[go: up one dir, main page]

TWI764297B - accumulator circuit - Google Patents

accumulator circuit

Info

Publication number
TWI764297B
TWI764297B TW109133293A TW109133293A TWI764297B TW I764297 B TWI764297 B TW I764297B TW 109133293 A TW109133293 A TW 109133293A TW 109133293 A TW109133293 A TW 109133293A TW I764297 B TWI764297 B TW I764297B
Authority
TW
Taiwan
Prior art keywords
circuit
bit
circuits
signal
memory
Prior art date
Application number
TW109133293A
Other languages
Chinese (zh)
Other versions
TW202121416A (en
Inventor
林金溪
Original Assignee
大陸商珠海南北極科技有限公司
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 大陸商珠海南北極科技有限公司 filed Critical 大陸商珠海南北極科技有限公司
Priority to TW109133293A priority Critical patent/TWI764297B/en
Publication of TW202121416A publication Critical patent/TW202121416A/en
Application granted granted Critical
Publication of TWI764297B publication Critical patent/TWI764297B/en

Links

Images

Landscapes

  • Dram (AREA)

Abstract

一種累加電路,包括一信號偵測器及計數器以及多個基本電路。該多個基本電路連接信號偵測器及計數器,並且分別接收多個信號。該多個基本電路的第M個基本電路所接收的信號為第一準位時,該第M個基本電路提供一預設的反應資料至該信號偵測器及計數器,使該信號偵測器及計數器的計數值增加以累計具有第一準位的信號的數量。An accumulating circuit includes a signal detector, a counter and a plurality of basic circuits. The plurality of basic circuits are connected to the signal detector and the counter, and respectively receive a plurality of signals. When the signal received by the Mth basic circuit of the plurality of basic circuits is at the first level, the Mth basic circuit provides a preset response data to the signal detector and the counter, so that the signal detector and the count value of the counter is increased to accumulate the number of signals having the first level.

Description

累加電路accumulator circuit

本發明係有關一種累加電路,特別是關於一種架構簡單且面積小的累加電路。The present invention relates to an accumulating circuit, in particular to an accumulating circuit with a simple structure and a small area.

在記憶體中,會因為在製造過程中的各種非理想因素而產生不佳記憶體位元,因此最初設計記憶體時,會利用內部的部分空間設置多餘記憶體位元,當記憶體經由測試發現具有不佳記憶體位元時,可以使用多餘記憶體位元取代這些不佳記憶體位元,以發揮修復效能,提升記憶體的製造良率。傳統的修復記憶體方式是在設計時加入多餘列(row redundancy)或多餘行(column redundancy)的輔助電路,以在發現記憶體中具有不佳記憶體位元時,可以取代包含該不佳記憶體位元的行或列。In the memory, bad memory bits will be generated due to various non-ideal factors in the manufacturing process. Therefore, when the memory is initially designed, some of the internal space is used to set up redundant memory bits. When the memory is tested and found to have When the bad memory bits are bad, the bad memory bits can be replaced by the surplus memory bits, so as to exert the repairing effect and improve the manufacturing yield of the memory. The traditional way of repairing memory is to add auxiliary circuits for row redundancy or column redundancy in the design, so that when a bad memory bit is found in the memory, it can replace the bad memory bit. A row or column of elements.

然而,傳統的方法是替換一整行或一整列的記憶體位元,因此除了替換不佳記憶體之外,也會替換一行或一列中的良好記憶體位元,造成浪費。此外隨著記憶體尺寸縮小及記憶體容量增大,記憶體位元陣列密度隨之提高,因此不佳記憶體位元的數量也隨之增多,設置多餘行或多餘列的空間需求也變多,導致若要配置足夠的多餘行或多餘列的記憶體位元進行替換,則所需額外的面積相當龐大。因此替換一整行或一整列的記憶體位元的方式已不適用在小體積且大容量的記憶體中。相對於傳統方法,本發明使用位元取代位元方法,在與傳統方法具有相同或更少多餘記憶體的情況下,達成更有效率的修補效果。However, the conventional method is to replace an entire row or column of memory bits, so in addition to replacing bad memory, good memory bits in a row or column are also replaced, causing waste. In addition, as the memory size shrinks and the memory capacity increases, the density of the memory bit array increases, so the number of bad memory bits also increases, and the space requirement for setting redundant rows or columns also increases, resulting in To allocate enough extra rows or extra columns of memory bits to replace, the extra area required is quite large. Therefore, the method of replacing a whole row or a whole column of memory bits is not suitable for a small-volume and large-capacity memory. Compared with the conventional method, the present invention uses the bit-to-bit method to achieve a more efficient repair effect under the condition of having the same or less redundant memory as the conventional method.

本發明的目的之一,在於提出一種累加電路。One of the objectives of the present invention is to provide an accumulating circuit.

根據本發明,一種累加電路,包括一信號偵測器及計數器以及多個基本電路。該多個基本電路連接信號偵測器及計數器,並且分別接收多個信號。該多個基本電路的第M個基本電路所接收的信號為第一準位時,該第M個基本電路提供一預設的反應資料至該信號偵測器及計數器,使該信號偵測器及計數器的計數值增加以累計具有第一準位的信號的數量。其中該多個基本電路是用邏輯閘組成,故累加電路的架構較簡單且面積較小。According to the present invention, an accumulating circuit includes a signal detector and a counter and a plurality of basic circuits. The plurality of basic circuits are connected to the signal detector and the counter, and respectively receive a plurality of signals. When the signal received by the Mth basic circuit of the plurality of basic circuits is at the first level, the Mth basic circuit provides a preset response data to the signal detector and the counter, so that the signal detector and the count value of the counter is increased to accumulate the number of signals having the first level. The plurality of basic circuits are composed of logic gates, so the structure of the accumulating circuit is relatively simple and the area is small.

圖1顯示應用本發明的記憶體10,其包括多個記憶體位元G1~Gm及R1~Rn、重新排序電路12及多個輸入輸出端I/O1~I/Om,其中m及n為正整數。記憶體位元G1~Gm為一般記憶體位元,而記憶體位元R1~Rn為多餘記憶體位元,資料DG1至資料DGm為記憶體位元G1~Gm中儲存的資料,資料DR1至資料DRn為記憶體位元R1~Rn中儲存的資料。在之後的說明中,為了方便說明及更容易理解本發明技術,資料DG1~DGm及DR1~DRn分別等同視為記憶體位元G1~Gm及R1~Rn。記憶體10在經過檢測後產生多個位元修補資料RPG1~RPGm及RPR1~RPRn,其中位元修補資料RPG1~RPGm分別對應一般記憶體位元G1~Gm,位元修補資料RPR1~RPRn分別對應多餘記憶體位元R1~Rn,多個位元修補資料RPG1~RPGm及RPR1~RPRn是用以標記對應的記憶體位元是否為不佳位元。重新排序電路12根據多個位元修補資料RPG1~RPGm及RPR1~RPRn從多個記憶體位元12中選取m個良好的記憶體位元依序耦接至記憶體10的多個輸入輸出端I/O1~I/Om。重新排序電路12會判斷一般記憶體位元G1~Gm是否為良好位元,若是則將其耦接至輸入輸出端I/O1~I/Om,倘若一般記憶體位元G1~Gm中存有不佳位元,則從多餘記憶體位元R1~Rn中選取良好位元來取代不佳的一般記憶體位元,將其耦接至輸入輸出端I/O1~I/Om。1 shows a memory 10 to which the present invention is applied, which includes a plurality of memory bits G1 ˜Gm and R1 ˜Rn, a reordering circuit 12 and a plurality of input and output terminals I/O1 ˜I/Om, wherein m and n are positive Integer. Memory bits G1~Gm are general memory bits, while memory bits R1~Rn are redundant memory bits, data DG1 to data DGm are data stored in memory bits G1~Gm, data DR1 to data DRn are memory bits Data stored in elements R1~Rn. In the following description, for convenience of description and easier understanding of the technology of the present invention, the data DG1 ˜DGm and DR1 ˜DRn are regarded as memory bits G1 ˜Gm and R1 ˜Rn, respectively. The memory 10 generates a plurality of bit repair data RPG1 to RPGm and RPR1 to RPRn after detection, wherein the bit repair data RPG1 to RPGm respectively correspond to the general memory bits G1 to Gm, and the bit repair data RPR1 to RPRn respectively correspond to the redundant bits. The memory bits R1~Rn, a plurality of bit repair data RPG1~RPGm and RPR1~RPRn are used to mark whether the corresponding memory bits are bad bits. The reordering circuit 12 selects m good memory bits from the plurality of memory bits 12 according to the plurality of bit patching data RPG1 ˜RPGm and RPR1 ˜ RPRn, and sequentially couples them to the plurality of input and output terminals I/ O1~I/Om. The reordering circuit 12 will determine whether the general memory bits G1~Gm are good bits, and if so, couple them to the input and output terminals I/O1~I/Om, if there are bad bits in the general memory bits G1~Gm For bits, good bits are selected from the redundant memory bits R1 ˜Rn to replace the bad general memory bits, and are coupled to the input and output terminals I/O1 ˜I/Om.

