Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
POWER DISTRIBUTION NETWORK (PDN) CONDITIONER
Document Type and Number:
WIPO Patent Application WO/2015/119971
Kind Code:
A1
Abstract:
Systems and methods for conditioning a power rail (230) (e.g., reducing voltage droops and/or voltage overshoots on the power rail) are described herein. In one embodiment, a power circuit comprises a capacitor (215) coupled to a high-voltage rail (210), and a droop slope limiter (DSL) coupled between the high-voltage rail and a power rail. The DSL is configured to detect a downward voltage slope on the power rail, and to control current flow from the high-voltage rail to the power rail through the DSL based on the detected downward voltage slope.

Inventors:
LI, Qing (5775 Morehouse Drive, San Diego, CA, 92121-1714, US)
Application Number:
US2015/014313
Publication Date:
August 13, 2015
Filing Date:
February 03, 2015
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
QUALCOMM INCORPORATED (Attn: International IP Administration, 5775 Morehouse DriveSan Diego, CA, 92121-1714, US)
International Classes:
H02J1/02; G11C5/14; H02H9/04; H02M1/15; G01R19/12
Foreign References:
US20090174262A12009-07-09
US20030085754A12003-05-08
US20130328536A12013-12-12
Other References:
YU-HUEI LEE ET AL: "A Low Quiescent Current Asynchronous Digital-LDO With PLL-Modulated Fast-DVS Power Management in 40 nm SoC for MIPS Performance Improvement", IEEE JOURNAL OF SOLID-STATE CIRCUITS, IEEE SERVICE CENTER, PISCATAWAY, NJ, USA, vol. 48, no. 4, 1 April 2013 (2013-04-01), pages 1018 - 1030, XP011497974, ISSN: 0018-9200, DOI: 10.1109/JSSC.2013.2237991
KISUN LEE ET AL: "Analysis and Design of Adaptive Bus Voltage Positioning System for Two-Stage Voltage Regulators", IEEE TRANSACTIONS ON POWER ELECTRONICS, INSTITUTE OF ELECTRICAL AND ELECTRONICS ENGINEERS, USA, vol. 24, no. 12, 1 December 2009 (2009-12-01), pages 2735 - 2745, XP011282604, ISSN: 0885-8993, DOI: 10.1109/TPEL.2009.2030806
Attorney, Agent or Firm:
MATTHEW, Benjamin, B. et al. (Novak Druce Connolly Bove + Quigg LLP, 333 S. Grand Ave. 23FLos Angeles, CA, 90071, US)
Download PDF:
Claims:
1. A power circuit, comprising:

a capacitor coupled to a high- voltage rail: and

a droop slope limiter (DSL) coupled between the high-voltage rail and a power rail, wherem the DSL is configured to detect a downward voltage slope on the power rail, and to control current flow from the hig -voltage rail to the power rail through the DSL based on the detected downward voltage slope.

2. The power circuit of claim 1, wherein the high-voltage rail is at a voltage that is at least 50 percent greater than the power rail

3. The power circuit of claim 1 , wherein the DSL is configured to control the current flow by blocking the current flow if a magnitude of the detected downward voltage slope is below a slope threshold, and allowing the current flow if the magnitude of the detected downward voltage slope exceeds the slope threshold.

4. The power circuit of claim 3, wherein, during a time that the DSL allows the current flow, the DSL is configured to control the current flo through the DSL such that the magnitude of the detected downward voltage slope is approximately limited to the slope threshold,

5. The power circuit of claim 1. wherein the DSL comprises:

a slope-detection circuit coupled to the power rail, wherein the slope-detection circuit is configured to detect the downward voltage slope on the power rail, and to output a slope signal based on the detected downward voltage slope;

a power transistor coupled between the high-voltage rail and the power rail; and a control circuit configured to compare the slope signal from the slope-detection circuit with a slope-threshold signal, and to control a conductivity of the power transistor based on the comparison.

6. The power circuit of claim 5, wherein the power transistor comprises a field effect transistor (FET), and the control circuit is configured to control the conductivity of the power transistor by controlling a gate voltage of the power transistor.

7. The power circuit of claim 1. further comprising an overshoot slope limiter (OSL) coupled between the power rail and a ground, wherein the OSL is configured to detect an upward voltage slope on the power rail, and to control current flow from the power rail to the ground through the OSL based on the detected upward voltage slope.

