KR20210037301A - Matrix switch and power converter using the same - Google Patents

Abstract

The present invention relates to a matrix switch and a power conversion device, which can be used to supply power to loads stacked in series or can be used as a general-purpose switch-capacitor converter. According to one aspect of the present invention, the matrix switch for selectively connecting between the M number of power sources and the N number of loads stacked in series comprises: a first switch group including the (N x M) number of switches selectively connecting a first terminal of each of the N number of loads and a first terminal of each of the M number of power sources; and a second switch group including the (N x M) number of switches selectively connecting a second terminal of each of the N number of loads and a second terminal of each of the M number of power sources.

Description

Matrix switch and power converter using the same {MATRIX SWITCH AND POWER CONVERTER USING THE SAME}

The present invention relates to a matrix switch and a power conversion device using the same. Specifically, the matrix switch and the power conversion device of the present invention may be used to supply power to a load stacked in series or may be used as a general-purpose switch-capacitor converter.

In recent years, high-performance digital circuits such as AP (Application Process) and GPU (Graphic Processing Unit) used in electronic devices such as smartphones, tablets, and notebooks have continuously lowered their operating voltage and their current consumption continues to increase.

In addition, high-performance digital circuits often employ parallel computing structures using a plurality of cores to improve performance. Referring to FIG. 1, while a plurality of cores 40 are connected in parallel and operated, power is supplied from the power converter 20 to operate. In this case, the power converter 20 must be able to supply a large current Ip corresponding to the sum of the currents consumed by a plurality of cores at a low voltage Vp in which one core operates. In this structure, the voltage Vp supplied by the power converter 20 is usually 1V or less, and the current Ip may reach about 10A. In terms of power supply, it is not efficient to supply a large current at such a low voltage.

As an alternative to this, as illustrated in FIG. 2, research is being conducted on a processor using a structure in which a plurality of cores 40 are stacked in series. In such a system, when the number of stacked cores 40 is N, the voltage Vs supplied by the power converter 20 becomes N times the voltage Vp of the parallel structure as the sum of the voltages of the plurality of cores, and the power converter The current Is supplied by (20) becomes 1/N times the current Ip of the parallel structure.

In this way, when the voltage supplied by the power converter 20 is N times and the current is 1/N times, the power loss is reduced due to the current reduction, and the equivalent impedance of the load is N ² times, so that it is difficult to stably supply power. It is advantageous, and there are advantages such as being able to reduce the number of I/O pads.

However, in the structure as illustrated in FIG. 2, since there is a possibility that the voltage Vs provided by the power converter 20 is not evenly distributed to each of the cores 40 connected in series, it is supplied to each core 40 A countermeasure is needed for balancing the voltage to be used.

As a conventional technique for voltage balancing of a plurality of loads stacked in series, Schaef proposes a method of using a DC-DC regulator including an inductor for each stack (C. Schaef and JT Stauth, "Efficient" Voltage Regulation for Microprocessor Cores Stacked in Vertical Voltage Domains," IEEE Transactions on Power Electronics, vol. 31, no. 2, pp. 1795-1808, Feb. 2016.). However, this method has the disadvantage of using a large number of large inductors.

As another prior art for voltage balancing of multiple loads stacked in series, Lee proposes a method of stably maintaining the intermediate node voltage of each stack using a switched-capacitor ladder converter ( SK Lee, T. Tong, X. Zhang, D. Brooks, and G. Wei, "A 16-Core Voltage-Stacked System With Adaptive Clocking and an Integrated Switched-Capacitor DC-DC Converter," IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 25, no. 4, pp. 1271-1284, Apr. 2017.). This method has an advantage in size as it does not use an inductor, but has a disadvantage in that the response speed to a change in load is slow.

As described above, there is a need for a method capable of stably supplying power while using a simple structure in a system in which a plurality of loads are stacked in series.

An object of the present invention is to stably supply power to a system in which a plurality of loads are stacked in series.

In addition, an object of the present invention is to enable voltage balancing with a simple structure in supplying power to a system in which a plurality of loads are stacked in series.

In addition, an object of the present invention is to enable communication with peripheral circuits in a multi-core processor stacked in series without taking a separate means such as a level shifter.

In addition, an object of the present invention is to provide a general-purpose step-down converter or step-up converter capable of easily changing the voltage conversion ratio.

One aspect of the present invention is a matrix switch that performs a selective connection between M power supplies and N loads stacked in series, the first terminal of each of the N loads and each of the M power supplies. A first switch group including N x M switches selectively connecting the first terminals; And a second switch group including N x M switches selectively connecting the second terminals of each of the N loads and the second terminals of each of the M power supplies.

The matrix switch may selectively connect a first terminal of each of the N loads to a first terminal of an arbitrary power source among M power sources, and connect a second terminal of each of the N loads to an arbitrary power source among M power sources. It can be selectively connected to the second terminal of.

The matrix switch may periodically or aperiodically change a connection relationship between the M power supplies and the N loads.

The matrix switch determines a load connected to each power source based on the voltage of the power source and/or the power consumption of the load, but is an independent switching method in which one load or a plurality of loads is arbitrarily connected to one power source. It can work.

In the matrix switch, N and M are the same, and the matrix switch determines a load connected to each power source based on the voltage of the power source and/or power consumption of the load, but one power source has one It can operate in a balanced switching method in which the load connects arbitrarily.

In the matrix switch, N and M are the same, and the matrix switch may operate in a rolling switching method in which only one load is connected to one power source and the connection between the power source and the load is changed in a predetermined order.

The matrix switch may periodically change a connection relationship between the M power sources and the N loads based on a first clock signal.