在圖1的實施例中,多個記憶體位元G1~Gm及R1~Rn中的位元G2、位元Gm-1及位元R1為不佳的記憶體位元,當重新排序電路12要將記憶體位元對應到輸入輸出端I/O1~I/Om時,其根據位元修補資料RPG1判斷第一個一般記憶體位元G1為良好記憶體位元,因此將位元G1耦接至第一輸入輸出端I/O1,因而可從輸入輸出端I/O1存取位元G1的資料DG1。接著重新排序電路12從位元修補資料RPG2得知第二個一般記憶體位元G2為不佳的記憶體位元,因此捨棄位元G2並從多餘記憶體位元R1~Rn中選擇一個來替換位元G2,此時重新排序電路12可由位元修補資料RPR1及RPR2得知第一個多餘記憶體位元R1為不佳的記憶體位元及第二個多餘記憶體位元R2為良好記憶體位元,因此重新排序電路12將選擇第二個多餘記憶體位元R2來取代位元G2,並將其耦接至第二個輸入輸出端I/O2,因而可從輸入輸出端I/O2存取位元R2的資料DR2。依此類推,重新排序電路12根據位元修補資料RPG3~RPGm依序判斷位元G3至位元Gm是否為良好記憶體位元,若是則耦接至輸入輸出端I/O3~I/Om,若為不佳的記憶體位元,例如位元Gm-1,則從多餘記憶體位元中選取良好且未被使用的記憶體位元R3,並將其耦接至輸入輸出端I/Om-1。重新排序電路12將輸入輸出端I/O1~I/Om與記憶體位元一一對應後,主機能透過輸入輸出端I/O1~I/Om存取對應的記憶體位元的資料。In the embodiment of FIG. 1 , the bit G2, the bit Gm-1 and the bit R1 among the plurality of memory bits G1-Gm and R1-Rn are bad memory bits. When the reordering circuit 12 wants to When the memory bit corresponds to the input and output terminals I/O1~I/Om, it determines that the first general memory bit G1 is a good memory bit according to the bit patching data RPG1, so the bit G1 is coupled to the first input The output terminal I/O1, so the data DG1 of the bit G1 can be accessed from the input and output terminal I/O1. Then, the reordering circuit 12 learns from the bit patch data RPG2 that the second general memory bit G2 is a bad memory bit, so it discards the bit G2 and selects one of the redundant memory bits R1-Rn to replace the bit G2, at this time, the reordering circuit 12 can know from the bit repair data RPR1 and RPR2 that the first redundant memory bit R1 is a bad memory bit and the second redundant memory bit R2 is a good memory bit, so reordering The sorting circuit 12 selects the second redundant memory bit R2 to replace the bit G2, and couples it to the second I/O terminal I/O2, so that the bit R2 can be accessed from the I/O terminal I/O2. Data DR2. By analogy, the reordering circuit 12 sequentially determines whether the bits G3 to Gm are good memory bits according to the bit patch data RPG3 to RPGm, and if so, the reordering circuit 12 is coupled to the input and output terminals I/O3 to I/Om. For bad memory bits, such as bit Gm-1, a good and unused memory bit R3 is selected from the surplus memory bits and coupled to the input and output terminals I/Om-1. After the reordering circuit 12 associates the input and output terminals I/O1 to I/Om with the memory bits one-to-one, the host can access the data of the corresponding memory bits through the input and output terminals I/O1 to I/Om.

圖2顯示圖1中重新排序電路12的實施例,其包括多個判斷電路20、22、24及26、多個多工器30、32、34及36及一多餘位元排序電路40。多個判斷電路20、22、24及26中的第一個判斷電路20根據位元修補資料RPG1產生一選擇信號Se1及一累計信號Sol。第二個判斷電路22根據位元修補資料RPG2及累計信號So1產生選擇信號Se2及累計信號So2。第三個判斷電路24根據位元修補資料RPG3及累計信號So2產生選擇信號Se3及累計信號So3。以此類推,多個判斷電路20、22、24及26各自接收一個位元修補資料RPG1~RPGm,並各自輸出一個選擇信號Se1~Sem,而除了第一個判斷電路20外,其餘的判斷電路22、24及26都是根據所接收的位元修補資料及前一個判斷電路輸出的累計信號So1~Som-1產生選擇信號Se2~Sem,其中該多個累計信號So1~Som-1記錄下一個可被使用的多餘記記憶體或記錄已被使用的多餘記憶體位元的數量。多餘位元排序電路40根據多個位元修補資料RPR1~RPRn將多餘記憶體位元R1~Rn(DR1~DRn)中的良好記憶體位元連接至每一個多工器30、32、34及36。參照圖1,在此實施例中,記憶體位元R1具有缺陷,因此多餘位元排序電路40會排除位元R1,使得多餘位元排序電路40的輸出RI為DR2~DRn(R2~Rn)。多個多工器30、32、34及36的輸出端各自連接一個輸入輸出端I/O1~I/Om。第一個多工器30根據位元修補資料RPG1及選擇信號Se1決定將一般記憶體位元G1或多餘記憶體位元R2至位元Rn(DG1或DR2~DRn)的其中一個耦接至第一輸入輸出端I/O1,在此實施例中,一般記憶體位元G1為良好位元,故多工器30選擇將一般記憶體位元G1(DG1)耦接到輸入輸出端I/O1。第二個多工器32根據位元修補資料RPG2及選擇信號Se2將一般記憶體位元G2或多餘記憶體位元R2至位元Rn的其中一個耦接至第二個輸入輸出端I/O2,在此實施例中,一般記憶體位元G2為不佳位元,故多工器32從多餘記憶體位元R2~Rn(DR2~DRn)中選取位元R2(DR2)耦接至輸入輸出端I/O2。以此類推,每一個多工器30、32、34及36都會將一個記憶體位元耦接至對應的輸入輸出端I/O1~I/Om。FIG. 2 shows an embodiment of the reordering circuit 12 in FIG. 1 , which includes a plurality of judgment circuits 20 , 22 , 24 and 26 , a plurality of multiplexers 30 , 32 , 34 and 36 , and a redundant bit sorting circuit 40 . The first judgment circuit 20 among the plurality of judgment circuits 20 , 22 , 24 and 26 generates a selection signal Se1 and an accumulation signal Sol according to the bit patch data RPG1 . The second judging circuit 22 generates the selection signal Se2 and the accumulation signal So2 according to the bit patch data RPG2 and the accumulation signal So1. The third judging circuit 24 generates the selection signal Se3 and the accumulation signal So3 according to the bit patch data RPG3 and the accumulation signal So2. By analogy, the plurality of judgment circuits 20 , 22 , 24 and 26 each receive a bit patch data RPG1 to RPGm and output a selection signal Se1 to Sem respectively. Except for the first judgment circuit 20 , the rest of the judgment circuits 22, 24 and 26 all generate selection signals Se2 to Sem according to the received bit patch data and the accumulated signals So1 to Som-1 output by the previous judgment circuit, wherein the plurality of accumulated signals So1 to Som-1 record the next one. The amount of spare memory that can be used or records the amount of spare memory bits that have been used. The redundant bit sorting circuit 40 connects the good memory bits among the redundant memory bits R1 ˜Rn (DR1 ˜DRn) to each of the multiplexers 30 , 32 , 34 and 36 according to the plurality of bit repair data RPR1 ˜RPRn. Referring to FIG. 1 , in this embodiment, the memory bit R1 has defects, so the redundant bit sorting circuit 40 excludes the bit R1, so that the output RI of the redundant bit sorting circuit 40 is DR2˜DRn (R2˜Rn). The output terminals of the multiplexers 30 , 32 , 34 and 36 are each connected to an input and output terminal I/O1 ˜I/Om. The first multiplexer 30 determines to couple the normal memory bit G1 or one of the redundant memory bits R2 to Rn (DG1 or DR2 to DRn) to the first input according to the bit patch data RPG1 and the selection signal Se1 The output terminal I/O1, in this embodiment, the general memory bit G1 is a good bit, so the multiplexer 30 selects to couple the general memory bit G1 (DG1) to the input and output terminal I/O1. The second multiplexer 32 couples one of the general memory bit G2 or the redundant memory bits R2 to Rn to the second input/output terminal I/O2 according to the bit patch data RPG2 and the selection signal Se2, In this embodiment, the general memory bit G2 is a bad bit, so the multiplexer 32 selects the bit R2 (DR2) from the redundant memory bits R2-Rn (DR2-DRn) to be coupled to the input and output terminals I/ O2. By analogy, each of the multiplexers 30 , 32 , 34 and 36 couples a memory bit to the corresponding input and output terminals I/O1 ˜I/Om.

