TW201035738A - Adaptive power control apparatus with delay estimation scheme - Google Patents

Adaptive power control apparatus with delay estimation scheme Download PDF

Info

Publication number
TW201035738A
TW201035738A TW98108462A TW98108462A TW201035738A TW 201035738 A TW201035738 A TW 201035738A TW 98108462 A TW98108462 A TW 98108462A TW 98108462 A TW98108462 A TW 98108462A TW 201035738 A TW201035738 A TW 201035738A
Authority
TW
Taiwan
Prior art keywords
voltage
vdd
circuit
power
virtual
Prior art date
Application number
TW98108462A
Other languages
Chinese (zh)
Other versions
TWI387871B (en
Inventor
Wei-Chih Hsieh
Wei Hwang
Original Assignee
Univ Nat Chiao Tung
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Univ Nat Chiao Tung filed Critical Univ Nat Chiao Tung
Priority to TW98108462A priority Critical patent/TWI387871B/en
Publication of TW201035738A publication Critical patent/TW201035738A/en
Application granted granted Critical
Publication of TWI387871B publication Critical patent/TWI387871B/en

Links

Abstract

An adaptive power control apparatus with delay estimation scheme comprises a multi-mode power gating network, a voltage sensor, a variable threshold comparator, a slack detection and a Bi-directional shift register. The multi-mode power gating network is serial connected between a voltage source and a circuit module to form a virtual VDD with the multi-mode power gating network therebetween. The multi-mode power gating network comprises multiple transistors that are selectively turned on to adjust the power into the circuit module, where the transistors are controlled by Bi-directional shift register to be turned on and off according to a detecting result of the slack detection.

Description

201035738 六、發明說明: 【發明所屬之技術領域】 本發明為一種電路延遲評估及適應性功率控制裝置 尤其是關於-種具環境適應性之動態功率㈣的裝置: 【先前技術】 近年來,由於半導體製程技術的突飛猛進,電晶體密 度跟系統複雜度快速成長,使控制並降低功率消耗之設計 ◎已為不容忽視之課題。目前,已有一些降低功率消耗的設 計’茲舉例說明如下: 美國專利第7276932號揭露一種利用虛擬功率閘胞元 (VPC,virtual power gating ce丨丨)之架構’其中該虛擬功率 閘胞元係由用來緩衝控制信鐃之控制電路及包含二個或更 多個NFETs與PFETs之功率閘區塊(PGB,p〇we「gating block)所構成。然而,該功率閘胞元只當作單純的開關,亦 即,只在所搭載之電路與電源之間進行開與關(連接或是斷 ◎ 開)兩種狀態’其除了在關閉的狀態下可節省靜態功率外, 並無動態功率控制之能力。 又例如,M. Nakai(M. Nakai, S. Akui > K > Seno ,Τ· Meguro,Τ· Seki,T. Kondo,A· Hashiguchi, H. Kawahara, K. Kumano,and M. Shimura * Dynamic201035738 VI. Description of the Invention: [Technical Field] The present invention relates to a circuit delay evaluation and adaptive power control device, and more particularly to a device for environmentally adaptive dynamic power (4): [Prior Art] In recent years, The rapid advancement of semiconductor process technology, the rapid growth of transistor density and system complexity, and the design of control and reduction of power consumption have become a topic that cannot be ignored. At present, there are some designs for reducing power consumption, which are exemplified as follows: U.S. Patent No. 7,276,932 discloses a virtual power gating ce (VPC) architecture in which the virtual power gate cell system It consists of a control circuit for buffering the control signal and a power gate block (PGB, p〇we "gating block" containing two or more NFETs and PFETs. However, the power gate cell is only used as a simple The switch, that is, only between the circuit and the power supply that is carried on and off (connected or disconnected), which can save static power in the closed state, and has no dynamic power control. For example, M. Nakai (M. Nakai, S. Akui > K > Seno, Τ· Meguro, Τ· Seki, T. Kondo, A· Hashiguchi, H. Kawahara, K. Kumano, and M Shimura * Dynamic

Voltage and Frequency Management for a Low-Power Embedded Microprocessor,IEEE Journal of Solid-State Circuits, vol· 40, no. 1, pp. 28-35, Jan. 2005)等 人所提出之一種利用動態電壓調整(DVS)與頻率調整之技術 201035738 ,作為有效降低功率消耗之方法。此技術係利用一種結合 閘極延遲、t阻電容交互連接延遲及上升/下&延遲之延遲 合成器,以達到較佳的關鍵路徑(critical path)之模擬 (ni u丨a t i ο η)。惟,此習知技術原理上也是採用延遲匹配電 路’因此同樣有前述缺失。此外’ M· Nakai在實現適應性 或者動態電壓調變時,需要一最差情況關鍵路徑延遲匹配 電路(wo「st case c「mca| path de|ay matching circujt),然 n而,取至情況實際上卻鮮少發生,因而大為低估降低功率 的可旎性,因此對於現今之高速電路來說並不適用。 【發明内容】 配合既有技術之功率閘等設置僅能單純作為開關之用 ’只在少數的狀態下能夠真正節省㈣,導致未能有效降 低功率消耗及達成適應性功率控制之要求 新的判斷電路運算狀態的機制,#由監控功率閑之 〇 ®源⑽tual_VDD,VDDV)的電壓而判定電路運算之 時機,而可以獲得電路之延遲資m,在透過所提之 電路運算結束時機的判斷方式’應用於功率控制,使戶" :之電路具有主動且具有環境變異適應效能的功率控制, 解決既有技術之技術問題。 配合前述之技術問胃’本發明提供—種具電路延遲評 估機制之適應性功率控制裝置,i . '、匕έ .一多模式功率閙 :路、-電壓感測ϋ、一可變門襤比較器、一冗餘 測電路以及一雙向位移暫存器, ' ° θ ' >如 °哀夕換式功率閘網路包合 设數個並聯之電晶冑,當該多模式功”網路連接於一; 5 201035738 源與'電路描 έΗ $ pq n't ^之^’各電晶體為p型電晶體且該多模 工n網路與該電路模組之連接節點視為—虛擬電壓源 ’:该多模式功率閉網路連接於一接地點與該電路模組之 :::各電晶體為N型電晶體且該多模式功率閘網路與該 “杈組之間的節點才見為一虛擬接地點,其中: 接祕I “ ^感心之輸人端連接於該虛擬電壓源或該虛擬 Ο Ο 八尋找4虛擬電壓源之—最低㈣(VL)或尋找該 雙擬接地點之一最高電壓(VH); ,^二變門捏比較器與該電壓感測器之輸出端電性連接 二:較該虛擬電壓源之一瞬間電壓(vv)與該最低電壓 (V L)疋否滿足一第— 、關係而輸出一第一比較結果狀態 二該可變門檻比較器比較該虛擬接地點之瞬間電壓 (:)㈣最高電壓⑽)是否滿足-第二特定關係而輸出一 弟二比較結果狀態; * 6亥几餘區間谓測電路$ #人办山、击μ 哭电路之輸入知連接於該可變門檻比較 :二據該可變門檻比較器之該第—或第二比較結果狀態 時脈邊緣則貞測-時脈週期内是否存在—冗餘區間; 該雙向位移暫存器之# λ Α山& # 輸入蝠與该冗餘區間偵測電路電 :妾,其依據該冗餘區間之是否存在而產生位移,使亨 又向位移暫存器之複數個控制訊號的輸出狀態改變;以及 號連: = =:?Γ雙向位移暫存器之各控制訊 訊號之二::=:::::W制 £間之存在狀況而改變該電源輸出至該電路模組之功率。 其中,該第一特定闕係滿足下列公式: 201035738 f VDD - F/;7 γ VDD-VL 、vl-m ) ~\Tm7 -VDD-Vth + -VL {VDD - VV) --?->Voltage and Frequency Management for a Low-Power Embedded Microprocessor, IEEE Journal of Solid-State Circuits, vol. 40, no. 1, pp. 28-35, Jan. 2005) et al. ) Technology with frequency adjustment 201035738, as a method to effectively reduce power consumption. This technique utilizes a delay synthesizer that combines gate delay, t-resistance interconnect delay, and rise/down & delay to achieve a better critical path simulation (ni u丨a t i ο η). However, this prior art technique also employs a delay matching circuit in principle, thus also having the aforementioned drawbacks. In addition, 'M·Nakai needs a worst case critical path delay matching circuit when implementing adaptive or dynamic voltage modulation (wo "st case c"mca| path de|ay matching circujt), then n, take the situation In fact, it rarely occurs, so the underestimation of power reduction is greatly underestimated, so it is not suitable for today's high-speed circuits. [Disclosed] The power brakes and other settings of the prior art can only be used as switches. 'There can be real savings in only a few states (4), resulting in a failure to effectively reduce power consumption and achieve adaptive power control requirements. A new mechanism for judging the operational state of the circuit, #by monitoring the power of the 〇® source (10)tual_VDD, VDDV) The voltage is used to determine the timing of the circuit operation, and the delay of the circuit can be obtained. The method of judging the timing of the end of the operation through the proposed circuit is applied to the power control, and the circuit of the user is active and has environmental variation adaptability. The power control solves the technical problems of the prior art. In conjunction with the aforementioned technology, the stomach is provided by the present invention. Adaptive power control device, i. ', 匕έ. A multi-mode power 閙: circuit, - voltage sensing ϋ, a variable threshold comparator, a redundant measuring circuit and a bidirectional displacement register, ' ° θ ' > such as ° 夕 换 换 换 换 换 换 换 换 换 换 换 换 换 换 换 换 换 换 换 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交 交Each transistor of the 't ^ ^ ^ is a p-type transistor and the connection node of the multi-mode n network and the circuit module is regarded as a - virtual voltage source ': the multi-mode power closed network is connected to a ground point And the circuit module::: each transistor is an N-type transistor and the multi-mode power gate network and the node between the "group" are seen as a virtual ground point, wherein: the secret I "^ sense The input end is connected to the virtual voltage source or the virtual Ο 八 to find 4 virtual voltage sources - the lowest (four) (VL) or find the highest voltage (VH) of the double pseudo grounding point; And the output end of the voltage sensor is electrically connected to the second: whether the instantaneous voltage (vv) and the minimum voltage (VL) of the virtual voltage source are full a first-to-relationship outputting a first comparison result state 2, the variable threshold comparator comparing the instantaneous voltage of the virtual ground point (:) (four) the highest voltage (10)) satisfies the second specific relationship and outputs a second comparison result State; * 6 hai interval interval pre-measure circuit $ #人办山, 击μ crying circuit input knowledge is connected to the variable threshold comparison: according to the variable threshold comparator of the first or second comparison result state The edge of the clock is speculative - whether there is a redundant interval in the clock cycle; the #λ Α山&# input bat and the redundant interval detection circuit of the bidirectional displacement register: 妾, according to the redundancy Whether the interval exists and the displacement is generated, so that the output state of the plurality of control signals of the shift register to the shift register is changed; and the number is connected: ==:?Γ The control signal of the two-way shift register is two::= The :::::W system changes the power output to the circuit module. Wherein, the first specific lanthanum satisfies the following formula: 201035738 f VDD - F/;7 γ VDD-VL , vl-m ) ~\Tm7 -VDD-Vth + -VL {VDD - VV) --?->

