TW503623B - Delay lock circuit using bisection algorithm and related method - Google Patents
Delay lock circuit using bisection algorithm and related method Download PDFInfo
- Publication number
- TW503623B TW503623B TW90119702A TW90119702A TW503623B TW 503623 B TW503623 B TW 503623B TW 90119702 A TW90119702 A TW 90119702A TW 90119702 A TW90119702 A TW 90119702A TW 503623 B TW503623 B TW 503623B
- Authority
- TW
- Taiwan
- Prior art keywords
- clock
- delay
- time
- period
- correction
- Prior art date
Links
Landscapes
- Stabilization Of Oscillater, Synchronisation, Frequency Synthesizers (AREA)
- Pulse Circuits (AREA)
Abstract
Description
發明之領域: 〜本發明係提供一種延遲鎖定的方法及相關電路,尤於 種以一分法將時脈鎖定至同步的方法及相關電路。 背景說明: 在資訊發達的今日社會,數位資訊的處理、交流與傳 助 夺層面,舉凡仃動電活、個人數位FIELD OF THE INVENTION: The present invention provides a method and a related circuit for delay locking, and more particularly, a method and a related circuit for locking a clock to synchronization in a one-point method. Background note: In today's information-developed society, digital information processing, communication, and assistance are gaining momentum.
Personal Digital Assistant)、可連接至網 的資 sfl 豕電(IA’ Information Applicant)乃至於個 電腦等的數位裝置,無不是用來方便數位資訊的處理與 1寻播。 人在以數位裝置傳播、交換與處理數位資訊時,都要配 又時脈之觸發’才能處理序列之數位訊號。舉例來說,個 電腦中的中央處理器就要以時脈觸發來協調中央處理器 各部份數位電路之資料存取及處理。另外,在行動電話 ’也要以時脈來觸發數位資訊的發送與傳輸;要以行動 ”治做為一接收端接收數位訊號的時候,要先根據接收到Personal Digital Assistant), Internet-connected digital devices (IA ’Information Applicant), and even digital devices such as computers are used to facilitate the processing and broadcasting of digital information. When people use digital devices to propagate, exchange, and process digital information, they must be equipped with a clock trigger to process digital signals in a sequence. For example, the central processing unit in a computer needs to trigger the clock to coordinate the data access and processing of the digital circuits of the central processing unit. In addition, when a mobile phone ’s clock is used to trigger the sending and transmission of digital information, it is necessary to use the mobile phone as a receiver to receive digital signals.
$數位資訊的時脈先在接收端這一方建立同步之時脈,才 能在接下來交換數位資訊的過程中,正確地收發數位資 &fL 〇The clock of the digital information is first established on the receiving side to synchronize the clock, so that the digital information can be sent and received correctly during the subsequent exchange of digital information & fL 〇
第5頁 503623Page 5 503623
用 鎖相電 方式之 請參考 圖。延 生一同 器16、 其中延 緩衝器 •端則 器22, 來根 路( — , 圖一 遲鎖 步之 一緩 遲器 2 0的 接收 控制 據一時脈產生另一同步時脈之電路通常稱Please refer to the figure for the phase-locked electric method. Delay synchronizer 16, delay buffer • end device 22, to the root (—, Figure 1 Delay lock Step 1 Delay 2 Receiving control The circuit that generates another synchronous clock according to one clock is usually called
Phase Lock Loop circuit)。鎖相電路實施 就是延遲鎖定(DLL,Delay-Lock Loop)電路 。圖一為一習知之延遲鎖定電路丨〇之功能方夫 疋電路10是用來根據一輸入之第一時脈12,) 第二時脈14。延遲鎖定電路10中包含有一延全 衝器(buffer)20、一比較器24與一控制器22 16電連於緩衝器20,具有複數個延遲單元u 輸出端回饋至比較器24的一端,比較器24的J 第一時脈1 2。比較器2 4的輸出端2 5電連至控, 器2 2則會控制延遲器1 6。Phase Lock Loop circuit). The implementation of the phase-locked circuit is a Delay-Lock Loop (DLL, Delay-Lock Loop) circuit. Figure 1 shows the function of a conventional delay lock circuit. The circuit 10 is used to input a first clock 12 according to an input, and a second clock 14. The delay lock circuit 10 includes a delay buffer 20, a comparator 24, and a controller 22 16 electrically connected to the buffer 20, and has a plurality of delay units. The output terminal of the delay unit u feeds back to one end of the comparator 24. The first clock of J of the device 24 1 2. The output terminal 2 5 of the comparator 2 4 is electrically connected to the controller, and the comparator 2 2 will control the delayer 16.
第6頁 503623 五、發明說明(3) 延遲鎖定電路1 〇之工作情形可描述如下。當第一時脈 1 2輸入至延遲器i 6後,延遲器1 6可將第_時脈丨2延遲一定 時間以產生另一個時脈,即為第二時脈1 4。由延遲器丨6產 生的第二時脈1 4輸入緩衝器2 0後,會由緩衝器2 〇增加其電 流輸出能力,並將其輸出。為了要使第二時脈丨4與第一時 脈1 2同步,延遲鎖定電路丨〇有一回饋修正之機制。在此機 制中,第一時脈1 2與第二時脈1 4會再輸入至比較器24,比 較器24會比較第一時脈與第二時脈對應週期是否同步,比 較之結果則輸出至控制器2 2。控制器2 2可依據比較器2 4比 較的結果,控制延遲器丨6以其中的延遲單元丨8改變^二時 脈14的延遲時間,以修正第二時脈H與第一時脈12間的延 遲,間。在延遲器16中的複數個延遲單元18中,每一個延 遲f兀1 8都能延遲第二時脈丨4一固定的延遲單位時間;延 遲裔1 6啟用不同數目的延遲單元1 8,就能改變第二脈i 4 延遲之時間。 、、 上述修正第二時脈14以使其與第一時脈12同步之方 法’,可用圖二之流程圖來進一步描述。請參考圖二。圖二 為習知延遲鎖定電路10改變第二時脈14之延遲時間以使其 _ 鎖定第一時脈丨2並與之同步的流程圖。圖二之流程有下列 步驟: 驟2 6、·開始延遲鎖定的流程。也就是開始調整第二時脈 1 4的延遲時間,使其鎖定第一時脈1 2並與之同步的過程。Page 6 503623 V. Description of the invention (3) The operation of the delay lock circuit 10 can be described as follows. After the first clock 12 is input to the delayer i 6, the delayer 16 can delay the _th clock 丨 2 by a certain time to generate another clock, which is the second clock 14. After the second clock 14 generated by the delayer 6 is input to the buffer 20, the current output capability of the buffer 20 will be increased and output. In order to synchronize the second clock 4 with the first clock 12, the delay lock circuit 丨 0 has a feedback correction mechanism. In this mechanism, the first clock 12 and the second clock 14 are input to the comparator 24 again, and the comparator 24 compares whether the corresponding periods of the first clock and the second clock are synchronized, and the comparison result is output. To the controller 2 2. The controller 22 can control the delayer 6 according to the comparison result of the comparator 2 4 to change the delay time of the second clock 14 with the delay unit 8 and modify the time between the second clock H and the first clock 12 The delay, between. In the plurality of delay units 18 in the delay unit 16, each delay f18 can delay the second clock 丨 4 a fixed delay unit time; the delay 16 enables a different number of delay units 18, so Can change the delay time of the second pulse i 4. The above method of modifying the second clock 14 to synchronize it with the first clock 12 'can be further described by using the flowchart of FIG. Please refer to Figure 2. FIG. 2 is a flowchart of the conventional delay lock circuit 10 changing the delay time of the second clock 14 so that it _ locks on the first clock 2 and synchronizes with it. The process in Figure 2 has the following steps: Step 2 6. Start the process of delayed lock. That is, the process of adjusting the delay time of the second clock 14 to lock the first clock 12 and synchronize with it.