圖3顯示圖2中的多工器30,其包括多個反相器3002、3003、3005、3007、3009、3011、3013、3015、3017及3019、多個開關3004、3008、3012、3016及3020以及多個及閘3006、3010、3014及3018,選擇信號Se1包括多個信號RENB11~RENB1n分別輸入至多個反相器3005、3009、3013、3017及3019,多個反相器3005、3009、3013、3017及3019的輸出分別連接多個及閘3006、3010、3014及2018的一輸入端,多個及閘3006、3010、3014及3018的另一輸入端接收位元修補資料RPG1。開關3004連接在一般記憶體位元G1(DG1)及輸入輸出端I/O1之間,位元修補資料RPG1透過反相器3002及3003控制開關3004導通或關閉。開關3008連接在多餘記憶體位元R2(DR2)及輸入輸出端I/O1之間,及閘3006根據位元修補資料RPG1及信號RENB11控制開關3008通導或關閉。開關3012連接在多餘記憶體位元R3(DR3)及輸入輸出端I/O1之間,及閘3010根據位元修補資料RPG1及信號RENB12控制開關3012通導或關閉。開關3016連接在多餘記憶體位元R4(DR4)及輸入輸出端I/O1之間,及閘3014根據位元修補資料RPG1及信號RENB13控制開關3016通導或關閉。開關3020連接在多餘記憶體位元Rn(DRn)及輸入輸出端I/O1之間,及閘3018根據位元修補資料RPG1及信號RENB1n控制開關3020通導或關閉。當位元修補資料RPG1為“0”時,開關3004導通,進而讓記憶體位元G1(DG1)耦接到輸入輸出端I/O1,而及閘3006、3010、3014及3018皆輸出低準位邏輯信號“0”使開關3008、3012、3016及3020關閉。當位元修補資料RPG1為“1”時,開關3004關閉,此時將由信號RENB11~RENB1n決定所要導通的開關,例如當信號RENB11為“0”而其餘信號REN12~REN1n 為“1”時,開關3008被導通以使多餘記憶體位元R2(DR2)耦接至輸入輸出端I/O1,當信號REN1n為“0”而其餘信號REN11~REN1n-1 為“1”時,開關3020被導通以使多餘記憶體位元Rn(DRn)耦接至輸入輸出端I/O1。FIG. 3 shows the multiplexer 30 of FIG. 2, which includes a plurality of inverters 3002, 3003, 3005, 3007, 3009, 3011, 3013, 3015, 3017 and 3019, a plurality of switches 3004, 3008, 3012, 3016 and 3020 and a plurality of AND gates 3006, 3010, 3014 and 3018, the selection signal Se1 includes a plurality of signals RENB11-RENB1n respectively input to a plurality of inverters 3005, 3009, 3013, 3017 and 3019, a plurality of inverters 3005, 3009, The outputs of 3013, 3017 and 3019 are respectively connected to one input terminal of the multiple AND gates 3006, 3010, 3014 and 2018, and the other input terminal of the multiple AND gates 3006, 3010, 3014 and 3018 receives the bit repair data RPG1. The switch 3004 is connected between the general memory bit G1 ( DG1 ) and the input/output terminal I/O1 , and the bit patch data RPG1 controls the switch 3004 to be turned on or off through the inverters 3002 and 3003 . The switch 3008 is connected between the redundant memory bit R2 (DR2) and the input/output terminal I/O1, and the gate 3006 controls the switch 3008 to be turned on or off according to the bit repair data RPG1 and the signal RENB11. The switch 3012 is connected between the redundant memory bit R3 (DR3) and the input/output terminal I/O1, and the gate 3010 controls the switch 3012 to be turned on or off according to the bit repair data RPG1 and the signal RENB12. The switch 3016 is connected between the redundant memory bit R4 (DR4) and the input/output terminal I/O1, and the gate 3014 controls the switch 3016 to be turned on or off according to the bit repair data RPG1 and the signal RENB13. The switch 3020 is connected between the redundant memory bit Rn (DRn) and the input/output terminal I/O1, and the gate 3018 controls the switch 3020 to be turned on or off according to the bit repair data RPG1 and the signal RENB1n. When the bit patch data RPG1 is "0", the switch 3004 is turned on, so that the memory bit G1 (DG1) is coupled to the input/output terminal I/O1, and the gates 3006, 3010, 3014 and 3018 all output a low level A logic signal "0" causes switches 3008, 3012, 3016 and 3020 to close. When the bit patch data RPG1 is "1", the switch 3004 is closed, and the switches to be turned on are determined by the signals RENB11~RENB1n. For example, when the signal RENB11 is "0" and the other signals REN12~REN1n are "1", the switch 3008 is turned on so that the redundant memory bit R2 (DR2) is coupled to the input/output terminal I/O1. When the signal REN1n is "0" and the remaining signals REN11~REN1n-1 are "1", the switch 3020 is turned on to make The redundant memory bits Rn (DRn) are coupled to the input and output terminals I/O1.

雖然圖3中僅顯示多工器30的電路,但其他多工器32、34及36的電路及操作類似於多工器30,從圖3的多工器30可以輕易推得其他多工器32、34及36的電路及操作。Although only the circuit of the multiplexer 30 is shown in FIG. 3 , the circuits and operations of the other multiplexers 32 , 34 and 36 are similar to the multiplexer 30 , and other multiplexers can be easily deduced from the multiplexer 30 of FIG. 3 Circuitry and operation of 32, 34 and 36.

圖4顯示構成圖2中判斷電路的基本電路50,每一個判斷電路20、22、24及26都是由多個基本電路50組成,如圖5所示。圖4的基本電路50是由邏輯閘構成的邏輯電路,其包括三個反及閘52、56及58與一個反相器54,反及閘52的輸出端連接至反及閘56及58的一輸入端,位元修補資料RP經反相器54輸入至反及閘56的另一輸入端,反及閘58的另一輸入端接收位元修補資料RP,反及閘58的輸出是用以組成判斷電路所輸出的選擇信號。當反及閘52的的其中一個輸入為“0”(反應資料)時,位元修補資料RP將決定反及閘52的輸入端上預設的反應資料成為反及閘56或58的輸出,在此實施例中,反應資料為“0”,當位元修補資料RP為“1”時,代表目前的一般記憶體位元為不佳位元,當反及閘58的輸出為反應資料“0”時,選取一個多餘記憶體位元來替換目前的一般記憶體位元,並耦接至記憶體10的輸入輸出端。當位元修補資料RP為“0”時,代表目前的一般記憶體位元為良好位元,反及閘56的輸出為反應資料“0”,使目前的一般記憶體位元耦接至記憶體10的輸入輸出端。圖6顯示圖4電路的概念,當位元修補資料RP為“1”時,相當於讓預設的反應資料“0”移動到右下的反及閘58的輸出,使反應資料“0”,出現於相對於輸入不同位置,而當位元修補資料RP為“0”,相當於讓反及閘52的輸出向右邊移動到的反及閘56的輸出,使所設計的資料,出現於相對於輸入的相同位置。FIG. 4 shows a basic circuit 50 constituting the judgment circuit in FIG. 2 . Each of the judgment circuits 20 , 22 , 24 and 26 is composed of a plurality of basic circuits 50 , as shown in FIG. 5 . The basic circuit 50 of FIG. 4 is a logic circuit composed of logic gates, which includes three inversion gates 52 , 56 and 58 and an inverter 54 , and the output terminal of the inversion gate 52 is connected to the inversion gates 56 and 58 . An input terminal, the bit repair data RP is input to the other input terminal of the inversion gate 56 through the inverter 54, and the other input terminal of the inversion gate 58 receives the bit repair data RP, and the output of the inversion gate 58 is used for to form the selection signal output by the judgment circuit. When one of the inputs of the inversion gate 52 is "0" (reaction data), the bit patch data RP will determine the preset reaction data on the input of the inversion gate 52 as the output of the inversion gate 56 or 58, In this embodiment, the response data is "0", when the bit repair data RP is "1", it means that the current general memory bit is a bad bit, and the output of the anti-and gate 58 is the response data "0" ”, select an extra memory bit to replace the current general memory bit, and is coupled to the input and output terminals of the memory 10 . When the bit patch data RP is "0", it means that the current general memory bit is a good bit, and the output of the anti-and gate 56 is the response data "0", so that the current general memory bit is coupled to the memory 10 input and output terminals. Fig. 6 shows the concept of the circuit of Fig. 4. When the bit repair data RP is "1", it is equivalent to move the preset response data "0" to the output of the inverse gate 58 at the lower right, so that the response data "0" , appear in different positions relative to the input, and when the bit patch data RP is "0", it is equivalent to moving the output of the inversion gate 52 to the right to the output of the inversion gate 56, so that the designed data appears in the The same position relative to the input.