• VDD - Vth + -VV• VDD - Vth + -VV

VDOVDO

VDD — Vlh —— VDQVDD — Vlh —— VDQ

VDSO m~2VDS〇 其中, VDD為該電源提供之電壓;VDSO m~2VDS〇 where VDD is the voltage supplied by the power supply;

Vth為該多模式功率閘網路及該電路模組内之p型電 晶體之一臨界電壓(threshold voltage); VD0為P型電晶體於VGS=VD[)時之一汲極飽和電 壓(drain saturation voltage); <2 為一速度飽和參數(ve|0Cjty saturation index); λ為p型電晶體通道長度調變參數(channe丨丨ength modulation parameter);及 VDS0是一預設臨界電壓值。 其中,該可變門檻比較器為一變型之史密特觸發電路 其中,δ亥電壓感測器包含以二極體連接方式串接之二 電晶體的感測電路。 其中’該VDS0為100mV時,該第一特定關係則滿 足下列二者公式之一: FF >0.5685+ 0.4568-PX ; ^ 心 0.0902 +1.7137 · KZ - 〇.81〇n 其中’該第二特定關係滿足下列公式:Vth is the threshold voltage of the multi-mode power gate network and the p-type transistor in the circuit module; VD0 is one of the gate-saturation voltage of the P-type transistor when VGS=VD[) Saturation voltage); <2 is a speed saturation parameter (ve|0Cjty saturation index); λ is a p-type transistor channel modulation parameter (channe丨丨ength modulation parameter); and VDS0 is a preset threshold voltage value. Wherein, the variable threshold comparator is a modified Schmitt trigger circuit, wherein the delta voltage sensor comprises a sensing circuit of a diode connected in series by a diode connection. Where 'VDS0 is 100mV, the first specific relationship satisfies one of the following two formulas: FF >0.5685+ 0.4568-PX ; ^ heart 0.0902 +1.7137 · KZ - 〇.81〇n where 'the second specific The relationship satisfies the following formula:

(VDD-Vth 〕 a VH VDD~Vth--VH --— 2 VG{vDD~Vth~~VG 、 l ? \ VDL) — V Η — Vi hJ 1 + A{VDD - VH) ( i λ VD〇\ VDD-Vlh--VD0 \ 2 J ----、 -/ ( ,\ VDSQ VDD-VG-Vth-^VDSQ L 2 J 其中, 201035738 VDD為該電源提供之電壓;(VDD-Vth) a VH VDD~Vth--VH --— 2 VG{vDD~Vth~~VG , l ? \ VDL) — V Η — Vi hJ 1 + A{VDD - VH) ( i λ VD〇 \ VDD-Vlh--VD0 \ 2 J ----, -/ ( , \ VDSQ VDD-VG-Vth-^VDSQ L 2 J where 201035738 VDD is the voltage supplied by the power supply;

Vth為該多模式功率閘網路及該電路模組内之n型電 晶體之一臨界電壓(threshold voltage); VD0為N型電晶體於VGS=VDD時之一汲極飽和電 壓(drain saturation voltage); ο:為一速度飽和參數(Ve|0Cjty saturatjori jnc|ex); 入為N型電日a體通道長度調變參數(channe丨丨ep)gth modulation parameter);及Vth is a threshold voltage of the multi-mode power gate network and the n-type transistor in the circuit module; VD0 is a drain saturation voltage of the N-type transistor at VGS=VDD (drain saturation voltage) ο: is a speed saturation parameter (Ve|0Cjty saturatjori jnc|ex); into the N-type electric day a body channel length modulation parameter (channe丨丨ep) gth modulation parameter);

V D S 0是一預設臨界電壓值。 其中’該VDS0 4 100mV時,該第二特定關係則滿 足下列二者公式之一: KG<-0.01729 +0.4424 ·Κ// ; ^ ⑽045 + 0•⑻0._咖2,電麼單位為伏特⑺。 藉此,本發明具有如下優點: 1.動態監控該虛擬電壓源’藉以判斷該電路模組之延 "及冗餘區間之存在結果,並據此控制輸出至該電路模組 力率達到降低功率消耗的技術效果。 2 ·本發明所提的 a ^ 且古^ 一 万法,與锿楗參數關連性低,因此, 八有彳艮咼的環境適應性。 3 ·本發明透過判斷y μ &、q 掉所有可料時脈長度可以讓電路模組盡量用 同的電路模組時,在:種以,可使本發明運用於各種不 耗達到最佳的能源效益。 刀牛肩 8 201035738 【實施方式】 首先,為了 S兒明本發明所提出之具電路延遲評估機制 之適應性功率控制裝置之一電路延遲評估機制(delay estimation scheme) ’請參考第一圖,其顯示一個16位元 乘法Is在工作中由電源端所汲取之一汲取電流(dra丨ned current)。該乘法器之狀態可分為一穩定態(stab|e)以及— 切換態(switching)。在穩定態中,該汲取電流僅包含微量 的漏電流,但在切換態中,該乘法器由電源端汲取大量的 〇忒汲取電流以執行乘法工作。由上述可知,透過判斷該汲 取電流的大小變化,可判斷該乘法器所處的工作狀態,而 當乘法器之工作狀態得以區分之後,即可獲取該乘法器於 工作時的一延遲時間評估(delay estimati〇n)。 為了更進一步將前述的範例應用於其他電路及便於解 說前述之延遲時間評估方法實際應用於其他電路模組,請 參考第二圖及第三圖,以一個最為簡單的CM〇s反向器 (inverter)作為說明範例:V D S 0 is a predetermined threshold voltage value. Where 'VDS0 4 100mV, the second specific relationship satisfies one of the following two formulas: KG<-0.01729 +0.4424 ·Κ// ; ^ (10)045 + 0•(8)0._Caf 2, the unit is volt (7) . Therefore, the present invention has the following advantages: 1. Dynamically monitoring the virtual voltage source 'by determining the delay of the circuit module" and the existence of the redundancy interval, and thereby controlling the output to the circuit module to reduce the power rate The technical effect of power consumption. 2 · The a ^ and the ancient ^ 10,000 method proposed by the present invention have low correlation with the 锿楗 parameter, and therefore, the environmental adaptability of the 彳艮咼 彳艮咼. 3 · The present invention can determine that y μ &, q can reduce the length of all available clocks, so that the circuit module can use the same circuit module as much as possible, so that the present invention can be applied to various types without optimal consumption. Energy efficiency. Knife shoulder 8 201035738 [Embodiment] First, a delay estimation scheme for an adaptive power control device with a circuit delay evaluation mechanism proposed by the present invention is referred to the first figure, Shows a 16-bit multiplication Is that draws current (dra丨ned current) from the power supply at work. The state of the multiplier can be divided into a steady state (stab|e) and a switching state. In the steady state, the current draw contains only a small amount of leakage current, but in the switching state, the multiplier draws a large amount of current drawn from the power supply terminal to perform multiplication. It can be seen from the above that by determining the change in the magnitude of the current drawn, the working state of the multiplier can be determined, and when the working state of the multiplier is distinguished, a delay time evaluation of the multiplier during operation can be obtained ( Delay estimati〇n). In order to further apply the foregoing examples to other circuits and to facilitate the explanation that the aforementioned delay time evaluation method is actually applied to other circuit modules, please refer to the second and third figures, with one of the simplest CM〇s inverters ( Inverter) as an example:

