes the step-down ratio of a HCXFR 36, for any practical value of C.sub.LOAD. In other words, the step-down ratio of CXFR 30 is also 2/3. Equation 18 holds for CXFR 30 because the absolute value of .beta. is less than 1 for 0.ltoreq..times.<.apprxeq..

The step-down ratio of CXFR 30 should remain unchanged at 2/3 even if unequal capacitor values are used. This is because the values of capacitors do not effect the ratio of .alpha./(1-.beta.) even though .alpha. and .beta. are changed from those given in Equations 30 and 31.

The transient response of VOUT exhibits three different regimes for CXFR 30, as with HCXFR 36. For CXFR 30, equation 31 dictates the boundaries of the regimes, whereas for HCXFR 36 the boundaries are dictated by equations 24 and 26. Thus, according to equation 31 it is anticipated that V.sub.OUT 34 will exhibit an oscillatory transient for values of C.sub.LOAD <2C. For values of C.sub.LOAD >2C, V.sub.OUT 34 displays a transient exponential decay characteristic. V.sub.OUT 34 displays an instantaneous transitory response at C.sub.LOAD 2C. Computer simulations of V.sub.OUT 34 bear out the transitory response predicted by equation 31. FIGS. 5A, 5B and 5C illustrates the computer simulations. FIG. 5A illustrates an oscillatory transient response for C.sub.LOAD /C=1. FIG. 5B illustrates an instantaneous response of V.sub.OUT 34 for C.sub.LOAD /C=2. Finally, FIG. 5C displays an exponential decay in the transitory response of V.sub.OUT 34 for C.sub.LOAD /C=3.

Capacitive transformers can be built that provide different step-down ratios than 2/3. In general, the step-down ratio, R, of a capacitive transformer is described by: ##EQU14## where N is an integer value representing the number of capacitors per half capacitive transformer. FIG. 6 illustrates schematically the equivalent half capacitive transformer 60 for an arbitrary number N. In one phase of CLK 37, HCXFR 60 couples all N capacitors in parallel between V.sub.in 32 and V.sub.OUT 34. In the other phase of CLK 37, HCXFR 60 couples all N capacitors in series between V.sub.OUT 34 and ground. The switching circuitry of HCXFR 60 in this phase couples the bottom plate of one capacitor to the top plate of the adjacent capacitor. Thus, for example, the bottom plate 62 of capacitor C1 is coupled to ground and its top plate 64 is coupled to the bottom plate 66 of capacitor C2. In other words, during its two phases CLK 37 switches HCXFR 60 between the two configurations illustrated in FIGS. 7A and 7B.

The range of step-down ratios possible using HCXFR 60 is 1>R.gtoreq.1/2. Step-down ratios smaller than 1/2 can be provided using alternative HCXFR 70 of FIG. 8. HCXFR 70 includes an arbitrary number, N, of capacitors and switching circuitry controlled by CLK 37. The switching circuitry switches HCXFR 70 between two configurations, which are illustrated in FIGS. 9A and 9B. While CLK 37 is active high, the switching circuitry couples all N capacitors in series between V.sub.in 32 and V.sub.out 34, as shown in FIG. 9A. This is because while CLK is active the bottom plate of each capacitor is coupled to the top plate of its adjacent upstream neighbor by the switching circuitry. Thus, the bottom plate 72 of capacitor C.sub.1 is coupled to the top plate 74 of capacitor C.sub.2, and so on. This is in contrast to HCXFR 60, in which all N capacitors are coupled in parallel between V.sub.in and V.sub.out while CLK 37 is active. FIG. 9B illustrates the configuration of HCXFR 70 while CLK 37 is inactive low. All N capacitors are coupled in parallel between V.sub.out 34 and ground. Recall that HCXFR 60 couples all N capacitors in series between V.sub.out 34 and ground while CLK is inactive.

HCXFR 70 provides a step-down ratio, R, given by: ##EQU15##

Thus, the range of step-down ratios possible using alternative HXCFR 70 is 1/2.gtoreq.R>0.

The discussion of capacitive transformer 30 has thus far been concerned only with step-down operation; i.e., stepping ratios of one or less. Capacitive transformer 30 can also operate as a step-up transformer by swapping the ports to which V.sub.in 32 is applied and from which V.sub.out 34 is sensed. FIG. 10 illustrates the coupling of V.sub.in 32 and V.sub.out 34 to HCXFR 36 for step-up operation. Thus, given a 3.3 volt input at port 2 35 HCXFR 36 outputs 5.0 volts at port 1 33. Note that C.sub.LOAD 39 is now connected at port 1 33 instead of port 2 35. Because the switching circuitry within HCXFR 36 requires 5 volts for proper operation, step up operation of HCXFR 36 is limited to those situations in which bootstrap operation of the capacitive transformer is acceptable. For example, the 5 volt swing necessary to operate the switching circuitry could be generated by switching control circuitry that is coupled to low-voltage input and the higher voltage output. As the output voltage rises, the output of the switching control circuitry would rise to approximately the output voltage level.

