Integrated magnetic converter circuit and method with improved filtering

[0003] This invention relates to switching power converter circuits and methods and, in particular, relates to a integrated magnetic converter circuit and method with improved filtering.

[0004] Switching power converters are used in a wide variety of applications to convert power from one form to another form. For example, dc/dc converters are used to convert dc power provided at one voltage level to dc power at another voltage level. Such dc/dc converters often use rectifier circuits, such as the well-known current-doubler rectifier (CDR) circuit. The CDR circuit has recently been rediscovered for use in high frequency dc/dc converter applications. The low current stresses in the transformer secondary, the inductors, and the rectifiers makes the circuit attractive for dc/dc converters with high output currents. The circuit can be used with different double-ended primary topologies, such as push-pull, half bridge and full bridge.

[0005] Early CDR circuits used three separate magnetic components, namely, one transformer and two inductors. However, this configuration results in interconnection losses which negatively impact efficiency, especially in high current applications. Additionally, this configuration results in increased size and cost. To address these problems, integrated magnetic implementations have been proposed that combine the transformer and the inductors into a single magnetic structure with one magnetic core, thereby reducing the size and the interconnection losses.

[0006] While this solution addresses some problems, further improvements are still needed. In particular, the effective filtering inductance of existing integrated magnetic implementations for CDR circuits is low due to the use of a single turn secondary for low conduction losses, resulting in high ripple currents and output ripple voltage. The circuit is also more likely to operate in a discontinuous conduction mode when the inductance is low, which is less efficient compared to a continuous conduction mode of operation. Accordingly, an improved rectifier circuit which overcomes one or more of the above problems would be highly advantageous.

[0007] According to a first preferred aspect, a converter and rectifier circuit comprises a magnetic core, a first secondary winding, a second secondary winding, a third secondary winding, a first rectifier, a second rectifier and an output capacitor. The first, second and third secondary windings are wound around the magnetic core and are connected to each other. The first rectifier is connected to the first secondary winding and the second rectifier is connected to the second secondary winding. The output capacitor is connected to the first and second rectifiers and is connected in series with the third secondary winding.

[0008] According to a second preferred aspect, a dc/dc conversion method comprises providing a voltage to a primary winding of a transformer. The voltage causes a first current to flow in a first secondary winding, a second current to flow in a second secondary winding, and a third current to flow in a third secondary winding, the third current being a summation of the first and second currents. The method further comprises rectifying the first current using a first rectifier, rectifying the second current using a second rectifier, and filtering the third current using inductance of the third secondary winding. The third winding is connected in series with an output capacitor. The method further comprises charging the output capacitor using the third current and providing dc power to a power-consuming device using the output capacitor.

[0009] FIG. 1 is a dc/dc converter circuit employing an integrated current-doubler rectifier according to a preferred aspect of the invention;

[0010] FIG. 2 is a schematic diagram of the circuit implemented by the integrated current doubler rectifier construction of FIG. 1;

[0011] FIG. 3 is a reluctance model of the magnetic component for the integrated current doubler rectifier of FIG. 1;

[0012] FIG. 4 is a timing diagram showing gate signals applied to a drive circuit of the converter circuit of FIG. 1;

[0013] FIG. 5 is an alternative preferred integrated magnetic construction for the current doubler rectifier circuit according to another preferred aspect of the invention; and

[0014] FIG. 6 is a graph of an output voltage of the converter circuit of FIG. 1 as a function of time.

[0015] Referring now to FIG. 1, a dc/dc converter circuit 10 that employs an integrated current doubler rectifier (CDR) circuit according to a preferred embodiment of the invention is illustrated. The system 10 comprises an drive circuit 14, an integrated CDR circuit 16, and a feedback control circuit 18.

[0016] The drive circuit 14 may have a variety of configurations. In FIG. 1, the drive circuit 14 is shown to have a double-ended, half-bridge topology which is capable of applying positive, negative, and zero voltage to the CDR circuit 16. To this end, the drive circuit 14 comprises first and second capacitors 20 and 22 and first and second switches 24 and 26. The capacitors 20 and 22 and switches 24 and 26 receive dc power from a dc voltage source at input terminals 28 and 30. The drive circuit 14 is connected across the windings 62 and 64 and provides a pulse width modulated voltage to the windings 62 and 64. The capacitors 20 and 22 may, for example, be electrolytic tantalum capacitors. The switches 24 and 26 may be transistors such as power MOSFETs.