Cl 如第二圖所示,該反向器連接至一理想的電源(vdd), 當該反向器之一輸入端(input node,丨N)由高準位切換至低 準位時,該反向器之P型金氧半電晶體(PM〇s,ρι)導通 而由該電源(VDD)擷取電流對該反向器之一輸出端(〇utput node,〇UT)充電,而使該輸出端切換至高準位,然而,習 用在輪出端之高、低準位之間是否完成切換的判斷,通常 採以一固定型判斷準則(static criteri〇n determinati〇n),即 設定該輸出端之電壓必須達到電源準位(即電源vdd所提 供之電壓準位)或為達到電源準位之9〇%等固定之電源準位 201035738 而所謂的「延遲時間」乃定Μ乱♦ * 義為電路模組或數位電路 中’自輸入端的切換開始,到給+ 士山丄 刮翰出端切換完成的時間。因Cl, as shown in the second figure, the inverter is connected to an ideal power supply (vdd). When one of the inverters (input node, 丨N) is switched from the high level to the low level, The P-type MOS transistor (PM〇s, ρι) of the inverter is turned on, and the current is drawn by the power source (VDD) to charge one output terminal (〇UT) of the inverter, so that The output is switched to a high level. However, the judgment of whether to switch between the high and low levels of the round-trip end is usually adopted by a fixed-type criterion (static criteri〇n determinati〇n), that is, setting the The voltage at the output must reach the power level (that is, the voltage level provided by the power supply vdd) or a fixed power supply level such as 9〇% of the power supply level. The so-called "delay time" is fixed. For the circuit module or the digital circuit, the switch from the input end starts to the time when the switch to the + 士山丄 is finished. because

此前例的反向器中’其延遲時間A 咬了间為自輸入端切換開始至輸 出端達到90% VDD之時間。依此中產 伋此疋義可知,此延遲時間受 到前述切換的判斷準則所影響。 而若該反向器與電源(VDD)之由u ;之間串接一個功率閘元件In the inverter of the previous example, the delay time A bites between the start of the input switching and the time when the output reaches 90% VDD. According to this definition, this delay time is affected by the above criteria for switching. And if the inverter is connected to the power supply (VDD) by a power gate component

(power gating device,PG),以作盔 a P A(power gating device, PG) for helmet a P A

)M作為後績電路元件與該電 源之間的切換開關’該功率閘元侏盔办+ & 干π兀件為與電路模組中採用之 Ρ型電晶體相同之元件。該功率閑元件(PG)可依據設計需 求採欲N型或P型電晶Μ ’若採用N型電晶體作為該功率 心WPCBm ’ _ Μ $電晶體之該功率閘元件(PG)連接 於§亥電路模組(此為反向器)與一接地點(G n d )之間。 如第三圖所示’假設該功率閘元件(PG)與該反向器之 P型金氧半電晶體(P1)之申接節點為一虛擬電壓源(Vir__ VDD’ VDDV)’該虛擬電壓源(VDDV)在該反向器之高低準 位切換過程之中,呈現先降後升(fa丨丨_then_「ise)之趨勢,使 該功率問兀件(PG)提供給反向器之電流也隨之改變,即該 ;:及取電流(drained current)在該功率閘元件(PG)存在的狀況 下,比該反向器直接連接至理想電源之電流相對較小,主 要是因為該功率閘元件(PG)增加了阻抗所導致。因此,輸 出端(OUT)之切換速率(switching slope)則比直接連接至理 想電源之反向益相對較慢’換言之,該輸出端(〇U T)的電壓 提高到理想電源(VDD)之90%而成為高準位之所需延遲時 間(delay time),最糟將可能是沒有連接功率閘元件(pG)時 】0 201035738 特定二::因此,單純用判斷輸出端(ουτ)電壓是否達到 22ΓΓΓ90%觸)對於目前高速切換需求之電 成,多餘的証、网* π為之狀恶其貫早已經切換完 '、,I战時間只是浪費在將# ψ M m I I τ、 及該虛擬電虔源(vdd^m )之電廢準位 ,、(DDV)之電壓提升到電源(VDD)所描彳 電壓準位。因此,Α A a 所如供之 本發明提出一個雷踗丁於士 十只】之問通, Ο Ο 述第電路工作中之延遲時間評估機制。沿用前 L第一圖為例,若假設M is used as a switch between the circuit components of the after-performance circuit and the power source. The power gate element is the same component as the 电-type transistor used in the circuit module. The power idle component (PG) can be selected according to the design requirements of the N-type or P-type electro-crystal Μ 'If the N-type transistor is used as the power core WPCBm ' _ 电 $ the transistor of the power gate element (PG) is connected to § The circuit module (this is the inverter) and a ground point (G nd ). As shown in the third figure, 'assuming that the power gate element (PG) and the P-type MOS transistor (P1) of the inverter are connected to a virtual voltage source (Vir__ VDD' VDDV) 'the virtual voltage The source (VDDV) exhibits a tendency to drop first and then rise (fa丨丨_then_“ise) during the switching of the high and low levels of the inverter, so that the power request element (PG) is supplied to the inverter. The current also changes, that is, and the drained current is relatively small in the presence of the power gate element (PG) than the inverter is directly connected to the ideal power source, mainly because The power gate element (PG) increases the impedance. Therefore, the switching slope of the output (OUT) is relatively slower than the reverse benefit of directly connecting to the ideal power supply. In other words, the output (〇UT) The voltage is increased to 90% of the ideal power supply (VDD) to become the high-latency delay time, and the worst would be when the power gate element (pG) is not connected.] 0 201035738 Specific two:: Therefore, Simply judge whether the output (ουτ) voltage reaches 22ΓΓΓ90%) In the current high-speed switching demand, the excess certificate, the network * π is the same as it has already been switched over, and the I war time is only wasted in the # ψ M m II τ, and the virtual power source ( Vdd^m) The electrical waste level, (DDV) voltage is raised to the voltage level indicated by the power supply (VDD). Therefore, ΑA a as provided by the present invention proposes a Thunder The question is , Ο The delay time evaluation mechanism in the circuit work. The first picture of the former L is taken as an example, if

门态凡成狀怨切換之判斷為一 B ’、中B點之選擇係藉由兮p剂人λ* 代、“ r 糟由°亥P型金氧半電晶體(P1 )之汲 極源極壓降(VDs)等於一値預执 < p 個預汉的臨界電壓(VDS0)。此預 叹臨界電壓可以經驗法則 、 j域相對小的值,如100mV,此 汲極源極壓降係由該p型全 ,玉氧+電日日體(ρυ在切換完成後 用來持續將輸出端(〇UT)之電壓進The judgment of the door state is a B', and the choice of the B point is by the 剂p agent λ* generation, "r 糟 由 ° ° P P P P P P P P P P P P P P The extreme voltage drop (VDs) is equal to the threshold voltage (VDS0) of the pre-existing <p pre-existing threshold. This pre-sighing threshold voltage can be a rule of thumb, a relatively small value of the j-domain, such as 100mV, this drain-source voltage drop It is used by the p-type full, jade oxygen + electric Japanese body (ρυ is used to continuously input the voltage of the output terminal (〇UT) after the switching is completed.

