[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

CN103345936A - Write-in circuits and read-out circuits of arbitrary-K-value DRAM and eight-value DRAM - Google Patents

Write-in circuits and read-out circuits of arbitrary-K-value DRAM and eight-value DRAM Download PDF

Info

Publication number
CN103345936A
CN103345936A CN2013102110232A CN201310211023A CN103345936A CN 103345936 A CN103345936 A CN 103345936A CN 2013102110232 A CN2013102110232 A CN 2013102110232A CN 201310211023 A CN201310211023 A CN 201310211023A CN 103345936 A CN103345936 A CN 103345936A
Authority
CN
China
Prior art keywords
circuit
value
output
input
dinj
Prior art date
Legal status (The legal status 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 status listed.)
Granted
Application number
CN2013102110232A
Other languages
Chinese (zh)
Other versions
CN103345936B (en
Inventor
方振贤
刘莹
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Heilongjiang University
Original Assignee
Heilongjiang University
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 Heilongjiang University filed Critical Heilongjiang University
Priority to CN201310211023.2A priority Critical patent/CN103345936B/en
Publication of CN103345936A publication Critical patent/CN103345936A/en
Application granted granted Critical
Publication of CN103345936B publication Critical patent/CN103345936B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Dram (AREA)

Abstract

The invention relates to write-in circuits and read-out circuits of an arbitrary-K-value DRAM and an eight-value DRAM. The write-in circuits and the read-out circuits have same structure characteristics, and the write-in circuits are each designed to provide a multi-value signal increased by delta than write-in circuit input; an output waveform of a memory cell is considered to be smaller than an input waveform or be of an unequal step shape, the read-out circuits are each designed for correction, and an irregular multi-value signal is converted into a regular (equal step) multi-value signal. The write-in circuits and the read-out circuits both have good quantified shaping effects; when a change of a Cj voltage is not greater than an upper new threshold or a lower new threshold, original multi-value information can be easily recovered, so the write-in circuits and the read-out circuits have anti-jamming capability and multi-value information recovery capability. The write-in circuits and the read-out circuits are mainly applied to the fields of VLSI and other digital IC technologies, such as an FPGA, a CPLD, a semi-custom or a full-custom ASIC, a memorizer and the like.

Description

任意K值和8值DRAM的写入电路和读出电路Write circuit and read circuit of arbitrary K value and 8-value DRAM

技术领域technical field

本发明是针对申请号:201110097206.7的分案申请,属于数字集成电路领域,具体地说是一种任意K值和8值DRAM的写入电路和读出电路。The present invention is directed to a divisional application with application number: 201110097206.7, and belongs to the field of digital integrated circuits, specifically a writing circuit and a reading circuit of an arbitrary K value and 8-value DRAM.

背景技术Background technique

随着MOS集成电路技术的飞速发展,集成规模越来越大,集成度越来越高,VLSI(超大规模集成电路)出现一些不足:①首先在VLSI基片上,布线却占用70%以上的硅片面积;在可编程逻辑器件(如FPGA和CPLD)中也需有大量可编程内部连线(包括可编程连接开关,如熔丝型开关、反熔丝型开关、浮栅编程元件等),将各逻辑功能块或输入/输出连接起来,完成特定功能的电路,布线(包括编程连接开关)占了材料很大的成本。减少布线成本的比重成为十分重要的问题。②从信息传输方面看,采用多值信号可减少连线数;对每根连线传输数字信息,二值信号是携带信息量最低的一种,多值信号携带信息量大于二值信号。③从信息存储方面看,采用多值信号可提高信息存储密度,特别是利用MOS管栅极电容存储信息(用于动态随机存取存储器DRAM中),因同一电容存储信息量多值比二值大,多值DRAM比二值DRAM可大大提高信息存储密度。目前多值器件的研制已广泛开展,东芝与Sandisk公司通过70nm的CMOS技术和2bit/单元的多值技术相配合,在146mm2的芯片上实现了8Gbit的存储容量;东芝与美国SanDisk发表了通过采用43nm工艺和2bit/单元多值技术实现的16gbitNAND闪存。三星开发的8Gbit产品采用63nm的CMOS技术和2bit/单元的多值技术。4值存储器的研制成功和商品化是多值研究的重要的一步,但需要控制或改变管的开关阈值Vtn,改变阈值方法是在半导体制造工艺中用多级离子注入技术,或控制浮游栅极存储的电子量等方法控制阈值。尚未发现有多于4值的DRAM的研制成功。With the rapid development of MOS integrated circuit technology, the integration scale is getting larger and higher, and VLSI (Very Large Scale Integration) has some shortcomings: ① First, on the VLSI substrate, the wiring takes up more than 70% of the silicon Chip area; in programmable logic devices (such as FPGA and CPLD), there are also a large number of programmable internal wiring (including programmable connection switches, such as fuse switches, anti-fuse switches, floating gate programming components, etc.), To connect each logic function block or input/output to complete a specific function circuit, wiring (including programming connection switches) accounts for a large cost of materials. Reducing the proportion of wiring costs becomes a very important issue. ② From the perspective of information transmission, the use of multi-valued signals can reduce the number of connections; for each connection to transmit digital information, binary signals carry the lowest amount of information, and multi-valued signals carry more information than binary signals. ③ From the perspective of information storage, the use of multi-valued signals can increase the density of information storage, especially the use of MOS transistor gate capacitance to store information (used in dynamic random access memory DRAM), because the same capacitor stores more information than binary Large, multi-valued DRAM can greatly increase information storage density than binary DRAM. At present, the development of multi-valued devices has been widely carried out. Toshiba and Sandisk have cooperated with 70nm CMOS technology and 2bit/unit multi-valued technology to realize a storage capacity of 8Gbit on a 146mm 2 chip; Toshiba and SanDisk of the United States have published a 16gbit NAND flash memory implemented with 43nm process and 2bit/cell multi-value technology. The 8Gbit product developed by Samsung adopts 63nm CMOS technology and 2bit/unit multi-value technology. The successful development and commercialization of 4-value memory is an important step in multi-value research, but it is necessary to control or change the switching threshold V tn of the tube. The method of changing the threshold is to use multi-level ion implantation technology in the semiconductor manufacturing process, or control the floating gate The threshold is controlled by methods such as the amount of electrons stored in the electrode. The development of DRAM with more than 4 values has not been found to be successful.

半导体存储器可以分为只读存储器ROM和随机存储器RAM。而RAM又分为双极型和MOS型两类。双极型RAM工作速度高,但制造工艺复杂、功耗大、集成度低,主要用于高速工作的场合。MOS型RAM又分为静态随机存取存储器SRAM和动态随机存取存储器DRAM(Dynamic Random Access Memory)两种。DRAM存储信息的原理是基于MOS管栅极电容的电荷存储效应。由于栅极存储电容的容量很小(通常仅为几皮法),而漏电流又不可能绝对等于零,所以电荷保存的时间有限;为了及时补充漏掉的电荷以避免存储的信号丢失,必须定时给栅极存储电容补充电荷,通常将这种操作称为刷新或再生,DRAM工作时必须辅以必要的刷新控制电路。DRAM是由大的矩形存储单元阵列与用来对阵列读和写的支持性逻辑电路,以及维持存储数据完整性的刷新电路等组成。在DRAM中最简单的可用单管动态存储单元。存储单元是按行、列排成矩阵式结构,用两个译码电路分别译码。X向译码称为行译码,其输出线称为字线,它选中存储矩阵中一行的所有存储单元。Y向译码又称为列译码,其输出线称为位线。因单管动态存储单元每次读出为破坏性读出,存储电容向位线上的电容CB提供电荷,使存储电容电荷减少,需立即恢复,在每根位线上接有灵敏度恢复/读出放大器,使用了灵敏度恢复/读出放大器之后,在每次读出数据的同时完成了对存储单元原来所存数据的恢复。一般将DRAM设计为字长n位(即一字有n位,如4位,8位或N位),对地址译码器译出每一字线输出有效时,有n个(如4个,8个或N个)存储单元同时被选中,使这些被选中的存储单元经读/写控制电路进行读写操作,DRAM读写控制电路控制数据信息输入输出。外界对存储器的控制信号有读信号RD、写信号WR和片选信号CS等等。DRAM的输入输出数据的位数有1位,2位,4位或N位。除多位输入输出外,为了提高集成度的同时减少器件引脚的数目,大容量DRAM常常采用1位输入、1位输出和地址分时输入的方式,相应的有输入缓冲器,输出缓冲器和输出锁存器等。Semiconductor memory can be divided into read-only memory ROM and random access memory RAM. RAM is divided into two types: bipolar type and MOS type. Bipolar RAM has a high working speed, but its manufacturing process is complicated, its power consumption is high, and its integration level is low. It is mainly used in high-speed working occasions. MOS-type RAM is divided into static random access memory SRAM and dynamic random access memory DRAM (Dynamic Random Access Memory) two. The principle of DRAM storing information is based on the charge storage effect of the gate capacitance of the MOS transistor. Since the capacity of the gate storage capacitor is very small (usually only a few picofarads), and the leakage current cannot be absolutely equal to zero, the charge storage time is limited; in order to replenish the missing charge in time to avoid the loss of the stored signal, it must be timed Supplementing charge to the gate storage capacitor is usually called refresh or regeneration, and the necessary refresh control circuit must be supplemented when DRAM works. DRAM is composed of a large rectangular array of memory cells, supporting logic circuits for reading and writing to the array, and refresh circuits to maintain the integrity of stored data. The simplest available single-transistor dynamic memory cell in DRAM. The memory cells are arranged in a matrix structure by rows and columns, and are decoded by two decoding circuits. X-direction decoding is called row decoding, and its output line is called word line, which selects all memory cells in a row in the memory matrix. Y-direction decoding is also called column decoding, and its output lines are called bit lines. Because each readout of the single-transistor dynamic memory cell is a destructive readout, the storage capacitor provides charges to the capacitor C B on the bit line, so that the charge of the storage capacitor is reduced and needs to be restored immediately. A sensitivity recovery/ The sense amplifier, after using the sensitivity restoration/sense amplifier, completes the restoration of the original stored data of the memory cell while reading out the data each time. Generally, DRAM is designed as a word length of n bits (that is, a word has n bits, such as 4 bits, 8 bits or N bits), and when the address decoder decodes each word line output is valid, there are n (such as 4 , 8 or N) memory cells are selected at the same time, so that these selected memory cells are read and written through the read/write control circuit, and the DRAM read and write control circuit controls the input and output of data information. External control signals to the memory include read signal RD , write signal WR , chip select signal CS and so on. The number of input and output data of DRAM is 1 bit, 2 bits, 4 bits or N bits. In addition to multi-bit input and output, in order to improve integration and reduce the number of device pins, large-capacity DRAM often adopts 1-bit input, 1-bit output, and address time-sharing input, and there are corresponding input buffers and output buffers. and output latches, etc.