[0017] The integrated CDR circuit 16 comprises a magnetic core 34, a primary winding arrangement 36, a secondary winding arrangement 40, an output capacitor 42, and first and second rectifiers 44 and 46. The magnetic core 34 comprises a center leg 54, a first outer leg 56, and a second outer leg 58. The outer legs 56 and 58 are disposed on opposite sides of the center leg 54.

[0018] The primary winding arrangement 36 comprises a primary winding that is formed of a first primary winding 62 which is wound around the outer leg 56 and a second primary winding 64 which is wound around the outer leg 58. The secondary winding arrangement 40 comprises first, second and third secondary windings 66, 68, 70. The winding 66 is wound around the outer leg 56 of the magnetic core 34, the winding 68 is wound around the outer leg 58 of the magnetic core 34, and the winding 70 is wound around the center leg 54 of the magnetic core 34. As described in greater detail below, the winding 70 increases the filtering inductance of the CDR circuit 16, and thereby reduces voltage and current ripple and improves efficiency.

[0019] The winding 70 is connected in series with the output capacitor 42. The output capacitor has first and second terminals 72 and 74 which form the output terminals of the integrated current-doubler rectifier 16 and the dc/dc converter circuit 10. The rectifier 44 is connected between the output capacitor 42 and the winding 66. The rectifier 46 is connected between the output capacitor 42 and the winding 68.

[0020] The magnetic core 34 is mounted on a multi-layer circuit board 76. The legs 54-58 of the magnetic core 34 extend through holes formed in the circuit board 76. The windings 62-70 are at least partially formed by traces on the multi-layer circuit board 76.

[0021] The feedback control circuit 18 comprises a voltage measurement circuit 78, a drive control circuit 80, and an isolation circuit 82. The voltage measurement circuit 78 is connected to the output capacitor 42 to measure an output voltage across the output terminals 72 and 74. The feedback control circuit 18 controls the switching of the drive circuit 14 to vary the duty cycle of the drive circuit 14 based on a voltage feedback signal from the voltage measurement circuit 78. The duty cycle is controlled so as to minimize deviation of the output voltage from a predetermined setpoint level. The feedback control circuit 18 also controls the rectifiers 44 and 46 when they are implemented by active switches, such as power MOSFETs. The isolation circuit 82, which may comprise an isolation transformer, provides electrical isolation between the electrical components on either side of the magnetic core 34.

[0022] In operation, a dc voltage is received at the input terminals 28 and 30 of the dc/dc converter circuit 10. The dc voltage is applied to the capacitors 20 and 22 and the switches 24 and 26. The switches 24 and 26 are controlled by the drive control circuit 80 such that at most only one switch 24, 26 is on at a time (see FIG. 4). The switches 24 and 26 apply a pulse-width modulated dc voltage to the primary winding formed of the subwindings 62 and 64. This causes a first current i₁to flow in the winding 66, a second current i₂to flow in the winding 68, and a third current i₃to flow in the winding 70 (where i₁+i₂=i₃), though ordinarily not all at the same time. The current i₁flowing the winding 66 is rectified by the rectifier 44 and the current i₂flowing in the winding 68 is rectified by the rectifier 46. Thus, whether current flows through the winding 66 or the winding 68 depends on the states of the switches 24 and 26. If the rectifiers 44 and 46 are diodes, then the rectifiers 44 and 46 are controlled by virtue of the bias voltages that are built up across the rectifiers 44 and 46. If the rectifiers 44 and 46 are synchronous rectifiers (implemented, for example, using MOSFETs), then the rectifiers 44 and 46 are controlled by the drive control circuit 80. Typically, synchronous rectifiers are controlled so as to operate in the same manner as diodes, but are used in situations where the converter circuit 10 produces a low output voltage. The converter circuit 10 uses the third current i₃to charge the output capacitor 42, which then provides power to a power-consuming device. The operation of the converter circuit 10 in different switch states is described in greater detail in connection with FIGS. 24.

[0023] Referring now also to FIGS. 2-4, FIG. 2 is a schematic diagram showing the circuit that is implemented by the construction of the integrated CDR circuit 16 of FIG. 1, and FIG. 3 is a reluctance model of the magnetic component of the integrated CDR circuit 16. As shown in FIG. 2, the resultant circuit comprises a primary winding 90 (implemented by the primary windings 62 and 64 in FIG. 1), a secondary winding 91 (implemented by the secondary windings 66 and 68 ), a first inductor 92 (implemented by the secondary winding 66), a second inductor 94 (implemented by the secondary winding 68), and a third inductor 96 (implemented by the secondary winding 70).