电&旱位铨升到電源(VD :供之電£準位。在B點時,該功率閘元件㈣)之電流 (d_CU_t)等…型金氧半電晶體"之 _ current) 〇 然而如前所述,該反向器電塵準位超過B點之後仍會 持續汲取電流以使該虛擬電壓源(VDDv)之電壓到達最高之 電£準位(VDD)為止。而隨著該虛擬電壓源(vddv)之電壓 上升’該功率間元件(PG)之汲極源極壓降(VDs)隨著該虛擬 電壓源(VDDV)之電壓上升而下降 ,.. 幵而下降,使更小的電流通過該功 率間元件(PG)°換言之’因為該功率閉元件(PG)在B點之 電流_叫等於該p型金氧半電晶體(ρι)之電流 (drain current) ’ 通過 B 點之接,,* 'a# , + 、。一之後,流過該功率閘元件(pG)之 電流持續減少。因此’當觀測到該反向器< p型金氧半電 201035738 晶體(P1)在B點的電流大於(或等於)該功率閉元件(pG)通 過B點之後遞減之電流(drain cu_t)之時,可稱該反向琴 已經完成狀態切換工作(其中該p型金氧半電晶體(p”、該 功率開元件(PG)均在線性區⑽reg|_〇n)工作)。據此, 當採用過去文獻之aipha p〇Wer _de|來描述電晶體的電 壓與電流行為後,前述的不等關係可以以下列式⑴描述之 〇 其中 …(1) 丨D0C為前述反向器電路(特指其p型金氧半電晶體 (P1))在 VGS = VDS = VDD 之狀況下之電流(drain current), 其與電晶體的尺寸參數有關(例如··電晶體之通道的寬度與 長度);Electric & dry position soared to the power supply (VD: for the electric charge level. At point B, the power gate element (four)) current (d_CU_t), etc. ... type gold oxide semi-transistor " _ current) 〇 However, as described above, after the inverter's electric dust level exceeds B, the current is continuously drawn to bring the voltage of the virtual voltage source (VDDv) to the highest voltage level (VDD). As the voltage of the virtual voltage source (vddv) rises, the drain-source voltage drop (VDs) of the inter-power element (PG) decreases as the voltage of the virtual voltage source (VDDV) rises, . Drop, so that a smaller current passes through the inter-power component (PG)° in other words, because the current of the power-off element (PG) at point B is equal to the current of the p-type MOS transistor (drain current) ) 'With the point B, * 'a# , + ,. After that, the current flowing through the power gate element (pG) continues to decrease. Therefore, 'when the inverter is observed, p-type MOS half-powered 201035738 crystal (P1) current at point B is greater than (or equal to) the current (drain cu_t) of the power-off element (pG) after passing point B (drain cu_t) At this time, it can be said that the reverse piano has completed the state switching operation (where the p-type MOS transistor (p", the power-on element (PG) operates in the linear region (10) reg|_〇n). When the aipha p〇Wer _de| of the past literature is used to describe the voltage and current behavior of the transistor, the aforementioned unequal relationship can be described by the following formula (1): (1) 丨D0C is the aforementioned inverter circuit ( Specifically refers to the current (drain current) of its p-type MOS transistor (P1) under the condition of VGS = VDS = VDD, which is related to the size parameter of the transistor (for example, the width and length of the channel of the transistor) );

ID0PG則為前述功率閘元件(pG)在vgs = VDS= VDD 之狀況下之電流(d r a i n c u ιτ 6 π t);ID0PG is the current (d r a i n c u ιτ 6 π t) of the aforementioned power gate element (pG) under the condition of vgs = VDS = VDD;

Vth疋%日日體之b界電壓(thresh〇|d v〇itage); VV是該虛擬電壓源(vddv)隨時間變化之瞬間電壓 (transient value);及 VDS0是前述之預設臨界電壓值。 前述與電晶體尺寸有關之參數的關連性則可由第三圖 中之A點獲取。該a點’其對應該虛擬電壓源(VDDV)之 電壓最低點(VL) ’代表通過該功率閘元件(pG)之電流最大 201035738 …5亥反向器進行切換的時刻;其中,該電壓最低點 )亚非不變常值’其隨該功率閘元件(pG)之電晶體尺寸 茶數、該功率閘元件(PG)串接之電路模組(例如本範例之反 向器)、電路操作條件·.等之不同而改變。而在A點時,通 過該功率閘元件(PG)與該p型金氧半電晶體(ρυ之電流 (drain⑽「e叫相同’且該卩型金氧半電晶體(ρι)、該功率 閘元件(PG)为別在飽和區(saturatj〇n regj〇n)及線性區 (linear region)工作。因此,在A點相等的電流可表示為式 〇 (2):Vth疋% b body voltage of the day body (thresh〇|d v〇itage); VV is the transient voltage of the virtual voltage source (vddv) as a function of time; and VDS0 is the aforementioned preset threshold voltage value. The correlation of the aforementioned parameters relating to the size of the transistor can be obtained from point A in the third figure. The point a of the voltage point (VL) corresponding to the virtual voltage source (VDDV) represents the time at which the current through the power gate element (pG) is switched to the maximum of 201035738 ... 5 hp reverser; wherein the voltage is the lowest Point) the non-invariant constant value of the transistor, the number of the crystals of the power gate element (pG), the circuit module of the power gate element (PG) connected in series (for example, the inverter of this example), circuit operation Conditions, etc. vary. At point A, the power gate element (PG) is passed through the p-type MOS transistor (the current of ρυ (drain (10) is the same as the 'e' and the 金-type MOS transistor (ρι), the power gate The component (PG) operates in the saturation region (saturatj〇n regj〇n) and the linear region. Therefore, the current equal to point A can be expressed as 〇(2):

f ID0C V VL-Vth、 VDD-Vth,f ID0C V VL-Vth, VDD-Vth,

(1 + XVL) = 1D0PG(1 + XVL) = 1D0PG

VDD-VL FDO VDD-Vth~~VD〇 I 一 VDD-Vth-VL u -(2) 其中: VDO為VGS = VDD時,本實施例中所用P型電晶體 的汲極飽和電壓(drain saturation voltage); Ο α 為速度飽和參數(velocity saturation index); λ為電晶體通道長度調變參數(channel length modulation parameter); 進一步,將式(1)之丨DOC以公式(2)替代,式(1)可改 寫為下列式(3):VDD-VL FDO VDD-Vth~~VD〇I VDD-Vth-VL u -(2) where: When VDO is VGS = VDD, the drain saturation voltage of the P-type transistor used in this embodiment (drain saturation voltage) Ο α is the velocity saturation index; λ is the channel length modulation parameter; further, the 丨DOC of equation (1) is replaced by the formula (2), ) can be rewritten as the following formula (3):

-VDD-Vth + -VL (F£>£) - VV){-VDD - Vth + \VV o 9 . ? 2 ( i 1 — f 1 ) VDO l 2 J VDSO VV - Vih --VS 0 l 2 J-VDD-Vth + -VL (F£>£) - VV){-VDD - Vth + \VV o 9 . ? 2 ( i 1 — f 1 ) VDO l 2 J VDSO VV - Vih --VS 0 l 2 J

^VDD-VthX VDD-VL 1 + IVL 13 .·· (3) 201035738 由上列式(3)可以看出,與電晶體之尺寸參數相依(有 關)的丨D0C、丨D0PG等參數在前述的簡化基礎下被消去, 代表前述的延遲時間評估機制與該功率間元件(pG)或電路^VDD-VthX VDD-VL 1 + IVL 13 .. (3) 201035738 It can be seen from the above formula (3) that parameters such as 丨D0C and 丨D0PG depend on the size parameter of the transistor (in the above). Simplified on the basis of elimination, representing the aforementioned delay time evaluation mechanism and the inter-power component (pG) or circuit

几件之尺寸參數無關,前述的延遲時間評估機制只與W 、VL有關,即與該虛擬電壓源(VDDV)之動態響應行為有 關而已’因此’本發明所指的判斷機制具有可容忍環境變 異的效能。 若當功率問元件(PG)為採用N型電晶體時,則稱該N 型電晶體與接地點(GND)之間的節點稱為一虛擬接地點 (V_al GND ’ VGND) ’與前述虛擬電壓點(vddv)不同的 是’該虛擬接地點(VGND)在該反向器高底準位切換過程中 呈現先升後降的㈣’也就是說,在該虛擬接地 〇 :有電壓最高點⑽);而前述依據p型電晶體所推導的公 式,於替換N型/p型之特性公4 ^ ν ^ Α式及所採取製程之Ν型/Ρ型 性參數後,可直接沿用前列所提之概念;舉例而言,相 對於上列式(3)’若該功率閘元 該式⑶可改為下式(3.1): )為採用Ν型電晶體, VDD-Vth ~VDD-VH-Vth)Regardless of the size parameters of several pieces, the aforementioned delay time evaluation mechanism is only related to W and VL, that is, related to the dynamic response behavior of the virtual voltage source (VDDV), and thus the judgment mechanism referred to in the present invention has tolerable environmental variation. Performance. If the power component (PG) is an N-type transistor, the node between the N-type transistor and the ground (GND) is called a virtual ground point (V_al GND 'VGND)' and the aforementioned dummy voltage. The difference between the points (vddv) is that the virtual ground point (VGND) is first rising and then falling during the high-level level switching of the inverter (four) 'that is, in the virtual ground 〇: the highest voltage point (10) The above-mentioned formula derived from the p-type transistor can be directly used in the front row after replacing the characteristic of the N-type/p-type and the Ν-type/Ρ-type parameter of the adopted process. The concept; for example, compared to the above formula (3)', if the power gating element, the formula (3) can be changed to the following formula (3.1): ) is a Ν-type transistor, VDD-Vth ~ VDD-VH-Vth )

VH 1 + _一所)了 VDD~V,h~\vD〇\ VDS〇 VG VDD-Vth-[VG S 2 , f 1 、 VDD - VG-m - ivDSQ V 2VH 1 + _ one) VDD~V, h~\vD〇\ VDS〇 VG VDD-Vth-[VG S 2 , f 1 , VDD - VG-m - ivDSQ V 2

J 14 201035738 其中: VDD為該電源提供之電壓; VG是該虛擬接地點(VGND)隨時間變化之瞬間電壓 (transient value);J 14 201035738 where: VDD is the voltage supplied by the power supply; VG is the transient value of the virtual ground point (VGND) as a function of time;