8. The power circuit of claim 7, wherein the OSL is configured to control the current flow from the power rail to the ground by blocking the curre t flow from the power rail to the ground if the detected upward voltage slope is below a slope threshold, and allowing the current flow from the power rail to the ground if the detected upward voltage slope exceeds the slope threshold.

9. The power circuit of claim 8, wherein, during a time that the OSL allows the current flow from the power rail to the ground, the OSL is configured to control the current flow from the power rail to the ground through the OSL such that the detected upward voltage slope is approximately limited to the slope threshold.

10. The power circuit of claim 1, wherein the power circuit is integrated on a single chip,

1 1. The power circuit of claim 10, further comprising a ead switch coupled between the power rail and a circuit on the chip, wherein the head switch is configured to power gate the circuit on the chip.

12. A method for conditioning a power rail, comprising:

detecting a downward voltage slope on the power rail; and

controlling current flow from a high-voltage rail to the power rail based on the detected downward voltage slope, wherein a capacitor is coupled to the high-voltage rail.

13. The method of claim 12. wherein the high-voltage rail is at a voltage that is at least 50 percent greater than the power rail. 4. The method of claim 12, wherein controlling the current flow further comprises: blocking the current flow if a magnitude of the detected downward voltage slope is below a slope threshold; and

allowing the current flow if the magnitude of the detected downward voltage slope exceeds the slope threshold.

15. The method of claim 14, wherein allowing the current flow further comprises adjusting the current flow such that the magnitude of the detected downward voltage slope is approximately limited to the slope threshold.

16. The method of claim 15, wherein a field effect transistor (FET) is coupled between the high-voltage rail and the power rail, and adjusting the current flow comprises adjusting a gate voltage of the FET. 7. The method of claim 12, further comprising:

detecting an upward voltage slope on the power rail: and

controlling current flow from the power rail to a ground based on the detected upward voltage slope.

18. The method of claim 17, wherein controlling the current flow from the powrer rail to the ground further comprises:

blocking the current flow from the power rail to the ground if the detected upward voltage slope is below a slope threshold; and

allowing the current flow from the power rail to the ground if the detected upward voltage slope exceeds the slope threshold.

19. The method of claim 18, wherein allowing the current flow- from the power rail to the ground further comprises adjusting the current flow from the power rail to the ground such that the detected upward voltage slope is approximately limited to the slope threshold.

20. The method of claim 12. wherein the detecting and the controlling are performed on a single chip.

2 . An apparatus for conditioning a power rail, comprising:

means for detecting a downward voltage slope on the power rail: and means for controlling current flow from a high-voltage rail to the power rail based on the detected downward voltage slope, wherein a capacitor is coupled to the high-voltage rail.

22. The apparatus of claim 21 , wherein the high-voltage rail is at a voltage that is at least 50 percent greater than the power rail.

23. The apparatus of claim 21, wherein the means for controlling the current flow further comprises:

means for blocking the current flow if a magnitude of the detected downward voltage slope is below a slope threshold; and

means for allowing the current flow if the magnitude of the detected downward voltage slope exceeds the slope threshold.

24. The apparatus of claim 23, wherein the means for allowing the current flow further comprises means for adj usting the current flow such that the magnitude of the detected downward voltage slope is approximately limited to the slope threshold.

25. The apparatus of claim 24, wherein a field effect transistor (FET) is coupled between the high-voltage rail and the power rail, and the means for adjusting the current flow comprises means for adjusting a gate voltage of the FET.

26. The apparatus of claim 21, further comprising:

means for detecting an upward voltage slope on the power rail; and

means for controlling current flow from the power rail to a ground based on the detected upward voltage slope.

27. The apparatus of claim 26, wherein the means for controlling the current flow from the power rail to the ground further comprises: means for blocking the current flow from the power rail to the ground if the detected upward voltage slope is below a slope threshold; and

means for allowmg the current flow from the power rail to the ground if the detected upward voltage slope exceeds the slope threshold.

28. The apparatus of claim 27, wherein the means for allowing the current flow from the power rail to the ground further comprises means for adjusting the current flow from the power rail to the ground such that the detected upward voltage slope is

approximately limited to the slope threshold.

29. The apparatus of claim 21 , wherein the apparatus is integrated on a single chip.

30. The apparatus of claim 29, further comprising a means for power gating a circuit on the chip, wherein the means for power gating is coupled between the power rail and the circuit on the chip.