在圖5中,每一個判斷電路20、22及24都是由多於n個基本電路50疊接構成,其中n為多餘記憶體位元數量,每一個基本電路50的輸出成下一個判斷電路的輸入。因預設的反應資料GND=“0”放在左上的基本電路50的輸入,使得相對位置右下的輸出成為位元修補的累加數量及所要的選擇信號,其中判斷電路20中的反及閘58所輸出的信號RENB11、RENB12及RENB13組成圖2中的選擇信號Se1,而反及閘56及58所輸出的信號Sa11、Sa12、Sa13、RENB11、RENB12及RENB13組成圖2中判斷電路20的累計信號So1。判斷電路22中的反及閘58所輸出的信號RENB21、RENB22及RENB23組成圖2中的選擇信號Se2,而反及閘56及58所輸出的信號Sa21、Sa22、Sa23、RENB21、RENB22及RENB23組成圖2中判斷電路22的輸出So2。判斷電路24中的反及閘58所輸出的信號RENB31、RENB32及RENB33組成圖2中的選擇信號Se2,而反及閘56及58所輸出的信號Sa31、Sa32、Sa33、RENB31、RENB32及RENB33組成圖2中判斷電路24的輸出So3。圖7顯示圖5電路的概念,參照圖5及圖7,在判斷電路22中,由於位元修補資料RPG2為“1”,因此 反應資料“0”向下移動,如圖7所示,使選擇信號Se2變為“011”,因此選擇將第一個良好的多餘記憶體位元R2耦接到輸入輸出端I/O2,選擇信號中反應資料“0”的位置決定所要選取的多餘記憶體位元,例如當選擇信號為“101”時,會選擇第二個良好的多餘記憶體位元R3耦接至輸入輸出端I/O,而當選擇信號為“110”時,會選擇第三個良好的多餘記憶體位元R4耦接至輸入輸出端I/O,依此類推。圖5中僅顯示判斷電路20、22及24的部分電路,本領域技術人員可從圖5所揭示的內容推得判斷電路20、22及24的完整電路。在此實施例中,判斷電路20、22及24是用多個反及閘52、56及58來實現,但判斷電路20、22及24並不只限於用反及閘來實現,也可以用多個反或閘或用多種不同邏輯閘元件來實現,例如預設的反應資料為“1(VDD)”時,反及閘52、56及58可用反或閘取代。In FIG. 5, each of the judgment circuits 20, 22 and 24 is formed by stacking more than n basic circuits 50, where n is the number of redundant memory bits, and the output of each basic circuit 50 becomes the output of the next judgment circuit. enter. Because the preset response data GND=“0” is placed on the input of the basic circuit 50 at the upper left, the output at the lower right of the relative position becomes the accumulated number of bit repairs and the desired selection signal, wherein the inversion gate in the judgment circuit 20 is The signals RENB11, RENB12 and RENB13 output by 58 constitute the selection signal Se1 in FIG. 2, and the signals Sa11, Sa12, Sa13, RENB11, RENB12 and RENB13 output by the reverse gates 56 and 58 constitute the accumulation of the judgment circuit 20 in FIG. 2. Signal So1. The signals RENB21, RENB22 and RENB23 output by the inversion gate 58 in the judgment circuit 22 constitute the selection signal Se2 in FIG. 2, and the signals Sa21, Sa22, Sa23, RENB21, RENB22 and RENB23 output by the inversion gates 56 and 58 constitute The output So2 of the judgment circuit 22 in FIG. 2 . The signals RENB31, RENB32 and RENB33 output by the inversion gate 58 in the judgment circuit 24 constitute the selection signal Se2 in FIG. 2, while the signals Sa31, Sa32, Sa33, RENB31, RENB32 and RENB33 output by the inversion gates 56 and 58 constitute The output So3 of the judgment circuit 24 in FIG. 2 . Figure 7 shows the concept of the circuit in Figure 5. Referring to Figures 5 and 7, in the judgment circuit 22, since the bit repair data RPG2 is "1", the response data "0" moves downward, as shown in Figure 7, so that The selection signal Se2 becomes "011", so the first good redundant memory bit R2 is selected to be coupled to the input and output terminals I/O2, and the position of the response data "0" in the selection signal determines the redundant memory bit to be selected For example, when the selection signal is "101", the second good redundant memory bit R3 will be selected to be coupled to the input and output I/O, and when the selection signal is "110", the third good memory bit R3 will be selected. The redundant memory bit R4 is coupled to the input and output terminals I/O, and so on. FIG. 5 only shows part of the circuits of the judgment circuits 20 , 22 and 24 , and those skilled in the art can deduce the complete circuits of the judgment circuits 20 , 22 and 24 from the content disclosed in FIG. 5 . In this embodiment, the judging circuits 20, 22 and 24 are realized by a plurality of inverting gates 52, 56 and 58, but the judging circuits 20, 22 and 24 are not limited to be realized by inverting and An inverse-OR gate can be implemented by using a variety of different logic gate elements. For example, when the default response data is "1 (VDD)", the inverse-OR gates 52, 56 and 58 can be replaced by inverse-OR gates.

在其他實施例中,圖4的基本電路50也可以修改為圖8的開關電路,其包括PMOS電晶體60及NMOS電晶體62,電晶體60及62的輸入端接收反應資料“1”或 “0”,控制端接收位元修補資料RP。當位元修補資料RP為“0”時,PMOS電晶體60導通而NMOS電晶體62關閉,因此反應資料“1”或 “0”由上方的PMOS電晶體60輸出。當位元修補資料RP為“1”時,PMOS電晶體60關閉而NMOS電晶體62導通,因此反應資料“1”或 “0”由下方的NMOS電晶體60輸出。In other embodiments, the basic circuit 50 of FIG. 4 can also be modified to the switch circuit of FIG. 8, which includes a PMOS transistor 60 and an NMOS transistor 62, and the input terminals of the transistors 60 and 62 receive the response data “1” or “ 0", the control end receives the bit repair data RP. When the bit repair data RP is "0", the PMOS transistor 60 is turned on and the NMOS transistor 62 is turned off, so the response data "1" or "0" is output from the upper PMOS transistor 60. When the bit repair data RP is "1", the PMOS transistor 60 is turned off and the NMOS transistor 62 is turned on, so the response data "1" or "0" is output from the NMOS transistor 60 below.

圖9顯示本發明重新排序記憶體位元的另一實施例,圖9的重新排序電路12是依序將多個記憶體位元G1~Gm及R1~Rn中的良好位元耦接至多個輸入輸出端I/O1~I/Om。重新排序電路12可透過位元修補資料RPG1~RPGm及RPR1~RPRn來判斷多個記憶體位元G1~Gm及R1~Rn是否為不佳位元,如圖9所示,重新排序電路12透過位元修補資料RPG1知道第一個記憶體位元G1為良好位元,故將記憶體位元G1耦接至第一個輸入輸出端I/O1,當重新排序電路12透過位元修補資料RPG2知道第二個記憶體位元G2為不佳位元時,捨棄記憶體位元G2,接著重新排序電路12透過位元修補資料RPG3知道第三個記憶體位元G3為良好位元時,將記憶體位元G3耦接至第二個輸入輸出端I/O2,依此類推,直至所有的輸入輸出端I/O1~I/Om都耦接一記憶體位元。FIG. 9 shows another embodiment of the present invention for reordering memory bits. The reordering circuit 12 in FIG. 9 sequentially couples the good bits in the plurality of memory bits G1 ˜Gm and R1 ˜Rn to a plurality of inputs and outputs Terminal I/O1~I/Om. The reordering circuit 12 can determine whether the plurality of memory bits G1~Gm and R1~Rn are bad bits through the bit repair data RPG1~RPGm and RPR1~RPRn. As shown in FIG. 9, the reordering circuit 12 passes the bits The meta-patching data RPG1 knows that the first memory bit G1 is a good bit, so the memory bit G1 is coupled to the first input/output terminal I/O1, when the reordering circuit 12 knows the second memory bit G1 through the bit-patching data RPG2 When one memory bit G2 is a bad bit, the memory bit G2 is discarded, and then the reordering circuit 12 couples the memory bit G3 when it knows the third memory bit G3 is a good bit through the bit patch data RPG3 to the second input/output terminal I/O2, and so on, until all the input/output terminals I/O1-I/Om are coupled to a memory bit.