Vth為該多模式功率閘網路及該電路模組内之n型電 日日體之一 ε品界電壓(threshold voltage); VD0為N型電晶體於vgS=VDD時之一汲極飽和電 壓(drain saturation voltage);Vth is the ε product threshold voltage of the multi-mode power gate network and the n-type electric solar field in the circuit module; VD0 is one of the N-type transistors when vgS=VDD (drain saturation voltage);

«為一速度飽和參數(vei〇cjty saturation index); 又為N型電日日體通道長度調變參數(channe丨丨ength modulation parameter);及 VDS0是一預設臨界電壓值。 °月參考第四圖,其為第三圖中參數VL、VV所繪製的 圖形解’式(3)的兩邊分別作為兩個獨立的多項式,而vl、 W分別為兩個多項式之變數。該式⑺之左、纟半邊式子 可分別緣成第四圖之圓形標號線(〇_,f(VL))及矩形標號線( f(VV))’其中’式(3)之VDs〇於此設定為1〇〇mV。透 過以參數^作為自變數,參數w之最小值可在滿足式(3) 之條件下於第四圖之中取 Τ取侍,重複改變VL之數值,可以獲 得較為完整的參數VL |« is a speed saturation parameter (vei〇cjty saturation index); is also a N-type electric day channel length modulation parameter (channe丨丨ength modulation parameter); and VDS0 is a preset threshold voltage value. The month of the reference is shown in the fourth figure, which is the two sides of the graph solution (3) drawn by the parameters VL and VV in the third figure, respectively, as two independent polynomials, and vl and W are variables of two polynomials respectively. The left and right half of the equation (7) can be respectively formed into the circular label lines (〇_, f(VL)) and the rectangular label lines (f(VV)) of the fourth graph, where the VDs of the equation (3)设定This is set to 1〇〇mV. By using the parameter ^ as the self-variable, the minimum value of the parameter w can be taken from the fourth picture under the condition that the formula (3) is satisfied, and the value of VL is repeatedly changed to obtain a relatively complete parameter VL |

對VV之關係數值組,藉由vl、VV 之關係數值組可以以—直線求律 ⑽ 列不等之式(4): 下 乃^2 0.5685 +0.4568. (4) 或者,也可以用二次方 fitted)得到下列不等之式(5): 程之曲線求律法(quadratic 15 201035738 ^>〇.〇9〇2 + 1.7137-FL-0.8109·^2 (5) 刖述式(4)、(5)之單位為伏特,其描述參數、w 之間的不等關係分別描繪於第五圖之中。 循前述之式(4)、(5)之類似推導過程,當功率閘元件 (PG)為採用N型電晶體,相對於(4)、(5)而用於n型電晶 體之不等式分如下列公式: ^<-0.01729 + 0.4424· W ... (4.1) Ο Ο VG < 0.0045 + 0.1810-F// +0.5810-F//2 (4 2) 其中,式(4.1)、(4.2)之各電壓單位為伏特(v)。 综言丽述内容及式(3〉所揭露的延遲時間評估機制 驟可包含: (i)開始/重置; (Π)擷取虛擬電壓源(VDDV)之電壓最低點vl . 间監控虛擬電壓源(卿v)與電壓最低點ν (4)或(5)之數值;及 (iv)完成延遲時間評估。 綜言之’該虛擬電壓源(VDDV)之電壓最低點Μ可以 在電路(於此為該反向器)之狀態切換過程被找到,而該虛擬 電壓源(VDDV)之電壓經過電壓最低點之後必須持續上升及 至滿足上述之式(4)或(5)以判定該雷故 。 j疋。g也路之切換狀態已經完成 :青參考第六圖’當在理想電源與_心元乘法器⑽ 」川丨PI叫連接-個該功率間元件(pG)時,也可以依據For the relational value group of VV, the relational value group of vl and VV can be expressed by the linear law (10). The equation is not equal (4): the lower is ^2 0.5685 +0.4568. (4) Alternatively, it can be used twice. The square fitted) obtains the following inequality (5): The curve of the rule of law (quadratic 15 201035738 ^>〇.〇9〇2 + 1.7137-FL-0.8109·^2 (5) Description (4) The unit of (5) is volt, and the unequal relationship between the description parameters and w is respectively depicted in the fifth figure. Following the similar derivation process of equations (4) and (5) above, when the power gate element ( PG) is an N-type transistor, and the inequality for the n-type transistor with respect to (4) and (5) is as follows: ^<-0.01729 + 0.4424· W ... (4.1) Ο Ο VG < 0.0045 + 0.1810-F// + 0.5810-F//2 (4 2) wherein the voltage units of the equations (4.1) and (4.2) are volts (v). The disclosed delay time evaluation mechanism may include: (i) start/reset; (Π) draw the virtual voltage source (VDDV) voltage lowest point vl. monitor the virtual voltage source (qing v) and the voltage lowest point ν (4) or (5) the value; and (iv) the completion Delay time evaluation. In summary, the voltage minimum point of the virtual voltage source (VDDV) can be found in the state switching process of the circuit (here, the inverter), and the voltage of the virtual voltage source (VDDV) passes. After the lowest voltage, it must continue to rise and meet the above formula (4) or (5) to determine the lightning. j疋.g also the switching state of the road has been completed: the green reference to the sixth figure 'when in the ideal power supply and _ heart Yuan multiplier (10) "When the Chuanxi PI is connected to a power component (pG), it can also be based on