[0024] A mathematical description of the preferred integrated CDR circuit 16 will now be given. Of course, it should be understood that other implementations may have different mathematical characterizations, and that the following description is merely exemplary for the construction shown.

[0025] First, the operation of the circuit 10 when the voltage applied to the primary winding 90 is greater than zero (V_ab>0) will be described. When V_ab>0, the rectifier 44 conducts, the rectifier 46 is reverse biased, and therefore the secondary currents i₁and i₂are as follows: i₁>0 and i₂=0. The flux in the three legs can be determined from FIG. 3 as follows: $\begin{array}{cc}{\phi }_1=\frac{i_1}{2{\Re }_c+{\Re }_o}·\left[N_s\left(1+\frac{{\Re }_c}{{\Re }_o}\right)+N_c\right]-\frac{N_p{i}_p}{{\Re }_o}& \left(1\right)\\ {\phi }_2=\frac{i_1}{2{\Re }_c+{\Re }_o}·\left[-N_s·\frac{{\Re }_c}{{\Re }_o}+N_c\right]+\frac{N_p{i}_p}{{\Re }_o}& \left(2\right)\\ {\phi }_c={\phi }_1+{\phi }_2=\frac{2N_c+N_s}{2{\Re }_c+{\Re }_o}·i_1& \left(3\right)\end{array}$

[0026] where φ_c, φ₁, and φ₂are the fluxes in the legs 54, 56, and 58, respectively, _c

[0027] is the reluctance of the center leg 54, _o

[0028] is the reluctance of the outer legs 56 and 58, N_cis the number of turns on the winding 70, N_sis the number of turns on each of the windings 66 and 68, and N_pis the number of turns on each of the windings 62 and 64.

[0029] The fluxes φ_c, φ₁, and φ₂meet the following differential equations: $\begin{array}{cc}\{\begin{array}{c}{N}_p\frac{d{\phi }_2}{dt}-N_p\frac{d{\phi }_1}{dt}=v_{\mathrm{ab}}\\ N_s\frac{d{\phi }_1}{dt}+N_c\frac{d{\phi }_c}{dt}=-v_0\end{array}& \left(4\right)\end{array}$

[0030] where V_abis the voltages applied to the primary winding 90 and the secondary winding 91, and V₀is the output voltage at the capacitor 42.

[0031] Substituting Eqs. (1)-(3) into Eq. (4), and solving for the primary and the secondary currents i₁and i_p, yields the following equations: $\begin{array}{cc}\frac{d{i}_1}{dt}=\frac{N_s{v}_{\mathrm{ab}}-2N_p{v}_o}{{N_p\left(2N_c+N_s\right)}^2}·\left(2{\Re }_c+{\Re }_o\right)& \left(5\right)\\ \frac{d{i}_p}{dt}=\frac{2N_c^2{\Re }_o+2N_c{N}_s{\Re }_o+N_s^2\left({\Re }_c+{\Re }_o\right)}{{N_p^2\left(2N_c+N_s\right)}^2}·v_{\mathrm{ab}}-\frac{N_s\left({\Re }_c+{\Re }_o\right)v_0}{{N_p\left(2N_c+N_s\right)}^2}& \left(6\right)\end{array}$

[0032] The filtering inductance L′ and magnetizing inductance L_m′ are defined as follows: $\begin{array}{cc}{L}^{\prime }=\frac{{\left(N_s+2N_c\right)}^2}{2{\Re }_c+{\Re }_o},L_m^{\prime }=\frac{2N_p^2}{{\Re }_o},& \left(7\right)\end{array}$

[0033] With these definitions, the above current equations (5)-(6) can be rewritten as follows: $\begin{array}{cc}\{\begin{array}{c}{L}^{\prime }\frac{d{i}_1}{dt}=\frac{N_s}{N_p}·v_{\mathrm{ab}}-2v_0\\ \frac{d{i}_p}{dt}=\left[\frac{1}{L_m^{\prime }}+\frac{1}{2L^{\prime }}·{\left(\frac{N_s}{N_p}\right)}^2\right]·v_{\mathrm{ab}}-\frac{N_s}{N_p}·\frac{v_o}{L^{\prime }}\end{array}& \left(8\right)\end{array}$