圖10顯示圖9中重新排序電路12的實施例,其包括多個判斷電路70、72、74及76以及多個多工器78、80、82及84。第一個判斷電路70接收多個位元修補資料RPG1~RPGm及RPR1~RPRn中的第一個位元修補資料RPG1以及其後的N個位元修補資料RPG2~RPGN+1,據以產生選擇信號Se1。第二個判斷電路72接收多個位元修補資料中的第二個位元修補資料RPG2以及其後的N個位元修補資料RPG3~RPGN+2,據以產生選擇信號Se2。依此類推,第M個判斷電路接收多個位元修補資料中的第M個位元修補資料以及其後的N個位元修補資料,據以產生第M個選擇信號SeM。在此實施例中,N等於n,即N等於多餘記憶體位元的數量,在其他實施例中,N也可以大於或小於n。第一個多工器78耦接判斷電路70以及對應位元修補資料RPG1~RPGN+1的記憶體位元G1~GN+1(DG1~DGN+1),根據選擇信號Se1從多個記憶體位元G1~GN+1(DG1~DGN+1)選擇其中一個良好位元G1(DG1)耦接至輸入輸出端I/O1。第二個多工器80耦接判斷電路72以及對應位元修補資料RPG2~RPGN+2的記憶體位元G2~GN+2(DG2~DGN+2),根據選擇信號Se2從多個記憶體位元G2~GN+2(DG2~DGN+2)選擇其中一個良好位元耦接至輸入輸出端I/O2,在此實施例中,由於記憶體位元G2(DG2)為不佳位元,故多工器80選擇將記憶體位元G3(DG3)耦接至輸入輸出端I/O2。第三個多工器82耦接判斷電路74以及對應位元修補資料RPG3~RPGN+3的記憶體位元G3~GN+3(DG3~DGN+3),根據選擇信號Se3從多個記憶體位元G3~GN+3(DG3~DGN+3)選擇其中一個良好位元耦接至輸入輸出端I/O2,在此實施例中,由於記憶體位元G3(DG3)已被選擇耦接至輸入輸出端I/O2,故多工器82選擇將記憶體位元G4(DG4)耦接至輸入輸出端I/O3。依此類推,第M個多工器耦接第M個判斷電路以及對應第M個至第M+N個位元修補資料的記憶體位元,根據該第M個判斷電路提供的選擇信號SeM從第M個至第M+N個記憶體位元中選擇其中一個良好且未被使用的記憶體位元耦接至第M個輸入輸出端I/OM。FIG. 10 shows an embodiment of the reordering circuit 12 in FIG. 9 , which includes a plurality of determination circuits 70 , 72 , 74 and 76 and a plurality of multiplexers 78 , 80 , 82 and 84 . The first judging circuit 70 receives the first bit repair data RPG1 and the following N bit repair data RPG2 to RPGN+1 among the plurality of bit repair data RPG1 to RPGm and RPR1 to RPRn, and generates a selection accordingly. Signal Se1. The second judging circuit 72 receives the second bit repair data RPG2 among the plurality of bit repair data and the following N bit repair data RPG3 to RPGN+2, and generates the selection signal Se2 accordingly. By analogy, the Mth judging circuit receives the Mth bit repair data and the N subsequent bit repair data among the plurality of bit repair data, and generates the Mth selection signal SeM accordingly. In this embodiment, N is equal to n, that is, N is equal to the number of redundant memory bits. In other embodiments, N may be larger or smaller than n. The first multiplexer 78 is coupled to the judgment circuit 70 and the memory bits G1 to GN+1 (DG1 to DGN+1) corresponding to the bit patch data RPG1 to RPGN+1. G1~GN+1(DG1~DGN+1) select one of the good bits G1(DG1) to be coupled to the input/output terminal I/O1. The second multiplexer 80 is coupled to the judgment circuit 72 and the memory bits G2 to GN+2 (DG2 to DGN+2) corresponding to the bit patch data RPG2 to RPGN+2, and from the plurality of memory bits according to the selection signal Se2 G2~GN+2 (DG2~DGN+2) select one of the good bits to be coupled to the input/output terminal I/O2. In this embodiment, since the memory bit G2 (DG2) is a bad bit, there are many The processor 80 selects to couple the memory bit G3 (DG3) to the input/output terminal I/O2. The third multiplexer 82 is coupled to the judgment circuit 74 and the memory bits G3 to GN+3 (DG3 to DGN+3) corresponding to the bit patch data RPG3 to RPGN+3, and selects the memory bits from the plurality of memory bits according to the selection signal Se3. G3~GN+3 (DG3~DGN+3) select one of the good bits to be coupled to the input/output terminal I/O2. In this embodiment, since the memory bit G3 (DG3) has been selected to be coupled to the input/output terminal I/O2, so the multiplexer 82 selects to couple the memory bit G4 (DG4) to the input and output terminal I/O3. By analogy, the M-th multiplexer is coupled to the M-th judgment circuit and the memory bits corresponding to the M-th to M+N-th bit repair data. According to the selection signal SeM provided by the M-th judgment circuit, from One of the good and unused memory bits selected from the Mth to M+Nth memory bits is coupled to the Mth input/output terminal I/OM.

圖11顯示圖10中的多工器78,其包括多個反相器7802、7803、7805、7807、7809、7811、7813、7815、7817及7819以及多個開關7804、7808、7812、7816及7820,選擇信號Se1包括多個信號RENB11~RENB1N+1分別輸入至多個反相器7802、7805、7809、7813、7817及7819。開關7804連接在記憶體位元G1(DG1)及輸入輸出端I/O1之間,信號RENB11經反相器7802及7803控制開關7804導通或關閉。開關7808連接在記憶體位元G2(DG2)及輸入輸出端I/O1之間,信號RENB12經反相器7805及7807控制開關7808通導或關閉。開關7812連接在記憶體位元G3(DG3)及輸入輸出端I/O1之間,信號RENB13經反相器7809及7811控制開關7812通導或關閉。開關7816連接在記憶體位元G4(DG4)及輸入輸出端I/O1之間,信號RENB14經反相器7813及7815控制開關7816通導或關閉。開關7820連接在記憶體位元GN(DGN)及輸入輸出端I/O1之間,信號RENB1N+1經反相器7817及7819控制開關7820通導或關閉。當信號RENB11為“0”時,開關7804導通,進而讓記憶體位元G1(DG1)耦接到輸入輸出端I/O1。當信號RENB12為“0”時,開關7808被導通以使記憶體位元G2(DG2)耦接至輸入輸出端I/O1。依此類推,當信號REN1N+1為“0”時,開關7820被導通以使記憶體位元GN+1(DRn)耦接至輸入輸出端I/O1。Figure 11 shows the multiplexer 78 of Figure 10, which includes a plurality of inverters 7802, 7803, 7805, 7807, 7809, 7811, 7813, 7815, 7817 and 7819 and a plurality of switches 7804, 7808, 7812, 7816 and 7820 , the selection signal Se1 includes a plurality of signals RENB11 ˜ RENB1N+1 , which are respectively input to the plurality of inverters 7802 , 7805 , 7809 , 7813 , 7817 and 7819 . The switch 7804 is connected between the memory bit G1 ( DG1 ) and the input/output terminal I/O1 , and the signal RENB11 controls the switch 7804 to be turned on or off through the inverters 7802 and 7803 . The switch 7808 is connected between the memory bit G2 (DG2) and the input/output terminal I/O1, and the signal RENB12 controls the switch 7808 to be turned on or off through the inverters 7805 and 7807. The switch 7812 is connected between the memory bit G3 (DG3) and the input/output terminal I/O1, and the signal RENB13 controls the switch 7812 to be turned on or off through the inverters 7809 and 7811. The switch 7816 is connected between the memory bit G4 (DG4) and the input/output terminal I/O1, and the signal RENB14 controls the switch 7816 to be turned on or off through the inverters 7813 and 7815. The switch 7820 is connected between the memory bit GN (DGN) and the input/output terminal I/O1, and the signal RENB1N+1 controls the switch 7820 to be turned on or off through the inverters 7817 and 7819. When the signal RENB11 is "0", the switch 7804 is turned on, so that the memory bit G1 (DG1) is coupled to the input/output terminal I/O1. When the signal RENB12 is "0", the switch 7808 is turned on so that the memory bit G2 (DG2) is coupled to the input/output terminal I/O1. By analogy, when the signal REN1N+1 is "0", the switch 7820 is turned on so that the memory bit GN+1 (DRn) is coupled to the input/output terminal I/O1.

雖然圖11中僅顯示多工器78的電路,但其他多工器80、82及84的電路及操作類似於多工器78,從圖11的多工器78可以輕易推得其他多工器80、82及84的電路及操作。Although only the circuit of multiplexer 78 is shown in FIG. 11, the circuits and operations of other multiplexers 80, 82, and 84 are similar to multiplexer 78, and other multiplexers can be easily deduced from multiplexer 78 of FIG. Circuitry and operation of 80, 82 and 84.

圖12顯示構成圖10中判斷電路的基本電路90,每一個判斷電路70、72、74及76都是由多個基本電路90組成。圖12的基本電路90是由邏輯閘構成的邏輯電路,其包括三個反及閘92、96及98與一個反相器94,反及閘92的輸出端連接至反及閘96及98的一輸入端,位元修補資料RP經反相器94輸入至反及閘98的另一輸入端,反及閘96的另一輸入端接收位元修補資料RP,反及閘98的輸出是用以組成判斷電路所輸出的選擇信號。當反及閘92的其中一個輸入為“0”(反應資料)時,位元修補資料RP將決定反及閘92輸入端上的反應資料“0”由反及閘96或98輸出,在其他實施例中,反應資料也可以是“1”。當位元修補資料RP為“1”時,代表其對應的記憶體位元為不佳位元,此時反及閘98的輸出為反應資料“0”,該位元修補資料RP所對應的記憶體位元將不會被耦接至輸入輸出端I/Om。圖13顯示圖12電路的概念,當位元修補資料RP為“1”時,相當於反應資料“0”移動到右方的反及閘96的輸出,使反應資料“0” 出現於相對於輸入的相同位置,而當位元修補資料RP為“0”,相當於讓反應資料“0”向右下移動到反及閘98的輸出,使反應資料“0”出現於相對於輸入不同位置。FIG. 12 shows a basic circuit 90 constituting the judgment circuit of FIG. 10 . Each of the judgment circuits 70 , 72 , 74 and 76 is composed of a plurality of basic circuits 90 . The basic circuit 90 of FIG. 12 is a logic circuit composed of logic gates, which includes three inversion gates 92, 96 and 98 and an inverter 94, and the output terminal of the inversion gate 92 is connected to the inversion gates 96 and 98. An input terminal, the bit repair data RP is input to the other input terminal of the inversion gate 98 through the inverter 94, and the other input terminal of the inversion gate 96 receives the bit repair data RP. The output of the inversion gate 98 is used for to form the selection signal output by the judgment circuit. When one of the inputs of the inversion gate 92 is "0" (reaction data), the bit repair data RP will determine that the response data "0" on the input terminal of the inversion gate 92 is output by the inversion gate 96 or 98, and in the other In the embodiment, the reaction data may also be "1". When the bit repair data RP is "1", it means that the corresponding memory bit is a bad bit. At this time, the output of the inverse gate 98 is the response data "0", and the memory corresponding to the bit repair data RP is a bad bit. The body element will not be coupled to the input and output terminals I/Om. Figure 13 shows the concept of the circuit in Figure 12. When the bit repair data RP is "1", it is equivalent to moving the response data "0" to the output of the inverse gate 96 on the right, so that the response data "0" appears relative to the The same position of the input, and when the bit repair data RP is "0", it is equivalent to moving the response data "0" to the lower right to the output of the anti-and gate 98, so that the response data "0" appears in a different position relative to the input .