In the matrix switch, the first clock signal may be synchronized with a second clock signal used by the load.

In the matrix switch, a rising or falling edge of the first clock signal is generated between a falling edge of the second clock signal and a next rising edge, and a change in the connection relationship between the power source and the load is performed by the first clock signal. It can be triggered by the edge of one clock signal.

In the matrix switch, the load may be a core used in a multi-core processor.

In the matrix switch, N and M may be equal to the number of stacks of cores serially stacked in the multi-core processor, respectively.

In the matrix switch, the second terminal of the i-th power supply (i = 1, 2, ... M-1) may be connected to the first terminal of the i+1-th power supply.

In the matrix switch, at least one of the switches included in the first switch group and the second switch group may operate as one switch by connecting a plurality of semiconductor switching elements in series and/or in parallel.

In the matrix switch, the M power supplies may be obtained by dividing a voltage provided from an external device by M series-connected capacitors.

Another aspect of the present invention is a power conversion device for supplying power to N loads, comprising: a power converter that receives power from the outside and provides an output voltage; M capacitors connected in series for dividing the output voltage provided by the power converter; And a matrix switch for selectively connecting the M capacitors and the N loads.

In the power conversion device, the matrix switch may selectively connect each of the N loads to an arbitrary capacitor among the M capacitors.

In the power conversion device, the matrix switch may selectively connect a first terminal of each of the N loads to a first terminal of an arbitrary capacitor among M capacitors, and a second terminal of each of the N loads It can be selectively connected to the second terminal of any capacitor among the M capacitors.

In the power conversion apparatus, the matrix switch comprises: a first switch group including N x M switches connecting between first terminals of each of the N loads and first terminals of each of the M capacitors; And a second switch group including N x M switches connecting the second terminals of each of the N loads and the second terminals of each of the M capacitors.

In the power conversion device, the matrix switch may periodically or aperiodically change a connection relationship between the M capacitors and the N loads.

Another aspect of the present invention, the power conversion device described above; A multi-core processor in which a plurality of cores are stacked in series; And a peripheral circuit communicating with the multi-core processor, wherein the N loads are cores of each stage of the multi-core processor stacked in series into N stages.

In the electronic device, a core receiving power from a capacitor connected to the ground among the M capacitors performs communication with the peripheral circuit, and a core receiving power from a capacitor not connected to the ground among the M capacitors is the Communication with peripheral circuits may not be performed.

Another aspect of the present invention is a converter that transfers power between a first voltage connected through a first terminal and a second terminal and a second voltage connected through a third terminal and a fourth terminal, the first terminal A plurality of capacitors connected in series between the and the second terminal; And a matrix switch selectively connecting the second voltage to an arbitrary capacitor among the plurality of capacitors.

In the converter, the matrix switch may selectively connect the third terminal to a first terminal of an arbitrary capacitor among the plurality of capacitors, and connect the fourth terminal to a second terminal of an arbitrary capacitor among the plurality of capacitors. Can be selectively connected to the terminal.

In the converter, the matrix switch comprises: a first switch group including the same number of switches as the capacitors, selectively connecting between the third terminal and the positive terminals of each of the plurality of capacitors; And a second switch group including the same number of switches as the capacitors, selectively connecting the fourth terminal to the negative terminal of each of the plurality of capacitors.

In the converter, the first voltage is an input voltage, the second voltage is an output voltage, the number of capacitors is M, and the converter changes a connection relationship between the second voltage and the M capacitors. Meanwhile, the input voltage and the output voltage may operate as a step-down converter having a voltage conversion ratio of M:1.

In the converter, the first voltage is an output voltage, the second voltage is an input voltage, the number of capacitors is M, and the converter changes a connection relationship between the second voltage and the M capacitors. Meanwhile, the input voltage and the output voltage may operate as a step-up converter having a voltage conversion ratio of 1:M.

According to the present invention, according to an embodiment, it is possible to stably supply power to a system in which a plurality of loads are stacked in series.

Further, according to the present invention, voltage balancing is possible with a simple structure in supplying power to a system in which a plurality of loads are stacked in series, according to an embodiment.

In addition, according to the present invention, in a multi-core processor stacked in series, communication with peripheral circuits is possible without taking a separate means such as a level shifter, according to an embodiment.

Further, according to the present invention, according to an embodiment, it is possible to provide a general-purpose step-down converter or step-up converter capable of easily changing the voltage conversion ratio.

1 illustrates a processing system in which a plurality of cores are connected in parallel and operated.
2 illustrates a processing system in which a plurality of cores are connected in series and operated.
3 illustrates a matrix switch and a power conversion device according to an embodiment of the present invention.
4 and 5 are views exemplarily explaining the operation of the matrix switch of FIG. 3.
6 is a diagram illustrating an independent switching method according to an embodiment of the present invention.
7 is a diagram illustrating a balance switching method according to an embodiment of the present invention.
8 is a diagram illustrating a rolling switching method according to an embodiment of the present invention.
9 is a diagram exemplarily illustrating a communication method with peripheral circuits in a structure in which a plurality of cores are connected in series.
10 and 11 are views exemplarily illustrating a communication method between a core and a peripheral circuit in a system using a power conversion device according to an embodiment of the present invention.
12 is a diagram exemplarily illustrating a relationship between a core clock signal and a core current consumption in a processing system.
13 is a diagram exemplarily illustrating a matrix switch clock signal used in a matrix switch according to an embodiment of the present invention.
14 illustrates a matrix switch and a power conversion device according to an embodiment of the present invention.
15 illustrates a step-down converter using a matrix switch according to an embodiment of the present invention.
16 to 19 are views exemplarily explaining the operation of the step-down converter of FIG. 15.
20 illustrates a step-up converter using a matrix switch according to an embodiment of the present invention.
21 to 24 are views exemplarily explaining the operation of the step-up converter of FIG. 20.