五、發明說明(4) 一開始的時候,控制 各時脈週期之觸發, 初始值)後,產生第 鎖定電路1 〇電路本身 再加上延遲時間的初 週期間的延遲時間也 同步。 步驟28·•產生第二時 脈1 4送到比較器2 4, 器2 4可判斷兩時脈對 脈1 21—週期與第二曰 係。比較之後,比較 器22會控制延遲器16根據第一時脈12 在延遲一固定時間(稱為延遲時間的 二時脈1 4的對應時脈週期。因為延遲 有未知的系統延遲(systen] dehyj, 始值,第一時脈12與第二時脈14對應 未知;第一時脈與第二時脈也還不能V. Description of the invention (4) At the beginning, after controlling the trigger of each clock cycle (initial value), the lock circuit 10 is generated, and the delay time during the initial cycle of the delay time is also synchronized. Step 28 · • Generates the second clock 14 and sends it to the comparator 24. The clock 24 can determine the two clock pairs 1 21-period and the second clock. After the comparison, the comparator 22 will control the delayer 16 to delay the corresponding clock period according to the first clock 12 by a fixed time (called the two clocks 14 of the delay time. Because the delay has an unknown system delay (systen) dehyj The initial value, the correspondence between the first clock 12 and the second clock 14 is unknown; neither can the first clock or the second clock
I 器22,表示第一時脈 應週期。本步驟即是 收1 4對應週期間的領 則到步驟3 0 ;若否, 脈1 4後,就能將第一時脈丨2與第二時 比較兩時脈對應週期間的關係。比較 應週期間的先後關係,也就是第一時 手脈1 4對應週期間的領先/落後之關 器2 4可發出一對應之比較訊號給控制 1 2—週期是否領先第二時脈丨4中之對 由比較器2 4比較第一時脈1 2與第二時 先/落後關係。若第一時脈1 2領先, 則到步驟3 2。 步驟30 :控制器22接收比較器24之比較訊號後,會判斷要 如何調整第二時脈1 4的延遲時間。由步驟3〇進行到此步 驟’控制器2 4會將啟動的延遲單元1 8減少一個。如前所 述’每個延遲單元1 8可延遲第二時脈丨4一延遲單位時間 (以下稱為d t;舉例來說,1 d t可以是一億分之一秒)。 少啟動一延遲單元1 8,就會使第二時脈1 4的延遲時間減少 一延遲單位時間(d t )。在完成本步驟,調整完第二時脈1 4 的延遲時間後,整個流程會遞回至步驟2 8,再重新比較一I device 22 indicates the first clock response period. This step is to receive the instructions for the week period corresponding to 14 and go to step 30; if not, after pulse 14, you can compare the relationship between the two clock periods corresponding to the first clock 2 and the second clock. Compare the sequential relationship during the week, that is, the first time hand pulse 1 4 corresponds to the leading / lagging gate 2 during the week 2 4 can send a corresponding comparison signal to the control 1 2-whether the cycle is ahead of the second clock 丨 4 The middle pair compares the first clock 12 and the second clock advance / lag relationship by the comparator 24. If the first clock 12 is leading, go to step 32. Step 30: After receiving the comparison signal from the comparator 24, the controller 22 determines how to adjust the delay time of the second clock 14. From step 30 to this step, the controller 24 will reduce the activated delay unit 18 by one. As mentioned previously, 'Each delay unit 18 can delay the second clock and delay a unit time (hereinafter referred to as d t; for example, 1 d t can be one hundred millionth of a second). Starting the delay unit 18 less often reduces the delay time of the second clock 14 by a delay unit time (d t). After completing this step and adjusting the delay time of the second clock 1 4, the whole process will be returned to step 2 8 and then compared again.
503623 五、發明說明⑸ —-—-- =第一時脈12與第二時脈14。 ί 5 It與步驟3〇相似,本步驟也是要依據步驟28比較的 φ枷f 一時脈1 4的延遲時間做出對應的調整。在此步驟 :^控制器22會控制延遲器1 6多啟用一個延遲單元,將第 :日太=的延遲時間增加一延遲單位時間。同步驟3 0,再完 =驟之後,整個流程會再度遞回至步驟2 8,重複「比 調整」的過程,逐次修正第二時脈14的延遲時間,最 κ曰使第二時脈1 4與第一時脈1 2同步。 一 以:將以一例來說明上述流程的執行過程。請參考圖 二。圖三為習知延遲鎖定電路丨〇使第二時脈同步於^ 一時 脈的過耘中,第一時脈丨2與第二時脈丨4的波形圖。圖三的 橫軸為時間,縱軸為波形幅度的大小。因為習知技術^定 1遲所需時間很長,第一時脈丨2與第二時脈丨4整個修正過 程的波形被分為三部份,故圖三中可見六個波形;其中波 形12a在標示A1處下接波形12b,波形ua在標示A2處下接 波形14b·,同理波形1213在B1處丁接波形i2c,波形Ub在標 示B2處下接波形14c,以此類推。分別在標示八卜M處組* 合波形12a、12b及12c就是第一時脈12在時間〇到時間28T 間的波形,同理波形1 4 a、1 4 b及1 4 c則完整表示了第一時 脈14在時間〇到時間28T間的波形。請先參考圖三中的波形 12a與14a。波形12a是第一時脈12在時間〇到時間1〇τ間的y 波形;其中T代表第一時脈1 2—週期的時間(如波形 時距1 2 p所標示)。請注意延遲鎖定電路1 〇是要根據第一503623 V. Description of the invention-------- = the first clock 12 and the second clock 14. ί 5 It is similar to step 30. This step is also to make corresponding adjustments according to the delay time of φ 枷 f clock 1 14 compared in step 28. In this step, the controller 22 will control the delayer 16 to enable one more delay unit, and increase the delay time of the first day of the day by a delay unit time. Same as step 30, after completing = step, the whole process will be re-transmitted to step 2 8 again, repeat the "ratio adjustment" process, and successively modify the delay time of the second clock 14 to make the second clock 1 4 is synchronized with the first clock 1 2. One: An example will be used to explain the implementation process of the above process. Please refer to Figure 2. FIG. 3 is a waveform diagram of a conventional delay lock circuit, which synchronizes the second clock with the clock of the first clock, the first clock, the second clock, and the second clock. The horizontal axis in Figure 3 is time, and the vertical axis is the magnitude of the waveform amplitude. Because the conventional technique takes a long time to set a delay, the waveforms of the entire correction process of the first clock 2 and the second clock 4 are divided into three parts, so six waveforms can be seen in Figure 3; Waveform 12a is followed by waveform 12b at mark A1, waveform ua is followed by waveform 14b · at mark A2, and waveform 1213 is followed by waveform i2c at B1, waveform Ub is followed by waveform 14c at mark B2, and so on. The combined waveforms 12a, 12b, and 12c at the marked position M are the waveforms of the first clock 12 between time 0 and time 28T. Similarly, the waveforms 1 4 a, 1 4 b, and 1 4 c are completely shown. The waveform of the first clock 14 between time 0 and time 28T. Please refer to the waveforms 12a and 14a in Figure 3. The waveform 12a is the y waveform of the first clock 12 between time 0 and time 10 τ; where T represents the time of the first clock 12-cycle (as indicated by the waveform time interval 12 p). Please note that the delay lock circuit 1 〇