則述的判斷機制取得該乘法考M 朱沄斋(11)之延遲時間評估,妷而 ,經過研究測試結果顯示,採i …' 抹用本發明所提之延遲評估機 16 201035738 制與實際電路之延遲約# 10%的差異,因此,必須加入約 10%〜15%之餘裕(margjn)於最終延遲評估結果。 利用前述所提的延遲評估機制,本發明之一適應性功 率控制(adaptive power contro卜APC)裝置之電路方塊圖 可如第七圖所示,該適應性功率控制裝置包含一多模式功 率閘網路(multi-mode power gating netwo「k,21)、一電壓 感測器(voltage sensor,23)、一可變門檻比較器(va「丨a — thresho丨d comparator,24)、一冗餘區間债測電路 ::;Detectlon,25)以及一雙向位移暫存器(B|__directi〇na|讣⑴ Register 27),其中,本貫施例之適應性功率控制裝置可 應用於各種不同的電路模組,如前述的乘法器、反向器等 各種邏輯電路,本實施例之該適應性功率控制裝置係連接 一互補式金氧半之電路模組(CMOS circuit,30丨。 該多模式功率閘網路(21)之輸出、輸入端分別電性連 接該電路模組(3〇)及該雙向位移暫存器(27),該電壓感測器 ^ (23)連接於該多模式功率閘網路(21)及該電路模組(3〇)連接 之一個该虛擬電壓源(VDDV),該電壓感測器(23)之輸出端 連接該可變門檻比較器(24),而該可變門檻比較器(24)、該 冗餘區間偵測器(25)、該雙向位移感測器(27)、該多模式功 率閘網路(21)則依序串接。 請參考第八圖,本實施例之該多模式功率閘網路(21) 包含相互並聯且串接於該理想電源(VDD)及該電路模組(3〇) 之間的複數個P型金氧半電晶體(PM〇s),其中,各p型金 氧半電晶體(PM〇S)之閘極連接至由該雙向位移感測器(27) 所储存之控制訊號(Ctrl4〜0)。依據前述之說明可知’加入 17 201035738 力率間S件(P G)則會4所連接之電路模組的虛擬電壓源產 生%壓下4的情形,因此,所加入的功率閘元件(ρ〇)之尺 寸/數越小,則电壓下降的狀況越大,這樣的電壓下降狀 .況可以被視為輸入該電路模組之電源雜訊(511叩丨乂叫丨%)。 而此、電£下降,則會反應在電路模組之延遲,因此,該 多模式功率閉網路(21)可以就由選擇性驅動該控制訊號 (Ctrl4〜0)而開啟該多模式功率閘網路(21)之p型金氧半電 晶體(PMOS),進-步影響或控制電路模組之延遲。 進一步地,該多模式功率閘網路(21)之各功率閘元件 (PG)可以具有不同的通道尺寸參數’電源經過該多模式功 率間網路(21)而輸出至該f路模組⑽)之功率,而可藉由開 關各功率閘元件(PG)而產生多種不同的組合。 曰歼 請參考第九圖,本實施例之該電壓感測器(23)包含以 二極體連接方式(di〇de-connected)連接之二電晶體(Μρι、 MP2)’其中’在該電路模組之工作切換過程中與該電晶 體(MP2)連接之- VL節點可以取得該虛擬電壓源(vddv)之 ^ 最低電壓。 請參考第十圖,本實施例之該可變門根比較器(2幻為 一變型之史密特觸發電路(modified Schmiu Trjgg叫嗜 史密特觸發電路由一時脈脈衝(pu|sed c丨〇ck)預先充電 (precharged)。該可變門檻比較器(24)係與該電壓感測器 (23)之電晶體(MP2)之VL節點連接,其—可變門; (thresho丨d)由該VL節點控制(為反向控制,如第十圖中 丨’—VL)。因此,該可變門檻比較器(24)可以判斷式或式^ 中的該虛擬電壓源(VDDV)之瞬間電壓(VV)飽和盥 "} /、ύ 。易言 18 201035738 之,該可變門檻比較器(24)比較該電壓感測器(23)之vl節 點(如第十圖中之i_VL)及該虛擬電壓源之瞬間電麼(第十圖 中之W)是否滿足式(4)或(5)。而當該可變門捏比較器(24) 之輸出端切換至低準位,形成一比較結果狀態(assert丨·〇η 〇f the comparator)而輸出時,代表該電路模組已經完成切換 工作。當該功率閘網路(21)所採用的電晶體為N型時,則 該電壓感測器(23)、該可變門檻比較器(24)之電路則需適應 性修改,而可以改為偵測該虛擬接地點(VGND)之電壓變化 特性;例如,該電壓感測器(23)除了必須改為連接於該虛 擬接地點(VGND)之外,也必須修改内部電路而可以感測該 虛擬接地點(VG N D)之電壓變化。 该冗餘區間偵測電路(25)透過比較該可變門檻比較器 (24)的比較結果狀態及一電路工作的時脈邊緣丨 以判斷時脈週期内是否存在一冗餘區間(unused s|ack)。該 =餘區間是指在時脈週期中,某些為了防止電路因工作環 ^兄菱化以成延遲時間的增加而加入的時間餘裕,或者是因 “路輸入資料不同造成運算時間的縮短而所剩餘的時脈 長度。 。月麥考第十一圖,該多模式功率閘網路(2彳)與該雙向 =私暫存器(27)之複數個控制訊號(Ctrl4〜0)連接,其中, 田發見過多冗餘區間存在時,該雙向位移暫存器(27)之狀 態二,右移以關掉該多模式功率閘網路(2彳)之功率開(即p 型金氧半電晶體(PM〇s));反之,當沒有足夠的冗餘區間 寺(比電路模組需要的時間餘裕⑴⑺丨叩ma「g jn)少時), 該雙向位移暫存器(27)之狀態,’〇,往左移而開啟更多的該多 19 201035738 模式功率開網路(21)之功率問。其中,該雙向位移暫存器 (27)可以設有一重置(reset)、一維持_d)、—電源切斷狀 態(Ρ_「_ng state)...等控制訊號,其分別作為將該雙 向位移暫存器(27)内的各位元歸零(打開所有的功率閉卜唯 持該雙向位移暫存器(27)内部的各位元之狀態,以及關閉 所有的功率閘等作用。 ο 一般目前既有之技術中,電路模組於工作時之速度規 格㈣…specmcation)通常需要考量製程、電壓及=度 (Process . Voltage, Temperature PVT)t,b^t,f 糕狀況而必須在時脈的規格之中加入許多的餘裕(贿g㈣ ,以確保電路工作的延遲時間不會因工作環境的變化而大 於時脈的規格,避免其他電路產生連鎖錯誤而影響最終之 運算結果。然而,電路模組在使用過程中,前述的最糟糕 狀況鮮少發生,因此,使得輸入該電路模組之大量功率卞 費在該些餘裕上。而前述的適應性功率控制方法及裳置2 可以判斷該電路槿組门η彳b尤时" 0 、、 疋否將所有的時脈餘裕加以用盡 〇 另’在某些大型電路模组中,其非關鍵路徑(non_ cmicai paths)的延遲時間相對於時脈規格亦存在一時脈餘 裕’因=相同的概念(判斷電路是否存在冗餘區間)亦可使用 於私路松组中的一非關鍵路徑(non-critical paths)。 另外’本實施例之適應性功率控制方法與裝置具 谷忍各種變显之辟柯 m /、特丨生因為所有製程電壓溫度之變昱均 映在電路的延遁眭P弓 ^ , π間上,而本實施例可以在電路模組工作 之中,取付電路之延遲狀況,同時主動使用電路之中的餘 20 201035738 裕區間’達到節省功率之技術效果。 以上所述者,僅為本發明—較佳 來限定本發明實施之範圍,故舉凡依::明:,並非用 所述之形狀、構造、特徵及精;:::利範圍 均應包括於本發明之申請專利範圍内。…化與修飾’ 【圖式簡單說明】 第一圖為一 弟二圖為一 第三圖為一* 意圖。The judgment mechanism described above obtains the delay time evaluation of the multiplication test M Zhu Xizhai (11), and after the research test results show that the delay of the delay evaluation machine 16 201035738 and the actual circuit are extracted by the invention Approximately #10% difference, therefore, it is necessary to add about 10% to 15% margin (margjn) in the final delay evaluation results. Using the foregoing delay evaluation mechanism, a circuit block diagram of an adaptive power control (APC) device of the present invention can be as shown in FIG. 7, the adaptive power control device includes a multi-mode power gate network. Multi-mode power gating netwo "k, 21", a voltage sensor (23), a variable threshold comparator (va "丨a - thresho丨d comparator, 24", a redundant interval Debt measurement circuit::;Detectlon, 25) and a bidirectional displacement register (B|__directi〇na|讣(1) Register 27), wherein the adaptive power control device of the present embodiment can be applied to various circuit modes. For example, the adaptive power control device of the present embodiment is connected to a complementary MOS circuit (CMOS circuit, 30 丨. The multi-mode power gate is connected to various logic circuits such as the multiplier and the inverter described above. The output and input ends of the network (21) are electrically connected to the circuit module (3〇) and the bidirectional displacement register (27), and the voltage sensor ^ (23) is connected to the multi-mode power gate network. Road (21) and the circuit module (3 〇) connecting one of the virtual voltage sources (VDDV), the output of the voltage sensor (23) is connected to the variable threshold comparator (24), and the variable threshold comparator (24), the redundant interval The detector (25), the bidirectional displacement sensor (27), and the multi-mode power gate network (21) are serially connected in series. Referring to the eighth figure, the multi-mode power gate network of the embodiment (21) comprising a plurality of P-type MOS transistors (PM〇s) connected in parallel with each other and connected in series between the ideal power supply (VDD) and the circuit module (3〇), wherein each p-type gold oxide The gate of the semi-transistor (PM〇S) is connected to the control signal (Ctrl4~0) stored by the bidirectional displacement sensor (27). According to the above description, it is known that 'join 17 201035738 between the force rate and the S piece (PG) The virtual voltage source of the four connected circuit modules generates a case where the % is depressed by 4, and therefore, the smaller the size/number of the added power gate elements (ρ〇), the larger the voltage drop is. The voltage drop can be regarded as the power noise (511 叩丨乂 丨%) input to the circuit module. The delay in the circuit module is reflected. Therefore, the multi-mode power closed network (21) can turn on the multi-mode power gate network (21) by selectively driving the control signal (Ctrl4~0). Type MOS semi-transistor (PMOS), step-in-effect or delay of the control circuit module. Further, each power gate element (PG) of the multi-mode power sluice network (21) may have different channel size parameters The power of the power supply to the f-channel module (10) through the multi-mode inter-power network (21) can be varied by switching the power gate elements (PG). Please refer to the ninth figure. The voltage sensor (23) of the embodiment includes two transistors (Μρι, MP2) connected in a diode-connected manner (wherein 'in the circuit The VL node connected to the transistor (MP2) during the working switching of the module can obtain the lowest voltage of the virtual voltage source (vddv). Please refer to the tenth figure, the variable gate root comparator of the embodiment (2 is a variant of the Schmitt trigger circuit (modified Schmiu Trjgg called the Schmidt trigger circuit by a clock pulse (pu|sed c丨〇ck) precharged. The variable threshold comparator (24) is connected to the VL node of the transistor (MP2) of the voltage sensor (23), which is a variable gate; (thresho丨d) Controlled by the VL node (for reverse control, such as 丨'-VL in the tenth figure). Therefore, the variable threshold comparator (24) can determine the instant of the virtual voltage source (VDDV) in the equation or the equation Voltage (VV) saturation 盥 "} /, ύ. 易言18 201035738, the variable threshold comparator (24) compares the vl node of the voltage sensor (23) (such as i_VL in the tenth figure) and Whether the instantaneous voltage of the virtual voltage source (W in the tenth figure) satisfies the formula (4) or (5), and when the output end of the variable gate pinch comparator (24) is switched to the low level, a Comparing the result state (assert丨·〇η 〇f the comparator) and outputting, it means that the circuit module has completed the switching work. When the power gate network 21) When the transistor used is N-type, the circuit of the voltage sensor (23) and the variable threshold comparator (24) needs to be adaptively modified, and the virtual ground point can be detected instead ( VGND) voltage variation characteristics; for example, the voltage sensor (23) must be modified to be connected to the virtual ground point (VGND), and the internal circuit must be modified to sense the virtual ground point (VG ND) The redundant interval detecting circuit (25) compares the comparison result state of the variable threshold comparator (24) with the clock edge of a circuit operation to determine whether there is a redundant interval in the clock period. (unused s|ack). This = residual interval refers to the time margin added in the clock cycle to prevent the circuit from being added due to the increase of the delay time due to the working ring, or because of the "input data" The length of the clock that is caused by the shortening of the calculation time. The eleventh figure of the monthly wheat test, the multi-mode power gate network (2彳) and the plurality of control signals of the two-way=private register (27) (Ctrl4~0) connection, where Tianfa sees too much When the remainder interval exists, the state of the bidirectional displacement register (27) is shifted to the right to turn off the power of the multimode power gate network (2彳) (ie, p-type MOS transistor (PM〇s) )); conversely, when there is not enough redundant interval temple (less than the time margin (1) (7) 丨叩 ma "g jn) required by the circuit module), the state of the bidirectional displacement register (27), '〇, toward Move left and turn on more of the power of the 19 201035738 mode power on the network (21). The bidirectional displacement register (27) may be provided with a reset signal, a sustain _d), a power off state (Ρ_“_ng state), etc., which are respectively used as the bidirectional The bits in the shift register (27) are reset to zero (open all power closures only to hold the state of each element inside the bidirectional displacement register (27), and to turn off all power gates. ο In the existing technology, the speed specification of the circuit module during operation (4)...specmcation usually needs to consider the process, voltage and degree (Process. Voltage, Temperature PVT)t, b^t, f, and must be in the clock. Many specifications are added to the specifications (bribe g (4) to ensure that the delay time of the circuit operation is not greater than the clock specification due to the change of the working environment, avoiding the chain error of other circuits and affecting the final operation result. However, the circuit mode During the use of the group, the aforementioned worst situation rarely occurs, so that a large amount of power input into the circuit module is spent on the margins. The aforementioned adaptive power control method and the skirt 2 can be judged. The circuit 槿 group gate η彳b especially time " 0, 疋 whether to exhaust all clock margins 〇 another 'in some large circuit modules, the non-critical path (non_ cmicai paths) delay time relative to There is also a clock margin in the clock specification. 'The same concept (determining whether there is a redundant interval in the circuit) can also be used in a non-critical path in the private branch group. In addition, this embodiment The adaptive power control method and device have a variety of changes, and the changes in the temperature of all process voltages are reflected in the delay of the circuit, π, and the implementation. For example, in the operation of the circuit module, the delay condition of the circuit can be taken, and at the same time, the remaining 20 201035738 margin in the circuit can be used to achieve the technical effect of saving power. The above is only the invention - preferably The scope of the present invention is not limited to the shapes, structures, features, and essences of the invention. The scope of the invention should be included in the scope of the patent application of the present invention. Schematic DESCRIPTION The first picture shows a picture shows a third picture shows the two brother a * intentions.