[0034] Using the same notations and procedure, the following equations can be derived for the case of v_ah<0(i₁=0, i₂>0): $\begin{array}{cc}\{\begin{array}{c}{L}^{\prime }\frac{d{i}_1}{dt}=-\frac{N_s}{N_p}·v_{\mathrm{ab}}-2v_0\\ \frac{d{i}_p}{dt}=\left[\frac{1}{L_m^{\prime }}+\frac{1}{2L^{\prime }}·{\left(\frac{N_s}{N_p}\right)}^2\right]·v_{\mathrm{ab}}+\frac{N_s}{N_p}·\frac{v_0}{L^{\prime }}\end{array}& \left(9\right)\end{array}$

[0035] Again, using the same notations and procedure, the following equations can be derived for the case of v_ah=0(i₁>0, i₂>0): $\begin{array}{cc}\{\begin{array}{c}{L}^{\prime }\frac{d\left(i_1+i_2\right)}{dt}=-2v_0\\ \frac{d{i}_p}{dt}=0\end{array}& \left(10\right)\end{array}$

[0036] Equations (8)-(10) confirm that the integrated CDR circuit 16 has the same functionality as a discrete CDR circuit with the effective filtering inductance L and the transformer magnetizing inductance L_mdetermined as follows: $\begin{array}{cc}\{\begin{array}{c}L=L^{\prime }\\ \frac{1}{L_m}=\frac{1}{L_m^{\prime }}-{\left(\frac{N_s}{N_p}\right)}^2·\frac{1}{2L^{\prime }}\end{array}& \left(11\right)\end{array}$

[0037] Using the flux equation (3), the flux density B_cin the center leg 54 is given by the equation (12): $\begin{array}{cc}{B}_c=\frac{\left(N_s+2N_c\right)\left(i_1+i_2\right)}{\left(2{\Re }_c+{\Re }_o\right)A_c}& \left(12\right)\end{array}$

[0038] where A_cis the cross-sectional area of the leg 54.

[0039] Assuming the flux density is kept under a maximum value, B_max, to avoid saturation of the magnetic core 34, then the following equation holds true: $\begin{array}{cc}2{\Re }_c+{\Re }_o\ge \frac{\left(N_s+2N_c\right)\left(i_1+i_2\right)}{A_c{B}_{\max }}\approx \frac{\left(N_s+2N_c\right)I_0{A}_c{B}_{\max }}{A_c{B}_{\max }}& \left(13\right)\end{array}$

[0040] where l_ois the load current, and the approximation holds when ignoring the ripple currents of i₁and i₂. This in turn defines a maximum inductance L′ achievable with the CDR circuit 16, as given by Eq. (14): $\begin{array}{cc}{L}^{\prime }\le \frac{\left(N_s+2N_c\right)A_c{B}_{\max }}{I_0}=\frac{N_s{A}_c{B}_{\max }}{I_0}+\frac{2N_c{A}_c{B}_{\max }}{I_0}& \left(14\right)\end{array}$

[0041] The last term in Eq. (14) is the additional inductance due to the winding 70 on the center leg 54. Assuming a single-turn secondary (N_s=1), it can be seen from Eq. (14) that adding a single turn to the winding 70 (N_c=1) triples the effective filtering inductance as compared to a center leg 54 with no winding.

[0042] In order to further explain the design and operation of the converter circuit 10, a numeric example is now given. Typically, the converter circuit 10 may be used to provide a low voltage (e.g., less than about 10V, and preferably less than about 5V), high current output (e.g., more than about 5 A, and preferably more than about 10 A). As an example, input and output specifications for the converter include V_in=48V, V_o=3.3V, l_o=30 A P_out=100 W.

[0043] The drive control circuit 80 is assumed to operate the converter circuit 10 in a symmetric mode of operation. The converter circuit 10 could also be operated in an asymmetric mode, which has the advantage of allowing the primary switches to turn on under zero drain-to-source voltages, resulting in low switching losses. However, asymmetric mode has the disadvantage of high conduction losses due to the asymmetric current stresses of the secondary components, including the magnetics and the rectifiers. The increase in conduction losses can be especially critical in low output-voltage, high output-current converters, which is the example presented. FIG. 4 shows the gate signals for the switches 24 and 26 under symmetric control. The switching frequency may, for example, be 400 kHz.