第9頁 ^UJ623 五、發明說明(6) K it子㊁一 f步的第二時脈1 4,<乍為依據的第-時脈 延i睡門疋从^的,延遲鎖定電路1 〇會調整第二時脈1 4的 rr f,故,、波形之週期不固定。波形14a則是第二時 脈1 4在時間〇到時間1 0 T間的波形。 f延遲鎖定電路1〇開始運作時,延遲器16會依據第一 的^、展性週期之波形升緣(rising edge)的觸發,在一定 i? 、、間後,產生第二時脈1 4中的對應週期。像在波形 Η ^\週^期2〇1&的升緣會觸發延遲器16在一定的延遲時 gf/ih門& $波形Ua中的對應週期2〇lb;週期201a與週期 理,i幵;12發中對/關係,在圖三中就用箭帛c來表示。同 第一時^ 週期2〇2a會觸發波形14a+的週期2 0 2b; 203b’等"等。、週期2〇3時觸發第二時脈14中的週期 期2〇=遲:上6依據週期201 a之觸發產生第二時脈中之週 』zuib後,比較器24會 士〆 Γ=‘ -Γ。相互 :產時::,2。城互間的關 第二時脈時,是以第一較為24在比較第一時脈與 的週期2 0 2a做為_參考週==^週期201a—個週期(1T) 2〇U之週期201b比較考;^較脈中對應於週期 同步。以圖三中的例子來乂 蚪脈與第二時脈是否 Λ 第一時脈中的週期2 〇 1 b落後 五、發明說明(7) :先^脈::週期2〇2a,所以比較器24會判定第-時脈1 2 - “。如前所述’延遲器16在-開始要依據第 因n!:n〇la之觸發產生Ϊ二時脈中週期2m時; 使i it期2 η ϊ沏延遲加上延遲态本身延遲時間之初始值, 像在圖三中的例子, 15a所指示f/Λ 2ΐΛ25(Η的時間差(如標示 的=ΐ:落後的關係,並不能量化地量測出兩者ΐ 的乙遲時間(即兩週期升緣間的時間差距)。 脈12ίίί 24i,較週期2〇2a與週期2〇ib後,判斷第二時 砗門ί 個延遲早兀18,以將第二時脈的延遲 :(之就觸γ產驟生3;) 遲:對T會因延遲⑽啟動 考…週期2_=’:會 以7 1二::當然’比較器24再度比較週期2〇3a與202b f (再度進仃步驟28),會發現第一時脈 ;ινΛί?〇) 24rr,m€iii6#^^:^^ 兀18(步驟30)。接下來延遲器16會以第一時脈 2 03a之觸發產生週期203b,而這兩個周期間的時間差又會 五、發明說明(8) 7 ί 3 ί延遲單元而減少ldt,相對地使週期204a |少id而變成2ϋ參考週期)與週期203b間升緣的時間差減 時S ΐ】2,遲鎖定電路10不斷比較一修正的過程,第 |逐次缩小:7二f參考周期間的時差也會以1dt為單位 時距Udt)來調W為習主知延遲鎖定電路1()每次都以一定的 l·中,延遲ί苐一時脈中各週期的延遲時間。如在圖 !ίίΐ參考週期2山間升緣的時間差距已縮:至?i ,週期220b與對應參考週期2仏之間的時 H知為6dt。等到時間26T時,週期226瞒發之週期 j即丄Ό ,此時第二時脈H的週期會和週f 7 Y。廷表示在逐次修正後,延遲器i 2所]-&同 數量,使其在由第一時脈週期升緣觸發 4單兀的 入之延遲時間剛好為一週期。如;=第時引 :時脈週期之升緣觸發第二時脈對應週期之二:二6在:第 為未知的系統延遲,加上延遲器1 6本身可 、’寸,曰因 延遲單元1 8,而使得第一時脈與第二時脈對2 ,數量的 延遲時間。一開始延遲器16無法得知系‘ 週期間有一 遲單元之數目也是初始猜測值,故此時苐—日士矿,用的延 脈無法同步。在接下來的過程中, ^脈與第二時 | 延鎖疋電路ίο會重複 503623 五、發明說明(9) 步驟2 8及3 0 (或3 2)來逐次修改第一時脈與第二時脈對應 週期間的延遲時間;最後使延遲器1 6啟用延遲單元1 8之數 量所造成的延遲時間,加上未知的系統延遲兩者共同造成 的延遲時間恰為第一時脈的一個週期γ。延遲器1 6後續再 依據第一時脈週期觸發所產生的第二時脈週期,就能和第 時脈同步了。接下來延遲鎖定電路1 Q仍會重複「比較一 修正」的過程,以隨時修正系統干擾的影響·,但此時延遲 器1 6多半只會再增加或減少一兩個啟用的延遲^元,以小 幅修正系統引入的干擾。 上述驾知技術的缺點,就是要花較長的時間才能修正 第二時脈使其與第一時脈同步。這是因為習知 中一次 僅能修正一單位延遲時間(即ldt)。為了要辦進時 脈鎖定於第一時脈的精確度,單位延遲時間必曰 夠小, 以避免量化,差(quantization err〇r)。承圖y三之例,若 第二時脈與第一時脈對應週期間的延遲時 麼不管如何修正’第二時脈與第一時脈】】會那因 这個0. fdt的無差~而無法完全同步。因為數位鎖 定電路每次修正第二時脈之延遲時間時,:处 延遲時間(dm修正的最小單位;而小於」;間 的時間差距(如0.5dt)就成為無法完全修正的誤差 了。換句話說,一單位延遲時間dt的時間 、里 量化誤差的大小。 長短可用來衡里 503623 五、發明說明(ίο) 為了要減少量化誤差,在設計像延遲鎖定電路丨〇這樣 的數位式延遲鎖定電路時,都會盡量縮小單位延遲時間 d t。舉例來说,若圖三中第一時脈一週期τ之時間相當於 10 0dt (即ldt = T/l〇〇),那麼造成第一時脈與第二時脈不 能完全同步之篁化誤差就會小於百分之一週期。若一單位 延遲時間dtlis小為千分之〜週期τ之時間,那麼量化誤差 就會進一步^低至千分之一週期。但單位延遲時間縮小, 就.會使修正第二時脈延遲時間的過程增加。由圖三中可看 出’習知技術在修正第二時脈延遲時間時,每一週期Τ只 能修正(增加或減少)一單位時間的誤差。單位時間越 小,每一週期Τ中修正延遲時間的量也越小;相對的,要 完成整個修正過程所需的時間也就越長(換句話說,需要 更多週期Τ才能將第二時脈修正至與第一時脈同步)。當Page 9 ^ UJ623 V. Description of the invention (6) The second clock 1 of step 1 in K it sub-clock, and the -clock-based delay i sleep gate 疋 from the first, delay lock circuit 1 〇 The rr f of the second clock 14 is adjusted, so the period of the waveform is not fixed. The waveform 14a is a waveform of the second clock 14 between time 0 and time 10 T. f When the delay lock circuit 10 starts to operate, the retarder 16 will generate the second clock 1 4 after a certain period of time, i. Corresponding period in. For example, the rising edge of the waveform ^ \ period ^ 2 will trigger the delayer 16 at a certain delay, gf / ih gate & $ Corresponding period 20b in the waveform Ua; period 201a and period theory, i幵; In 12 rounds, the relationship is represented by the arrow 帛 c in Figure 3. At the same time, the period 002a will trigger the period 2 0 2b; 203b ', etc. of the waveform 14a +. Trigger cycle period 2 in the second clock 14 at cycle 203 = 20: Late: the previous 6 generates a week in the second clock according to the trigger of the cycle 201 a "After the zuib, the comparator 24 will be 〆Γ = ' -Γ. Mutual: Delivery time::, 2. When the second clock is closed between the cities, the first 2 is compared with the period 2 0 2a in the first clock. The reference period == ^ period 201a—a period (1T) and a period of 20U. 201b comparative test; ^ comparative pulse corresponds to cycle synchronization. Take the example in Figure 3 to determine whether the pulse and the second clock are Λ. The period 2 in the first clock is 2 〇1 b. 5. The description of the invention (7): The first pulse:: the period 202a, so the comparator 24 will determine the -clock 1 2-". As mentioned earlier, the" delayer 16 will start at the time of the second clock with a period of 2m according to the trigger of the factor n!: N〇la; make it it phase 2 The initial value of the delay time plus the delay time of the delay state itself, like the example in Figure 3, the time difference of f / Λ 2ΐΛ25 (Η indicated by 15a (as indicated by == ΐ: backward relationship, cannot be quantified) Measure the second delay time between the two (the time difference between the rising edge of the two cycles). Pulse 12ίί 24i, compared with the cycle 2202a and the cycle 20b, determine that the second door is delayed by 18, In order to delay the second clock: (it will touch γ to produce a sudden 3;) Late: the test will be started due to the delay ... Cycle 2_ = ': will be 7 1 2: 2: Of course, the comparator 24 compares again Cycles 203a and 202b f (step 28 again), the first clock will be found; ινΛί? 〇) 24rr, m € iii6 # ^^: ^^ Wu 18 (step 30). Next, the retarder 16 will With the first clock 2 The trigger of 03a generates period 203b, and the time difference between these two weeks will be five. Invention description (8) 7 ί 3 ί delay unit and reduce ldt, relatively make period 204a | less id into 2ϋ reference period) and period The time difference between the rising edges between 203b is reduced by S ΐ] 2. The late-lock circuit 10 continuously compares a correction process, and it is gradually reduced: 7 The time difference during the reference period of 2f will also be adjusted in units of 1dt and time interval Udt). As I know, the delay lock circuit 1 () delays the delay time of each cycle in a clock by a certain l · times. As shown in the picture, the time gap between the rising edge of the mountain between the reference cycle 2 has been reduced to: ? I, the time H between the period 220b and the corresponding reference period 2 仏 is known as 6dt. When the time 26T is reached, the period j which the period 226 conceals is 丄 Ό. At this time, the period of the second clock H will be equal to the period f 7 Y. Ting said that after successive corrections, the retarder i 2]-& has the same number, so that the delay time of triggering the 4-unit entry by the rising edge of the first clock cycle is exactly one cycle. For example, the == Timing: The rising edge of the clock cycle triggers the second clock corresponding cycle: 2: 6 at: the unknown system delay In addition, the retarder 16 itself can be used, "inch," because the delay unit 18 makes the first clock and the second clock pair 2 delay time. The delay 16 can not know the system at the beginning. During this period, the number of late units is also the initial guess. Therefore, at this time, the delay pulses used in the 苐 -Rishi mine cannot be synchronized. In the next process, the ^ pulse and the second time | delay lock circuit will repeat 503623. Description of the invention (9) Steps 2 8 and 30 (or 32) to modify the delay time of the corresponding period of the first clock and the second clock successively; finally, the delayer 16 is enabled by the number