圖14顯示圖10中判斷電路70、72及74的實施例,每一個判斷電路70、72及74都是由N個基本電路90串接構成。因為預設的反應資料GND“0”放在左上的基本電路90的輸入,使得相對位置右下的輸出成為所要的選擇信號。在圖14中,判斷電路70的反及閘98所輸出的信號RENB11、RENB12、RENB13及RENB14組成圖10中的選擇信號Se1。判斷電路72中的反及閘98所輸出的信號RENB21、RENB22及RENB23組成圖10中的選擇信號Se2。判斷電路74中的反及閘98所輸出的信號RENB31及RENB32組成圖10中的選擇信號Se3。圖14中僅顯示判斷電路70、72及74的部分電路,本領域技術人員可從圖14所揭示的內容推得判斷電路70、72及74的完整電路。圖15顯示圖14電路的概念,參照圖11、圖14及圖15,在判斷電路70中,第一個基本電路90接收反應資料“0(GND)”且位元修補資料RPG1為“0”,因此反應資料“0”由反及閘98輸出,而第二個以後的基本電路90的反及閘92的輸入信號均為“1”,因此不論位元修補資料RPG2、RPG3、RPG4為何,反及閘98的輸出皆為“1”,如圖15所示,使選擇信號Se1變為“0111”,因此多工器78選擇將記憶體位元G1(DG1)耦接到輸入輸出端I/O1,選擇信號中反應資料“0”的位置決定所要選取的多餘記憶體位元。在判斷電路72中,第一個基本電路90接收反應資料“0”且位元修補資料RPG2為“1”,故反及閘98輸出信號“0”,而反應資料“0”將由反及閘96傳送到第二個基本電路90,由於第二個基本電路90的位元修補資料RPG3為“0”,因此反應資料“0”會由反及閘98輸出,第三個以後的基本電路90的輸入信號均為“1” ,因此不論位元修補資料RPG4為何,反及閘98的輸出皆為“1”,故可得到選擇信號Se2為“101”,以使多工器80選擇其所耦接的記憶體位元中的第二個記憶體位元G3(DG3)耦接至輸入輸出端I/O2。選擇信號Se3~Sem也是以相同方式產生,故不再贅述。在此實施例中,判斷電路70、72及74是用多個反及閘92、96及98來實現,但判斷電路70、72及74並不只限於用反及閘來實現,也可以用多個反或閘或用多種不同邏輯閘元件來實現,例如預設的反應資料為“1”時,反及閘92、96及98可用反或閘取代。FIG. 14 shows an embodiment of the judgment circuits 70 , 72 and 74 in FIG. 10 , and each judgment circuit 70 , 72 and 74 is formed by connecting N basic circuits 90 in series. Because the preset response data GND "0" is placed at the input of the basic circuit 90 at the upper left, the output at the lower right at the relative position becomes the desired selection signal. In FIG. 14 , the signals RENB11 , RENB12 , RENB13 and RENB14 output from the inverting gate 98 of the judgment circuit 70 constitute the selection signal Se1 in FIG. 10 . The signals RENB21 , RENB22 and RENB23 output from the inverting gate 98 in the judgment circuit 72 constitute the selection signal Se2 in FIG. 10 . The signals RENB31 and RENB32 output by the inversion gate 98 in the judgment circuit 74 constitute the selection signal Se3 in FIG. 10 . FIG. 14 only shows part of the circuits of the judgment circuits 70 , 72 and 74 , and those skilled in the art can deduce the complete circuits of the judgment circuits 70 , 72 and 74 from the content disclosed in FIG. 14 . FIG. 15 shows the concept of the circuit of FIG. 14. Referring to FIG. 11, FIG. 14 and FIG. 15, in the judgment circuit 70, the first basic circuit 90 receives the response data "0 (GND)" and the bit repair data RPG1 is "0" , so the response data "0" is output by the inversion gate 98, and the input signals of the inversion and gate 92 of the second and subsequent basic circuits 90 are all "1", so no matter what the bit repair data RPG2, RPG3, RPG4 are, The outputs of the inverter gate 98 are all “1”, as shown in FIG. 15, the selection signal Se1 becomes “0111”, so the multiplexer 78 selects to couple the memory bit G1 (DG1) to the input and output terminals I/ O1, the position of the response data "0" in the selection signal determines the redundant memory bits to be selected. In the judgment circuit 72, the first basic circuit 90 receives the response data "0" and the bit repair data RPG2 is "1", so the inversion gate 98 outputs a signal "0", and the response data "0" will be transmitted by the inversion gate 96 is sent to the second basic circuit 90. Since the bit repair data RPG3 of the second basic circuit 90 is "0", the response data "0" will be output by the inversion gate 98, and the third and subsequent basic circuits 90 The input signals of s are all "1", so no matter what the bit patch data RPG4 is, the output of the inverter gate 98 is "1", so the selection signal Se2 can be obtained as "101", so that the multiplexer 80 selects its The second memory bit G3 ( DG3 ) of the coupled memory bits is coupled to the input/output terminal I/O2 . The selection signals Se3 to Sem are also generated in the same way, so they will not be described again. In this embodiment, the judging circuits 70, 72 and 74 are realized by a plurality of inverting gates 92, 96 and 98, but the judging circuits 70, 72 and 74 are not limited to be realized by inverting and An inverse-OR gate can be implemented by a variety of different logic gate elements. For example, when the preset response data is "1", the inverse-OR gates 92, 96 and 98 can be replaced by inverse-OR gates.

在其他實施例中,圖12的基本電路90也可以修改為圖16的開關電路,其包括NMOS電晶體100及PMOS電晶體102,電晶體100及102的輸入端接收反應資料“1”或 “0”,控制端接收位元修補資料RP。當位元修補資料RP為“0”時,NMOS電晶體100關閉導通而PMOS電晶體102,因此反應資料“1”或 “0”由下方的PMOS電晶體102輸出。當位元修補資料RP為“1”時,NMOS電晶體100導通PMOS電晶體102關閉而,因此反應資料“1”或 “0”由上方的NMOS電晶體100輸出。In other embodiments, the basic circuit 90 of FIG. 12 can also be modified to the switch circuit of FIG. 16, which includes an NMOS transistor 100 and a PMOS transistor 102, and the input terminals of the transistors 100 and 102 receive the response data "1" or "" 0", the control end receives the bit repair data RP. When the bit repair data RP is "0", the NMOS transistor 100 is turned off and the PMOS transistor 102 is turned on, so the response data "1" or "0" is output from the PMOS transistor 102 below. When the bit repair data RP is "1", the NMOS transistor 100 turns on the PMOS transistor 102 and turns off, so the response data "1" or "0" is output from the upper NMOS transistor 100.

圖5的電路可以作為累加電路來使用。參照圖5及圖7,如前所述,當位元修補資料RPGm為“1”時代表對應的記憶體位元為不佳位元,同時所設定的資料“0”在選擇信號中的位置會跟著改變,如圖7所示,每出現一個不佳位元,資料“0”就向下移動一個位置,因此可根據反應資料“0”在選擇信號中的位置來判斷不佳記憶體位元的數量,例如,當選擇信號為“011”時,資料“0”出現在第一個位置,這代表有一個不佳位元,若選擇信號為“110”時,資料“0”出現在第三個位置,這代表有三個不佳位元。本發明的判斷電路20、22、24及26不只可以用來判斷所對應的記憶體位元是否為不佳位元,還可以作為不佳位元累加電路來累計不佳位元的數量。圖5的累加電路不只限於應用在記憶體中,其也可以應用在記憶體以外的電路中,判斷多個信號(RPG1~RPGm)中準位為“0”或“1”的數量。同理,圖14的電路也可以作為累加電路來使用。The circuit of Figure 5 can be used as an accumulation circuit. Referring to FIG. 5 and FIG. 7, as mentioned above, when the bit repair data RPGm is "1", it means that the corresponding memory bit is a bad bit, and the position of the set data "0" in the selection signal will be changed. Following the change, as shown in Figure 7, each time a bad bit occurs, the data "0" moves down one position, so the position of the bad memory bit can be judged according to the position of the response data "0" in the selection signal. For example, when the selection signal is "011", the data "0" appears in the first position, which means that there is a bad bit, if the selection signal is "110", the data "0" appears in the third position position, which means there are three bad bits. The judging circuits 20 , 22 , 24 and 26 of the present invention can not only be used for judging whether the corresponding memory bit is a bad bit, but also can be used as a bad bit accumulating circuit to accumulate the number of bad bits. The accumulating circuit shown in FIG. 5 is not limited to be applied to the memory, but can also be applied to circuits other than the memory to determine the number of “0” or “1” levels in a plurality of signals (RPG1 to RPGm). Similarly, the circuit of FIG. 14 can also be used as an accumulator circuit.