Description:
POWER DISTRIBUTION NETWORK (PDN) CONDITIONER

BACKGROUND

Field

[0001] Aspects of the present disclosure relate generally to power distribution networks, and. more particularly, to power distribution network (PDN) conditioners.

Background

[0002] A power distribution network (PDN) is used to distribute power to various circuits in a system on a chip (SoC) from an off-chip power source (e.g., power management integrated circuit (PMIC)). The PDN may employ power gating to conserve power, in which the PDN selectively connects circuits in the SoC that are active to the power source and disconnects circuits in the SoC that are inactive from the power source. The PDN typically has a large inductance in the lead connecting the SoC to the off-chip power source (e.g., PMIC). The inductance induces a ripple on the power rail when the load on the power rail suddenly changes (e.g., due to power gating).

SUMMARY

[0003] The following presents a simplified summary of one or more embodiments in order to provide a basic understanding of such embodiments. This summary is not an extensive overview of all contemplated embodiments, and is intended to neither identify key or critical elements of all embodiments nor delineate the scope of any or all embodiments. Its sole purpose is to present some concepts of one or more embodiments in a simplified form as a prelude to the more detailed description that is presented later.

[0004] According to a first aspect, a power circuit is described herein. The power circuit comprises a capacitor coupled to a high-voltage rail, and a droop slope limiter (DSL) coupled between the high-voltage rail and a power rail. The DSL is configured to detect a downward voltage slope on the power rail, and to control current flow from the high-voltage rail to the power rail through the DSL based on the detected downward voltage slope.

[0ΘΘ5] A second aspect relates to a method for conditioning a power rail. The method, comprises detecting a downward voltage slope on the power rail, and controlling current flow from a high-voltage rail to the power rail based on the detected downward voltage slope, wherein a capacitor is coupled to the high-voltage rail. [0006] A third, aspect relates to an apparatus for conditioning a power rail. The apparatus comprises means for detecting a downward voltage slope on the power rail, and means for controlling current flow from a high-voltage rail to the power rail based on the detected downward voltage slope, wherein a capacitor is coupled to the high- voltage rail

[0007] To the accomplishment of the foregoing and related ends, the one or more embodiments comprise the features hereinafter fully described and. particularly pointed out in the claims. The following description and the annexed drawings set forth in detail certain illustrative aspects of the one or more embodiments. These aspects are indicative, however, of but a few of the various ways in which the principles of various embodiments may be employed and the described embodiments are intended to include all such aspects and. their equivalents.

BRIEF DESCRIPTION OF THE DRAWINGS

[0008] FIG. 1 shows an example of active charge for a decoupling capacitor placed directly on a rail and active charge for a decoupling capacitor placed behind a low- dropout regulator (LDO) coupled to the rail.

[0009] FIG. 2 is a block diagram of a power distribution network (PDN) conditioner according to an embodiment of the present disclosure.

[0010] FIG. 3 shows an example of active charge for a capacitor coupled to a high- voltage rail according to an embodiment of the present disclosure.

[0011] FIG. 4 shows an example of a ripple without a droop slope limiter (DSL) and a ripple with the DSL according to an embodiment of the present disclosure.

[0012] FIG, 5 shows an example of a ripple without an overshoot slope limiter (OSL) and a ripple with the OSL according to an embodiment of the present disclosure.

[0013] FIG. 6 shows an example in which the PDN conditioner sources inrush current to an on-chip circuit according to an embodiment of the present disclosure.

[0014] FIG. 7 shows an example of a PDN conditioner comprising a separate DSL and a separate OSL for each one of a plurality of power rails according to an embodiment of the present disclosure.

[0015] FIG, 8 shows an implementation of the DSL according to an embodiment of the present disclosure.

[0016] FIG. 9 shows an implementation of the OSL according to an embodiment of the present disclosure. [0017] FIG, 10 is a flow diagram of a method for conditioning a power rail according to an embodiment of the present disclosure.

DETAILED DESCRIPTION

[0018] The detailed description set forth below, in connection with the appended drawings, is intended as a description of various configurations and is not intended to represent the only configurations in which the concepts described herein may be practiced. The detailed description includes specific details for the purpose of providing a thorough understanding of the various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. In some instances, well-known structures and components are shown in block diagram form in order to avoid obscuring such concepts.