现有技术和存在问题:Existing technology and existing problems:

1.对存储在DRAM的存储电容中的多值信号,读出数据是困难重重的(二值数据是按存储电容的电荷的有和无来决定的,很容易读出;多值信号读出要区分出量级,而且多值信号在传输中可能出现衰减和变形,常规放大器对多值信号容易形成严重失真,得不到DRAM规范的等阶梯多值信号输入输出,常规读出放大器方法不能读出多值信号,不能实现任意K值和8值DRAM的存储单元电路,尚未发现有多于4值的DRAM的研制成功。为克服此困难,不能按传统方法单纯考虑多值存储单元,必须同时考虑与多值存储单元相配合的多值写入电路和多值读出电路。对字长4位,8位或N位数据,则相应的写入电路和读出电路有4个,8个或N个。写入电路和读出电路以要求得到DRAM的规范的等阶梯多值输入输出信号为前提,多值DRAM存储单元电路,写入电路和读出电路应按一个总的发明构思来设计这三种电路,该三种电路是密切相关,但实用时三种电路数量各不相同(不能组成一种电路的整体),按一个总的发明构思来设计可克服读出存储在电容多值信号数据的困难。1. For the multi-value signal stored in the storage capacitor of DRAM, it is difficult to read the data (binary data is determined by the presence or absence of the charge of the storage capacitor, and it is easy to read; multi-value signal readout It is necessary to distinguish the magnitude, and multi-valued signals may be attenuated and deformed during transmission. Conventional amplifiers are prone to serious distortion of multi-valued signals, and the input and output of equal-step multi-valued signals in DRAM specifications cannot be obtained. Conventional sense amplifier methods cannot Read multi-valued signal, can not realize the storage unit circuit of arbitrary K value and 8-value DRAM, have not yet found to have the successful development of the DRAM of more than 4 values.For overcoming this difficulty, can not simply consider multi-valued storage unit by traditional method, must Consider simultaneously the multi-value write circuit and the multi-value read circuit that cooperate with multi-value storage unit.To word length 4, 8 or N data, then corresponding write circuit and read-out circuit have 4, 8 One or N. The write-in circuit and the read-out circuit are based on the premise that the equal ladder multi-value input and output signals of the DRAM are required to be obtained, and the multi-value DRAM storage unit circuit, the write-in circuit and the read-out circuit should be according to a general inventive concept To design these three kinds of circuits, these three kinds of circuits are closely related, but when practical, the quantities of the three kinds of circuits are different (the whole of a kind of circuit cannot be formed), and designing according to a general inventive concept can overcome the problem of reading out stored in the capacitor. Difficulties with multivalued signal data.

2.在实现多值电路中,已有技术控制MOS管阈值有很大的缺点:①控制阈值的幅度有限(因离子注入浓度是有限的),开启分辨率低;而且工艺中控制阈值幅度常会改变MOS管的性能,例如阈值电压的降低回导致切断电流的剧增,阈值电压的调整对管的性能和稳定性有影响,稳定的Vtn非常重要。对多值记忆,注入浮游栅极的电子量是连续变化的,需极精细地控制,各门槛电压电平尚达不到准稳定状态。因此目前实用的电压型多值电路不大于4值电路,更多值电路应用较困难。②只能控制阈值的幅度,不能改变MOS管开启性质(如变≥t导通为<t导通),而多值逻辑门须有二种开启性质的MOS管,才能使组合电路结构最简,例如多值非门、多值右移门和多值跟随器的电路结构本应完全相同,只是阈值电压及其开启性质不同。然而目前只控制阈值幅度的工艺,使上述多值门结构差别很大,结构复杂,影响其实现。③需要增加离子注入额外的工序,只能在半导体制造工艺中控制阈值,既增加工艺复杂性,又不能后由用户来控制阈值,或对阈值用户不可编程。2. In the realization of multi-valued circuits, the existing technology to control the threshold of MOS transistors has great disadvantages: ① the amplitude of the control threshold is limited (because the ion implantation concentration is limited), and the resolution of opening is low; and the amplitude of the control threshold in the process is often Changing the performance of the MOS tube, for example, the reduction of the threshold voltage will lead to a sharp increase in the cut-off current, and the adjustment of the threshold voltage will affect the performance and stability of the tube, and a stable V tn is very important. For multi-valued memory, the amount of electrons injected into the floating gate is continuously changing and needs to be controlled very finely, and the voltage levels of each threshold cannot reach a quasi-stable state. Therefore, the current practical voltage-type multi-valued circuits are not larger than 4-valued circuits, and the application of more-valued circuits is difficult. ② Only the amplitude of the threshold value can be controlled, and the turn-on nature of the MOS tube cannot be changed (such as changing ≥t conduction to <t conduction), and the multi-valued logic gate must have two kinds of MOS tubes with turn-on properties to make the structure of the combination circuit the simplest , For example, the circuit structures of multi-valued NOT gate, multi-valued right shift gate and multi-valued follower should be exactly the same, but the threshold voltage and its opening properties are different. However, the current technology that only controls the threshold amplitude makes the structure of the above-mentioned multi-value gate very different and complex, which affects its realization. ③ It is necessary to add an additional process of ion implantation, and the threshold can only be controlled in the semiconductor manufacturing process, which not only increases the complexity of the process, but also cannot be controlled by the user later, or the threshold cannot be programmed by the user.

K值DRAM(K>2)习惯通称为多值DRAM,但在设计存储单元电路,写入和读出电路中,电路结构常与K有关,这时写明K值(称呼K值DRAM等)较为方便,而一些内容介绍、非结构性描述或与K值无关的名词常可沿用习惯称呼(如多值信号,多值门)。K-value DRAM (K>2) is commonly referred to as multi-value DRAM, but in designing storage unit circuits, writing and reading circuits, the circuit structure is often related to K, and at this time write the K value (called K-value DRAM, etc.) It is more convenient, and some content introductions, non-structural descriptions, or nouns that have nothing to do with K values can often use customary names (such as multi-valued signals, multi-valued gates).

发明内容Contents of the invention

本发明目的是公开一种任意K值和8值DRAM的写入电路和读出电路。The purpose of the invention is to disclose a writing circuit and a reading circuit of any K value and 8-value DRAM.

上述的目的通过以下的技术方案实现:Above-mentioned purpose realizes by following technical scheme:

1.本发明的一种任意K值DRAM的写入电路是这样实现的:该写入电路如图4所示,在所述的K值DRAM的写入电路中,设K=3,4,5,……;采用K-1=L个变阈型PMOS管Qak,k=1,2,3,……,L,管Qak的栅极经变阈电路连接到写入电路的输入Dinj,变阈型PMOS管Qak的新阈值为tak,管Qak导通时源极漏极间压降为0;管Qak的源极接电源Vdd,选取Vdd的电压比写入电路输入和读出电路输出的最大逻辑电平VDinj(L)和VDouj(L)高Δ,Δ是电压跟随器F输入输出间向下的直流电平偏移;采用L-1个二极管Dan,n=2,3,……,L,二极管Dan的导通电压是VDon;Dan的正极和负极分别连接到变阈型PMOS管Qan-1的漏极和管Qan的漏极;管QaL的漏极经过恒流源Ij接地,管QaL的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管QaL的漏极形成写入电路的输出Gwrij,写入电路输出Gwrij接到存储单元电路的写位线输入;选取tak为写入电路输入Dinj的K值信号的相邻逻辑电平VDinj(k)和VDinj(k-1)的平均值(VDinj(k)+VDinj(k-1))/2,即tak为VDinj(k)和VDinj(k-1)的中间值,VDinj(k)>VDinj(k-1);写入电路输入Dinj的K值信号和读出电路输出Doutj及DRAM输入输出的规范的等阶梯的K值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,也即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,L,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;写入电路的输出Gwrij除0电平外比写入电路输入的K值信号高Δ,0电平仍为0,该K值写入电路又称为K值写入增高电路。1. the write-in circuit of a kind of arbitrary K value DRAM of the present invention is realized like this: this write-in circuit as shown in Figure 4, in the write-in circuit of described K value DRAM, establish K=3,4, 5,...; K-1=L threshold-variable PMOS transistors Qak are adopted, k=1, 2, 3,..., L, and the gate of the transistor Qak is connected to the input of the writing circuit via a variable threshold circuit D inj , the new threshold value of variable-threshold PMOS transistor Qak k is ta k , the voltage drop between the source and drain is 0 when the transistor Qak k is turned on; the source of the transistor Qak k is connected to the power supply V dd , and the voltage ratio of V dd is selected The maximum logic levels V Dinj (L) and V Douj (L) of the write circuit input and read circuit output are high Δ, Δ is the downward DC level shift between the input and output of the voltage follower F; using L-1 Diode Dan , n=2, 3,..., L, the conduction voltage of the diode Dan is V Don ; the positive pole and the negative pole of Dan are connected to the drain of the threshold-variable PMOS transistor Qa n-1 and the transistor Qa respectively The drain of n ; the drain of the tube Qa L is grounded through the constant current source Ij , and the drain of the tube Qa L is connected to the constant current source I j to keep the current flowing through the conduction diode at the same fixed value, and the drain of the tube Qa L Pole forms the output G wrij of the write circuit, and the output G wrij of the write circuit is connected to the write bit line input of the memory cell circuit; ta k is selected as the adjacent logic level V Dinj of the K value signal of the write circuit input D inj ( The average value of k) and V Dinj (k-1) (V Dinj (k)+V Dinj (k-1))/2, that is, ta k is the middle of V Dinj (k) and V Dinj (k-1) value, V Dinj (k)>V Dinj (k-1); the K value signal of the input circuit D inj is the same as the K value signal of the output D outj of the readout circuit and the K value signal of the norm of DRAM input and output are the same The difference between adjacent logic levels of input D inj is equal, the difference of adjacent logic levels of output D outj is equal, and the step voltages of input D inj and output D out j are the same, and the step voltage is V Don , that is Satisfy V Dinj (m)-V Dinj (m-1)=V Doutj (m)-V Doutj (m-1)=V Don , m=1, 2, 3,..., L, V Dinj (m) and V Doutj (m) are respectively the logic level of the input of the write circuit and the output logic value of the readout circuit; the output G wrij of the write circuit is Δ higher than the K value signal input by the write circuit except for 0 level, The 0 level is still 0, and the K value writing circuit is also called the K value writing boosting circuit.