Hereinafter, some embodiments of the present invention will be described in detail through exemplary drawings. In adding reference numerals to elements of each drawing, it should be noted that the same elements are assigned the same numerals as possible, even if they are indicated on different drawings. In addition, in describing the present invention, when it is determined that a detailed description of a related known configuration or function may obscure the subject matter of the present invention, a detailed description thereof will be omitted.

In addition, in describing the constituent elements of the present invention, terms such as first, second, A, B, (a) and (b) may be used. These terms are for distinguishing the constituent element from other constituent elements, and the nature, order, or order of the constituent element is not limited by the term. When a component is described as being "connected", "coupled" or "connected" to another component, the component may be directly connected or connected to that other component, but another component between each component It will be understood that elements may be “connected”, “coupled” or “connected”.

3 illustrates a matrix switch 310 and a power conversion device 300 according to an embodiment of the present invention.

The power conversion device 300 may include a matrix switch 310, a power converter 320, and a plurality of capacitors C1, C2, C3, and C4.

The power conversion device 300 may receive power from the outside through the power converter 320 and may supply power to a plurality of loads R1 to R4. The number of loads may be arbitrarily set, but for convenience of explanation, a case of supplying power to the four loads R1 to R4 will be described as an example. Here, the loads R1 to R4 may be in a serially stacked state or a floating state.

Depending on the embodiment, the power conversion device 300 may be included in an electronic device including a multi-core processor in which a plurality of cores are stacked in series and used to supply power to a plurality of cores. In this case, the four loads R1 to R4 may be cores of each stage in a multi-core processor having four stacks. That is, the number of loads may be the same as the number of stacks in a multi-core processor stacked in series.

According to an embodiment, the power conversion device 300 may be used as a voltage regulator as a part of a power management IC (PMIC) that supplies power to a processor included in an electronic device. Depending on the embodiment, the power conversion device 300 may be used as an integrated voltage regulator (IVR) built into a system on chip (SOC). Depending on the embodiment, the power conversion device 300 may be used to convert power in a system such as a data center and renewable energy. In addition, the power conversion device 300 may be used to convert power in various systems.

In FIG. 3, the loads R1 to R4 are shown as resistances according to a general convention, and the loads R1 to R4 may be any device that consumes power. It should be understood that just because the loads R1 to R4 are shown as resistance, it is not intended to exclude inductive or capacitive loads.

Meanwhile, in FIG. 3, it is shown that the loads R1 to R4 are included in the power conversion device 300, but this is for convenience of explanation, and the loads R1 to R4 are the targets for power supply of the power conversion device 300. However, it should be understood that the loads R1 to R4 are not included in the power conversion device 300, and this may be equally applied to other drawings of the present specification.

The power converter 320 may receive power from the outside and provide an output voltage Va to the matrix switch 310. Power converter 320 may be any converter that processes power. For example, the power converter 320 may be any one of a buck converter, a boost converter, a buck-boost converter, a switch-capacitor converter, or a linear regulator, or a combination thereof, but is not limited thereto.

The plurality of capacitors C1, C2, C3, and C4 may be stacked in series to divide the output voltage Va provided by the power converter 320. In this case, a voltage obtained by dividing the output voltage Va provided by the power converter 320 may be formed in each of the capacitors C1, C2, C3, and C4. Each of the capacitors C1, C2, C3, and C4 may receive power from the power converter 320 and may supply power to the matrix switch 310. From this point of view, each of the capacitors C1, C2, C3, and C4 may be understood as a separate power supply having a voltage obtained by dividing the output voltage Va provided by the power converter 320. The number of capacitors may be arbitrarily set as needed, but for convenience of explanation, a case in which four capacitors C1 to C4 are used will be described as an example. When the power conversion device 300 is included in an electronic device including a multi-core processor in which a plurality of cores are stacked in series and is used to supply power to a plurality of cores, the number of capacitors is the same as the number of stacks of the multi-core processor. can do.

The matrix switch 310 may perform selective connection between the four capacitors C1 to C4 and the four loads R1 to R4 through an on/off switching operation. For example, the matrix switch may selectively connect each of the four loads R1 to R4 to any of the four capacitors C1 to C4.

To this end, the matrix switch 310 may selectively connect the first terminal of each of the four loads R1 to R4 to the first terminal of an arbitrary capacitor among the four capacitors C1 to C4, and the four loads Each of the second terminals (R1 to R4) may be selectively connected to a second terminal of an arbitrary capacitor among the four capacitors (C1 to C4). For example, the first terminal may be understood as a positive (+) terminal, and the second terminal may be understood as a negative (-) terminal. For example, the first terminal may be understood as a high potential terminal, and the second terminal may be understood as a low potential terminal.

According to an embodiment, the matrix switch 310 includes 4 x 4 switches connecting between the first terminals of each of the four loads R1 to R4 and the first terminals of each of the four capacitors C1 to C4. It may include a first switch group (S11p, S12p, S13p, S14p, S21p, ~ S44p). In addition, the matrix switch 310 is a second switch including 4 x 4 switches connecting between the second terminals of each of the four loads R1 to R4 and the second terminals of each of the four capacitors C1 to C4. A second switch group including groups S11n, S12n, S13n, S14n, S21n, and S44n may be included.