乘法器之隸電流表;見示意圖。 反向器切換之輸出入電壓狀態示意圖。 Μ Μ反Μ電路之輸出人電壓狀態示 第四圖為一以虛擬電壓湄Α m .. ΑΛ ι泉為威測對象之延遲時間評估 機制的圖形解示意圖。 圖 第五圖為一延遲時間評估機制 之直/曲線求律結果示意The current meter of the multiplier; see the schematic. Schematic diagram of the input and output voltage state of the inverter switching. The output voltage state of the Μ Μ Μ circuit is shown in the figure. The fourth figure is a graphical solution of the delay time evaluation mechanism with the virtual voltage 湄Α m .. ΑΛ ι泉 as the test object. Figure 5 is a schematic diagram showing the results of the straight/curve law of a delay time evaluation mechanism.

第六圖為-包含功率閘之乘法器電路圖。 第七圖為一適應性功率控制装置電路方塊圖。 =八圖為一多模式功率閘網路之電路示意圖。 第九圖為一電壓感測器之電路示意圖。 第十圖為一可變門檻比較器示意圖。 第十—圖為一雙向位移暫存器示意圖。 【主要元件符號說明】 (21)多模式功率閘網路 21 201035738 (23) 電壓感測器 (24) 可變門檻比較器 (25) 冗餘區間偵測電路 (27)雙向位移暫存器 (30)電路模組The sixth picture is a circuit diagram of the multiplier containing the power gate. The seventh figure is a block diagram of an adaptive power control device circuit. = Figure 8 is a circuit diagram of a multi-mode power gate network. The ninth figure is a circuit diagram of a voltage sensor. The tenth figure is a schematic diagram of a variable threshold comparator. The tenth-figure is a schematic diagram of a bidirectional displacement register. [Main component symbol description] (21) Multi-mode power gate network 21 201035738 (23) Voltage sensor (24) Variable threshold comparator (25) Redundant interval detection circuit (27) Bidirectional displacement register ( 30) Circuit module

Claims (1)