2.本发明的一种任意K值DRAM的读出电路是这样实现的:读出电路如图5所示,该任意K值DRAM的读出电路是按照上述1所述的任意K值DRAM的写入电路的相同结构特征而形成的,在所述的K值DRAM的读出电路中,设K=3,4,5,……;采用K-1=L个变阈型PMOS管Qbk,k=1,2,3,……,L,管Qbk的栅极经变阈电路连接到读出电路的输入Grdij,变阈型PMOS管Qbk的新阈值为tbk,管Qbk导通时源极漏极间压降为0;Grdij接到存储单元电路的读位线输出,管Qbk的源极接电源Vdc,选取Vdc的电压等于写入电路输入和读出电路输出的最大逻辑电平VDinj(L)和VDouj(L)(即Vdc=VDinj(L)=VDouj(L));采用L-1个二极管Dbn,n=2,3,……,L,二极管Dbn的导通电压是VDon;Dbn的正极和负极分别连接变阈型PMOS管Qbn-1的漏极和管Qbn的漏极;变阈型PMOS管QbL的漏极经过恒流源Ij接地,管QbL的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管QbL的漏极形成读出电路的输出Doutj;选取tbk为读出电路输入Grdij的K值信号的相邻逻辑电平VGrdij(k)和VGrdij(k-1)的平均值(VGrdij(k)+VGrdij(k-1))/2,即tbk为VGrdij(k)和VGrdij(k-1)的中间值,VGrdij(k)>VGrdij(k-1);读出电路输出Doutj的信号和写入电路输入Dinj及DRAM输入输出的规范的等阶梯的K值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,也即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,L,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;读出电路输入Grdij是来自存储单元电路输出的不规范的K值信号,所述不规范的K值信号就是对比DRAM输入输出和写入电路输入及读出电路输出的规范的等阶梯的K值信号为逻辑电平幅度不一致;读出电路输出Doutj为规范的等阶梯的K值信号,即读出电路将不规范的K值信号输入Grdij转换为规范的等阶梯的K值信号输出Doutj,该K值读出电路又称为K值读出校正电路。2. the readout circuit of a kind of arbitrary K value DRAM of the present invention is realized like this: readout circuit as shown in Figure 5, the readout circuit of this arbitrary K value DRAM is according to the described arbitrary K value DRAM of above-mentioned 1 Write in the same structural feature of the circuit and form, in the readout circuit of described K value DRAM, set K=3,4,5,...; Adopt K-1=L variable threshold type PMOS transistor Qb k , k=1, 2, 3,..., L, the gate of tube Qb k is connected to the input G rdij of the readout circuit through the variable threshold circuit, the new threshold value of variable threshold PMOS tube Qb k is tb k , tube Qb When k is turned on, the voltage drop between the source and the drain is 0; G rdij is connected to the output of the read bit line of the memory cell circuit, the source of the tube Qb k is connected to the power supply V dc , and the voltage of V dc is selected to be equal to the input and read of the write circuit The maximum logic levels V Dinj (L) and V Douj (L) output by the output circuit (that is, V dc = V Dinj (L) = V Douj (L)); use L-1 diodes Db n , n=2, 3, ..., L, the turn-on voltage of the diode Db n is V Don ; the anode and cathode of Db n are respectively connected to the drain of the threshold-variable PMOS transistor Qb n-1 and the drain of the transistor Qb n ; the threshold-variable PMOS The drain of the tube Qb L is grounded through the constant current source Ij , and the drain of the tube Qb L is connected to the constant current source I j so that the current flowing through the conduction diode remains at the same fixed value, and a readout circuit is formed at the drain of the tube Qb L The output D outj of; choose tb k to be the average value ( V Grdij ( k)+ V Grdij (k-1))/2, that is, tb k is the intermediate value of V Grdij (k) and V Grdij (k-1), V Grdij (k)>V Grdij (k-1); the readout circuit outputs D outj The characteristics of the signal written into the circuit input D inj and the standard K value signal of the DRAM input and output are the same: the difference between the adjacent logic levels of the input D inj is equal, and the adjacent logic levels of the output D outj The differences are equal, and the step voltages of input D inj and output D outj are the same, and the step voltage is V Don , that is, V Dinj (m)-V Dinj (m-1)=V Doutj (m)-V Doutj (m -1)=V Don , m=1, 2, 3,..., L, V Dinj (m) and V Doutj (m) are the logic levels of the input of the writing circuit and the output logic value of the reading circuit of m respectively ; The input G rdij of the readout circuit is an irregular K value signal from the output of the storage unit circuit, and the irregular K value signal is compared to the standard equal steps of the input and output of the DRAM, the input of the write circuit, and the output of the readout circuit The K value signal is inconsistency in the logic level amplitude; the output D outj of the readout circuit is a standardized K value signal of equal steps, that is, the readout circuit converts the non-standard K value signal input G rdij into a standardized K value signal of equal steps output D outj , and the K value readout circuit is also called the K value readout correction circuit.

注:本发明的一种任意K值DRAM的存储单元电路是这样实现的:如图1所示,所述的K值DRAM的存储单元电路是由电压跟随器F,F的栅极存储电容Cj和二个CMOS传输门G1和G2组成,用电容Cj存储K值信号,电压跟随器F包括NMOS管Qm1和NPN管Qm2,管Qm1的栅极接电容Cj的一端Cmij,即Cmij为电压跟随器F的输入,Cj的另一端接地,管Qm1的源极接管Qm2的栅极和电阻Rm1,Rm1的另一端接地,管Qm2的射极Folij经恒流源Ij接地,Qm2射极接恒流源Ij使管Qm2的射极负载为恒流源,管Qm2的射极Folij为F的输出,管Qm1的漏极和管Qm2的集电极都接电源Vdd,选取Vdd的电压比写入电路输入和读出电路输出的K值逻辑电平的最大值高Δ,Δ为电压跟随器F输入输出间向下的直流电平偏移;传输门G1的输入接写位线Gwrij,传输门G1的输出接F的输入Cmij,传输门G1的控制输入接写入脉冲wri,传输门G2的输入接F的输出Folij,传输门G2的输出接读位线Grdij,传输门G2的控制输入接读出脉冲rdi,写入脉冲wri和读出脉冲rdi来自DRAM的控制电路;写入脉冲wri来到时,传输门G1导通,将写位线Gwrij的K值信号传送到存储电容Cj,电容Cj接收写位线Gwrij的K值信号,电容Cj的K值信号就是F输入Cmij的K值信号;写入脉冲wri未来到时,传输门G1截止,存储电容Cj与外界为直流开路,电容Cj存储的K值信号保持不变,即具有记忆功能;读出脉冲rdi来到时,传输门G2导通,将F输出Folij的K值信号传送到读位线Grdij;写位线Gwrij和读位线Grdij各自是存储单元电路的输入和输出;存储单元电路输入和输出各自接到写入电路输出和读出电路输入;F输出的K值信号必须是与F输入信号相对应的K值信号,F输入输出信息相同,即F输出无信息丢失,F输出无信息丢失要求Cj存储的K值信号是增高的K值信号,所述增高的K值信号就是除0电平外比写入电路输入的K值信号高Δ的信号,其中0电平仍为0;Cj存储的增高的K值信号是来自写入电路的输出,即提供给Cj存储信号的写入电路输出也是增高的K值信号;Cj存储的增高K值信号经过F传送到读位线Grdij,在Grdij上形成不规范的K值信号,也即存储单元电路输出是不规范的K值信号,所述不规范的K值信号就是对比DRAM输入输出和写入电路输入及读出电路输出的规范的等阶梯的K值信号为逻辑电平幅度不一致;读出电路输入信号是来自存储单元电路输出Grdij的不规范的K值信号,读出电路输出是对存储单元电路输出不规范的K值信号校正得出的规范的等阶梯的K值信号,该校正得出的规范的等阶梯的K值信号作为对存储单元电路存储信息的校正读出。Note: the storage unit circuit of a kind of arbitrary K value DRAM of the present invention is realized like this: As shown in Figure 1, the storage unit circuit of described K value DRAM is made up of voltage follower F, the gate storage capacitance C of F j and two CMOS transmission gates G 1 and G 2 , the capacitor C j is used to store the K value signal, the voltage follower F includes NMOS transistor Q m1 and NPN transistor Q m2 , the gate of transistor Q m1 is connected to one end of capacitor C j C mij , that is, C mij is the input of voltage follower F, the other end of C j is grounded, the source of tube Q m1 takes over the gate of Q m2 and resistor R m1 , the other end of R m1 is grounded, and the emitter of tube Q m2 The electrode F olij is grounded through the constant current source I j , the emitter of Q m2 is connected to the constant current source I j so that the emitter load of the tube Q m2 is a constant current source, the emitter F olij of the tube Q m2 is the output of F, and the tube Q m1 Both the drain and the collector of the tube Q m2 are connected to the power supply V dd , and the voltage of V dd is selected to be Δ higher than the maximum value of the K value logic level of the write circuit input and the read circuit output, and Δ is the input of the voltage follower F Downward DC level shift between outputs; the input of the transmission gate G 1 is connected to the write bit line G wrij , the output of the transmission gate G 1 is connected to the input C mij of F, the control input of the transmission gate G 1 is connected to the write pulse w ri , and the transmission The input of the gate G2 is connected to the output F olij of F, the output of the transmission gate G2 is connected to the read bit line G rdij , the control input of the transmission gate G2 is connected to the read pulse r di , the write pulse w ri and the read pulse r di From the control circuit of DRAM; when the write pulse w ri comes, the transmission gate G 1 is turned on, and the K value signal of the write bit line G wrij is transmitted to the storage capacitor C j , and the capacitor C j receives the K value of the write bit line G wrij Value signal, the K value signal of capacitor C j is the K value signal of F input C mij ; when the write pulse w ri arrives in the future, the transmission gate G1 will be cut off, the storage capacitor C j and the outside world will be DC open circuit, and the capacitor C j will store The K value signal remains unchanged, that is, it has a memory function; when the read pulse r di comes, the transmission gate G2 is turned on, and the K value signal of the F output F olij is transmitted to the read bit line G rdij ; the write bit line G wrij and the read bit line G rdij are the input and output of the storage unit circuit respectively; the input and output of the storage unit circuit are respectively connected to the output of the write circuit and the input of the read circuit; the K value signal output by F must be corresponding to the F input signal K value signal, F input and output information are the same, that is, F output has no information loss, and F output has no information loss. The K value signal stored by C j is an increased K value signal, and the increased K value signal is except for 0 level. A signal Δ higher than the K value signal input by the write circuit, where the 0 level is still 0; the increased K value signal stored by C j is the output from the write circuit, that is, the write circuit that provides the C j storage signal The output is also an increased K value signal; the increased K value signal stored by C j is transmitted through F To the read bit line G rdij , an irregular K value signal is formed on G rdij , that is, the output of the storage unit circuit is an irregular K value signal, and the irregular K value signal is compared with the DRAM input and output and the write circuit The K-value signals of the standard equal steps of the input and readout circuit output are inconsistent in logic level amplitude; the input signal of the readout circuit is an irregular K-value signal from the output Grdij of the storage unit circuit, and the output of the readout circuit is for storage The unit circuit outputs a standardized equal-step K value signal obtained by correcting the non-standard K value signal, and the corrected standardized equal-step K value signal is used as a corrected readout of the information stored in the memory unit circuit.

在所述的K值DRAM存储单元电路中,恒流源Ij可由电阻Rj取代。In the K-value DRAM storage unit circuit, the constant current source Ij can be replaced by a resistor Rj .