FIG. 11 illustrates schematically an implementation 36a of HCXFR 36 suitable for fabrication as part of an integrated circuit. Switching circuitry 54 is realized using a number of transistors 80, 82, 84, 86, 88, 90 and 92 that function as switches. In other words, when transistors are conducting the voltage drop between their drain and source is close to 0 volts, and the switch is closed. Conversely, when the transistors are not conducting they are analogous to an open switch. Transistors P.sub.1 80 and P.sub.2 82 are P-type transistors, which conduct when the voltage applied to their gates is low. In contrast, transistor N.sub.1 84, N.sub.2 86, N.sub.3 88, N.sub.4 90 and N.sub.5 92 are N-type transistors, which conduct when the voltage applied to their gates is high. In one embodiment, the fabricated width of the transistors in microns is:

These transistor widths are much larger than generally used in CMOS devices. However, there is no reason such widths cannot be implemented provided the devices fit into the available silicon area. Other transistor widths may be appropriate using other technologies, such as junction field effect transistors (JFET).

Operating HCXFR 36a so that it can assume only the configurations of FIGS. 3A and 3B requires completely breaking all connections associated with one configuration before making any of the connections of another configuration. Implementing this "break before make" strategy dictates that transistors coupled to CLK port 100 and CLKB port 102 must never be active simultaneously with transistors coupled to CLKC port 104. The signals coupled to CLK port 100 and CLK B port 102 must be complementary because port 100 controls N-type transistors while port 102 controls P-type transistors. The signals applied to ports 100 and 104 must be non-overlapping complements. FIG. 12 is a timing diagram illustrating the relationship between phase signal PHl1 120 applied to port 100, phase signal PHl1B 122 applied to port 102 and phase signal PHl2 124 applied to port 104, as well as PHl2B 126. Let us call phase pair 120 and 122 and phase pair 124 and 126 "active" when the transistors controlled by the phase pair are conducting. Thus, signals 120 and 122 are active when PHl1 120 is high and PHl1B 122 is low, and signals 124 and 126 are active when PHl2 124 is high and PHl2B 126 is low. Analogously, phase pair signals 120 and 122 are inactive when PHl1 120 is low and PHl1B 122 is high. Note that active periods of phase pair 120 and 122 are separated from active periods of phase pair 124 and 126 by dead time 130, a period of time during which both pairs of phase signals are simultaneously inactive. As a result, transistors turned on by one phase pair are turned off before transistors controlled by the other phase pair are turned on.

Analysis of FIG. 12 confirms that HCXFR 36a operates as described. Consider first the operation of HCXFR 36a while phase pair 120 and 122 are active. During this period the high voltage level applied to port 100 causes transistors N.sub.1 84 and N.sub.2 86 to conduct, tying one plate of both capacitors to V.sub.out 34. Similarly, during this period the low voltage level applied to port 102 causes transistors P.sub.1 80 and P.sub.2 82 to conduct. As a result, one plate of capacitors 50 and 52 is coupled to V.sub.in 32. Because PHl2 is low during this phase transistors N.sub.3 88, N.sub.4 90 and N.sub.5 92 do not conduct and have no effect upon the configuration of HCXFR 36a. As a result, HCXFR 36a operates in the configuration illustrated in FIG. 3A while phase pair 120 and 122 is active.

During dead time 130 all transistors are turned off. Afterward, phase pair 124 and 126 becomes active. In response to the high voltage applied to port 104 transistors N.sub.3 88, N.sub.4 90 and N.sub.5 92 begin to conduct. This couples C.sub.1 50 between ground and capacitor C.sub.2 52 and couples capacitor C.sub.2 52 to V.sub.OUT 34. The voltage levels applied to ports 100 and 102 keep transistors 80, 82, 84 and 86 from conducting during this phase. Consequently, while phase pair 124 and 126 is active HCXFR 36a operates in the configuration of FIG. 3B.

FIG. 13 illustrates in block diagram form a capacitive transformer 30a that uses two HCXFRs 36a operating in antiphase. The response of CXFR 30a to a 2.9 amp change in load current lasting 20 ns was simulated for two different values of C.sub.LOAD 39: 1 .mu.F and 20 .mu.F. In both simulations the value of capacitors C.sub.1 and C.sub.2 was set to 10 .mu.F. Line 140 of FIG. 14B shows that for C.sub.LOAD 39 equal to 1 .mu.F V.sub.OUT 34 dips approximately 0.18 volts in response to the change in load current. Line 140 also displays some small amplitude ringing, approximately 0.05 volts. In contrast line 142 shows no ringing in V.sub.out 34 for C.sub.LOAD 39 equal to 20 .mu.F, even though the dip in output voltage remains. This voltage dip is caused by the on resistance of the transistors. The smaller the on resistance, the smaller the dip. Ideal switches, which have zero on resistance, would not produce a voltage dip.

In the foregoing specification, the invention has been described with reference to specific exemplary embodiments thereof. It will, however, be evident that various modifications and changes may be made thereto without departing from the broader spirit and scope of the invention as set forth in the appended claims. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.