[0019] A power distribution network (PDN) is used to distribute power to various circuits in a system on a chip (SoC) from an off-chip power source (e.g., power management integrated circuit (PMIC)). The PDN may employ power gating to conserve power, in which the PDN selectively connects circuits in the SoC that are active to the power source and disconnects circuits in the SoC that are inactive from the power source. This prevents power leakage from circuits that are inactive (e.g.. idle), thereby conserving power,

[0020] The PDN typically has a large inductance in the lead connecting the SoC to the off-chip power source (e.g., PMIC). For example, the inductance may include board, inductance, package inductance, etc. The inductance induces a ripple on the power rail when the load on the power rail suddenly changes (e.g., due to power gating). As the voltage on the power rail decreases (e.g., below 0.9 V) and the load of the SoC increases (e.g., due to higher integration), the ripple effect becomes so severe that it greatly affects the operation of digital circuits coupled to the power rail. For example, the ripple effect may cause logic errors in the SoC and/or memory cells to flip states in the SoC. Accordingly, it is desirable to reduce the ripple effect.

[0021] One approach to reduce the ripple effect is to directly connect an on-chip decoupling capacitor to the power rail. However, fabricating a large on-chip capacitor is very expensive. Further, only a small portion of the capacitor's charge storage capacity can be used to reduce the ripple effect. This can be demonstrated with reference to FIG. 1. [0022] In the example in FIG. 1. the supply voltage on the power rail is approximately

1.0V, and the maximum voltage droop on the power rail that can be tolerated is approximately ΔΥ, which is much smaller than 1.0V (e.g., 0.1V or less). The total charge stored on the decoupling capacitor (denoted CAP in FIG. 1) is approximately equal to C-V, where C is the capacitance of the capacitor and V is the supply voltage (e.g., 1.0V in FIG. 1). However, only a small portion of this charge approximately equal to C-AV (referred, to as active charge) is available to reduce the ripple effect on the power rail. I other words, only a small portion of the capacitor's charge storage capacity (referred to as active capacity) is available to reduce the ripple effect. As shown in FIG, 1, the active charge (shaded region) makes up only a small portion of the total charge stored on the decoupling capacitor.

[0023] Another approach to reduce the ripple effect is to place the on-chip decoupling capacitor behind an on-chip low-dropout regulator (LDO) coupled to the power rail In this approach, the decoupling capacitor is connected to a higher voltage (e.g., 1.2V) and supplies charge to the power rail through the LDO to reduce the ripple effect. The active charge is larger in this approach compared with the previous approach (as shown in FIG. 1 ), but not by much due to energy loss in the LDO.

[0024] Embodiments of the present disclosure substantially increase the active capacity of a capacitor for reducing the ripple effect compared with the previous two approaches, as discussed further below.

[0025] FIG. 2 shows a PD conditioner 205 according to an embodiment of the present disclosure. The PDN conditioner 205 may be integrated on a chip 200 with a power rail 230 connected to an off-chip PMIC 250 for supplying power from the PMIC 250 to various circuits (not shown in FIG. 2) on the chip. The lead connecting the PMIC 250 to the power rail 230 may include inductance (e.g., board and/or package inductance) that produces a ripple on the power rail 230 when the load on the power rail changes (e.g., due to power gating),

[0026] The PDN conditioner 205 comprises an on-chip capacitor 215 coupled between a high-voltage rail 210 and ground. The high-voltage rail 210 has a higher voltage than the power rail 230 used to supply power to the various circuits (not shown in FIG, 2) on the chip from the PMIC 250. For example, the high-voltage rail 210 may have a nominal voltage of approximately 2.4V while the power rail 230 may have a supply voltage of approximately 1 ,0V. In general, the high-voltage rail 210 may have a voltage that is at least 50% higher than the supply voltage of the power rail 230. The PDM conditioner 205 also comprises a droop slope limiter (DSL) 220 coupled between the high-voltage rail 210 and the power rail 230, and an overshoot slope limiter (OSL) 240 coupled, between the power rail 230 and ground. The DSL 220 and OSL 240 are discussed, in greater detail below.

[0Θ27] The active charge of the on-chip capacitor 215 is much larger (e.g., 5~10x higher) compared with the previous two approaches discussed above. This is because the on-chip capacitor 215 is connected to the high-voltage rail 210, and the active charge is proportional to the voltage difference between the high- voltage rail 210 and the power rail 230, which is relatively large. This can be demonstrated with reference to FIG. 3, which shows an example of the active charge for the on-chip capacitor 215 compared with the previous two approaches shown in FIG. 1. As shown in FIG. 3, the active charge (shaded region) of the on-chip capacitor 215 is significantly larger than the active charge for the capacitors in the previous two approaches for a given capacitor size. Because the capacitor 215 has a much larger active charge, the capacitor 215 is able to supply significantly more charge to reduce ripple on the power rail 230.