of delay units 18 The delay time combined with the unknown system delay is exactly one cycle γ of the first clock. The retarder 16 can then be synchronized with the second clock by triggering the second clock cycle based on the first clock cycle. Next, the delay lock circuit 1 Q will still repeat the "compare one correction" process to correct the impact of system interference at any time, but at this time, the retarder 16 will most likely only increase or decrease one or two enabled delays. Correct the interference introduced by the system by a small amount. The disadvantage of the aforementioned driving technology is that it takes a long time to modify the second clock to synchronize it with the first clock. This is because in practice, only one unit of delay time (ie, ldt) can be corrected at a time. In order to make the clock locked to the accuracy of the first clock, the unit delay time must be small enough to avoid quantization err0r. Following the example of figure y3, if the second clock and the first clock correspond to the delay time in the week period, no matter how to correct the 'second clock and the first clock]] it will be because of this 0. fdt's no difference ~ Without full synchronization. Because each time the digital lock circuit corrects the delay time of the second clock, the delay time (the smallest unit of dm correction; less than ""), the time difference (such as 0.5dt) becomes an error that cannot be completely corrected. In other words, the time of a unit delay time dt, the magnitude of the quantization error. The length can be used to balance 503623. V. Description of the invention (ίο) In order to reduce the quantization error, a digital delay lock such as a delay lock circuit is designed. When the circuit is used, the unit delay time dt will be minimized. For example, if the time of the period τ of the first clock in FIG. 3 is equivalent to 100 dt (that is, ldt = T / 100), then the first clock and If the second clock is not completely synchronized, the error will be less than one hundredth of a cycle. If a unit delay time dtlis is as small as one thousandth to a period τ, then the quantization error will be further reduced to one thousandth Cycle, but the reduction of the unit delay time will increase the process of correcting the second clock delay time. From Figure 3, it can be seen that when the conventional technique is used to modify the second clock delay time, every week The period T can only correct (increase or decrease) the error of a unit time. The smaller the unit time, the smaller the amount of correction delay time in each period of T; relatively, the more time it takes to complete the entire correction process Long (in other words, it takes more cycles T to correct the second clock to synchronize with the first clock). When
然’在現代的貧訊工業中,不僅希望延遲鎖定電路的量化 誤差要小(以精確鎖相),更希望延遲鎖定電路能快速地 將第二時脈鎖定與第〜時脈同步,以提昇資訊處理的效 率。 發明概述: 因此,本發明夕I ^ 主要目的在於提供一種延遲鎖定的方 法及相關電路,能在不秘士晉於Μ , ^ ^ τ ^ ^ 仕不增加里化誤差的情況下,加快延遲 鎖疋之〇 耘,以解決習知技術之缺點。However, in the modern lean industry, not only does the quantization error of the delay lock circuit need to be small (to accurately phase lock), but also the delay lock circuit can quickly synchronize the second clock lock to the first clock to improve Information processing efficiency. Summary of the Invention: Therefore, the main purpose of the present invention is to provide a method for delay locking and related circuits, which can speed up the delay lock without increasing the secret error without increasing the secretion error.疋 之 〇 Yun to solve the shortcomings of conventional technology.
第14頁 503623 五、發明說明(π) 發明之詳細說明 請 塊圖。 之緩衝 數個延 dt。控 相似, 以延遲 裔6 0加 一時脈 64,以 發出一 據接收 產生的 參考圖四。圖四為本發明延 延遲鎖定電路5 0有一延遲器 器6 0、一比較器6 4與一控制 遲單元58,每個延遲單元可 制器62中則有一暫存器65。 本發明之延遲鎖定電路5 〇也 益5 6來產生一第二時脈5 6。 強其電流驅動能力並由緩衝 52與第二時脈54同步,兩時 比較兩時脈是否同步。比較 對應的比較訊號輸出至控制 到的比較訊號來控制延遲器 第二鸣脈5 4,使其能與第一 遲鎖定電 5 6、—電 器6 2。延 弓 I 與習知延 是依據一 第二時脈 器6 0輸出 脈都會再 器6 4會依 器6 2。控 5 6,以調 時脈52同 路5 0之功能方 連於延遲器5 6 遲器56中有複 遲單位時間 遲鎖定電路1 0 第一時脈5 2, 5 6會經過緩衝 。為了要使第 回授至比較器 據比較的結果 制器6 2便可依 整延遲器5 6所 步〇Page 14 503623 V. Description of the invention (π) Detailed description of the invention Please block diagram. It buffers several delays dt. The control is similar to delaying the clock 60 plus one clock 64 to send a data reception. Refer to Figure 4. Fig. 4 shows that the delay lock circuit 50 of the present invention has a delay device 60, a comparator 64 and a control delay unit 58, and each delay unit controller 62 has a register 65. The delay lock circuit 50 of the present invention also benefits 56 to generate a second clock 56. The current driving ability is enhanced and the buffer 52 is synchronized with the second clock 54. At two times, it is compared whether the two clocks are synchronized. The corresponding comparison signal is output to the control comparison signal to control the second pulse 5 4 of the retarder, so that it can be locked with the first delay 5 6,-the electric 62 2. The extension of bow I and the extension of knowledge is based on a second clock device 60 which outputs pulses to device 64 and device 4 2. Control 5 6 in order to adjust the function of clock 52 and 50 0 in the same way. Connected to the delay unit 5 6 delay unit 56 has a delay unit time delay lock circuit 1 0 the first clock 5 2, 5 6 will be buffered. In order to make the first feedback to the comparator, according to the comparison result, the controller 62 can follow the steps of the delayer 56.
I 候,ΐ:” 過的延遲鎖定電路-開始運作的時 所以态所產生的第二妗脈是無法和第一時脈同步的。 戶^ = S知之延遲鎖定電路要以每次修正一單位延遲時間的 T: 逐次修正第二時脈的延遲時間。為了要減少上述修 k程所要花的時間,本發明之延遲鎖定電路在開始運^ 广欲會先進行一初始鎖相程序,以更快的程序,將第二時 二二t f與第一時脈同步。請參考圖五。圖五為就是本發 進行初始鎖相程序之流程圖。初始鎖相程序有下列步I wait, ΐ: "Passed delay lock circuit-when it starts to operate, the second pulse generated by the state cannot be synchronized with the first clock. The user knows that the delay lock circuit must be corrected by one unit at a time. T of the delay time: The delay time of the second clock is modified one by one. In order to reduce the time required for the above repair process, the delay lock circuit of the present invention starts an initial phase lock procedure first. The fast program synchronizes the second clock two two tf with the first clock. Please refer to Figure 5. Figure 5 is the flow chart of the initial phase lock procedure of the present invention. The initial phase lock procedure has the following steps