本发明还有以下技术特征:The present invention also has the following technical characteristics:

(1)根据上述1所述的一种任意K值DRAM的写入电路,取K=8,得出8值DRAM的写入电路,如图2所示,其中采用7个变阈型PMOS管Qak,k=1,2,3,……,7,管Qak的栅极经变阈电路连接到写入电路的输入Dinj,变阈型PMOS管Qak的新阈值为tak,管Qak导通时源极漏极间压降为0;管Qak的源极接电源Vdd,选取Vdd的电压比写入电路输入和读出电路输出的最大逻辑电平VDinj(7)和VDouj(7)高Δ,Δ是电压跟随器F的输入输出间向下的直流电平偏移;采用6个二极管Dan,n=2,3,……,7,二极管Dan的导通电压是VDon;Dan的正极和负极分别连接到变阈型PMOS管Qan-1的漏极和管Qan的漏极;管Qa7的漏极经过恒流源Ij接地,管Qa7的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管Qa7的漏极形成写入电路的输出Gwrij,写入电路输出Gwrij接到存储单元电路的写位线输入;选取tak为写入电路输入Dinj的8值信号的相邻逻辑电平VDinj(k)和VDinj(k-1)的平均值(VDinj(k)+VDinj(k-1))/2,即选取tak为VDinj(k)和VDinj(k-1)的中间值,VDinj(k)>VDinj(k-1);写入电路输入Dinj的8值信号和读出电路输出Doutj及DRAM输入输出的规范的等阶梯的8值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,也即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,7,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;写入电路输出Gwrij除0电平外比写入电路输入的8值信号高Δ,0电平仍为0,该8值写入电路又称为8值写入增高电路。(1) According to the write-in circuit of a kind of arbitrary K-value DRAM described in above-mentioned 1, get K=8, draw the write-in circuit of 8-value DRAM, as shown in Figure 2, wherein adopt 7 variable-threshold type PMOS transistors Qak k , k=1, 2, 3, ..., 7, the gate of the transistor Qak k is connected to the input D inj of the writing circuit via the variable threshold circuit, the new threshold value of the threshold variable PMOS transistor Qak k is ta k , When the tube Qak is turned on, the voltage drop between the source and the drain is 0; the source of the tube Qak is connected to the power supply V dd , and the voltage of V dd is selected to be higher than the maximum logic level V Dinj ( 7) and V Douj (7) high Δ, Δ is the downward DC level shift between the input and output of the voltage follower F; adopt 6 diodes Dan, n =2, 3, ..., 7, diode Dan The conduction voltage of V Don is V Don ; the positive pole and negative pole of Da n are respectively connected to the drain of variable threshold PMOS transistor Qa n-1 and the drain of transistor Qa n ; the drain of transistor Qa 7 is grounded through constant current source I j , the drain of the tube Qa 7 is connected to the constant current source I j to keep the current flowing through the conduction diode at the same fixed value, the output G wrij of the writing circuit is formed on the drain of the tube Qa 7 , and the output G wrij of the writing circuit is connected to The write bit line input of the memory cell circuit; selecting ta k is the average value (V Dinj (k) of adjacent logic levels V Dinj (k) and V Dinj (k-1) of the 8-value signal of the write circuit input D inj )+V Dinj (k-1))/2, that is, select ta k as the intermediate value of V Dinj (k) and V Dinj (k-1), V Dinj (k)>V Dinj (k-1); write The characteristics of the 8-value signal of the input circuit input D inj and the output D outj of the readout circuit and the standard equal-level 8-value signal of the DRAM input and output are the same: the difference of each adjacent logic level of the input D inj is equal, and the output D The differences between the adjacent logic levels of outj are equal, and the step voltages of the input D inj and the output D outj are the same, and the step voltage is V Don , that is, V Dinj (m)-V Dinj (m-1)=V Doutj ( m)-V Doutj (m-1)=V Don , m=1, 2, 3,..., 7, V Dinj (m) and V Doutj (m) are respectively the input logic of the write circuit and the output logic of the read circuit The value is the logic level of m; the write circuit output G wrij is Δ higher than the 8-value signal input by the write circuit except for the 0 level, and the 0 level is still 0. The 8-value write circuit is also called 8-value write into the booster circuit.

(2)根据上述2所述的任意K值DRAM的读出电路中取K=8,得出8值DRAM的读出电路,如图3所示,该8值DRAM的读出电路是按照上述(1)所述的8值DRAM的写入电路的相同结构特征而形成的,其中采用7个变阈型PMOS管Qbk,k=1,2,3,……,7,管Qbk的栅极经变阈电路连接到读出电路的输入Grdij,变阈型PMOS管Qbk的新阈值为tbk;管Qbk导通时源极漏极间压降为0;Grdij接到存储单元电路的读位线输出,管Qbk的源极接电源Vdc,选取Vdc的电压等于写入电路输入和读出电路输出的最大逻辑电平VDinj(7)和VDouj(7);采用6个二极管Dbn,n=2,3,……,7,二极管Dbn的导通电压是VDon;Dbn的正极和负极分别连接变阈型PMOS管Qbn-1的漏极和管Qbn的漏极;变阈型PMOS管Qb7的漏极经过恒流源Ij接地,管Qb7的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管Qb7的漏极形成读出电路的输出Doutj;选取tbk为读出电路输入Grdij的8值信号的相邻逻辑电平VGrdij(k)和VGrdij(k-1)的平均值(VGrdij(k)+VGrdij(k-1))/2,即选取tbk为VGrdij(k)和VGrdij(k-1)的中间值,VGrdij(k)>VGrdij(k-1);读出电路输出Doutj的信号和写入电路输入Dinj及DRAM输入输出的规范的等阶梯的8值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,7,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;读出电路输入Grdij是来自存储单元电路输出的不规范的8值信号,所述不规范的8值信号就是对比DRAM输入输出和写入电路输入及读出电路输出的规范的等阶梯的8值信号为逻辑电平幅度不一致;读出电路输出Doutj为规范的等阶梯的8值信号,即读出电路将不规范的8值信号Grdij转换为规范的等阶梯的8值信号Doutj,该8值读出电路又称为8值读出校正电路。(2) Get K=8 in the readout circuit of arbitrary K value DRAM according to above-mentioned 2, draw the readout circuit of 8-value DRAM, as shown in Figure 3, the readout circuit of this 8-value DRAM is according to above-mentioned (1) Formed by the same structural feature of the write-in circuit of the 8-value DRAM, wherein 7 threshold-variable PMOS transistors Qb k are adopted, k=1, 2, 3, ..., 7, the Qb k transistor The gate is connected to the input G rdij of the readout circuit through the variable threshold circuit, and the new threshold value of the variable threshold PMOS transistor Qb k is tb k ; when the transistor Qb k is turned on, the voltage drop between the source and the drain is 0; G rdij is connected to The read bit line output of the memory cell circuit, the source of the tube Qb k is connected to the power supply V dc , and the voltage of V dc is selected to be equal to the maximum logic level V Dinj (7) and V Douj (7) of the write circuit input and the read circuit output ); adopt 6 diodes Db n , n=2, 3, ..., 7, the conduction voltage of diode Db n is V Don ; the anode and cathode of Db n are respectively connected to the drain of variable threshold PMOS transistor Qb n-1 electrode and the drain of the tube Qb n ; the drain of the variable-threshold PMOS transistor Qb 7 is grounded through the constant current source I j , and the drain of the tube Qb 7 is connected to the constant current source I j to keep the current flowing through the conduction diode at the same constant value, forms the output D outj of the readout circuit at the drain of the tube Qb 7 ; selects tb k to be the adjacent logic levels V Grdij ( k) and V Grdij (k-1 ) (V Grdij (k)+V Grdij (k-1))/2, that is, select tb k as the middle value of V Grdij (k) and V Grdij (k-1), V Grdij (k)> V Grdij (k-1); the characteristics of the 8-valued signal of the signal of the output D outj of the readout circuit and the standards of the input D inj of the write circuit and the input and output of the DRAM are the same: each adjacent logic circuit of the input D inj The differences of the levels are equal, the difference between the adjacent logic levels of the output D outj is equal, and the step voltages of the input D inj and the output D outj are the same, and the step voltage is V Don , which satisfies V Dinj (m)-V Dinj (m- 1)=V Doutj (m)-V Doutj (m-1)=V Don , m=1, 2, 3,..., 7, V Dinj (m) and V Doutj (m) are the inputs of the write circuit respectively And readout circuit output logical value is the logic level of m; Readout circuit input G rdij is the non-standard 8-value signal from memory unit circuit output, and described non-standard 8-value signal is exactly contrasting DRAM input and output and writing The 8-value signal of the standard equal ladder of the circuit input and the output of the readout circuit is inconsistent with the logic level amplitude; the output D outj of the readout circuit is a standard 8-value signal of the equal ladder, that is, the readout circuit will not standardize the 8-value signal The signal G rdij is converted into a standard equal-level 8-value signal D outj , and the 8-value readout circuit is also called an 8-value readout correction circuit.