[0028] Referring back to FIG. 2, the DSL 220 is configured to control the flow of current from the capacitor 215 to the power rail 230 based, on a downward (negative) voltage slope on the power rail 230. More particularly, the DSL 220 is configured to detect a downward voltage slope on the power rail 230, and to turn on/off based on the detected, downward voltage slope. When the DSL 220 is turned on, the DSL 220 allows current to flow from the capacitor 215 to the power rail 230 through the DSL 220, and when the DSL 220 is turned off, the DSL 220 blocks current flow from the capacitor 215 to the power rail 230.

[0029] In one embodiment, the DSL 220 is configured, to turn off if the magnitude

(absolute value) of a detected downward voltage slope is below a slope threshold or no downward voltage slope is detected. The DSL 220 is configured to turn on if the magnitude of a detected downward voltage slope exceeds (crosses) the slope threshold. As a result, the DSL 220 allows current to flow from the capacitor 215 to the power rail 230 when a ripple on the power rail 230 produces a downward voltage slope on the power rail 230 having a magnitude that exceeds the slope threshold. The current from the capacitor 215 limits the downward voltage slope, thereby reducing the voltage droop of the ripple. [0030] FIG, 4 shows a comparison of a ripple 405 without the DSL 220 and a ripple

410 with the DSL 220. As shown in FIG. 4, the ripple 405 without the DSL 220 has a relatively large voltage droop 415. The ripple 405 may be caused by an increase in the load coupled to power rail 230 (e.g., due to power gating). The ripple 405 dampens over time (e.g., due to resistance in the FDN), causing the amplitude of the ripple 405 to decrease over time. The ripple 410 with the DSL 220 has a much smaller voltage droop 420. This is because the DSL 220 turns on when the magnitude of a downward voltage slope crosses the slope threshold, allowing current to flow from the capacitor 215 to the power rail 230 through the DSL 220. The current from the capacitor 215 reduces the size of the voltage droop 420, thereby reducing the amplitude of the ripple 410.

[0031] When the DSL 220 turns on, the DSL, 220 may control (regulate) the amount of current flow from the capacitor 215 to the power rail 230 such that the magnitude of the downward voltage slope on the power rail 230 is approximately limited to the slope threshold. When the magnitude of the downward voltage slope on the power rail 230 falls below the slope threshold, the DSL 220 may turn itself off. As a result, the DSL 220 may only turn on for a relatively short time duration (e.g., less than 100 ns) to reduce a ripple on the power rail 230. An example of this is shown in FIG . 4.

[0032] In the example in FIG. 4, the DSL 220 turns on when the magnitude of the downward voltage slope on the power rail 230 crosses the slope threshold, reducing the size of the voltage droop 420 on the power rail 230. The DSL 220 turns off when the magnitude of the downward voltage slope falls below the slope threshold, which may- occur just before the bottom of the voltage droop 420 is reached, as shown in FIG. 4, During the time that the DSL 220 is turned on, the DSL 220 may control (regulate) the downward voltage slope to a value set by the slope threshold. By reducing the size of the voltage droop 420, the DSL 220 significantly reduces the amplitude of the ripple 410 compared to the ripple 405 without the DSL 220. Because the DSL 220 only turns on for a short time duration to reduce the ripple, the buck efficiency of the DSL 220 is not a problem, and therefore the DSL 220 can operate on the high-voltage rail 210.

[0033] The OSL 240 operates in a similar manner as the DSL 220 except that the OSL

240 controls the flow of current from the power rail 230 to ground based on an upward, (positive) voltage slope on the power rail 230. More particularly, the OSL 240 is configured to detect an upward voltage slope on the power rail 230, and to turn on/off based on the detected upward voltage slope. When the OSL 240 is turned on, the OSL 240 allows current to flow from the power rail 230 to ground through the OSL 240, and when the OSL 240 is turned off, the OSL 240 blocks current flow from the power rail 230 to ground.