503623 五、發明說明(12) 驟: 步驟66 ··開始初始鎖相程序。在狂、麗辦a s ^ 作的時候,延遲器56依據第4= J電謂 ,… 砰脈52所產生的第二時脈54 尚未同步,此時便可開始初始鎖相程序。 步驟6 8 :設定一校正時距d t之值a 4 u ^ ^ 進行校正程序71。 之值為一初始值。接下來就要 校正程序7 1 ··校正程序7 1有下列步驟: 步驟7 0 :先以延遲器5 6依據第一眛脱 上 係 » 定的延遲時間Η生對應乂第=發再= 較器、,較第「時脈52與第二時脈54相互=落:的關 成的延遲時間 果,若第一時脒 則進行步驟7 2 Β 步驟7 2 A :將第 間。換句話說, 時脈週期時, 湖與對應之第二 ί Ϊ 5 3遲Ϊ f,ί Ζ延遲器56啟用之延遲單元所造 ^ M⑽避吋間减少一校正時距Dt的時 在延遲益5 6根據第一時脈週期 會少啟用數個延遲單元58,使第一 時脈周期間的延遲時間ig — I. 、义週 的時間1為各延遲單元58可增加—時距Dt 將=間減少一校正時距Dt,延遲化要 兀58之數目就是(Dt/dt)。 夕啟用的延遲單 步驟72B:與步驟m之目的相反,此步 的延遲時間增加一校正時距Dt的時間。也=B將第二時脈 器56根據第—時脈週期產生對應之第二時吐在延遲 了脈週期時,會多 503623 條件來決定是否要 ,就是檢查校正時 校正程序遞回進行 ,就會滿足符合條 正時距Dt的資訊。 定值,則結束條件 例,結束條件一旦 結束整個初始鎖相 則遞回到步驟7 0, 56,使第一時脈週期與對應之第二時脒 增加一校正時距Dt的時間。要增加_ 1 延遲器56要額外多啟動(Dt/dt)個延遲又 時距Dt之值,使其變為原來的一半。與 資訊(例如校正時距Dt的長短,或是^交 次數),也會同時更新,並存入控制^ 五、發明說明(13) 啟用數個延遲單元 週期間的延遲時間 正時距D t的時間, 單元58。 步驟7 4 :更新校正 校正時距Dt有關的 正時距Dt被更新的 62中的暫存器65。 步驟7 6 :以一結束 條件的實施例之一 為校正時距Dt會隨 減短到某一程度時 暫存器65中有關校 的次數已經超過一 件是上述何種實施 步驟78 (也就是要 束條件若未滿足, 7卜 ^滿足結束 鎖相程序也會結束 結束校正程序。結束 距Dt的時間長短。因 而減短,校正時距D七 件。另外,也可檢查 若校正時距Dt被更新 已滿足。不管結束條 已滿足則繼續進行至 程序)。相反的,姓 再次進行校正程序503623 V. Description of the invention (12) Step: Step 66 ·· Start the initial phase lock procedure. When the mad and beautiful office as s ^ work, the delayer 56 according to the 4th = J electric predicate, ... The second clock 54 generated by the bang pulse 52 is not yet synchronized, and the initial phase lock procedure can be started at this time. Step 68: Set a correction time interval d t to a 4 u ^ ^ and perform the correction procedure 71. The value is an initial value. Next, the calibration procedure 7 1 ·· The calibration procedure 7 1 has the following steps: Step 7 0: First use the delay device 5 6 to generate the correspondence according to the predetermined delay time. Compared with the second clock 54 and the second clock 54, the delay time of Guan Cheng is the same as the delay time. If it is the first time, go to Step 7 2 Β Step 7 2 A: Put the first time. In other words In the clock cycle, the lake and the corresponding second ί 5 3 Ϊ f, ί △ delay unit 56 enabled by the delay unit ^ M ⑽ avoidance by a correction time interval Dt in the delay benefit 5 6 according to In the first clock cycle, a few delay units 58 will be enabled, so that the delay time during the first clock cycle ig — I., the time 1 of the Yi Zhou can be increased for each delay unit 58 — the time interval Dt will be reduced by one. When correcting the time interval Dt, the number of delays required to be 58 is (Dt / dt). The delayed single step 72B that is enabled: contrary to the purpose of step m, the delay time of this step is increased by the time of the correction time Dt. Also = B generates the second clock according to the first clock cycle according to the first clock cycle. When the clock cycle is delayed, there will be 503623 more conditions. The decision is whether or not to check the calibration process recursively during the calibration, and it will meet the information that meets the timing interval Dt. If the value is set, the end condition is exemplified. Once the end condition ends the entire initial phase lock, it returns to step 70. 56, increase the time between the first clock cycle and the corresponding second clock by a correction time interval Dt. To increase the _ 1 delay 56 additional delay (Dt / dt) delay and time interval Dt value, so that It becomes half of the original. The information (such as the length of the correction time Dt or the number of crossovers) will also be updated at the same time and stored in the control ^ V. Description of the invention (13) Enable several delay unit cycles. The delay time is the time from the time Dt, unit 58. Step 7 4: Update the register 65 in the time interval Dt in which the correction time Dt is updated. Step 7 6: Implementation with an end condition One example is when the correction time interval Dt will be shortened to a certain degree. The number of relevant calibrations in the register 65 has exceeded one. What is the implementation step 78 above (that is, if the requirements are not met, 7? ^ Satisfying the end of the phase lock program will also end the end of the school Program. The length of time from the end to Dt. Therefore, it is shortened, and the correction time is seven pieces. In addition, you can also check if the correction time Dt is updated and satisfied. No matter the end bar is satisfied, continue to the program). Conversely, Surname process again
條件後,校正程序7 1便停止,整個初始 Q 參考圖六:步說明本發明中初始鎖相程序進行的過程,請 5 2與第1、L圖六為本發明中初始鎖相程序進行時第一時^ /、 一時脈54之波形的時序圖。圖六之橫軸為時間After the conditions, the correction procedure 71 stops, and the entire initial Q is referred to FIG. 6: Steps to explain the process of the initial phase-locking procedure in the present invention. Please refer to FIG. 5 and FIG. 1 and L for the initial phase-locking procedure in the present invention. Timing chart of the waveform of the first clock ^ /, one clock 54. The horizontal axis of Figure 6 is time
第17頁 503623 五、發明說明(14) » 軸為波形幅度之大小。為求圖示清晰,第一時脈5 2盥 時脈54之波形各被分為兩部份^皮形52a是第一脈、5 時間0T到時間9T之間的波形;波形52b為第一時脈52在時 =0T到時間1 5T之間的波形(請注意,為求技術揭露之完 整,波形52a與波形52b有部份週期之波形重複)。將波形 52a與波形52b在標示A1處銜接,就是第—時脈52從時間〇τ 到時間/1 5Τ間的完整波形。同理,波形54a與波形54b在標 不A2處銜接就是第二時脈54的在時間〇τ到時間} 5τ間的完 整波形。類似習知技術之情況,本發明也是以第一時脈之 觸發來產生-第二時脈;第一時脈5 2的週期固定為τ,如時 距52ρ所;b不。第二時脈54的延遲時間會由控制器62控制 延遲器56加以改變,所以第二時脈54的週期會改變。Page 17 503623 V. Description of the invention (14) »The axis is the magnitude of the waveform amplitude. For the sake of clarity, the waveform of the first clock 5 2 and the clock 54 are each divided into two parts. The skin shape 52a is the waveform of the first pulse, 5 time 0T to 9T; the waveform 52b is the first Waveform of clock 52 between time = 0T and time 15T (please note that for complete technical disclosure, waveform 52a and waveform 52b have partial periodic waveform repetitions). The waveform 52a and the waveform 52b are connected at the mark A1, which is the complete waveform of the first clock 52 from time τ to time / 15T. Similarly, the connection of the waveform 54a and the waveform 54b at the mark A2 is the complete waveform of the second clock 54 between time 0τ to time} 5τ. Similar to the case of the conventional technology, the present invention also uses the trigger of the first clock to generate the second clock; the period of the first clock 52 is fixed to τ, as indicated by the time interval 52ρ; b does not. The delay time of the second clock 54 is changed by the controller 62 to control the delay 56 so that the period of the second clock 54 is changed.