Here, the reference symbol Sijp of the switch denotes a switch connecting the ith capacitor Ci and the first terminal (positive) of the j-th load Rj, and the reference numeral Sijn of the switch denotes the ith capacitor Ci and the j It can be understood to mean a switch connecting the second terminal (negative) of the load (Rj). That is, it can be understood that the first switch group selectively connects an arbitrary capacitor and a first terminal of an arbitrary load, and the second switch group selectively connects an arbitrary capacitor and the second terminal of an arbitrary load. .

Depending on the embodiment, the matrix switch 310 may periodically or aperiodically change the connection relationship between the four capacitors C1 to C4 and the four loads R1 to R4 through a switching operation. For example, the matrix switch 310 may periodically change the connection relationship between the capacitors C1 to C4 and the loads R1 to R4 in synchronization with a clock signal. Alternatively, as an example, the matrix switch 310 may change a connection relationship between the capacitors C1 to C4 and the loads R1 to R4 when an event occurs. For example, the event may occur when it is determined that the connection relationship needs to be changed according to the state of the voltage and current of the capacitors C1 to C4 and/or the voltage and current of the loads R1 to R4. have.

3 illustrates the case of using four loads (R1 to R4) and four capacitors (C1 to C4), but the number of loads can be set to an arbitrary number, N, and the number of capacitors is also arbitrary. It can be set to the number of M. In this case, the matrix switch 310 includes a first switch group including N x M switches connecting the first terminals of each of the N loads and the first terminals of each of the M capacitors, and each of the N loads. A second switch group including N x M switches connecting the second terminals and the second terminals of each of the M capacitors may be included. According to an embodiment, when the power conversion device 300 is used in a multi-core processor having L stacks, the number of loads N and the number of capacitors M may be equal to the number of stacks L, respectively.

In FIG. 3, at least one of the switches included in the first switch group (S11p, S12p, S13p, S14p, S21p, ~ S44p) and the second switch group (S11n, S12n, S13n, S14n, S21n, ~ S44n) is a plurality of The semiconductor switching elements may be connected in series and/or in parallel to be configured to operate as a single switch. In addition, at least one of the capacitors C1 to C4 may be configured such that a plurality of capacitors are connected in series and/or in parallel to operate as one capacitor. In referring to the number of switches or capacitors in the present specification, when a plurality of switches or capacitors operate as one switch or capacitor, it may be understood that one switch or capacitor is used.

4 and 5 are views exemplarily explaining the operation of the matrix switch 310 of FIG. 3.

4, a matrix switch 310 connects a first capacitor C1 and a first load R1, a second capacitor C2 and a second load R2, and a third capacitor ( C3) and the third load R3 may be connected, and the fourth capacitor C4 and the fourth load R4 may be connected. To this end, S11p, S11n, S22p, S22n, S33p, S33n, S44p, and S44n of the matrix switch 310 may be turned on and the other switches may be turned off.

5, the matrix switch 310 connects the first capacitor C1 and the fourth load R4, connects the second capacitor C2 and the third load R3, and connects the third capacitor ( C3) and the second load R2 may be connected, and the fourth capacitor C4 and the first load R1 may be connected. To this end, S14p, S14n, S23p, S23n, S32p, S32n, S41p, and S41n of the matrix switch 310 may be turned on and the other switches may be turned off.

4 and 5 illustrate two connection relationships, in this way, the matrix switch 310 transfers an arbitrary load among the four loads R1 to R4 to an arbitrary capacitor among the four capacitors C1 to C4. Can be connected selectively. Each of the loads R1 to R4 may operate by receiving power from a connected capacitor.

In this way, the matrix switch 310 may perform voltage balancing of capacitors using a method such as connecting a load having a high power consumption to a capacitor having a high voltage while changing the connection relationship between a plurality of loads and a plurality of capacitors. . That is, in the case of using the matrix switch 310 of this embodiment, voltage balancing is possible by simply changing the connection relationship between the power source and the load, so that a regulator is added for each capacitor to adjust the voltage of each capacitor. Compared to the technology, it is possible to easily perform voltage balancing with a simple structure.

Next, with reference to FIGS. 6 to 8, a description will be given of a manner in which the matrix switch changes the connection relationship between the capacitor and the load.

6 is a diagram illustrating an independent switching method according to an embodiment of the present invention.

6(a) shows the connection between the first capacitor C1 and the first load R1, the connection between the second capacitor C2 and the second load R2, and the third capacitor C3 and the third load. A state in which (R3) is connected and the fourth capacitor (C4) and the fourth load (R4) are connected is shown.

6(b) shows that the positions of the second load R2 and the third load R3 in FIG. 6(a) are changed to each other to connect the first capacitor C1 and the first load R1, 2 A state in which the capacitor C2 and the third load R3 are connected, the third capacitor C3 and the second load R2 are connected, and the fourth capacitor C4 and the fourth load R4 are connected. Is shown.

6(c) is a change in which the fourth load R4 is connected to the first capacitor C1 in FIG. 6(b), and the first capacitor C1 is connected to the first load R1 and the fourth load ( It is connected to R4) at the same time, the second capacitor C2 and the third load R3 are connected, and the third capacitor C3 and the second load R2 are connected.

Referring to FIG. 6, one load or a plurality of loads may be arbitrarily connected to one capacitor (or power source), and a load may not be connected to some of the power sources. This connection scheme can be referred to as an independent switching scheme. The independent switching method has a disadvantage in that the control is relatively complex, but has an advantage in that the response speed is fast and the voltage balancing performance between capacitors is excellent. Here, the load connected to each capacitor may be determined so that voltage balancing between the capacitors is efficient based on the voltage of each capacitor (or power source) and/or power consumption of each of the loads.