[0034] In one embodiment, the OSL 240 is configured to turn off if a detected, upward slope is below a slope threshold or no upward voltage slope is detected. The OSL 240 is configured to turn on if a detected upward slope exceeds the slope threshold. As a result, the OSL 240 allows current to flow from the power rail 230 to ground when a ripple on the power rail 230 produces an upward, voltage slope on the pow r er rail 230 exceeding the slope threshold. The current flow to ground limits the upward voltage slope, thereby reducing the voltage overshoot of the ripple. The slope thresholds for the DSL 220 and the OSL 240 may be the same or different.

[0035] FIG. 5 shows a comparison of a ripple 505 without the OSL 240 and a ripple

510 with the OSL 240. As shown in FIG. 5, the ripple 505 without the OSL 240 has a relatively large voltage overshoot 515. The ripple 505 may be caused by a decrease in the load, coupled to power rail 230 (e.g., due to power gating). The ripple 505 dampens over time (e.g., due to resistance in the PD ), causing the amplitude of the ripple to decrease over time. The ripple 510 with the OSL 240 has a much smaller overshoot 520. This is because the OSL 240 turns on when an upward voltage slope on the power rail 230 crosses the slope threshold, allowing current to flow r from the power rail 230 to ground. The current flow to ground reduces the size of the overshoot 520, thereby reducing the amplitude of the ripple 510.

[0036] When the OSL 240 turns on, the OSL 240 may control (regulate) the amount of current flow from the pow r er rail 230 to ground such that the upward voltage slope on the power rail 230 is approximately limited to the slope threshold. When the upward voltage slope on the power rail 230 falls below the slope threshold, the OSL 240 may turn itself off. As a result, the OSL 240 may only turn on for a relatively short time (less than 100 ns) to reduce a ripple. An example of this is shown in FIG. 5.

[0037] In the example in FIG. 5, the OSL 240 turns on when the upward voltage slope on the power rail 230 crosses the slope threshold, reducing the size the overshoot 520. The OSL 240 turns off when the upward voltage slope falls below the slope threshold, which may occur just before the top of the o vershoot 520 is reached, as shown in FIG. 5. During the time that the OSL 240 is turned on, the OSL 240 may control (regulate) the upward, voltage slope to a value set by the slope threshold. By reducing the size of the overshoot 520, the OSL 240 significantly reduces the amplitude of the ripple 510 compared to the ripple 505 without the OSL 240.

[0038] FIG, 6 shows an example in which the chip 200 comprises a circuit 620 that receives power from the power rail 230 through a power-gating switch 610 (e.g., head switch). A power management circuit (not shown) may connect the circuit 620 to the power rail 230 when the circuit 620 is active by turning on the power-gating switch 610. The power management circuit may disconnect the circuit 620 from the power rail 230 when the circuit 620 is inactive (e.g., idle) by turning off the power-gating switch 610. Although one circuit 620 is shown in FIG. 6 for ease of illustration, it is to be appreciated that the chip 200 may comprise a plurality of circuits, in which each circuit may be selectively connected to the power rail 230 by a separate power-gating switch. This allows the circuits to be power gated independently.

[0Θ39] When the power-gating switch 610 disconnects the circuit 620 from the power rail 230, capacitors in the circuit 620 are discharged due to current leakage. When the power-gating switch 610 initially connects the circuit 620 to the power rail 230 to power up the circuit 620 to the active state, a large capacitive load, is suddenly placed on the power rail 230 due to the capacitors in the circuit 620. The capacitors in the circuit 620 drain current from the power rail 230 in order to charge up, causing the voltage on the power rail 230 to droop. The PMIC 250 may not be able to supply current fast enough to prevent the droop due the large inductance (e.g., board and/or package inductance) in the lead connecting the PMIC 250 to the power rail 230. Thus, without the DSL 220, a large voltage droop may appear on the power rail 230 when the circuit 620 is initially connected to the power rail 230 after being in the inactive state.

[00401 With the DSL 220, when the voltage on the power rail 230 starts drooping, a downward, voltage slope appears on the power rail 230, which is detected by the DSL 220. When the magnitude of the downward voltage slope crosses the slope threshold, of the DSL 220, the DSL 220 turns on allowing the on-chip capacitor 215 to source current to the circuit 620 through the DSL 220, which reduces the voltage droop, as discussed above. The on-chip capacitor 215 is able to source current to the circuit 620 much faster than the PMIC 250 to reduce the droop. This is because the path between the on-chip capacitor 215 and the circuit 620 through the DSL 220 has a much smaller inductance than the lead, connecting the PMIC 250 to the power rail 230. Both the on-chip capacitor 215 and the circuit 620 are located on the chip 200. As a result, the path between them is much shorter (and therefore has less inductance) than the lead connecting the power rail 230 to the PMIC 250, which is located off-chip. Further, sourcing current from the on-chip capacitor 215 to the circuit 620 reduces the current change in the inductance in the lead connecting the power rail 230 to the PMIC 250, thereby reducing noise induced by the inductance.