[0044] With reference to equations (8)-(10), the output voltage V_oof the converter 10 in steady state satisfies the following volt-second balance relation: $\begin{array}{cc}\left(\frac{N_s}{N_p}·\frac{V_{\mathrm{in}}}{2}-2V_0\right)D=2V_0\left(1-D\right)& \left(15\right)\end{array}$

[0045] Hence, the steady state output voltage V_ois given by the following equation: $\begin{array}{cc}{V}_0=\frac{N_s{V}_{\mathrm{in}}D}{4N_p}& \left(16\right)\end{array}$

[0046] The turns ratio (N_p:N_s) is preferably selected such that the duty ratio is approximately 50% under nominal input. With an input voltage V_inof 48V and an output voltage V_oof 3.3V as set forth above, a 2:1 turns ratio may be used, which results in 55% duty ratio under ideal conditions. Furthermore, to reduce winding conduction losses, the center-leg winding 70 on the center leg 54 and the two secondary windings 66 and 68 on the outer legs 56 and 58 preferably have a single turn (i.e., N_p=2, N_s=1, N_c=1).

[0047] The magnetic core 34 may be selected based on predicted core and winding losses. In one embodiment, a combination of E22/6/16 (planar E core, 3F 3 material) and PLT22/16/2.5 plate is used to form the magnetic core 34 (Data Handbook, MA01, Phillips Components 1996). Since the center-leg winding 70 carries all the load, current and the dc components of the fluxes on the two outer legs 56 and 58 are additive on the center leg 54, gapping of the center leg 54 is desirable in order to prevent saturation of the magnetic core 34. The length of the air gap x on the center leg 54 can be determined as follows. First, reluctance _c

[0048] and _o

[0049] are expressed by Eqs. (17) and (18): $\begin{array}{cc}{\Re }_c=\frac{x}{4\pi {10}^{-7}{A}_c}+\frac{\left(h-x\right)}{4\pi {10}^{-7}{\mu }_r{A}_c},& \left(17\right)\\ {\Re }_o=\frac{l_o}{4\pi {10}^{-7}{\mu }_r{A}_o},& \left(18\right)\end{array}$

[0050] where h=3.2 mm is the height of the core window area, A_c=78.5 mm²is the cross-sectional area of the center leg, A_o=39.25 mm²is the cross-sectional area of the outer legs, l_ois the length of the outer leg including the top and bottom portion of the core structure, and μ_ris the relative permeability of the core 34 which is assumed to be 1800. The steady state output current l_oat full power is about 30 A (producing 100 W output at 3.3V). Based on these parameters and Eq. (12), the desired air gap length can be solved for any given maximum flux density, B_max, from the following equation: $\begin{array}{cc}\frac{30\left(N_s+2N_c\right)}{\left(2{\Re }_c+{\Re }_o\right)A_c}=B_{\max }& \left(19\right)\end{array}$

[0051] For B_max=150 mT, the solution is x=0.362 mm. Then, using Eq. (7), the corresponding inductance L′ is found to be L′=1.17 μH.

[0052] FIG. 5 is a graph of an output voltage V_o(waveform 110) of the converter circuit 10 of FIGS. 1-5 as a function of time. The switching-frequency ripple in the output voltage V_ois reduced by about a factor of three in the converter circuit 10 as compared to a converter circuit without winding on the center leg 54. The winding 70 increases the effective filtering inductance L′ by a factor of three. The high frequency noise caused by device switching and parasitic ringing is also reduced by roughly the same factor. Additionally, the winding 70 significantly improves the efficiency at low switching frequency for a given output power, or at low output power under a given switching frequency. This is because of the increased filtering inductance that significantly reduces the secondary ripple and RMS current at low switching frequencies and/or under low output power, resulting in lower overall conduction losses despite the additional resistance presented by the center-leg winding.

[0053] FIG. 6 shows an alternative construction of an integrated CDR circuit 116. The construction of the integrated CDR circuit 116 is similar to the construction of the integrated CDR circuit 16 of FIG. 1, except that the winding 70 is replaced with two windings 170 and 172 wound on the outer legs 56 and 58, each of which has the same number of turns as the winding 70 and carries the total secondary current. The analysis presented in connection with FIGS. 1-5 applies to this construction as well. A disadvantage of this construction is the higher winding resistance due and the associated conduction losses to the series connection of two windings.

[0054] Although the invention has been described in conjunction with specific embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those of ordinary skilled in the art. Accordingly, it is intended to embrace all such alternatives, modifications and variations that fall within the spirit and scope of the appended claims.