7 is a diagram illustrating a balance switching method according to an embodiment of the present invention.

7(a) shows a connection between the first capacitor C1 and the first load R1, the connection between the second capacitor C2 and the second load R2, and the third capacitor C3 and the third load. A state in which (R3) is connected and the fourth capacitor (C4) and the fourth load (R4) are connected is shown.

7(b) shows that the positions of the third load R3 and the fourth load R4 in FIG. 7(a) are changed to each other to connect the first capacitor C1 and the first load R1. 2 A state in which the capacitor C2 and the second load R2 are connected, the third capacitor C3 and the fourth load R4 are connected, and the fourth capacitor C4 and the third load R3 are connected. Is shown.

7(c) shows that the positions of the first load R1 and the third load R3 are changed from each other in FIG. 7(b), and the positions of the second load R2 and the fourth load R4 are changed from each other. As one, the first capacitor C1 and the third load R3 are connected, the second capacitor C2 and the fourth load R4 are connected, and the third capacitor C3 and the second load R2 are connected. And the fourth capacitor C4 and the first load R1 are connected.

Referring to FIG. 7, one load may be arbitrarily connected to one capacitor (or power source) (that is, a plurality of loads are connected to any one of the capacitors or a capacitor to which no load is connected does not exist). Therefore, in this method, the number of capacitors and the number of loads may be the same. However, there is no special limitation on the change of the connection relationship between the capacitor and the load. For example, two arbitrary loads may change positions with each other, or three or four arbitrary loads may change positions with each other, and so on. This connection method may be referred to as a balanced switching method. In the balanced switching method, the control complexity, response speed, and voltage balancing performance may be intermediate compared to the above-described independent switching method and a rolling switching method to be described later. Here, the load connected to each capacitor may be determined so that voltage balancing between the capacitors is efficient based on the voltage of each capacitor (or power source) and/or power consumption of each of the loads.

8 is a diagram illustrating a rolling switching method according to an embodiment of the present invention.

8(a) shows a first capacitor C1 and a first load R1, a second capacitor C2 and a second load R2, and a third capacitor C3 and a third load. A state in which (R3) is connected and the fourth capacitor (C4) and the fourth load (R4) are connected is shown.

Fig. 8(b) shows that all four loads R1 to R4 are moved one by one in Fig. 8(a), and connects the first capacitor C1 and the fourth load R4, and the second capacitor C2 And the first load R1 are connected, the third capacitor C3 and the second load R2 are connected, and the fourth capacitor C4 and the third load R3 are connected.

Fig. 8(c) shows that all four loads R1 to R4 are moved one by one again in Fig. 8(b). The first capacitor C1 and the third load R3 are connected, and the second capacitor C2 is ) And the fourth load R4, the third capacitor C3 and the first load R1 are connected, and the fourth capacitor C4 and the second load R2 are connected. .

Referring to FIG. 8, one load is connected to one capacitor (or power source) (the number of capacitors and the number of loads are the same), but the connection relationship between the capacitor and the load is not arbitrarily changed, but the power supply and the power supply in a predetermined order It is a method of changing the connection of the load. This connection method may be referred to as a rolling switching method. The rolling switching method has the advantage of being simple to implement because it is simple to control, but has a disadvantage in that the response speed is slow compared to the above-described independent switching method or balanced switching method. However, even in the rolling switching method, the voltage balancing function can be sufficiently performed. In other words, if the rolling switching method is used in a situation where there is a difference in the current consumed by each of the plurality of loads, the current supplied to the load by each of the plurality of capacitors becomes the average of the currents consumed by the plurality of loads and is equal to each other. (After the switching operation is repeated enough), the voltage balancing performance may be sufficient to apply to the actual system.

As described above, as a method of connecting a capacitor and a load using a matrix switch, three examples of independent switching, balanced switching, and rolling switching have been exemplified, but are not limited thereto. Although any of the above-described three methods may be used, the voltage balancing function may be sufficiently performed, but there may be differences in control complexity and response speed depending on the method.

9 is a diagram illustrating a communication method between a core and a peripheral circuit (peri) in an electronic device using a multi-core processor in which a plurality of cores are connected in series. Here, the peripheral circuit peri may be a memory, a sensor, a display, an input/output device, or the like.

In a multi-core processor in which a plurality of cores are connected in series, some of the plurality of cores may not be connected to a reference potential (eg, ground or ground). On the other hand, it is common to process signals while the peripheral circuit peri is connected to the reference potential. In this way, the core not connected to the reference potential and the peripheral circuit peri connected to the reference potential need to have separate means for data communication. For example, the peripheral circuit peri is also configured in a serially stacked state, or an element such as a level shifter needs to be used between the core and the peripheral circuit peri. This may be a disadvantage of a multi-core processor in which a plurality of cores are connected in series.

10 and 11 are views exemplarily illustrating a communication method between a core and a peripheral circuit peri in a system using the power conversion device 300 according to an embodiment of the present invention.

10 illustrates a state in which the first core core1 receives power from the fourth capacitor C4 by the switching operation of the matrix switch 310. In this case, since the first core (core1) is connected to the reference potential, it is possible to communicate with the peripheral circuit (peri) without any other means.

11 illustrates a state in which the fourth core core1 receives power from the fourth capacitor C4 by the switching operation of the matrix switch 310. In this case, since the fourth core (core4) is connected to the reference potential, it is possible to communicate with the peripheral circuit (peri) without any other means.