々在本發明延遲鎖定電路5〇開始運作時,延遲器56會依 照第一時脈中週期301a (請參考波形52a)之升緣的觸 發,在延遲一定時間後,產生第二時脈中的對應週期 j01b=週期jOla與週期301b間的對應關係,就用圖六中的 成頭^表示。類似習知技術中的情況,在週期3 〇丨a與週 期301b升緣間的時間差距(也就是延遲時間),會包括未 知,二統延遲以及延遲器5 6一開始啟用之延遲單元5 8的數 I i ί兩因素的影響。請注意,為了要和習知技術在圖三 中的^正過程比較,在圖六中週期30 la與30 lb間的延遲時 間:與圖三^週期2〇1碘2〇lb間的延遲時間是相同的(也 就是說,圖三與圖六中開始修正前的條件是一樣的);且々 When the delay lock circuit 50 of the present invention starts to operate, the retarder 56 will trigger on the rising edge of the period 301a (please refer to waveform 52a) in the first clock, and after a certain time delay, it will generate the signal in the second clock. Corresponding period j01b = The corresponding relationship between the period jOla and the period 301b is represented by the header ^ in FIG. Similar to the situation in the conventional technology, the time difference (ie, the delay time) between the rising edge of the period 3 0a and the period 301b will include unknown, secondary delay, and the delay unit 5 8 which is initially activated. The number I i ί affects two factors. Please note that in order to compare with the positive process of the conventional technique in FIG. 3, the delay time between the periods 30a and 30 lb in FIG. 6: Are the same (that is, the conditions before the correction in Figure 3 and Figure 6 are the same); and
第18頁 503623 五、發明綱(⑸ " — 圖六中ί 一延遲單元58之延遲單位時間以的長度也和圖三 中延遲單位時間dt之長度相同。圖六中第一時脈一週期的 時間T也和圖三中第一時脈丨2—週期的時間了長度相同^ 在圖 (d t是一 第一時脈 5 2中落後 來和第二 五中的步 六開始初始鎖 延遲單元58的 與第二時脈領 週期3 0 1 a兩個 時脈5 4中與週 驟7 0)。由圖 時間領先參考 以15dt的 二時脈領先第一時脈 6 2。以落 是因為要 後週期3 0 1 a兩 等週期3 0 1 b之 第一時脈52中的其他週 I較。與習知延遲鎖定電 時脈中週期與第一時 並不量化彼此間延遲時 相程序後’枝正時距D t會設為3 2 d t 單位延遲時間)。當比較器64比較 先/落後的關係時,是以第一時脈 週期之週期3 0 3 a做為一參考週期, 期3 0 1 a對應之週期3 〇丨b比較(即圖 六中可看出,週期301b 週期303a;所以比較器64會判斷第 並發出對應之比較訊號給控制器 個f期的週期303a作為參考週期, 狀態穩定。者麸,;、a a 1 β &田 也可視情況選用 玖二蛑、多考週期’來和週期3 01 b比 r f f況近似,比較器64只比較第 =:j考週期領先/落後的 間的長短。 控制器6 2在接到比較器6 4之 器56在下一次產生第二時脈週龙軏f f後,會控制延遲 ’·並更新校正時距之值匕;尤=丄日寺距_時間長度 65中與校正時距Dt長短有關的同:改變暫存器 貝针(以上即步驟723與74 503623 五、發明說明(16) )。·在本實施例中,結束條件是校正時距之 ldt;故校正程序會遞回至步驟7〇。 、" 在接下來的過程中,延遲哭 期3 0 3 a之觸恭而立山贫 遲态5 6會以弟一時脈5 2中的遇 太用杰失去消如1 r、…〜w I延造日f間3 2 d t,故原 之延f i P1 °只—的f二時脈週期,會因額外增加32dt 之延遲時間,反而落後參考週期17dt。 出,對應於週期3 〇 3 a的週期3 0 3 b,落後夂老^细古 17dt的時間。比較器64比較週 又週』3 0 53有 會判斷第二時脈54落後第_時脈52。一週期3 0 5a後’ 由圖五之流程圖可知 52中的週期30 5a產生對應 動16個延遲單元58,以將 少一校正時距D t的時間, 距Dt已經是前一校正程序 正程序原本是32dt的校正 時間減少1 6 d t後,第二時 17dt延遲時間,會縮減至 期305a之週期305b,和第 時間為1 d t。 9 田延遲器5 6再一次以第一時脈 ,週期30 5b時,延遲器56會少啟 第^時脈對應週期之延遲時間減 也就是16dt。請注意此時校正時 ^校正時距Dt的一半;在前一校 ^距、,現在已經變成l6dt。延遲 脈中週期原本和參考週期間的 idt。由圖六中可知,對應於週 一時脈中參考週期30 7a間的延遲 當延遲器56再次以週期3〇7a之觸發產生第二時脈州Page 18 503623 V. Outline of Invention (⑸) — The length of the delay unit time of a delay unit 58 in FIG. 6 is also the same as the length of the delay unit time dt in FIG. 3. The first clock is one cycle in FIG. The time T is also the same as the length of the first clock in the third cycle in Figure 3. The length of the cycle is the same. ^ In the figure (dt is a first clock 5 2 lags behind and the sixth step in the second fifth starts the initial lock delay unit. 58 and the second clock collar period 3 0 1 a two clocks 5 4 middle and week 7 0). According to the chart time leading reference, the second clock 15dt leads the first clock 6 2. The reason is that The post-cycle 3 0 1 a is equal to the other cycles I in the first clock 52 of the two-equal cycle 3 0 1 b. Compared with the conventional delay-locked electrical clock, the cycle and the first time do not quantify the delay phase program between each other The back branch timing D t is set to 3 2 dt unit delay time). When the comparator 64 compares the first / lagging relationship, the period 3 0 3 a of the first clock cycle is used as a reference period, and the period 3 0 1 a corresponds to the period 3 〇 丨 b comparison (that is, the It can be seen that the period 301b is the period 303a; so the comparator 64 judges and sends the corresponding comparison signal to the controller for the period f period 303a as a reference period, and the state is stable. Bran, aa 1 β & Tian is also visible In the case, “玖 二 蛑, multi-test period” is used to approximate period 3 01 b to rff. Comparator 64 only compares the length of the leading / lagging period of the test period j: controller 6 2 is connected to comparator 6 After the second clock 56 is generated next week, the 4th device 56 will control the delay and update the value of the correction time interval; especially = the time interval of the day of the day_65 is related to the length of the correction time interval Dt Same: Change register buffer (the above is step 723 and 74 503623 V. Description of the invention (16)). · In this embodiment, the end condition is the ldt of the calibration time interval; therefore, the calibration procedure will return to step 7 〇, " In the next process, the delay of crying period of 3 0 3 a respectfully stands up The late state 5 6 will disappear with the younger one in the clock 5 2 and the loss will disappear as 1 r, ... ~ w I delay the day f 3 2 dt, so the original extension fi P1 ° only-f two clock cycle , It will be behind the reference period by 17dt because of an additional delay time of 32dt. Out, it corresponds to the period of 3 003 a and the period of 3 0 3 b, which is behind the age of the old ^ fine ancient by 17 dt. Comparator 64 compares week by week. " 3 0 53 will judge that the second clock 54 is behind the _ clock 52. After one cycle 3 0 5a 'From the flowchart in Figure 5, we can see that the cycle 30 5a in 52 generates 16 delay units 58 to reduce The time between the correction time Dt, the time Dt is already the previous correction program, the original program was originally 32dt, the correction time is reduced by 16 dt, and the second 17dt delay time will be reduced to the period 305a of the period 305a, and the second time It is 1 dt. 9 Tian delayer 5 6 uses the first clock again. At period 30 5b, the delayer 56 will lessen the delay time of the period corresponding to the ^ clock minus 16 dt. Please note that when correcting at this time ^ Corrected the half of the time interval Dt; in the previous school, the distance has now become 16dt. The original period of the delay pulse and the reference period The idt. Seen from FIG six, corresponding to the delay between the periphery 30 7a with reference to a clock cycle delay when the periodic trigger 56 to generate a second clock once again 3〇7a state
503623503623
之對應週期307b時,因為延遲時間你 3〇7b再度以7dt的時間領先參 二,二8疒,使週期 時間又減少-校正時距(此時考二 隨著校正程序71的遞回,當延遲哭 觸發第二時脈54中之對應週期3〇9b時;請及緣 ,开二541〇 ,週期3〇9b會因為延遲時間再度增加二(一校 ΐ 的/μ間),和參考週期3〇9綱的延遲時間又會拉近 至12。週期311a與週期311b的延遲時間會再增加At the corresponding period of 307b, because of the delay time, 307b leads you again by 7dt, and the second time is 8 二, which reduces the cycle time again-the correction time interval (At this time, the test 2 is returned with the correction program 71, when Delayed crying triggers the corresponding period of 309b in the second clock 54; please be in touch, open the second 5410, the period 309b will increase by two (one school / μ interval) due to the delay time, and the reference period The delay time of class 309 will be closer to 12. The delay time of cycle 311a and cycle 311b will increase again.