圖17顯示一種根據圖5電路概念的累加電路110,其包括多個基本電路112、多個開關116以及一信號偵測器及計數器118,多個基本電路112分別接收多個信號S1~Sm,當累加電路110應用在記憶體中計數不佳位元的數量時,信號S1~Sm為位元修補資料。在此實施例中,基本電路112的架構如同圖5的基本電路50,每一個基本電路112都包括三個反及閘1121、1123及1124以及一個反相器1122,在其他實施例中,基本電路112也可以用多個反或閘來實現,或由多種邏輯閘元件的組合來實現。由於基本電路112是由簡單的邏輯閘組成,故累加電路110的架構簡單且面積小。在第一個基本電路112中,反及閘1121的第一輸入端接收信號GND=“0”或VDD=“1”,反及閘1121的第二輸入端接收信號VDD=“1”,其中信號GND=“0”為預設的反應資料,反及閘1121的輸出端連接反及閘1123及1124的一輸入端,信號S1連接反及閘1124的另一輸入端以及經反相器1122連接反及閘1123的另一輸入端。當第一個基本電路112的反及閘1121的第一輸入端接收的信號VDD時,代表累加電路110為關閉狀態,當反及閘1121的第一輸入端接收的信號為GND時,代表累加電路110為啟動狀態。當第一個基本電路112的反及閘1121的第一輸入端的信號由VDD變為GND時,若信號S1為“1”,,此時反及閘1123輸出信號“1”而反及閘1124輸出信號“0”,信號偵測器及計數器118偵測到反及閘1124所輸出的信號為反應資料“0”時,信號偵測器及計數器118的計數值加1,同時信號偵測器及計數器118會輸出控制信號Sc。相反的,若信號S1為“0”,此時反及閘1123輸出信號“0”而反及閘1124輸出信號“1”,信號偵測器及計數器118偵測到反及閘1124所輸出的信號並非反應資料“0”時,計數器不動作。前一級的基本電路112的反及閘1123的輸出端連接後一級的基本電路112的反及閘1121的第一輸入端,開關116連接後一級的基本電路112的反及閘1121的第二輸入端,開關116是受控於信號偵測器及計數器118輸出的控制信號Sc,在初始狀態下,後一級的基本電路112的反及閘1121的第二輸入端的初始準位被設定為VDD(圖中未示),當反及閘1124所輸出的信號為反應資料“0”時,信號偵測器及計數器118輸出控制信號Sc,使開關116將前一級的基本電路112的反及閘1124的輸出端連接至後一級的基本電路112的反及閘1121的第二輸入端,進而使預設的反應資料GND=“0”輸入至後一級的基本電路112。簡單來說,在圖17的實施例中,第M個基本電路112接收第M個信號SM,當第M個信號SM的準位為“1”時,第M個基本電路提供預設的反應資料GND=“0”至信號偵測器及計數器118,使信號偵測器及計數器118的計數值增加以累計具有準位“1”的信號的數量。在其他實施例中,信號偵測器及計數器118也可以累計信號S1~Sm中,準位為“0”的信號的數量。在一實施例中,基本電路112也可以用圖8所示的電路來實現。FIG. 17 shows an accumulating circuit 110 according to the circuit concept of FIG. 5, which includes a plurality of basic circuits 112, a plurality of switches 116, a signal detector and a counter 118, and the plurality of basic circuits 112 respectively receive a plurality of signals S1-Sm, When the accumulating circuit 110 is used to count the number of bad bits in the memory, the signals S1 ˜ Sm are bit repair data. In this embodiment, the structure of the basic circuit 112 is similar to the basic circuit 50 in FIG. 5 , and each basic circuit 112 includes three inverting gates 1121 , 1123 and 1124 and an inverter 1122 . In other embodiments, the basic circuit 112 is basically Circuit 112 may also be implemented with multiple inverting OR gates, or a combination of multiple logic gate elements. Since the basic circuit 112 is composed of simple logic gates, the structure of the accumulating circuit 110 is simple and the area is small. In the first basic circuit 112, the first input terminal of the inversion gate 1121 receives the signal GND=“0” or VDD=“1”, and the second input terminal of the inversion gate 1121 receives the signal VDD=“1”, wherein The signal GND=“0” is the default response data, the output terminal of the inverter gate 1121 is connected to an input terminal of the inverter gate 1123 and 1124 , the signal S1 is connected to the other input terminal of the inverter gate 1124 and the inverter 1122 Connect the other input terminal of the inverter gate 1123. When the signal VDD received by the first input terminal of the inversion gate 1121 of the first basic circuit 112 indicates that the accumulating circuit 110 is in a closed state, when the signal received by the first input terminal of the inversion gate 1121 is GND, it indicates that the accumulation circuit 110 is in a closed state. Circuit 110 is in an enabled state. When the signal of the first input terminal of the inversion gate 1121 of the first basic circuit 112 changes from VDD to GND, if the signal S1 is "1", then the inversion gate 1123 outputs the signal "1" and the inversion gate 1124 The output signal "0", the signal detector and the counter 118 detect that the signal output by the inversion gate 1124 is the response data "0", the count value of the signal detector and the counter 118 is increased by 1, and the signal detector And the counter 118 outputs the control signal Sc. Conversely, if the signal S1 is "0", the inversion gate 1123 outputs a signal "0" and the inversion gate 1124 outputs a signal "1", and the signal detector and counter 118 detect the output signal of the inversion gate 1124. When the signal is not the response data "0", the counter does not operate. The output terminal of the inverting and gate 1123 of the basic circuit 112 of the previous stage is connected to the first input terminal of the inverting and gate 1121 of the basic circuit 112 of the subsequent stage, and the switch 116 is connected to the second input of the inverting and gate 1121 of the basic circuit 112 of the subsequent stage. The switch 116 is controlled by the signal detector and the control signal Sc output by the counter 118. In the initial state, the initial level of the second input terminal of the inverting gate 1121 of the basic circuit 112 of the subsequent stage is set to VDD ( (not shown in the figure), when the signal output by the inverting gate 1124 is the response data "0", the signal detector and the counter 118 output the control signal Sc, so that the switch 116 switches the inverting gate 1124 of the basic circuit 112 of the previous stage The output terminal of φ is connected to the second input terminal of the inverting gate 1121 of the basic circuit 112 of the subsequent stage, so that the preset response data GND=“0” is input to the basic circuit 112 of the subsequent stage. To put it simply, in the embodiment of FIG. 17 , the M-th basic circuit 112 receives the M-th signal SM, and when the level of the M-th signal SM is “1”, the M-th basic circuit provides a preset response The data GND=“0” to the signal detector and counter 118 increases the count value of the signal detector and counter 118 to accumulate the number of signals with a level of “1”. In other embodiments, the signal detector and counter 118 may also accumulate the number of signals whose level is "0" in the signals S1 to Sm. In one embodiment, the basic circuit 112 may also be implemented with the circuit shown in FIG. 8 .

10:記憶體 12:重新排序電路 20:判斷電路 22:判斷電路 24:判斷電路 26:判斷電路 30:多工器 3002:反相器 3003:反相器 3004:開關 3005:反相器 3006:及閘 3007:反相器 3008:開關 3009:反相器 3010:及閘 3011:反相器 3012:開關 3013:反相器 3014:及閘 3015:反相器 3016:開關 3017:反相器 3018:及閘 3019:反相器 3020:開關 32:多工器 34:多工器 36:多工器 40:多餘位元排序電路 50:基本電路 52:反及閘 54:反相器 56:反及閘 58:反及閘 60:PMOS電晶體 62:NMOS電晶體 70:判斷電路 72:判斷電路 74:判斷電路 76:判斷電路 78:多工器 7802:反相器 7804:開關 7803:反相器 7805:反相器 7807:反相器 7808:開關 7809:反相器 7811:反相器 7812:開關 7813:反相器 7815:反相器 7816:開關 7817:反相器 7819:反相器 7820:開關 80:多工器 82:多工器 84:多工器 90:基本電路 92:反及閘 94:反相器 96:反及閘 98:反及閘 100:NMOS電晶體 102:PMOS電晶體 110:累加電路 112:基本電路 116:開關 118:信號偵測器及計數器10: Memory 12: Reordering the circuit 20: Judgment circuit 22: Judgment circuit 24: Judgment circuit 26: Judgment circuit 30: Multiplexer 3002: Inverter 3003: Inverter 3004: switch 3005: Inverter 3006: and gate 3007: Inverter 3008: switch 3009: Inverter 3010: and gate 3011: Inverter 3012: switch 3013: Inverter 3014: and gate 3015: Inverter 3016: Switch 3017: Inverter 3018: and gate 3019: Inverter 3020: Switch 32: Multiplexer 34: Multiplexer 36: Multiplexer 40: redundant bit sorting circuit 50: Basic Circuit 52: Reverse and gate 54: Inverter 56: Reverse and gate 58: Reverse and gate 60: PMOS transistor 62: NMOS transistor 70: Judgment circuit 72: Judgment circuit 74: Judgment circuit 76: Judgment circuit 78: Multiplexer 7802: Inverter 7804: Switch 7803: Inverter 7805: Inverter 7807: Inverter 7808: Switch 7809: Inverter 7811: Inverter 7812: Switch 7813: Inverter 7815: Inverter 7816: switch 7817: Inverter 7819: Inverter 7820: switch 80: Multiplexer 82: Multiplexer 84: Multiplexer 90: Basic Circuits 92: Reverse and gate 94: Inverter 96: Reverse and gate 98: Reverse and gate 100: NMOS transistor 102: PMOS transistor 110: Accumulator circuit 112: Basic Circuits 116: switch 118: Signal detectors and counters