In this way, when the power conversion device 300 illustrated in FIG. 3 is used to supply power to an electronic device including a multi-core processor, each of the plurality of cores is from the capacitor C4 connected to the reference potential. You can have a section to receive power (even if not all cores are simultaneously powered from capacitors connected to the reference potential). That is, since the power conversion device 300 according to the embodiment of the present invention can change the core to be supplied with power from the capacitor C4 connected to the reference potential through the matrix switch 310 at any time, a plurality of cores ( Each of the core) may have a section connected to the reference potential and operating. Therefore, each of the plurality of cores can communicate with the peripheral circuit peri when connected to the reference potential, and in this case, a separate means such as a level shifter between the core and the peripheral circuit peri You don't have to go for it. Such a communication method may be referred to as a single channel communication method.

As described above, when using the power conversion device 300 according to an embodiment of the present invention, even in a multi-core processor stacked in series, a single channel communication method without having to serially stack peripheral circuits or devise a separate means such as a level shifter. Can be used to communicate with peripheral circuits.

Next, a method of generating a clock signal for a matrix switch will be described with reference to FIGS. 12 and 13. 12 is a diagram exemplarily illustrating a relationship between a core clock signal CK_c and a current consumption I_c of a core in a processing system, and FIG. 13 is a diagram illustrating a matrix switch clock signal CK_MS according to an embodiment of the present invention. It is a figure explaining the method of generation by way of example.

12 illustrates a core clock signal CK_c used by the core and a current I_c consumed by the core. Since the semiconductor devices inside the core generally operate in synchronization with the rising edge or the falling edge of the core clock signal CK_c, the current consumption I_c of the core is in the period where the core clock signal CK_c is present. It may have a tendency to increase and decrease in a period in which the core clock signal CK_c is not present.

The matrix switch according to the embodiment of the present invention operates in a manner that temporarily cuts off the connection between the capacitor (power) and the core and then changes the connection relationship again, as illustrated in FIGS. 4 and 5. When supplying and blocking are repeated, voltage spikes or voltage ringing may occur due to parasitic inductances present in wiring or the like. The voltage spike depends on the magnitude of the current. If the voltage spike is large, it may act as a voltage stress on each device inside the system, affecting the stable operation of the system. Therefore, it is preferable that the switching of the matrix switch is performed when the current is small. From this point of view, referring to FIG. 12, it is preferable that the switching of the matrix switch is performed during a transition of MS in which the core clock signal CK_c is not present, for example, between t2 and t3 and between t4 and t5. will be.

Referring to FIG. 13, the matrix switch may periodically change a connection relationship between a plurality of power sources and a plurality of cores based on a matrix switch clock signal CK_MS. According to an embodiment, the matrix switch clock signal CK_MS may be synchronized with the core clock signal CK_c used by the core. Depending on the embodiment, the rising or falling edge of the matrix switch clock signal CK_MS is generated between the falling edge of the core clock signal CK_c and the next rising edge (e.g., between t2 and t3 and between t4 and t5). ), a change in a connection relationship between a plurality of power sources and cores (ie, switching of a matrix switch) may be triggered by an edge of the matrix switch clock signal CK_MS.

On the other hand, the core clock signal (CK_c) is usually as high as several hundred MHz, and when it is not necessary to perform the switching of the matrix switch at the same frequency as the frequency of the core clock signal (CK_c), according to the embodiment, the matrix switch clock signal (CK_MS) ) May be generated at a lower frequency than the frequency of the core clock signal CK_c. For example, whenever a period between the falling edge of the core clock signal CK_c and the next rising edge occurs several times, a pulse of the matrix switch clock signal CK_MS may be generated once.

In this way, when switching of the matrix switch is performed in synchronization with the core clock signal (CK_c), the switching of the matrix switch may be performed in a section in which the core current (I_c) is small, thereby reducing voltage spikes or ringing due to parasitic inductance. have.

14 illustrates a matrix switch 1410 and a power conversion device 1400 according to an embodiment of the present invention.

In the power conversion device 1400 of FIG. 14, the power converter 1420 generates a plurality of power sources Vb1, Vb2, Vb3, and Vb4 and provides them to the matrix switch 1410. ), the power converter 320 provides one output Va and generates a plurality of power sources using a plurality of capacitors C1 to C4.

According to an embodiment, when the plurality of power supplies provided by the power converter 1420 is M, the second terminal of the i-th (i = 1, 2, ... M-1) power supply is the second terminal of the i+1-th power supply. It can be provided in a state connected to the 1 terminal. Here, the first terminal may be a positive terminal or a high potential terminal, and the second terminal may be a negative terminal or a low potential terminal.

Since the contents described with reference to FIGS. 3 to 13 may be similarly applied to the power conversion device 1400, a redundant description will be omitted.

15 illustrates a step-down converter 1500 using the matrix switch 1510 according to an embodiment of the present invention. The step-down converter 1500 may include a plurality of capacitors C1 to C4 and a matrix switch 1510.

The step-down converter 1500 may transfer power between a first voltage Vin connected through the first and second terminals and a second voltage Vo connected through the third and fourth terminals. Here, the first terminal may be a positive terminal or a high potential terminal of the first voltage Vin, and the second terminal may be a negative terminal or a low potential terminal of the first voltage Vin. The third terminal may be a positive terminal or a high potential terminal of the second voltage Vo, and the fourth terminal may be a negative terminal or a low potential terminal of the second voltage Vo.

The plurality of capacitors C1 to C4 may be connected in series between the first terminal and the second terminal. The plurality of capacitors C1 to C4 may divide the first voltage Vin and provide it to the matrix switch 1510.