[0041] Although the example in FIG. 2 shows one power rail 230, it is to be appreciated that, since the on-chip capacitor 215 has a large active charge, the on-chip capacitor 215 may be shared by multiple power rails in the SoC. In this regard, FIG. 7 shows an example of a FDN conditioner 705 comprising a plurality of power rails 230-1 to 230-3. The power rails 230- 1 to 230-3 are coupled to the PMIC 250, and may receive the same supply voltage from the PMIC 250 or different supply voltages from the PMIC 250. In either case, the supply voltage for each power rail 230- 1 to 230-3 is lower than the voltage of the high- voltage rail 210.

[0042] The PDN conditioner 705 also comprises a separate DSL 220-1 to 220-3 for each power rail 230-1 to 230-3, and a separate OSL 240-1 to 240-3 for each power rail 230-1 to 230-3. Each DSL 220- 1 to 220-3 is coupled between the high-voltage rail 210 and the respective power rail 230-1 to 230-3, and each OSL 240-1 to 240-3 is coupled between the respective power rail 230-1 to 230-3 and ground. Each DSL 220- 1 to 220- 3 is configured, to reduce a voltage droop of a ripple (and hence the amplitude of the ripple) on the respective power rail 230-1 to 230-3 by limiting the downward voltage slope on the respective power rail 230-1 to 230-3, as discussed above. Similarly, each OSL 240-1 to 240-3 is configured to reduce a voltage overshoot of a ripple (and hence the amplitude of the ripple) on the respective power rail 230-1 to 230-3 by limiting the upward voltage slope on the respective power rail 230-1 to 230-3, as discussed above.

[0043] FIG. 8 shows an exemplary implementation of the DSL 220 according to an embodiment of the present disclosure. For ease of illustration, the OSL 240 is not shown in FIG. 8. In this embodiment, the DSL 220 comprises a power field effect transistor (FET) 810, a control circuit 820, and a slope-detection circuit 830. The power FET 810 may be a P-type FET (PFET) or an N-type FET ( FET), The source and drain of the power FET 810 are coupled between the high-voltage rail 210 and the power rail 230. For example, if the power FET 810 is a PEFT, then the source of the power FET 810 is coupled to the high-voltage rail 210 and the drain of the power FET 810 is coupled to the power rail 230. The gate of power FET 810 is coupled to the control circuit 820. As discussed further below, the control circuit 820 controls the conducti vity between the source and drain of the power FET 810 (and hence the current flow from the high-voltage rail 210 to the power rail 230) by controlling the gate voltage of the power FET 810.

[0044J The slope-detection circuit 830 is coupled to the power rail 230, and configured, to detect a downward (negative) voltage slope on the power rail 230. For example, the slope-detection circuit 830 may detect a downward voltage slope by detecting a time rate of change of voltage (i.e., AdV/AdT) on the power rail 230 that is negative. The slope-detection circuit 830 may output a voltage (denoted Vjslope) that is proportional to the magnitude of the detected downward voltage slope to the control circuit 820,

[0045] The control circuit 820 compares the voltage (V slope) from the slope -detection circuit 830 with a slope-threshold voltage (denoted V threshold) that is set according to a desired downward voltage slope. If V_slope is below V jhreshold, then the control circuit 820 turns off the power FET 810. For the example in which the power FET 810 is a PFET. the control circuit 820 may turn off the power FET 810 by outputting a gate voltage that is approximately equal to the voltage of the high-voltage rail 210.

[0046] If V jslope exceeds V lireshold, then the control circuit 820 may turn on the power FET 810 and adjust the gate voltage of the power FET 810 in a direction that reduces the voltage difference between V slope and V threshold. For example, the control circuit 820 may reduce the voltage difference by adjusting the gate voltage of the power FET 810 in a direction that increases the conductivity of the power FET 810. The increased conductivity allows more current to flow from the on-chip capacitor 215 to the power rail 230 through the power FET 810, which reduces the downward voltage slope on the power rail 230, and. therefore reduces the voltage difference between

V slope and. V threshold.. Thus, the control circuit 820 employs negative feedback to limit the downward voltage slope on the power rail 230 to a value set by V threshold. In FIG. 8, the negative feedback is indicated by negative feedback loop 835. When

V slope falls below V threshold, the control circuit 820 may turn the power FET 810 back off.