注:与8值DRAM的写入电路和读出电路相应的8值存储单元电路得出如下;在任意K值DRAM的存储单元电路中取K=8,选取电源Vdd的电压比写入电路的输入和读出电路的输出逻辑值为7的逻辑电平高Δ,Δ为F输入输出间向下的直流电平偏移,得出8值DRAM的存储单元电路,示如图1,该8值DRAM存储单元电路由电压跟随器F,F的栅极存储电容Cj和二个CMOS传输门G1和G2组成,用电容Cj存储8值信号,电压跟随器F包括NMOS管Qm1和NPN管Qm2,管Qm1的栅极接电容Cj的一端Cmij,即Cmij为电压跟随器F的输入,Cj的另一端接地,管Qm1的源极接管Qm2的栅极和电阻Rm1,Rm1的另一端接地,管Qm2的射极Folij经恒流源Ij接地,Qm2射极接恒流源Ij使管Qm2的射极负载为恒流源,管Qm2的射极Folij为F的输出,管Qm1的漏极和管Qm2的集电极都接电源Vdd;传输门G1的输入接写位线Gwrij,传输门G1的输出接F的输入Cmij,传输门G1的控制输入接写入脉冲wri,传输门G2的输入接F的输出Folij,传输门G2的输出接读位线Grdij,传输门G2的控制输入接读出脉冲rdi,写入脉冲wri和读出脉冲rdi来自DRAM的控制电路;写入脉冲wri来到时,传输门G1导通,将写位线Gwrij的8值信号传送到存储电容Cj,电容Cj接收Gwrij的8值信号,电容Cj的8值信号就是F输入Cmij的8值信号;写入脉冲wri未来到时,传输门G1截止,存储电容Cj与外界为直流开路,电容Cj存储的8值信号保持不变,即具有记忆功能;读出脉冲rdi来到时,传输门G2导通,将F输出Folij的8值信号传送到读位线Grdij;写位线Gwrij和读位线Grdij各自是存储单元电路的输入和输出;存储单元电路输入和输出各自接到写入电路输出和读出电路输入;电压跟随器F输出的8值信号必须是与F输入相对应的8值信号,F输入输出信息相同,即F输出无信息丢失,F输出无信息丢失要求Cj存储的8值信号是增高的8值信号,所述增高的8值信号就是除0电平外比写入电路输入的8值信号高Δ的信号,其中0电平仍为0;存储单元电路的输入信号是来自写入电路的输出,写入电路输出提供给存储单元电路输入的信号是增高的8值信号;增高8值信号经过F传送到读位线Grdij是不规范的8值信号,也即存储单元电路输出是不规范的8值信号,所述不规范的8值信号就是对比DRAM输入输出和写入电路输入及读出电路输出的规范的等阶梯的8值信号为逻辑电平幅度不一致;读出电路输入信号是来自存储单元电路输出的不规范的8值信号,读出电路输出是对存储单元电路输出不规范的8值信号校正得出的规范的等阶梯的8值信号,该校正得出的规范的等阶梯的8值信号作为对存储单元电路存储信息的校正读出。Note: The 8-value storage unit circuit corresponding to the write-in circuit and read-out circuit of the 8-value DRAM is obtained as follows; take K=8 in the storage unit circuit of any K-value DRAM, and select the voltage ratio of the power supply V dd to the write-in circuit The output logic value of the input and readout circuit is 7, the logic level is high Δ, and Δ is the downward DC level offset between the input and output of F, and the storage unit circuit of the 8-value DRAM is obtained, as shown in Figure 1, the 8 The value DRAM storage unit circuit is composed of a voltage follower F, a gate storage capacitor C j of F, and two CMOS transmission gates G1 and G2 . The capacitor C j is used to store 8-value signals. The voltage follower F includes an NMOS transistor Q m1 and NPN tube Q m2 , the gate of tube Q m1 is connected to one end C mij of capacitor C j , that is, C mij is the input of voltage follower F, the other end of C j is grounded, and the source of tube Q m1 is connected to the gate of Q m2 electrode and resistance R m1 , the other end of R m1 is grounded, the emitter F olij of tube Q m2 is grounded through constant current source I j , and the emitter of Q m2 is connected to constant current source I j so that the emitter load of tube Q m2 is constant current The source, the emitter F olij of the tube Q m2 is the output of F, the drain of the tube Q m1 and the collector of the tube Q m2 are both connected to the power supply V dd ; the input of the transmission gate G 1 is connected to the write bit line G wrij , and the transmission gate G 1 The output of the transmission gate G 1 is connected to the input C mij of F, the control input of the transmission gate G 1 is connected to the write pulse w ri , the input of the transmission gate G 2 is connected to the output F olij of F, the output of the transmission gate G 2 is connected to the read bit line G rdij , and the transmission The control input of the gate G2 is connected to the read pulse r di , the write pulse w ri and the read pulse r di come from the control circuit of the DRAM; when the write pulse w ri comes, the transfer gate G 1 is turned on, and the write bit line The 8-value signal of G wrij is transmitted to the storage capacitor C j , and the capacitor C j receives the 8-value signal of G wrij , and the 8-value signal of capacitor C j is the 8-value signal of F input C mij ; when the write pulse w ri arrives in the future, The transmission gate G 1 is cut off, the storage capacitor C j is open to the outside world, and the 8-value signal stored in the capacitor C j remains unchanged, that is, it has a memory function; when the readout pulse r di comes, the transmission gate G 2 is turned on, and the The 8-value signal of F output F olij is transmitted to the read bit line G rdij ; the write bit line G wrij and the read bit line G rdij are the input and output of the storage unit circuit respectively; the input and output of the storage unit circuit are respectively connected to the output of the write circuit and the input of the readout circuit; the 8-value signal output by the voltage follower F must be an 8-value signal corresponding to the F input, and the F input and output information are the same, that is, there is no information loss at the F output, and no information loss at the F output requires C j to be stored The 8-value signal is an increased 8-value signal, and the increased 8-value signal is a signal that is Δ higher than the 8-value signal input by the write circuit except for the 0 level, wherein the 0 level is still 0; the input of the storage unit circuit signal is the output from the write circuit, and the write circuit output is provided to the storage The signal input by the unit circuit is an increased 8-value signal; the increased 8-value signal is transmitted to the read bit line G rdij is an irregular 8-value signal, that is, the output of the storage unit circuit is an irregular 8-value signal. The standard 8-value signal is the standard 8-value signal that compares the DRAM input and output with the write circuit input and read circuit output. The logic level amplitude is inconsistent; the read circuit input signal is not standardized from the memory unit circuit output. 8-value signal, the output of the readout circuit is a standardized 8-value signal obtained by correcting the non-standard 8-value signal output by the storage unit circuit, and the standardized 8-valued signal obtained by the correction is used as a pair of storage Corrective readout of information stored in the unit circuit.

本发明具体的内容说明如下:The concrete content of the present invention is described as follows:

(A)本发明的存储单元电路,写入电路和读出电路的优点。①存储单元电路的优点:电路结构简单和成本极低。因为K>2,每个电容Cj存储K值信息比存储二值信息的信息量要大,显然K越大存储单元存储信息量越多,而Cj是MOS管栅极电容,成本极低,另外,存储单元电路只用G1和G2和F组成,电路结构简单,对多值DRAM很有利;一般要求DRAM存储的信息量越多越好,即要求存储单元电路的数量越大越好,要求每个电容Cj存储的信息量越多越好,要求电路结构简单使占硅片面积少越好,本发明的存储单元电路满足这个要求;②读出电路的优点:具有良好的量化整形作用,即具有恢复原多值信息能力,此能力用于抗干扰和刷新。用正弦波连续信号输入Grdij经读出电路后得出的输出Doutj曲线为不连续的(校正为规范的等阶梯的)多值信号,表明读出电路具有良好的类似4舍5入的量化整形作用,当输入Grdij电压上升或下降(如漏电和干扰影响)不越过上下二新阈值时,输出Doutj仍为规范的等阶梯的多值信息(恢复原信息),即具有恢复原多值信息能力,此能力用于提高抗干扰性能和刷新;③写入电路的优点:具有良好的量化整形作用(类似4舍5入的量化作用),得出稳定的满足要求的增高的多值信息;正弦波连续信号输入Dinj经写入电路后得出的输出Gwrij曲线为不连续的多值信息(增高的多值信息),当输入Dinj电压上升或下降(如干扰影响)不越过上下二新阈值时,输出Gwrij仍恢复原(增高的)多值信息,即具有恢复原增高的多值信息能力,此能力也可用于抗干扰和刷新。恢复原多值信息和刷新都是针对信息而言的。DRAM存储量很大且所用硅片面积小就必须要求:存储单元电路数量很大,写入电路和读出电路的数量尽量少;它们数量不等。多值DRAM存储单元(包括常规二值DRAM存储单元)是按行、列排成矩阵式结构,行译码输出线即字线(行选择线)选中存储矩阵中一行(一字线)的所有存储单元。对字长n位(如4位,8位或N位),行地址译码器译出每一字线输出有效时,有n个(如4个,8个或N个)存储单元同时被选中(通过该单元的CMOS传输门使其存储的多值信息与外接通,进行信息交换),有n根写位线和读位线各自接被选中的该位上的一存储单元,在每根写位线上接有写入电路,每根读位线上接有读出电路,使这些被选中的存储单元经过读出电路和写入电路,由读/写控制电路等进行读写操作。多值存储单元电路,多值读出电路和多值写入电路是十分重要的部件,任意K值和8值DRAM的存储单元电路保持非常简单的结构,仅写入和读出电路结构随K值大小有所不同,其优点十分显著的。除多位输入输出外,为了提高集成度同时减少器件引脚的数目,大容量DRAM常常采用1位输入、1位输出和地址分时输入的方式,此时存储单元仍保持按行、列排成矩阵式结构,而由DRAM输入缓冲、输出缓冲(注:DRAM输入输出指的是DRAM数据输入输出)和地址输入缓冲部分和控制电路部分等来完成数据串行输入串行输出任务,有时甚至地址也可用串行输入。如果将存储单元电路、写入电路和读出电路归为一个整体电路,则是成本极高的和不实用的,实用中存储单元电路,写入电路和读出电路的数量是大不相同的;存储单元电路,写入电路和读出电路在数量上在空间上不相同,按信息特征它们是按一个总的发明构思统一设计的紧密相关的三种电路。(A) Advantages of the memory cell circuit, write circuit and read circuit of the present invention. ①Advantages of the storage unit circuit: the circuit structure is simple and the cost is extremely low. Because K>2, each capacitor C j stores K value information is larger than the amount of information stored in binary information. Obviously, the larger the K, the more information the storage unit stores, and C j is the gate capacitance of the MOS transistor, and the cost is extremely low , in addition, the storage unit circuit is only composed of G1 , G2 and F, and the circuit structure is simple, which is very beneficial to multi-valued DRAM; generally, it is required that the amount of information stored in DRAM is as large as possible, that is, the number of storage unit circuits is required to be as large as possible , it is required that the amount of information stored by each capacitor C j is as much as possible, and the circuit structure is required to be simple so that the area occupied by the silicon chip is as small as possible. The memory cell circuit of the present invention meets this requirement; 2. The advantage of the readout circuit: it has a good quantization Shaping function, that is, it has the ability to restore the original multi-valued information, which is used for anti-jamming and refreshing. The output D outj curve obtained by inputting G rdij with a sine wave continuous signal and passing through the readout circuit is a discontinuous (corrected to a standard equal-step) multi-valued signal, which shows that the readout circuit has a good similarity to 4 rounding and 5 rounding Quantization shaping effect, when the input G rdij voltage rises or falls (such as leakage and interference effects) does not exceed the upper and lower thresholds, the output D outj is still the multi-valued information of the standard equal steps (restoring the original information), that is, it has the restoring original Multi-valued information capability, this capability is used to improve anti-interference performance and refresh; ③Advantages of writing into the circuit: it has a good quantization and shaping effect (similar to the quantization effect of rounding to 5), and obtains a stable increase in multiples that meets the requirements value information; the output G wrij curve obtained after the sine wave continuous signal input D inj is written into the circuit is discontinuous multi-valued information (increased multi-valued information), when the input D inj voltage rises or falls (such as interference effects) When the upper and lower new thresholds are not exceeded, the output G wrij still restores the original (increased) multi-value information, that is, it has the ability to restore the original increased multi-value information, and this ability can also be used for anti-jamming and refreshing. Restoring the original multi-valued information and refreshing are both for information. DRAM has a large storage capacity and a small area of silicon wafers, so it must be required: the number of memory cell circuits is large, and the number of write circuits and read circuits is as small as possible; their numbers are not equal. Multi-valued DRAM storage units (including conventional binary DRAM storage units) are arranged in a matrix structure by rows and columns, and the row decoding output line, that is, the word line (row selection line) selects all of the rows (one word line) in the storage matrix. storage unit. For the word length of n bits (such as 4 bits, 8 bits or N bits), when the row address decoder decodes that the output of each word line is valid, there are n (such as 4, 8 or N) memory cells simultaneously Selected (through the CMOS transmission gate of this unit, the multi-valued information stored in it is connected with the outside, and information exchange is carried out), n root write bit lines and read bit lines are respectively connected to a memory cell on the selected bit, at each A write circuit is connected to the root write bit line, and a read circuit is connected to each read bit line, so that these selected memory cells can be read and written by the read/write control circuit through the read circuit and the write circuit. . The multi-value storage unit circuit, the multi-value read circuit and the multi-value write circuit are very important components. The storage unit circuit of any K value and 8-value DRAM maintains a very simple structure, and only the write and read circuit structure varies with K The value size is different, and the advantages are quite significant. In addition to multi-bit input and output, in order to improve integration and reduce the number of device pins, large-capacity DRAM often adopts 1-bit input, 1-bit output and address time-sharing input. At this time, the memory cells are still arranged in rows and columns. In a matrix structure, the data serial input and serial output tasks are completed by the DRAM input buffer, output buffer (note: DRAM input and output refers to DRAM data input and output), address input buffer part and control circuit part, and sometimes even Addresses can also be entered serially. If the storage unit circuit, write circuit and readout circuit are classified as a whole circuit, it is extremely expensive and impractical. In practical storage unit circuits, the number of write circuits and readout circuits is very different. ; The storage unit circuit, the write circuit and the read circuit are different in quantity and space, and they are three closely related circuits that are uniformly designed according to a general inventive concept according to information characteristics.