間),將週期311b與對應參考週期Wa間升 d: Ϊ”至1“。最後在以週期313a產生對應週 ΪΓΡί, 時距會進一步縮減為ldt,使週期313b與 對應參考週期31 5a間的時間差距減為零,第一時脈與第二 ,脈也能同步了。請注意此時校正程序及整個初始鎖相程 序也會隨之結束,因為接下來校正時距已經小於1 dt而滿 足結束條件了。 結束初始鎖相程序後,延遲器5 6所啟用的延遲單元5 8 造成的延遲時間,加上系統延遲的時間,其總和就是.週期 T之整數倍(如圖六中之例即為2 T)。接下來本發明之延 遲鎖定電路5 0會繼續以每個第一時脈週期之升緣觸發第二 日^脈中對應週期的產生(請注意在初始鎖相程序中,延遲 鎖定電路5 0每兩個週期τ才會觸發一次第二時脈中的週期 ),因為延遲器5 6之延遲時間已經調整為週期τ之整數Time), the period 311b and the corresponding reference period Wa are raised by d: Ϊ "to 1". Finally, the corresponding cycle ΪΓΡί is generated with the cycle 313a, and the time interval will be further reduced to ldt, so that the time difference between the cycle 313b and the corresponding reference cycle 31 5a is reduced to zero, and the first clock and the second clock can be synchronized. Please note that the calibration procedure and the entire initial phase-locking procedure will also end at this time, because the next calibration interval is less than 1 dt and the end condition is met. After the initial phase-locking process is completed, the delay time caused by the delay unit 5 8 enabled by the delay unit 5 6 plus the system delay time, the sum of which is an integer multiple of the period T (as shown in the example in Figure 6 is 2 T ). Next, the delay lock circuit 50 of the present invention will continue to trigger the generation of the corresponding period in the second day pulse with the rising edge of each first clock cycle (please note that in the initial phase lock program, the delay lock circuit 50 Only two periods τ will trigger a period in the second clock), because the delay time of the retarder 56 has been adjusted to an integer of the period τ
第21頁 503623 五、發明說明(18) 倍,第二時脈54也得以和第一時脈52同步。去妙 也還是會持續的運作,以使控制器6 2得以修正^ ^ ^ 正 引起的延遲時間誤差;而此種誤差通常僅須小幅度的^ 總結初始鎖相程 圖七為本發明進行初 逐次修正延遲時間之 形E 3外,由上而下排 是週期 301b、3 03b、 波形。週期E 1榡示的 緣的時間;週期E 2是 形E 3則是第一時脈三 30 lb來說,週期E1就 週期3 0 3 a。以此類推 是週期3 0 3a之升緣; 最後,對週期3 1 3 b來 週期3 1 3 a,週期e 2則 出校正時距Dt隨校正 一時脈與第二時脈也 序修正的過程,可進一步參考圖七。 始鎖相程序中,第二時脈54中各週期 不意圖。圖七之橫軸為時間;除了波 列的是第二時脈54中的各週期( 30 7b、3 0 9b、3m及 31;二 疋觸發以上各週期之第一時脈週期升 上述各週期對應參考週期之時間;波 ,期間的波形。換句話說,對週期 是週期301a,週期E2則是對應之參考 ,對週期3 0 3b來說,週期E1的升緣就 ,期E2之升緣就是週期3〇53之升緣。 週期E1會對應至觸發週期3l3b之 是參考週期315a。由圖七中可明顯看 程序之進行而逐次縮小的情形;而第 終於能夠同步。 明在知技術在圖三中修正延遲時間的過程,本 相同週二、#下(相同初始狀況、相同單位延遲時間di 。J T)進行初始鎖相程序而鎖定延遲的過程(示方Page 21 503623 V. Description of the invention (18) times, the second clock 54 is also synchronized with the first clock 52. Demystifying will still continue to operate, so that the controller 62 can correct the delay time error caused by ^ ^ ^; and this error usually requires only a small amount of ^ Summary Initial phase-locking process Figure 7 shows the initial stage of the invention. In addition to the successive correction of the delay time E3, the cycles from the top to the bottom are the cycles 301b, 303b, and the waveform. Period E 1 shows the time of the edge; period E 2 is the shape of E 3 and the first clock 3 is 30 lb. Period E 1 is the period 3 0 3 a. The analogy is the rising edge of the cycle 3 0 3a. Finally, for the cycle 3 1 3 b, the cycle 3 1 3 a, and the cycle e 2 is a process in which the correction time interval Dt is sequentially corrected with the correction of one clock and the second clock. For further reference, refer to Figure 7. In the initial phase lock procedure, each cycle in the second clock 54 is not intended. The horizontal axis of Figure 7 is time; except for the wave train, each cycle in the second clock 54 (30 7b, 3 0 9b, 3m, and 31; the first clock cycle that triggers each of the above cycles rises to each of the above cycles). Corresponds to the time of the reference period; wave, the waveform of the period. In other words, for the period is the period 301a, and the period E2 is the corresponding reference. For the period 3 0 3b, the rising edge of the period E1 is the rising edge of the period E2. It is the rising edge of cycle 3053. The cycle E1 will correspond to the trigger cycle 3l3b and the reference cycle 315a. As can be clearly seen in Figure 7, the progress of the program is reduced one by one; and finally, the technology can finally be synchronized. The process of correcting the delay time in Figure 3. This process of the same Tuesday and # (the same initial condition and the same unit delay time di. JT) is used to perform the initial phase lock procedure and lock the delay (shown in the side
503623 五、發明說明(19) 圖六)顯然快速的多 一及第二 到同樣的 開始便以 正時距來 程序所要 所时論過的,在 是一樣的(即_ 的過程 量化誤 >不但 圖六中 是會以 增加量化誤差 ,就必須 差。相對 能加速鎖 的例子可 一單位延 時間才 的時間 相程序 再逐次 縮短初 如前面 幅度都 定延遲 會增加 距減半 差。由 最後還 故不會 能使第 就能達 中,一 縮小校 始鎖相 習知技 時脈同 效果。 較大的 進行細 花的時 習知技 單位延 增長單 的,本 定延遲 知,在 遲時間 術要以 步,本 這是因 校正時 微的調 間,也 術中, 遲時間 位延遲 發明以 的過程 本發明 dt的幅 26個週期(26T)的 發明之方法只要14τ 為本發明中的初使I貞 距來調整延遲時間, 整,這樣一來不僅能^ 不會增加量化誤差。 每次修正延遲時間的 1 d t),若要加快鎖 時間;但這樣一來就 二分法逐次將校正時 ,也不會增加量化誤 之初始鎖相程序中, 度來修正延遲時間, 以較為定量的顴點也& 就更能凸顯本發明之严^比較習知技術與本發明之技術, 時間的丰t f ^彳t 假設一單位延遲時間為一週期 =:四分之-(即1…mm),則習知技 遲時pU : ϊ 1 i個週期的時間才能修正第二時脈之延 間=其與第-時脈同步。相對地,在本發明中’因為 i=T g刀法逐次改變調整延遲時間的幅度,大約只要10個 中讓第:為底之對數值)京尤能完成初始鎖相程序 ΐ r時與苐一時脈同步之過程。若一單位延遲時間 _減為一週期時間的兩千零四十八分n即503623 V. Description of the invention (19) Figure 6) Obviously fast one and the second to the same start will be discussed with the timing of the program at the same time, which is the same (that is, the process of _ quantization error > Not only in Figure 6 will increase the quantization error, it must be bad. For the example that can accelerate the lock, the time phase program that is delayed by one unit can be successively shortened. Initially, the delay will increase the distance by half the difference. It will not be able to achieve the first time, the first time to reduce the timing and effect of phase-locking skills in the school. Larger fine-practice units with delayed skills will have to delay the growth of orders. Time technology has to be stepped. This is due to the fine tuning between corrections. In the operation, the time delay is delayed. The invention is based on the method of 26 cycles (26T) of dt. As long as 14τ is in the invention, Initially adjust the delay time by adjusting the I time interval, so that not only will not increase the quantization error. Each time you modify the delay time by 1 dt), if you want to speed up the lock time; but in this way, the dichotomy is successively performed. During the correction, the quantization error will not be increased in the initial phase-locking program, and the delay time will be modified. A more quantitative threshold will also highlight the strictness of the present invention. ^ Comparing the conventional technology with the technology of the present invention, The abundance of time tf ^ 彳 t Assume that a unit delay time is one cycle =: one-fourth-(that is, 1 ... mm), then the conventional technique is delayed pU: ϊ 1 i cycle time to correct the second clock Delay = It is synchronized with the -clock. In contrast, in the present invention, 'since the i = T g knife method successively changes the magnitude of the adjustment delay time, it only takes about 10 out of 10 logarithmic values.] Jingyou can complete the initial phase-locking procedure 时 r and 苐The process of clock synchronization. If a unit delay time _ is reduced to 2048 minutes of a cycle time n, then
第23頁 503623Page 23 503623