[0047] FIG. 9 shows an exemplary implementation of the OSL 240 according to an embodiment of the present disclosure. For ease of illustration, the DSL 220 is not shown in FIG. 9. In this embodiment, the OSL 240 comprises a power FET 910, a control circuit 920, and a slope-detection circuit 930. The power FET 910 may be a P- type FET (PFET) or an N-type FET (NFET). The drain and source of the power FET 910 are coupled between the power rail 230 and ground. For example, if the power FET 910 is an EFT, then the drain of the power FET 910 is coupled to the power rail 230 and the source of the power FET 9 0 is coupled to ground. The gate of the power FET 910 is coupled to the control circuit 920.

[0Θ48] The slope-detection circuit 930 is coupled to the power rail 230, and configured to detect an upward (positive) voltage slope on the power rail 230. For example, the slope-detection circuit 930 may detect an upward voltage slope by detecting a time rate of change of voltage (i.e.. AdVVAdT) on the power rail 230 that is positive. The slope- detection circuit 930 may output a voltage (denoted. V_slope) that is proportional to the detected upward voltage slope to the control circuit 920.

[0049] The control circuit 920 compares the voltage (V slope) from the slope-detection circuit 930 with a slope- threshold voltage (denoted Vjhreshold.) that is set according to a desired upward voltage slope. If V slope is below V threshold, then the control circuit 920 turns off the power FET 910. For the example in which the power FET 910 is an NFET, the control circuit 920 may turn off the power FET 910 by grounding the gate of the power FET 910.

[0050] If V slope exceeds V threshold, then the control circuit 920 may turn on the power FET 910 and adjust the gate voltage of the power FET 910 in a direction that reduces the voltage difference between V_slope and Vjhreshold. For example, the control circuit 920 may reduce the voltage difference by adjusting the gate voltage of the power FET 910 in a direction that increases the conductivit of the power FET 910. The increased conductivity allows more current to flow from the power rail 230 to ground through the power FET 910, which reduces the upward voltage slope on the power rail 230, and. therefore reduces the voltage difference between V slope and V threshold. Thus, the control circuit 920 employs negative feedback to limit the upward voltage slope on the power rail 230 to a value set by Vjhreshold. In FIG. 9, the negative feedback is indicated by negative feedback loop 935. When V slope falls below V threshold, the control circuit 920 may turn the power FET 910 back off.

[0051] FIG. 10 shows a method 1000 for conditioning a power rail 230. For example, the method may condition the power rail 230 by reducing voltage droops and/or voltage overshoots on the power rail 230. [0052] In step 1010, a downward voltage slope is detected on the power rail. For example, the downward (negative) voltage slope may be detected by a slope-detection circuit (e.g., slope-detection circuit 830) coupled to the power rail (e.g., power rail 230).

[0053] In step 1020, current flow is controlled from a high-voltage rail to the power rail based on the detected downward voltage slope, wherein a capacitor is coupled to the high-voltage rail. For example, the current flow may be controlled by adjusting a gate voltage of a power FET (e.g., power FET 810) based on the detected downward voltage slope.

[0054] Those skilled in the art will appreciate that the circuits described herein may be realized using a variety of transistor types, and are therefore not limited to field effect transistors. For example, transistor types such as bipolar junction transistors, junction field effect transistor or any other transistor type may be used in place of the power FE T 810. For the example of a bipolar junction transistor, the collector and emitter of the bipolar transistor may be coupled between the high-voltage rail 210 and the power rail 230. In this example, the control circuit 820 may control the conductivity of the bipolar junction transistor (and hence the current flow from the on-chip capacitor 215 to the power rail 210) by controlling the base current of the bipolar junction transistor. Those skilled in the art will also appreciate that the circuits described herein may be fabricated with various IC process technologies such as CMOS, bipolar junction transistor (BJT), bipolar-CMOS (BiCMOS), silicon germanium (SiGe), gallium arsenide (GaAs), etc.

[0055] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be readily apparent to those skilied in the art, and the generic principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples described, herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.