201035738 七、申請專利範圍: 其二==評估機制之適應性功率控制裝置, 栌比較哭 ]、,同路、—電壓感測器、一可變門 才皿比季乂态、—冗餘區間偵 j义π 咳多槎十说玄 、、電路以及一雙向位移暫存哭, 口亥夕松式功率閑網路包含 《存口口 , ^ nB 双個亚聯之電晶體,杏兮炙掠 式功率閘網路連接於— 田5玄^杈 电々、與—電路槿4 Ο201035738 VII. Patent application scope: The second == adaptive power control device of the evaluation mechanism, 栌 relatively crying], the same way, the voltage sensor, a variable door, the seasonal state, the redundant interval Detective j π cough more than ten said Xuan, circuit and a two-way displacement temporary crying, mouth Hai Xisong power idle network contains "storage mouth, ^ nB two sub-connected crystals, apricots The power gate network is connected to - Tian 5 Xuan ^ 杈 々, and - circuit 槿 4 Ο 體為P型電晶體且該多模文…、、' 之間%,各電晶 接節點視為-虛擬购, 、广路桓組之連 -接地點與該電路模植之門昧一式功率間網路連接於 該多模式功率閘網路盥兮φ 勹N i電日日體且 接地點,其中: jπ即點視為一虛擬 接地電則益之輪入端連接於該虛擬電壓源或該虛擬 产擬接地f:寻找:亥虛擬電壓源之—最低電壓(VL)或尋找該 虛擬接地點之一最高電壓(VH); 該:變門檀比較器與該電壓感測器之輸出端電性連接 其曰比較δ亥虛擬電壓源之一瞬間電壓(vv)與該最低電壓 ()疋否滿足第-特定關係而輸出―第—比較結果狀態 ’或者該可變門檻比較器比較該虛擬接地點之瞬間電壓 (VG)與該最高電壓(VH)是否滿足一第二特定關係而輸出一 第二比較結果狀態; 4几餘區間偵測電路之輸入端連接於該可變門檻比較 器,依據該可變門檀比較器之該第—或第二比較結果狀態 及一時脈邊緣以偵測—時脈週期内是否存在—冗餘區間; 该雙向位移暫存器之輸入端與該冗餘區間偵測電路電 ί"生連接,其依據έ亥几餘區間之是否存在而產生位移,使該 23 201035738 雙向位移暫存器之複數在 ^ ^ t制訊唬的輪出狀態改變 忒多杈式功率閘網 文以及 號連接,纟電晶體分別= 暫存器之各控制訊 訊號之狀態開啟或關閉 :之各““空制 ^ . , U| 使^夕拉式功率閘網路依據冗铨 S間之存在狀況而改變該锞几餘 2.如申請專利範圍第彳 半 图弟1項所述之具電路延遲評估檣制 之適應性功率控制裝置,並 歲制 公式·· 〃、宁6亥弟—特定關係滿足下列 Ο 歷:十z (VDD~vv{l_-m+W VD〇\, VDD-Vth~~VD〇 VL-Vth ) l+AVL 7~~ --r->—-_V2 2 , VDSO 其中 VV — Vth—2V_ VDD為该電源提供之電壓; Vih為該多模式功率閘網路及該電路模組内之p型電 晶體之一臨界電壓(threshold voltage);The body is a P-type transistor and the multi-mode text ...,, '% between, each of the electro-glyph nodes is regarded as - virtual purchase, the connection of the Guangluo group - the grounding point and the gate of the circuit molded power The inter-network is connected to the multi-mode power gate network 盥兮φ 勹N i electric day and body and the grounding point, wherein: jπ is the point that a virtual ground power is connected to the virtual voltage source or The virtual product is grounded f: looking for: the virtual voltage source of the sea - the lowest voltage (VL) or looking for one of the virtual ground points (VH); the: the gate comparator and the output of the voltage sensor Electrically connecting the transient voltage (vv) of one of the ΔH virtual voltage sources to the minimum voltage () 满足 whether the first-specific relationship is satisfied and outputting the “first comparison result state” or the variable threshold comparator comparing the virtual The instantaneous voltage (VG) of the grounding point and the highest voltage (VH) satisfy a second specific relationship and output a second comparison result state; 4 the input of the interval detection circuit is connected to the variable threshold comparator, According to the first or second comparison of the variable gate comparator The state and the edge of the clock are detected - whether there is a redundant interval in the clock cycle; the input end of the bidirectional displacement register is electrically connected with the redundant interval detecting circuit, and the basis is Whether the interval exists or not, the displacement of the 23 201035738 bidirectional displacement register is changed in the turn-off state of the ^^ t system, and the multi-turn power gate network and the number connection are connected, and the transistors are respectively temporarily stored. The status of each control signal of the device is turned on or off: each of the "empty system ^ , U | makes the ^ Xi pull power gate network change according to the existence of the redundant S. 2. If the application The adaptive power control device with the circuit delay evaluation system described in the first part of the patent scope, and the formula of the age system··〃,宁六海弟—the specific relationship satisfies the following calendar: ten z (VDD~ Vv{l_-m+W VD〇\, VDD-Vth~~VD〇VL-Vth) l+AVL 7~~ --r->--_V2 2 , VDSO where VV — Vth—2V_ VDD is the power supply The voltage supplied; Vih is the multi-mode power gate network and one of the p-type transistors in the circuit module Voltage (threshold voltage); VD0為P型電晶體於vgs=VDD 壓(drain saturation voltage); 時之一沒極飽和電 α為一速度飽和參數(ve|0Cjty ίΓ1(:^χ丨; 入為p型電晶體通道長度調變參數(channe丨丨ength modulation parameter);及 VDS0是一預設臨界電壓值。 3.如申請專利範圍第1項所述之具電路延遲評估機 制之適應性功率控制裝置,其中,該第二特定關係滿足下 列公式: 24 201035738 VDD-Vih VDD-VH-Vih VH VDD-Vth-^VH VG 1 + k{VDD ~ VH) VDO VDD-Vth ——VDO VDSO VDD~Vt}卜 tyG 2 Γ 、 VDD — VG-Vth 二 VDSQ 2 y 其中 VDD為該電源提供之電壓; Vth為該多模式功率閘網路及該電路模組内之N型電 晶體之一臨界電壓(thresh0ld v〇|tage); VDO為N型電晶體於VGS=VDD時之—汲極飽和電 ® (drain saturation voltage); α 為一速度飽和參數(ve|0Cjty saturation index); λ為N型電晶體通道長度調變參數(channe丨丨ength modulation parameter);及 VDS0是一預設臨界電壓值。 4. 如申請專利範圍第1或2或3項所述之具電路延遲 。平估機制之適應性功率控制裝置,該可變門檻比較器為一 變型之史密特觸發電路。 5. 如申請專利範圍第1或2或3項所述之具電路延遲 §平估機制之適應性功率控制裝置’該電壓感測器包含以二 極體連接方式串接之二電晶體的感測電路。 6 ·如申請專利範圍第2項所述之具電路延遲評估機 制之適應性功率控制裝置,該X/DS0為1 〇〇mV時,該第 一特定關係則滿足下列二者公式之一: )^> 0.5685 +0.4568.KL ;及 KKk〇.〇9〇2 + l.7137.FL-0.81〇9.Ki:2,電壓單位為伏特(V)。 7 _如申請專利範圍第3項所述之具電路延遲評估機 25 201035738 制之適應性功率控制裝置,該VDS0為10OmV時,該第 二特定關係則滿足下列二者公式之一: FGS-0.01729 +0.4424. K// ;及 KGS0.0045 + 0.1810.K// + 0.5810.K//2,電壓單位為伏特(V)。 八、圖式:(如次頁)VD0 is a P-type transistor at vgs=VDD saturation voltage; one of the non-polar saturation electric α is a speed saturation parameter (ve|0Cjty ίΓ1(:^χ丨; into the p-type transistor channel length adjustment) The variable parameter (channe丨丨ength modulation parameter); and VDS0 is a predetermined threshold voltage value. 3. The adaptive power control device with a circuit delay evaluation mechanism according to claim 1, wherein the second The specific relationship satisfies the following formula: 24 201035738 VDD-Vih VDD-VH-Vih VH VDD-Vth-^VH VG 1 + k{VDD ~ VH) VDO VDD-Vth ——VDO VDSO VDD~Vt} tyG 2 Γ , VDD — VG-Vth 2 VDSQ 2 y where VDD is the voltage supplied by the power supply; Vth is the threshold voltage of the multimode power gate network and the N-type transistor in the circuit module (thresh0ld v〇|tage); VDO For the N-type transistor at VGS=VDD, the drain saturation voltage is used; α is a velocity saturation parameter (ve|0Cyty saturation index); λ is the N-type transistor channel length modulation parameter (channe)丨丨ength modulation parameter); and VDS0 is a preset Boundary voltage value 4. An adaptive power control device with a circuit delay as described in claim 1 or 2 or 3. The variable threshold comparator is a modified Schmitt trigger circuit. 5. An adaptive power control device with a circuit delay § evaluation mechanism as described in claim 1 or 2 or 3 'The voltage sensor comprises a diode connected in series by a diode connection 6. The adaptive power control device with the circuit delay evaluation mechanism described in claim 2, when the X/DS0 is 1 〇〇mV, the first specific relationship satisfies the following two formulas. One: )^> 0.5685 +0.4568.KL; and KKk〇.〇9〇2 + l.7137.FL-0.81〇9.Ki:2, the voltage unit is volt (V). 7 _ The adaptive power control device of the circuit delay evaluation machine 25 201035738, as described in claim 3, wherein the second specific relationship satisfies one of the following two formulas when the VDS0 is 10OmV: FGS-0.01729 +0.4424. K// ; and KGS0.0045 + 0.1810.K// + 0.5810.K//2, the voltage unit is volt (V). Eight, schema: (such as the next page) 2626
TW98108462A 2009-03-16 2009-03-16 Adaptive power control apparatus with delay estimation scheme TWI387871B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW98108462A TWI387871B (en) 2009-03-16 2009-03-16 Adaptive power control apparatus with delay estimation scheme

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW98108462A TWI387871B (en) 2009-03-16 2009-03-16 Adaptive power control apparatus with delay estimation scheme

Publications (2)

Publication Number Publication Date
TW201035738A true TW201035738A (en) 2010-10-01
TWI387871B TWI387871B (en) 2013-03-01

Family

ID=44855967

Family Applications (1)

Application Number Title Priority Date Filing Date
TW98108462A TWI387871B (en) 2009-03-16 2009-03-16 Adaptive power control apparatus with delay estimation scheme

Country Status (1)

Country Link
TW (1) TWI387871B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI470395B (en) * 2012-12-21 2015-01-21 Nat Univ Chung Cheng Dynamic voltage modulation system with pre-set time margin and localized voltage increase
TWI804338B (en) * 2022-06-02 2023-06-01 國立中山大學 Voltage and temperature variation sensing detector

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI470395B (en) * 2012-12-21 2015-01-21 Nat Univ Chung Cheng Dynamic voltage modulation system with pre-set time margin and localized voltage increase
TWI804338B (en) * 2022-06-02 2023-06-01 國立中山大學 Voltage and temperature variation sensing detector

Also Published As

Publication number Publication date
TWI387871B (en) 2013-03-01

Similar Documents

Publication Publication Date Title
Elgebaly et al. Variation-aware adaptive voltage scaling system
CN101278248B (en) Semiconductor integrated circuit having current leakage reduction scheme
TWI451104B (en) Operating parameter monitoring circuit,system,and method
TWI546526B (en) Thermal sensor, combination current generator and thermal sensor system
CN203588103U (en) Circuit device
Mukundrajan et al. Ultra low power circuit design using tunnel FETs
TW200925819A (en) Self-aware adaptive power control system and a method for determining the circuit state
Padmavathi et al. Low power design techniques and implementation strategies adopted in VLSI circuits
Gupta et al. Transistor gating: A technique for leakage power reduction in CMOS circuits
TW201035738A (en) Adaptive power control apparatus with delay estimation scheme
Kim et al. Ultralow-voltage power gating structure using low threshold voltage
Jang et al. On-chip aging prediction circuit in nanometer digital circuits
Xu et al. Run-time active leakage reduction by power gating and reverse body biasing: An energy view
Gopalakrishnan et al. Leakage power reduction using self-bias transistor in VLSI circuits [digital circuits]
Miyazaki et al. Observation of one-fifth-of-a-clock wake-up time of power-gated circuit
Kaffashian et al. An optimization method for NBTI-aware design of domino logic circuits in nano-scale CMOS
Arumugam et al. SAPON approach: A new technique for Low Power VLSI Design
Dadoria et al. New leakage reduction techniques for FinFET technology with its application
Sarma High Speed Dual Mode Logic
Kim et al. Analytical delay and variations modeling in the subthreshold region
Prakash et al. Design and analysis of low power energy efficient, domino logic circuit for high speed application
Rjoub et al. The Influence of the Nanometer Technology on Performance of CPL Full Adders.
Kassa et al. Forced stack sleep transistor (FORTRAN): a new leakage current reduction approach in CMOS based circuit designing
Chouhan et al. A novel delay minimization technique for low leakagewide fan-in domino logic gates
Marranghello et al. Delay model for static CMOS complex gates

Legal Events

Date Code Title Description
MM4A Annulment or lapse of patent due to non-payment of fees