The matrix switch 1510 may selectively connect the second voltage Vo to an arbitrary capacitor among the plurality of capacitors C1 to C4. According to an embodiment, the matrix switch 1510 may selectively connect a third terminal to a first terminal of an arbitrary capacitor among a plurality of capacitors C1 to C4, and a fourth terminal to a plurality of capacitors C1 to C4. It can be selectively connected to the second terminal of any of the capacitors.

According to an embodiment, the matrix switch 1510 includes a plurality of switches S11p, S21p, S31p, and S41p selectively connecting a third terminal and a first terminal of each of the plurality of capacitors C1 to C4. It may include one switch group S11p, S21p, S31p, and S41p. Here, the number of switches in the first switch group may be the same as the number of capacitors. In addition, the matrix switch 1510 may include a second switch group including a plurality of switches S11n, S21n, S31n, and S41n selectively connecting a fourth terminal and a second terminal of each of the plurality of capacitors. . Here, the number of switches in the second switch group may be the same as the number of capacitors.

Depending on the embodiment, the first voltage may be the input voltage Vin, the second voltage may be the output voltage Vo, and the number of capacitors may be M. The step-down converter 1500 changes the connection relationship between the second voltage Vo and the M capacitors, while the step-down converter 1500 has a voltage conversion ratio of M:1 in the ratio of the input voltage Vin and the output voltage Vo. -down) Can operate as a converter.

Although four capacitors C1 to C4 are illustrated in FIG. 15 as being used, the number of capacitors may be variously set as needed, and the voltage conversion ratio may vary accordingly.

16 to 19 are views exemplarily explaining the operation of the step-down converter 1500 of FIG. 15.

16 to 19, the step-down converter 1500 may connect the second voltage Vo to the first capacitor C1 in the first section (Fig. 16), and the second voltage Vo in the second section. ) Can be connected to the second capacitor C2 (Fig. 17), the second voltage Vo can be connected to the third capacitor C3 in the third section (Fig. 18), and the second voltage in the fourth section (Vo) may be connected to the fourth capacitor C4 (FIG. 19). Assuming that each of the first to fourth capacitors C1 to C4 divides the input voltage Vin equally, the voltage of each of the capacitors C1 to C4 is Vin/4. When the first to fourth sections are repeated as shown in FIGS. 16 to 19, the output voltage Vo may be Vin/4 in any of the first to fourth sections. That is, the step-down converter 1500 may operate as a step-down converter having a voltage conversion ratio of 4:1 in which the ratio of the input voltage Vin and the output voltage Vo is 4:1. According to an embodiment, when the number of capacitors is changed to M, the step-down converter 1500 may operate as a step-down converter having a voltage conversion ratio of M:1.

20 illustrates a step-up converter 2000 using the matrix switch 9210 according to an embodiment of the present invention.

The boost converter 2000 may include a plurality of capacitors C1 to C4 and a matrix switch 2010.

The step-up converter 2000 of FIG. 20 is different from the step-down converter 1500 of FIG. 15 in that the positions of the input voltage Vin and the output voltage Vo are changed. That is, in the boost converter 2000, the sum of voltages formed in the plurality of capacitors C1 to C4 may form the output voltage Vo.

Depending on the embodiment, the first voltage may be the output voltage Vo, the second voltage may be the input voltage Vin, and the number of capacitors may be M. The step-up converter 2000 is a step-up converter having a voltage conversion ratio of 1:M in a ratio of the input voltage Vin and the output voltage Vo while changing the connection relationship between the second voltage Vin and the M capacitors. It can work.

Although four capacitors C1 to C4 are illustrated in FIG. 20 as being used, the number of capacitors may be variously set as necessary, and a voltage conversion ratio may vary accordingly.

21 to 24 are views exemplarily explaining the operation of the step-up converter 2000 of FIG. 20.

21 to 24, the step-up converter 2000 may connect the second voltage Vin to the first capacitor C1 in the first section (FIG. 21), and the second voltage Vin in the second section. ) Can be connected to the second capacitor C2 (Fig. 22), the second voltage Vin can be connected to the third capacitor C3 in the third section (Fig. 23), and the second voltage in the fourth section (Vin) may be connected to the fourth capacitor C4 (FIG. 24). By such an operation, the voltage of each of the capacitors C1 to C4 may be Vin. When the first to fourth periods are repeated as shown in FIGS. 21 to 24, the output voltage Vo is the sum of the voltages of the four capacitors C1 to C4 and may be four times the input voltage Vin. That is, the step-up converter 2000 may operate as a step-up converter having a voltage conversion ratio of 1:4 in which the ratio of the input voltage Vin and the output voltage Vo is 1:4. According to an embodiment, when the number of capacitors is changed to M, the boost converter 2000 may operate as a boost converter having a voltage conversion ratio of 1:M.

As described above, the matrix switch according to an embodiment of the present invention may be used to implement a general-purpose step-down converter having a voltage conversion ratio of M:1 or a general-purpose step-down converter having a voltage conversion ratio of 1:M. In addition, the step-down converter or step-up converter of the present embodiment has the advantage of being able to simply change the voltage conversion ratio only by changing the number of capacitors.

As described above, according to the present invention, according to an embodiment, power can be stably supplied to a system in which a plurality of loads are stacked in series. Further, according to the present invention, voltage balancing is possible with a simple structure in supplying power to a system in which a plurality of loads are stacked in series, according to an embodiment. In addition, according to the present invention, in a multi-core processor stacked in series, communication with peripheral circuits is possible without taking a separate means such as a level shifter, according to an embodiment. Further, according to the present invention, it is possible to provide a general-purpose step-down converter or step-up converter capable of easily changing a voltage conversion ratio according to an embodiment.