(B)三种电路重在信息特征,即存储单元电路,写入电路和读出电路重在信息:信息存储,信息写入和信息读出。存储单元电路所存储的信息无信息丢失要求:写入电路送到存储单元电路输入的是增高的多值信息。存储单元电路输出不规范的多值信息要求:读出电路将不规范的多值信息校正为规范的等阶梯的多值信息输出。存储单元电路也是重在信息特征,存储单元电路具有多值信息存储,多值信息接收和多值信息发出三个信息特征:①信息接收:写位线Gwrij的多值信息传送到存储电容Cj,使电容Cj接收多值信息Cmij;②信息存储:G1截止时电容Cj与外界直流电阻几乎为无穷大,用电容Cj可以很好的存储多值信号;③信息发出:传输门G2导通,电压跟随器F输出的多值信息传送到读位线Grdij。存储单元电路,写入电路和读出电路有相同的重要信息特征:存储单元电路输入和写入电路输出的信息特征一致,它们都是增高的多值信息;存储单元电路输出和读出电路输入的信息特征一致,它们都是不规范的多值信息;写入电路输入和读出电路输出的信息特征一致,它们都是规范的等阶梯多值信息(注:该规范的等阶梯多值信息和DRAM输入输出的信息特征一致,DRAM输入输出就是DRAM数据输入输出)。(B) Three types of circuits focus on information features, that is, memory cell circuits, write circuits and read circuits focus on information: information storage, information writing and information reading. There is no information loss requirement for the information stored in the storage unit circuit: what the write circuit sends to the storage unit circuit is increased multi-value information. Requirements for outputting non-standard multi-valued information by the storage unit circuit: the readout circuit corrects the non-standard multi-valued information into a standardized multi-valued information output with equal steps. The storage unit circuit also focuses on information features. The storage unit circuit has three information features: multi-value information storage, multi-value information reception and multi-value information sending: ①Information reception: multi-value information written on the bit line G wrij is transmitted to the storage capacitor C j , so that capacitor C j can receive multi-valued information C mij ; ② information storage: when G 1 is cut off, capacitor C j and the external DC resistance are almost infinite, and capacitor C j can store multi-valued signals very well; ③ information sending: transmission The gate G 2 is turned on, and the multi-value information output by the voltage follower F is transmitted to the read bit line G rdij . The storage unit circuit, the writing circuit and the reading circuit have the same important information characteristics: the information characteristics of the input of the storage unit circuit and the output of the writing circuit are consistent, and they are all multi-valued information that is increased; the output of the storage unit circuit and the input of the reading circuit The information characteristics of the information are consistent, and they are all non-standard multi-valued information; the information characteristics of the input circuit input and the output of the read circuit are consistent, and they are all standardized multi-valued information with equal steps (Note: the multi-valued information with equal steps in this specification It is consistent with the information characteristics of the DRAM input and output, and the DRAM input and output is the DRAM data input and output).

(C)按一个总的发明构思统一设计二种紧密相关的写入电路和读出电路,任意K值DRAM的读出电路和任意K值DRAM的写入电路的结构特征是相同的,对照图4和图5,或对照图2和图3,立即看出:写入电路和读出电路的结构特征相同(二电路的变阈型PMOS管和二极管的管数相同,各管间的连接方式相同等),但参数不同,功能不同。(C) According to a general inventive conception, two kinds of closely related write-in circuits and read-out circuits are uniformly designed. The structural features of the read-out circuit of any K-value DRAM and the write-in circuit of any K-value DRAM are the same, as shown in Fig. 4 and Fig. 5, or comparing Fig. 2 and Fig. 3, it can be seen immediately that the structural features of the write-in circuit and the read-out circuit are the same (the variable-threshold PMOS transistors and diodes of the two circuits have the same number of tubes, and the connection mode between the tubes same, etc.), but with different parameters and different functions.

按任意K值DRAM存储单元,写入和读出电路的相同信息特征,设计中所用的一个总思路表述如下:According to any K-value DRAM storage unit, the same information characteristics of the write and read circuits, a general idea used in the design is expressed as follows:

DRAM输入输出是规范的多值信号,如果按常规思路,电容Cj直接接收DRAM输入的规范的多值信号并保存下来,即电容Cj存储的是规范的多值信号,Cj存储的规范的多值信号经过电压跟随器F,在F的输出就会有丢失信息的情况发生,因此常规思路不可用。现在换一个思路设计,为保证F的输出不发生丢失信息的情况,F输入必然与规范的信息特征不相同,可改用增高的多值信号传送到电容Cj(即F输入),该增高的多值信号要求满足:既能保证F输出不会有丢失信息的情况发生,而且能保证能将F输出不规范的多值信号校正为规范的多值信号,作为信息校正读出,使得从外面看读出仍是正确的多值信号,表明经读出电路能读出与DRAM输入输出相同特性的规范的多值信号,即读出电路将不规范的多值信号校正为规范的多值信号;由此可见,按上述信息特征,及写入电路输入和读出电路输出是规范的多值信号的要求来设计写入电路和读出电路,则读数困难就克服了。The input and output of DRAM is a standardized multi-valued signal. If the conventional thinking is followed, the capacitor C j directly receives the standardized multi-valued signal input by the DRAM and saves it, that is, the capacitor C j stores a standardized multi-valued signal, and the stored standard The multi-valued signal of the voltage follower F passes through the voltage follower F, and there will be a loss of information at the output of F, so the conventional thinking is not available. Now change the idea of design, in order to ensure that the output of F does not lose information, the input of F must be different from the information characteristics of the specification, and the multi-valued signal with increased height can be used to transmit to the capacitor C j (namely, the input of F). The requirements of the multi-valued signal are met: it can not only ensure that the F output will not lose information, but also ensure that the non-standard multi-valued signal of the F output can be corrected into a standardized multi-valued signal, which can be read out as information correction, so that from From the outside, the readout is still a correct multi-value signal, indicating that the readout circuit can read a standardized multi-value signal with the same characteristics as the DRAM input and output, that is, the readout circuit corrects the non-standard multi-value signal to a standardized multi-value signal Signal; It can be seen that, according to the above-mentioned information characteristics, and the requirement that the input circuit input and the output circuit of the readout circuit are multi-valued signals designed to design the write circuit and the readout circuit, then the difficulty in reading is overcome.

附图说明Description of drawings

图1.为本发明相关的一种任意K值和8值DRAM的存储单元电路图;Fig. 1. is the storage cell circuit diagram of a kind of arbitrary K value and 8-value DRAM relevant to the present invention;

图2.为本发明的一种8值DRAM的写入电路图;Fig. 2. is the writing circuit diagram of a kind of 8-value DRAM of the present invention;

图3.为本发明的一种8值DRAM的读出电路图;Fig. 3. is the readout circuit diagram of a kind of 8-value DRAM of the present invention;

图4.为本发明的一种K值DRAM的写入电路图;Fig. 4. is the writing circuit diagram of a kind of K value DRAM of the present invention;

图5.为本发明的一种K值DRAM的读出电路图;Fig. 5. is the readout circuit diagram of a kind of K value DRAM of the present invention;

图6.为本发明相关的第一种PMOS管变阈电路图和变阈型PMOS管符号图;Fig. 6. is the first kind of PMOS tube variable threshold circuit diagram and the variable threshold type PMOS tube symbol diagram related to the present invention;

图7.为本发明相关的第二种PMOS管变阈电路图和变阈型PMOS管符号图;Fig. 7. is the second kind of PMOS tube variable threshold circuit diagram and the variable threshold type PMOS tube symbol diagram related to the present invention;

图8.为图6中用Vdc代替Vdd的第一种PMOS管变阈电路图和变阈型PMOS管符号图;Fig. 8. replaces the first kind of PMOS transistor variable threshold circuit diagram and variable threshold type PMOS transistor symbol diagram with V dc replacing V dd in Fig. 6;

图9.为图7中用Vdc代替Vdd的第二种PMOS管变阈电路图和变阈型PMOS管符号图;Fig. 9. is the second kind of PMOS transistor variable threshold circuit diagram and the variable threshold type PMOS transistor symbol diagram with V dc instead of V dd in Fig. 7;

图10.为已有的一种多输出精密镜像恒流源电路图和符号图;Figure 10 is a circuit diagram and symbol diagram of an existing multi-output precision mirror constant current source;

图11.为本发明写入电路和读出电路及相关的8值DRAM的存储单元电路,在wri和rdi依次作用下的wri、rdi、Dinj、Gwrij、Cmij、Folij、Grdij和Doutj的先后上下分立的波形图;Fig. 11 is the storage unit circuit of the writing circuit and reading circuit and the related 8-value DRAM of the present invention, w ri , r di , D inj , G wrij , C mij , F under the sequential action of w ri and r di Waveform diagrams of olij , G rdij and D outj separated up and down successively;

图12.为本发明8值DRAM的写入电路输入Dinj和读出电路输出Doutj在wri和rdi依次作用下的波形图;Fig. 12. is the oscillogram of the writing circuit input Dinj and the readout circuit output Doutj of the 8-value DRAM of the present invention under the action of wri and rdi in sequence;

图13.为本发明8值DRAM的写入电路输入Dinj和输出Gwrij在wri作用下的波形图;Fig. 13. is the oscillogram of input D inj and output G wrij of the writing circuit of 8-value DRAM of the present invention under the effect of w ri ;

图14.为本发明8值DRAM的读出电路输入Grdij和输出Doutj在rdi作用下的波形图;Fig. 14. is the oscillogram of input G rdij and output D outj of the readout circuit of 8-value DRAM of the present invention under the effect of r di ;