圖1顯示應用本發明重新排序電路的記憶體。 圖2顯示圖1中重新排序電路的實施例。 圖3顯示圖2中多工器的實施例。 圖4顯示應用在圖2中判斷電路的基本電路。 圖5顯示圖2中判斷電路的實施例。 圖6顯示圖4的基本電路的概念圖。 圖7是圖5電路的概念圖。 圖8顯示圖4的基本電路的另一種實施例。 圖9顯示本發明重新排序記憶體位元的方法的另一實施例。 圖10顯示圖9中重新排序電路的實施例。 圖11顯示圖10中的多工器。 圖12顯示構成圖10中判斷電路的基本電路。 圖13是圖12電路的概念圖。 圖14顯示圖10中判斷電路的實施例。 圖15是圖14電路的概念圖。 圖16顯示圖12中基本電路的另一實施例。 圖17顯示一種累加電路的實施例。FIG. 1 shows a memory to which the reordering circuit of the present invention is applied. FIG. 2 shows an embodiment of the reordering circuit of FIG. 1 . FIG. 3 shows an embodiment of the multiplexer of FIG. 2 . FIG. 4 shows the basic circuit applied to the judgment circuit in FIG. 2 . FIG. 5 shows an embodiment of the judgment circuit in FIG. 2 . FIG. 6 shows a conceptual diagram of the basic circuit of FIG. 4 . FIG. 7 is a conceptual diagram of the circuit of FIG. 5 . FIG. 8 shows another embodiment of the basic circuit of FIG. 4 . FIG. 9 shows another embodiment of the method for reordering memory bits according to the present invention. FIG. 10 shows an embodiment of the reordering circuit of FIG. 9 . FIG. 11 shows the multiplexer of FIG. 10 . FIG. 12 shows a basic circuit constituting the judgment circuit in FIG. 10 . FIG. 13 is a conceptual diagram of the circuit of FIG. 12 . FIG. 14 shows an embodiment of the judgment circuit in FIG. 10 . FIG. 15 is a conceptual diagram of the circuit of FIG. 14 . FIG. 16 shows another embodiment of the basic circuit of FIG. 12 . Figure 17 shows an embodiment of an accumulation circuit.

110:累加電路110: Accumulator circuit

112:基本電路112: Basic Circuits

116:開關116: switch

118:信號偵測器及計數器118: Signal detectors and counters

Claims (8)

一種累加電路,包括: 一信號偵測器及計數器; 以及 多個基本電路,連接信號偵測器及計數器,分別接收多個信號; 其中,該多個基本電路的第M個基本電路所接收的信號為第一準位時,該第M個基本電路提供一預設的反應資料至該信號偵測器及計數器,使該信號偵測器及計數器的計數值增加以累計具有第一準位的信號的數量。An accumulating circuit comprising: a signal detector and counter; and A plurality of basic circuits are connected to the signal detector and the counter to receive a plurality of signals respectively; Wherein, when the signal received by the Mth basic circuit of the plurality of basic circuits is at the first level, the Mth basic circuit provides a preset response data to the signal detector and the counter, so that the signal detection The count values of the detector and the counter are increased to accumulate the number of signals having the first level. 如請求項1的累加電路,其中該多個基本電路包括多個反及閘或反或閘。The accumulating circuit of claim 1, wherein the plurality of basic circuits comprise a plurality of inverting or inverting or gates. 如請求項1的累加電路,其中該多個基本電路包括MOS電晶體。The accumulation circuit of claim 1, wherein the plurality of basic circuits include MOS transistors. 一種累加電路,包括: 多個判斷電路,用以接收多個信號以及用以計數該多個信號中具有第一準位的信號的數量,每一該判斷電路具有多個基本電路; 其中,每一該判斷電路接收一選擇信號,且該選擇信號具有多個位元,該多個位元的其中一個為一反應資料; 其中,每一該判斷電路對應該多個信號的其中一個,且每一該判斷電路判斷對應的信號是否為第一準位; 其中,當該對應的信號為第一準位時,每一該判斷電路移動該反應資料的位置,並將調整後的選擇信號傳送給下一個判斷電路; 其中,當該對應的信號為第二準位時,每一該判斷電路將該選擇信號傳送給下一個判斷電路; 其中,在該多個判斷電路的最後一個判斷電路所輸出的選擇信號中,該反應資料的位置代表該多個信號中具有第一準位的信號的數量。An accumulating circuit comprising: a plurality of judging circuits for receiving a plurality of signals and for counting the number of signals having a first level in the plurality of signals, each of the judging circuits having a plurality of basic circuits; wherein, each of the judging circuits receives a selection signal, and the selection signal has a plurality of bits, and one of the plurality of bits is a response data; Wherein, each of the judging circuits corresponds to one of the plurality of signals, and each of the judging circuits judges whether the corresponding signal is a first level; Wherein, when the corresponding signal is at the first level, each of the judgment circuits moves the position of the response data, and transmits the adjusted selection signal to the next judgment circuit; Wherein, when the corresponding signal is at the second level, each of the judging circuits transmits the selection signal to the next judging circuit; Wherein, in the selection signal output by the last judgment circuit of the plurality of judgment circuits, the position of the response data represents the number of signals having the first level among the plurality of signals. 如請求項4的累加電路,其中該多個基本電路接收該對應的信號。The accumulation circuit of claim 4, wherein the plurality of base circuits receive the corresponding signals. 如請求項4的累加電路,其中該多個基本電路的其中一個接收該對應的信號。The accumulation circuit of claim 4, wherein one of the plurality of basic circuits receives the corresponding signal. 如請求項4的累加電路,其中該多個基本電路包括多個反及閘或反或閘。The accumulating circuit of claim 4, wherein the plurality of basic circuits comprise a plurality of inverting or inverting or gates. 如請求項4的累加電路,其中該多個基本電路包括MOS電晶體。The accumulation circuit of claim 4, wherein the plurality of basic circuits include MOS transistors.
TW109133293A 2019-11-20 2019-11-20 accumulator circuit TWI764297B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW109133293A TWI764297B (en) 2019-11-20 2019-11-20 accumulator circuit

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW109133293A TWI764297B (en) 2019-11-20 2019-11-20 accumulator circuit

Publications (2)

Publication Number Publication Date
TW202121416A TW202121416A (en) 2021-06-01
TWI764297B true TWI764297B (en) 2022-05-11

Family

ID=77517025

Family Applications (1)

Application Number Title Priority Date Filing Date
TW109133293A TWI764297B (en) 2019-11-20 2019-11-20 accumulator circuit

Country Status (1)

Country Link
TW (1) TWI764297B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20080052015A1 (en) * 2006-04-06 2008-02-28 Advantest Corporation Test apparatus and test method
US7444566B2 (en) * 2006-03-06 2008-10-28 Verigy (Singapore) Pte. Ltd. Memory device fail summary data reduction for improved redundancy analysis
TW201916060A (en) * 2017-09-18 2019-04-16 塔普思科技公司 Memory bit repair method capable of reducing the amount of redundant memory bits used for repairing
US20190114140A1 (en) * 2018-07-12 2019-04-18 Intel Corporation Adder circuitry for very large integers
US10348320B1 (en) * 2018-06-26 2019-07-09 International Business Machines Corporation Charge-scaling adder circuit

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7444566B2 (en) * 2006-03-06 2008-10-28 Verigy (Singapore) Pte. Ltd. Memory device fail summary data reduction for improved redundancy analysis
US20080052015A1 (en) * 2006-04-06 2008-02-28 Advantest Corporation Test apparatus and test method
TW201916060A (en) * 2017-09-18 2019-04-16 塔普思科技公司 Memory bit repair method capable of reducing the amount of redundant memory bits used for repairing
US10348320B1 (en) * 2018-06-26 2019-07-09 International Business Machines Corporation Charge-scaling adder circuit
US20190114140A1 (en) * 2018-07-12 2019-04-18 Intel Corporation Adder circuitry for very large integers

Also Published As

Publication number Publication date
TW202121416A (en) 2021-06-01

Similar Documents

Publication Publication Date Title
US6940765B2 (en) Repair apparatus and method for semiconductor memory device to be selectively programmed for wafer-level test or post package test
US7602660B2 (en) Redundancy circuit semiconductor memory device
US6947322B2 (en) Semiconductor memory device
JP2009176384A (en) Semiconductor memory device
KR101034967B1 (en) Data input / output control circuit of semiconductor memory device and data input / output method using the same
US7227782B2 (en) NAND flash memory device capable of improving read speed
US6822912B2 (en) Semiconductor device
US6868021B2 (en) Rapidly testable semiconductor memory device
TWI764297B (en) accumulator circuit
TWI709143B (en) Memory reordering circuit and method for reordering memory bits
US6535436B2 (en) Redundant circuit and method for replacing defective memory cells in a memory device
JP7502396B2 (en) Semiconductor memory device
US7681095B2 (en) Methods and apparatus for testing integrated circuits
US7548479B2 (en) Semiconductor memory device and manufacturing method thereof
US20040076042A1 (en) High performance memory column group repair scheme with small area penalty
US6314033B1 (en) Semiconductor memory device with redundancy circuit
TWI711036B (en) Repair circuit of memory and method thereof
KR20020025030A (en) Semiconductor memory device and method for replacing redundancy circuit
US6928009B2 (en) Redundancy circuit for memory array and method for disabling non-redundant wordlines and for enabling redundant wordlines
JPH0581899A (en) Semiconductor memory
JP4872299B2 (en) Semiconductor memory device
JPH0373500A (en) Semiconductor storage device
JPH05234396A (en) Semiconductor memory device
JP2001229696A (en) Semiconductor integrated circuit and its test method