五、發明說明(20) ldt = T/2048),習知技術會需要大約1 0 24個週期的時間 能使第二時脈與第一時脈同步;在相同情況下,使用二分 法的本發明僅須11個週期(2048以2為底之對數值)就_能^ 完成初始鎖相程序,而且也不會增加量化誤差。總而:匕 之’本發明能在不增加量化誤差的情況下,大幅加速鎖定 延遲的過程;不但能確保數位式延遲鎖相電路鎖相的精疋 確,更能提昇效率,減少浪費在鎖相過程反覆修正 = 的時間。 因為本發明中的初始鎖相程序會在每一次校正程 $新校正時距之值,本發明中之延遲鎖定電路5 〇會以 器6 5,儲存校正時距的相關資訊,並據以判斷校正程 f 否應,結束。既然本發明是以二分法來控制校正時距= 縮減(即每次將校正時距縮為原來的二分之一),就 ^ 直接用二進位的方式來暫存有關校正時距的資訊。^j 中的例子來說,一開始校正時距Dt = 32dt,可以在 ": 個位元(Μί)記為〇〇〇〇〇;校正時距Dt減半而 8dt後/智’户暫存器65之值可記為1〇 0 0 0。校正時距Dt變成 4dt^ t 春: 成11110 π為1 1100,权正時距為2dt時,暫存器65之值變 ίϋ校t時距Dt減半為,暫存器65之值 而此時校正程序也將要結束。以上述這種 方法,就能以暫存器65來管理本發明中的初始鎖相V. Description of the invention (20) ldt = T / 2048), the conventional technique will take about 10 24 cycles to synchronize the second clock with the first clock; in the same case, the bisection method is used. The invention only needs 11 cycles (the logarithm of 2048 to 2) to complete the initial phase-locking procedure, and it will not increase the quantization error. In summary: the invention can significantly accelerate the lock delay process without increasing the quantization error; it can not only ensure the accuracy of the phase lock of the digital delay phase lock circuit, but also improve the efficiency and reduce the waste of lock Phase process repeated correction = time. Because the initial phase-locking procedure in the present invention will calculate the value of the new time interval in each correction process, the delay lock circuit 5 in the present invention will store the information about the time interval of the correction and judge it accordingly If the calibration process f is no, the process ends. Since the present invention uses a dichotomy to control the correction time interval = reduction (that is, each time the correction time interval is reduced to one-half of the original), ^ directly uses the binary method to temporarily store the information about the correction time interval. ^ j For example, at the beginning of the correction time interval Dt = 32dt, can be recorded in the "quot :: bits (Μί) as 〇〇〇〇〇〇; the correction time interval Dt is halved and after 8dt The value of the register 65 can be recorded as 1000 00. The correction time interval Dt becomes 4dt ^ t spring: when 11110 π is 1 1100 and the weighted time interval is 2dt, the value of the register 65 becomes halved, and the value of the register time Dt is halved to the value of the register 65. The time calibration procedure will also end. In this way, the initial phase lock in the present invention can be managed by the register 65
503623 五、發明說明(21) 以上所述僅為本發明之較佳實施例,凡依本發明申請 專利範圍所做之均等變化與修飾,皆應屬本發明專利之涵 蓋範圍。例如本發明中校正程序7 1的數個步驟(步驟7 0、 74、76等)在合理修改後可互換執行之順序,皆應屬於本 發明之均等變化。503623 V. Description of the invention (21) The above description is only a preferred embodiment of the present invention. Any equivalent changes and modifications made in accordance with the scope of the patent application for the present invention shall fall within the scope of the invention patent. For example, the order in which several steps (steps 70, 74, 76, etc.) of the calibration program 71 in the present invention can be executed interchangeably after being reasonably modified should all belong to the equal change of the present invention.
第25頁 503623 圖式簡單說明 圖示之簡單說明: 圖一為一習知延遲鎖定電路之功能方塊圖。 圖二為用於圖一延遲鎖定電路之方法的流程圖。 圖三為圖一中延遲鎖定電路之第一時脈與第二時脈波 形之時序圖。 圖四為本發明延遲鎖定電路之功能方塊圖。 圖五為用於圖四中延遲鎖定電路之方法的流程圖。 圖六為圖四中延遲鎖定電路之第一時脈與第二時脈波 形之時序圖。 圖七為本發明方法修正延遲時間之示意圖。 圖示之符號說明: 50 本發明之延遲鎖定電路 52 第一時脈 54 第二時脈 56 延遲器 58 延遲單元 60 緩衝器 62 控制器 64 比較器 65 暫存器 71 校正程序 C 箭頭 A卜A2 標示 T 週期 d t 單位延遲時間 52a 、52b、 54a、 54b、 E3 波形 6 6^ 68、 70、 72A、 72B、 74、 76 〜78 步驟 豪Page 25 503623 Brief description of the diagram Brief description of the diagram: Figure 1 is a functional block diagram of a conventional delay lock circuit. FIG. 2 is a flowchart of a method for the delay lock circuit of FIG. 1. FIG. Figure 3 is a timing diagram of the first clock and the second clock waveforms of the delay lock circuit in Figure 1. FIG. 4 is a functional block diagram of the delay lock circuit of the present invention. FIG. 5 is a flowchart of a method for the delay lock circuit in FIG. 4. Figure 6 is a timing diagram of the first clock and second clock waveforms of the delay lock circuit in Figure 4. FIG. 7 is a schematic diagram of a modified delay time of the method of the present invention. Explanation of the symbols in the figure: 50 the delay lock circuit 52 of the present invention 52 the first clock 54 the second clock 56 the delay device 58 the delay unit 60 the buffer 62 the controller 64 the comparator 65 the temporary register 71 the calibration program C arrow A and A2 Mark T period dt Unit delay time 52a, 52b, 54a, 54b, E3 waveform 6 6 ^ 68, 70, 72A, 72B, 74, 76 ~ 78 Step How
第26頁 503623 圖式簡單說明 301a 313a 3 0 9b 303a、 305a、 307a、 309a、 311a 315a、 301b、 303b、 305b、 307b 311b、 313b、 E卜 E2 週期Page 26 503623 Schematic description 301a 313a 3 0 9b 303a, 305a, 307a, 309a, 311a 315a, 301b, 303b, 305b, 307b 311b, 313b, E2 cycle
第27頁Page 27
Claims (1)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW90119702A TW503623B (en) | 2001-08-10 | 2001-08-10 | Delay lock circuit using bisection algorithm and related method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
TW90119702A TW503623B (en) | 2001-08-10 | 2001-08-10 | Delay lock circuit using bisection algorithm and related method |
Publications (1)
Publication Number | Publication Date |
---|---|
TW503623B true TW503623B (en) | 2002-09-21 |
Family
ID=27607895
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
TW90119702A TW503623B (en) | 2001-08-10 | 2001-08-10 | Delay lock circuit using bisection algorithm and related method |
Country Status (1)
Country | Link |
---|---|
TW (1) | TW503623B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11288724B2 (en) | 2016-02-26 | 2022-03-29 | Nike, Inc. | Method of customizing stability in articles of footwear |
-
2001
- 2001-08-10 TW TW90119702A patent/TW503623B/en not_active IP Right Cessation
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11288724B2 (en) | 2016-02-26 | 2022-03-29 | Nike, Inc. | Method of customizing stability in articles of footwear |
US12008617B2 (en) | 2016-02-26 | 2024-06-11 | Nike, Inc. | Systems for producing articles of footwear having customized stability |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7970092B2 (en) | Phase comparator and regulation circuit | |
TWI448081B (en) | All-digital clock correction circuit and method thereof | |
TW455871B (en) | Apparatus and method for accelerating initial lock time of delayed locked loop | |
JP2002100982A (en) | Dll circuit | |
KR100408727B1 (en) | Clock synchronization device | |
TWI304685B (en) | Clock generation circuit and clock generation method | |
US8847641B2 (en) | Phase comparison device and DLL circuit | |
JP2976198B2 (en) | Clock synchronization circuit | |
KR20170094036A (en) | Duty correction circuit | |
US7667510B2 (en) | Delay locked loop circuit and method thereof | |
TW503623B (en) | Delay lock circuit using bisection algorithm and related method | |
KR100839499B1 (en) | Apparatus of controlling a delay and method thereof | |
KR100925393B1 (en) | Domain Crossing Circuit of Semiconductor Memory Apparatus | |
TW201316150A (en) | Multi-phase clock generation system and clock calibration thereof | |
JP2013005050A (en) | Clock generation device and electronic apparatus | |
KR20110133781A (en) | Duty correcting circuit having a duty detector, delay-locked loop circuit having the duty correcting circuit and method of duty correction | |
KR101013920B1 (en) | Frequency Multiplication System and Control Method of the Same | |
KR20080002590A (en) | Delay locked loop circuit | |
KR20170027913A (en) | Integrated circuit | |
US20240106444A1 (en) | Low jitter clock multiplier circuit and method with arbitary frequency acquisition | |
KR20090121469A (en) | Semiconductor memory device | |
KR20240145647A (en) | Apparatus for correcting time skew | |
JP2692071B2 (en) | Phase-locked pulse generator | |
JPH07273648A (en) | Pll circuit | |
JP3505479B2 (en) | Multiple synchronizers and clock branching / dividing devices |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
GD4A | Issue of patent certificate for granted invention patent | ||
MK4A | Expiration of patent term of an invention patent |