图15.为本发明8值DRAM的存储单元电路的电压跟随器F的输入Cmij和输出Folij在有wri作用时的波形图;Fig. 15. is the input C mij and the output F olij of the voltage follower F of the storage unit circuit of 8-value DRAM of the present invention and the waveform diagram when there is w ri effect;

图16.为本发明写入电路和读出电路及相关的8值DRAM的存储单元电路,在wri和rdi依次作用下的Dinj、Gwrij、Cmij、Folij、Grdij和Doutj的上下不分立的波形图;Fig. 16 is the storage unit circuit of the writing circuit and reading circuit of the present invention and the related 8-value DRAM, D inj , G wrij , C mij , F olij , G rdij and D under the sequential action of w ri and r di Waveform diagram of up and down of outj ;

图17.为本发明8值DRAM的写入电路在wri=0传输门G1截止时并且将写入电路输入Dinj改为正弦波时写入电路输入Dinj和写入电路输出Gwrij的波形图;Fig. 17 is the writing circuit of the 8-value DRAM of the present invention. When w ri = 0, the transmission gate G 1 is cut off and the writing circuit input D inj is changed to a sine wave, the writing circuit input D inj and the writing circuit output G wrij Waveform diagram;

图18.为本发明8值DRAM的读出电路在rdi=0传输门G2截止时并且将读出电路输入Grdij改为正弦波时读出电路输入Grdij和读出电路输出Doutj的波形图;Fig. 18. is the readout circuit of the 8-value DRAM of the present invention when rdi =0 transmission gate G2 is cut off and when the readout circuit input Grdij is changed to a sine wave, the readout circuit input Grdij and the readout circuit output Doutj Waveform diagram;

图19.为常用CMOS传输门的电路和符号图。Figure 19. Circuit and symbol diagram for commonly used CMOS transmission gates.

具体实施方式Detailed ways

下面具体对本发明作进一步的说明:The present invention is specifically described further below:

实施例1:存储单元电路信息功能的说明。Embodiment 1: Description of the circuit information function of the storage unit.

存储单元电路具有多值信息存储,多值信息接收和多值信息发出三个信息功能:①信息接收:由图1看出,写入脉冲wri来到时,传输门G1导通,将写位线Gwrij的多值信息传送到存储电容Cj,使电容Cj接收多值信息Cmij;电容Cj接收是个充放电过程,是充电还是放电取决于电容Cj原存储的信息和现接收的信息,充放电时间常数与是Cj电容量有关,Cj通常仅为几皮法,不能再大;②信息存储:写入脉冲wri未来到时,传输门G1截止,由图1看出,存储电容Cj仅与NMOS管Qm1的栅极和CMOS传输门G1的输出相连,管Qm1的栅极输入电阻接近开路,G1截止时也接近开路,此时电容Cj与外界直流电阻几乎为无穷大,用电容Cj可以很好的存储多值信号;③信息发出:由图1看出,读出脉冲rdi来到时,传输门G2导通,将电压跟随器F输出的多值信息传送到读位线Grdij,既使电容Cj有微弱漏电和F不完善,影响F输出传送到读位线Grdij的多值信息,但读出电路仍有能力校正为正确的规范的多值信息。The storage unit circuit has three information functions of multi-valued information storage, multi-valued information reception and multi-valued information sending: ①Information reception: It can be seen from Figure 1 that when the write pulse w ri arrives, the transmission gate G1 is turned on, and the The multi-valued information of the write bit line G wrij is transmitted to the storage capacitor C j , so that the capacitor C j receives the multi-valued information C mij ; the reception of the capacitor C j is a charging and discharging process, and whether charging or discharging depends on the original stored information of the capacitor C j and For the received information, the charging and discharging time constant is related to the capacitance of C j , and C j is usually only a few picofarads and cannot be larger; ②Information storage: when the write pulse w ri arrives in the future, the transmission gate G 1 will be cut off, and the transmission gate G 1 will be cut off by It can be seen from Figure 1 that the storage capacitor Cj is only connected to the gate of the NMOS transistor Qm1 and the output of the CMOS transmission gate G1 , the gate input resistance of the transistor Qm1 is close to an open circuit, and G1 is also close to an open circuit when it is cut off. At this time, the capacitance The resistance between C j and the external DC is almost infinite, and multi-valued signals can be well stored with capacitor C j ; ③Information sending: It can be seen from Figure 1 that when the readout pulse r di arrives, the transmission gate G 2 is turned on, and the The multi-valued information output by the voltage follower F is transmitted to the read bit line G rdij , even if the capacitor C j has a weak leakage and F is not perfect, which affects the multi-valued information output by F and transmitted to the read bit line G rdij , but the read circuit is still Ability to correct multi-valued information to the correct specification.

注意:传输门G1截止时,理论上要求Cj与外界为直流开路,即直流电阻为无穷大,实际上Cj与外界为高阻(几乎无直流通路),仍然实际有微弱漏电,Cj存储的多值信号只能保持一定时间,因此所有DRAM需要另加一个刷新电路,定时刷新使其恢复原存信息。Note: When the transmission gate G1 is cut off, it is theoretically required that C j and the outside world be a DC open circuit, that is, the DC resistance is infinite. In fact, C j and the outside world are high resistance (almost no DC path), and there is still a weak leakage current in reality, C j The stored multi-valued signal can only be kept for a certain period of time, so all DRAMs need to add a refresh circuit to restore the original stored information by regular refreshing.

存储单元电路中电压跟随器F很重要,但若F输入信息使用不当,则实际F输出有丢失多值信息的情况发生,用电容Cj(F输入)存储多值信息是成本极低的,常规DRAM用电容Cj存储二值信息,二值信息的信息量最低,多值信息的信息量比二值的高,用电容Cj存储多值信息比存储二值信息当然更划算。F输出有丢失多值信息发生的原因描述如下:如果电容Cj接到一个理想电压跟随器FA的输入,理想电压跟随器FA的电压放大倍数恒为1,无直流偏移,则FA的输出电压和电容Cj电压完全相同,即FA的输出和电容Cj存储的多值信息完全相同。实际中没有理想电压跟随器,实际电压跟随器(本发明中的F)的电压放大倍数小于1,且有直流偏移Δ,当电容Cj上的电压小于Δ时(如逻辑值为1,其逻辑电平=VDon<Δ),F输出为0,即Cj存储信号逻辑值不为0,而F输出信号逻辑值却为0;为克服直流偏移的缺点,避免F输出信息丢失,写入电路提供给存储单元电路输入的是增高Δ的多值信号,而增高Δ的多值信号经F后输出的是非规范的多值信号,进一步用读出电路将此校正为规范的等阶梯的多值信号。考虑输入Dinj和输出Doutj的阶梯电压相同,阶梯电压等于是二极管的导通电压VDon,输入Dinj和输出Doutj的最大逻辑电平等于VDon的L倍(LVDon),所以电源电压Vdd比上述最大逻辑电平高Δ,Δ容易由F算出或实测出,Cj通常仅为几皮法。The voltage follower F in the storage unit circuit is very important, but if the F input information is used improperly, the actual F output will lose multi-valued information, and the cost of storing multi-valued information with the capacitor C j (F input) is extremely low. Conventional DRAM uses capacitor C j to store binary information. The information content of binary information is the lowest, and the information content of multi-valued information is higher than that of binary information. Of course, it is more cost-effective to store multi-valued information with capacitor C j than to store binary information. The reasons for the loss of multi-valued information at the output of F are described as follows: If the capacitor C j is connected to the input of an ideal voltage follower FA , the voltage amplification factor of the ideal voltage follower FA is always 1, and there is no DC offset, then F The output voltage of A is exactly the same as the voltage of capacitor C j , that is, the output of F A is exactly the same as the multi-valued information stored in capacitor C j . In reality, there is no ideal voltage follower, the voltage magnification of the actual voltage follower (F in the present invention) is less than 1, and there is a DC offset Δ, when the voltage on the capacitor C j is less than Δ (such as logic value 1, Its logic level = V Don < Δ), the F output is 0, that is, the logic value of the stored signal of C j is not 0, but the logic value of the F output signal is 0; in order to overcome the shortcoming of DC offset and avoid the loss of F output information , what the writing circuit provides to the storage unit circuit is a multi-valued signal with an increase in Δ, and the multi-valued signal with an increase in Δ after passing through F will output a non-standard multi-valued signal, and further use the readout circuit to correct this to a standard, etc. Ladder multivalued signal. Considering that the ladder voltage of the input D inj and the output D outj are the same, the ladder voltage is equal to the conduction voltage V Don of the diode, and the maximum logic level of the input D inj and the output D outj is equal to L times of V Don (LV Don ), so the power supply The voltage V dd is Δ higher than the above maximum logic level, Δ is easily calculated or measured from F, and C j is usually only a few picofarads.

存储单元电路中CMOS传输门示如图19,即图19为常用CMOS传输门的电路和符号图,CMOS传输门由一个P沟道和一个N沟道增强型MOSFET(即NMOS管QG1和PMOS管QG2)并联而成,该CMOS传输门本身带有一个CMOS反相器(即NMOS管QG4和PMOS管QG3)。CMOS传输门电路简单,可双向传输信号,常用作模拟开关。The CMOS transmission gate in the storage unit circuit is shown in Figure 19, that is, Figure 19 is the circuit and symbol diagram of a common CMOS transmission gate. The CMOS transmission gate is composed of a P-channel and an N-channel enhancement MOSFET (ie NMOS transistor Q G1 and PMOS The transistor Q G2 ) is connected in parallel, and the CMOS transmission gate itself has a CMOS inverter (that is, the NMOS transistor Q G4 and the PMOS transistor Q G3 ). The CMOS transmission gate circuit is simple, can transmit signals in both directions, and is often used as an analog switch.

实施例2:任意K值和8值DRAM的写入电路满足设计要求的证明。Embodiment 2: Proof that the writing circuit of arbitrary K value and 8-value DRAM satisfies design requirements.