Terms such as "comprise", "comprise", or "have" described above mean that the corresponding component may be embedded, unless otherwise stated, and therefore, other components are not excluded. It should be interpreted as being able to further include other components. All terms, including technical or scientific terms, unless otherwise defined, have the same meaning as commonly understood by one of ordinary skill in the art to which the present invention belongs. Terms generally used, such as terms defined in the dictionary, should be interpreted as being consistent with the meaning in the context of the related technology, and are not interpreted as ideal or excessively formal meanings unless explicitly defined in the present invention.

The above description is merely illustrative of the technical idea of the present invention, and those of ordinary skill in the art to which the present invention pertains will be able to make various modifications and variations without departing from the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but to explain the technical idea, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

Claims (20)

As a matrix switch that performs selective connection between M power supplies and N loads stacked in series,
A first switch group including N x M switches selectively connecting a first terminal of each of the N loads and a first terminal of each of the M power supplies; And
Matrix switch comprising: a second switch group including N x M switches selectively connecting between the second terminals of each of the N loads and the second terminals of each of the M power supplies. The method according to claim 1,
The matrix switch may selectively connect a first terminal of each of the N loads to a first terminal of an arbitrary power source among M power sources, and connect a second terminal of each of the N loads to an arbitrary power source among M power sources. Matrix switch, characterized in that it can be selectively connected to the second terminal of. The method according to claim 1,
Wherein the matrix switch periodically or aperiodically changes a connection relationship between the M power supplies and the N loads. The method of claim 3,
The matrix switch determines a load connected to each power source based on the voltage of the power source and/or the power consumption of the load, but is an independent switching method in which one load or a plurality of loads is arbitrarily connected to one power source. Matrix switch, characterized in that to operate. The method of claim 3,
N and M are the same,
The matrix switch determines a load connected to each power source based on the voltage of the power source and/or the power consumption of the load, and operates in a balanced switching method in which one load is arbitrarily connected to one power source. Matrix switch. The method of claim 3,
N and M are the same,
The matrix switch is a matrix switch, characterized in that operating in a rolling switching method in which only one load is connected to one power source and the connection between the power source and the load is changed in a predetermined order. The method of claim 3,
The matrix switch periodically changes the connection relationship between the M power supplies and the N loads based on a first clock signal,
The matrix switch, wherein the first clock signal is synchronized with a second clock signal used by the load. The method of claim 7,
A rising or falling edge of the first clock signal is generated between a falling edge of the second clock signal and a next rising edge,
The change in the connection relationship between the power source and the load is triggered by an edge of the first clock signal. The method according to claim 1,
The load is a core used in a multi-core processor,
Wherein N and M are the same as the number of stacks of cores stacked in series in the multi-core processor, respectively.
As a power conversion device that supplies power to N loads,
A power converter that receives power from the outside and provides an output voltage;
M capacitors connected in series for dividing the output voltage provided by the power converter; And
And a matrix switch for selectively connecting the M capacitors and the N loads. The method of claim 10,
Wherein the matrix switch is capable of selectively connecting each of the N loads to an arbitrary capacitor among the M capacitors. The method of claim 10,
The matrix switch may selectively connect a first terminal of each of the N loads to a first terminal of an arbitrary capacitor among M capacitors, and a second terminal of each of the N loads may be connected to an arbitrary capacitor among M capacitors. Power conversion device, characterized in that it can be selectively connected to the second terminal of. The method of claim 10,
The matrix switch,
A first switch group including N x M switches connecting the first terminals of each of the N loads and the first terminals of each of the M capacitors; And
And a second switch group including N x M switches connecting the second terminals of each of the N loads and the second terminals of each of the M capacitors.
A power conversion device according to claim 10;
A multi-core processor in which a plurality of cores are stacked in series; And
Including; peripheral circuits in communication with the multi-core processor,
Wherein the N loads are cores of each stage in the multi-core processor stacked in series in N stages. The method of claim 14,
Among the M capacitors, a core receiving power from a capacitor connected to the ground performs communication with the peripheral circuit,
And a core receiving power from a capacitor not connected to the ground among the M capacitors does not communicate with the peripheral circuit.
A converter that transfers power between a first voltage connected through a first terminal and a second terminal and a second voltage connected through a third terminal and a fourth terminal,
A plurality of capacitors connected in series between the first terminal and the second terminal; And
A matrix switch selectively connecting the second voltage to any of the plurality of capacitors;
Converter comprising a. The method of claim 16,
The matrix switch,
The third terminal may be selectively connected to a first terminal of any capacitor among the plurality of capacitors, and the fourth terminal may be selectively connected to a second terminal of any capacitor among the plurality of capacitors. Converter. The method of claim 16,
The matrix switch,
A first switch group including the same number of switches as the capacitors, selectively connecting between the third terminal and the positive terminals of each of the plurality of capacitors; And
And a second switch group including the same number of switches as the capacitors, selectively connecting between the fourth terminal and the negative terminal of each of the plurality of capacitors. The method of claim 16,
The first voltage is an input voltage,
The second voltage is an output voltage,
The number of capacitors is M,
The converter, while changing a connection relationship between the second voltage and the M capacitors, operates as a step-down converter having a voltage conversion ratio of M:1 in a ratio of the input voltage and the output voltage. Converter characterized by. The method of claim 16,
The first voltage is an output voltage,
The second voltage is an input voltage,
The number of capacitors is M,
The converter, while changing a connection relationship between the second voltage and the M capacitors, operates as a step-up converter having a ratio of the input voltage and the output voltage to a voltage conversion ratio of 1:M. Converter characterized by.

Images