任意K值DRAM的写入电路示如图4,需证明满足设计要求:当写入电路输入Dinj逻辑值为0、1、2、3、4、……、L-2、L-1、L时,写入电路输出Gwrij逻辑值仍依次为0、1、2、3、4、……、L-2、L-1、L;但Gwrij逻辑值对应逻辑电平VGwrij(n)除0电平外比Dinj逻辑值对应逻辑电平VDinj(n)高Δ(n=1~L),0电平仍为0,即VGwrij(0)=VDinj(0)=0V、VGwrij(1)=VDinj(1)+Δ、VGwrij(2)=VDinj(2)+Δ、……、VGwrij(L-1)=VDinj(L-1)+Δ、VGwrij(L)=VDinj(L)+Δ,VDinj(k)>VDinj(k-1);tak为写入电路输入的相邻逻辑电平VDinj(k)和VDinj(k-1)的中间值,满足VDinj(k-1)<tak<VDinj(k),k=1、2、3、4、……、L-2、L-1、L,即满足不等式0<ta1<VDinj(1)<ta2<VDinj(2)<ta3<VDinj(3)<ta4<……<taL-2<VDinj(L-2)<taL-1<VDinj(L-1)<taL<VDinj(L);因为管Qak导通电压为0V(或近0V),管Qak导通电压就是管Qak导通时源极和漏极间的压降,管Qak导通电流取较小的值(如30μA),也即恒流源Ij电流取较小的值(如30μA),记VDinj和VGwrij各自为写入电路输入Dinj和输出Gwrij的电压(瞬时值),二极管导通电压VDon等于输入VDinj的K值信号的阶梯电压,也即VDon等于VDinj各相邻逻辑电平的差VDinj(m)-VDinj(m-1),所以VDinj(m)=mVDon,电源电压Vdd比写入电路的输入和读出电路的输出的K值逻辑电平的最大值高Δ,即Vdd=VDoutj(L)+Δ=VDinj(L)+Δ=LVDon+Δ。The writing circuit of DRAM with arbitrary K value is shown in Figure 4, which needs to be proved to meet the design requirements: when the logic value of the writing circuit input D inj is 0, 1, 2, 3, 4, ..., L-2, L-1, When L, the logic value of the write circuit output G wrij is still sequentially 0, 1, 2, 3, 4, ..., L-2, L-1, L; but the logic value of G wrij corresponds to the logic level V Gwrij (n ) is higher than the logic level V Dinj (n) corresponding to the logic value of D inj by Δ(n=1~L) except the 0 level, and the 0 level is still 0, that is, V Gwrij (0)=V Dinj (0)= 0V, V Gwrij (1) = V Dinj (1) + Δ, V Gwrij (2) = V Dinj (2) + Δ, ..., V Gwrij (L-1) = V Dinj (L-1) + Δ , V Gwrij (L) = V Dinj (L) + Δ, V Dinj (k) > V Dinj (k-1); ta k is the adjacent logic level V Dinj (k) and V Dinj input by the write circuit The intermediate value of (k-1), satisfying V Dinj (k-1)<ta k <V Dinj (k), k=1, 2, 3, 4,..., L-2, L-1, L, That is, satisfy the inequality 0<ta 1 <V Dinj (1)<ta 2 <V Dinj (2)<ta 3 <V Dinj (3)<ta 4 <...<ta L-2 <V Dinj (L-2) <ta L-1 <V Dinj (L-1)<ta L <V Dinj (L); because the conduction voltage of the tube Qa k is 0V (or nearly 0V), the conduction voltage of the tube Qa k is the conduction voltage of the tube Qa k When the voltage drop between the source and the drain, the conduction current of the tube Qak takes a smaller value (such as 30μA), that is, the current of the constant current source I j takes a smaller value (such as 30μA), record V Dinj and V Gwrij are the voltages (instantaneous values) of the input D inj and output G wrij of the writing circuit, and the diode conduction voltage V Don is equal to the step voltage of the K value signal input to V Dinj , that is, V Don is equal to the adjacent logic voltages of V Dinj The flat difference V Dinj (m)-V Dinj (m-1), so V Dinj (m)=mV Don , the power supply voltage V dd is more than the input of the writing circuit and the K value of the output of the reading circuit logic level The maximum value is higher by Δ, that is, V dd =V Doutj (L)+Δ=V Dinj (L)+Δ=LV Don +Δ.

写入电路的输入输出关系证明如下:根据上述不等式0<ta1<VDinj(1)<ta2<VDinj(2)<ta3<VDinj(3)<ta4<……<taL-2<VDinj(L-2)<taL-1<VDinj(L-1)<taL<VDinj(L),由图4看出:①当Dinj输入0电平时,VDinj(0)<ta1,所有管Qa1~QaL都截止,写入电路输出电压VGwrij=VDinj(0)=0V,②当Dinj输入1电平时,ta1<VDinj(1)<ta2,管Qa1导通,管Qa2~QaL截止,(L-1)个二极管Da2~DaL导通,VGwrij=Vdd-(L-1)VDon=LVDon+Δ-(L-1)VDon=VDon+Δ=VDinj(1)+Δ,③当Dinj输入2电平时,ta2<VDinj(2)<ta3,管Qa1和Qa2导通,管Qa3~QaL截止,(L-2)个二极管Da3~DaL导通,VGwrij=Vdd-(L-2)VDon=LVDon+Δ-(L-2)VDon=2VDon+Δ=VDinj(2)+Δ,④当Dinj输入3电平时,ta3<VDinj(3)<ta4,管Qa1~Qa3导通,管Qa4~QaL截止,(L-3)个二极管Da4~DaL导通,VGwrij=Vdd-(L-3)VDon=LVDon+Δ-(L-3)VDon=3VDon+Δ=VDinj(3)+Δ,……⑤当Dinj输入L-2电平时,taL-2<VDinj(L-2)<taL-1,管Qa1~QaL-2导通,管QaL-1和QaL截止,2个二极管DaL-1和DaL导通,VGwrij=Vdd-2VDon=LVDon+Δ-2VDon=(L-2)VDon+Δ=VDinj(L-2)+Δ,⑥当Dinj输入L-1电平时,taL-1<VDinj(L-1)<taL,管Qa1~QaL-1导通,管QaL截止,1个二极管DaL导通,VGwrij=Vdd-VDon=LVDon+Δ-VDon=(L-1)VDon+Δ=VDinj(L-1)+Δ,⑦当Dinj输入L电平时,taL<VDinj(L),管Qa1~QaL导通,VGwrij=Vdd=LVDon+Δ=LVDon+Δ=VDinj(L)+Δ。由此得出写入电路的输出Gwrij除0电平外比写入电路输入的K值信号高Δ,0电平仍为0,输出Gwrij是增高的K值信号,克服存储单元电路中电压跟随器F的输入输出间有直流电平偏移Δ的不足。附图中的Pspice计算机模拟波形也证实它的正确性。所用二极管为硅二极管,导通电流取较小值,也即恒流源Ij电流取较小的值,Δ很容易由F计算出或实测出。实用中Δ也可取稍大的值,由此所得的输出Gwrij是Δ稍大的增高信号,Δ稍大的增高信号经F得出实际Grdij输入的非规范的K值信号,但并不影响结果,这是因为读出电路可将任何非规范的非等阶梯的K值信号Grdij转换为规范的等阶梯的K值信号Doutj,此时只要读出电路按实际Grdij输入的非规范的K值信号(对应Δ稍大)来设计即可。The input-output relationship of the writing circuit is proved as follows: according to the above inequality 0<ta 1 <V Dinj (1)<ta 2 <V Dinj (2)<ta 3 <V Dinj (3)<ta 4 <...<ta L -2 <V Dinj (L-2)<ta L-1 <V Dinj (L-1)<ta L <V Dinj (L), it can be seen from Figure 4: ①When D inj input 0 level, V Dinj (0)<ta 1 , all transistors Qa 1 to Qa L are cut off, the write circuit output voltage V Gwrij =V Dinj (0)=0V, ②When D inj input 1 level, ta 1 <V Dinj (1) <ta 2 , the tube Qa 1 is turned on, the tubes Qa 2 ~Qa L are off, (L-1) diodes Da 2 ~Da L are turned on, V Gwrij =V dd -(L-1)V Don =LV Don + Δ-(L-1)V Don =V Don +Δ=V Dinj (1)+Δ, ③When D inj input 2 levels, ta 2 <V Dinj (2)<ta 3 , tubes Qa 1 and Qa 2 turn on, the tubes Qa 3 ~Qa L are cut off, (L-2) diodes Da 3 ~Da L are turned on, V Gwrij =V dd -(L-2)V Don =LV Don +Δ-(L-2) V Don =2V Don +Δ=V Dinj (2)+Δ, ④When D inj input 3 levels, ta 3 <V Dinj (3)<ta 4 , tubes Qa 1 ~Qa 3 are turned on, tubes Qa 4 ~ Qa L is off, (L-3) diodes Da 4 ~Da L are on, V Gwrij =V dd -(L-3)V Don =LV Don +Δ-(L-3)V Don =3V Don +Δ =V Dinj (3)+Δ,...⑤ When D inj input L-2 level, ta L-2 <V Dinj (L-2)<ta L-1 , tubes Qa 1 ~Qa L-2 are turned on , tube Qa L-1 and Qa L are cut off, two diodes Da L-1 and Da L are turned on, V Gwrij =V dd -2V Don =LV Don +Δ-2V Don =(L-2)V Don +Δ =V Dinj (L-2)+Δ, ⑥When D inj input L-1 level, ta L-1 <V Dinj (L-1)<ta L , tubes Qa 1 ~Qa L-1 are turned on, tubes Qa L is off, one diode Da L is on, V Gwrij =V dd -V Don =LV Don +Δ-V Don =(L-1)V Don +Δ=V D inj (L-1)+Δ, ⑦When D inj input L level, ta L <V Dinj (L), tubes Qa 1 ~Qa L conduction, V Gwrij =V dd =LV Don +Δ=LV Don + Δ=V Dinj (L)+Δ. Therefore, the output G wrij of the write circuit is Δ higher than the K value signal input by the write circuit except for the 0 level, and the 0 level is still 0, and the output G wrij is an increased K value signal, which overcomes the problem in the memory cell circuit There is a shortage of DC level offset Δ between the input and output of the voltage follower F. The Pspice computer simulation waveform in the accompanying drawing also confirms its correctness. The diode used is a silicon diode, and the conduction current takes a smaller value, that is, the current of the constant current source I j takes a smaller value, and Δ is easily calculated or measured by F. In practice, Δ can also take a slightly larger value, and the resulting output G wrij is an increased signal with a slightly larger Δ, and the increased signal with a slightly larger Δ is passed through F to obtain a non-standard K value signal input by the actual G rdij , but it does not This is because the readout circuit can convert any non-standard non-equal-level K-value signal G rdij into a standardized equal-level K-value signal D outj . The standard K value signal (corresponding to a slightly larger Δ) can be designed.

任意K值DRAM的写入电路中取K=8,则得出8值DRAM的写入电路,示如图2,同样方法证明8值DRAM的写入电路满足设计要求。图10为已有的一种多输出精密镜像电流源(恒流源)电路图和符号图,为降低功耗和提高性能等,其恒流源Ij电流取较小值。Take K=8 in the write-in circuit of any K-value DRAM, then the write-in circuit of 8-value DRAM is obtained, as shown in Figure 2, and the same method proves that the write-in circuit of 8-value DRAM meets the design requirements. Figure 10 is a circuit diagram and symbol diagram of an existing multi-output precision mirror current source (constant current source). In order to reduce power consumption and improve performance, etc., the current of the constant current source I j takes a smaller value.

实施例3:任意K值和8值DRAM的读出电路满足设计要求的证明。Embodiment 3: Proof that the readout circuit of any K value and 8-value DRAM satisfies the design requirements.

任意K值DRAM的读出电路示如图5,需证明满足设计要求:当读出电路的输入Grdij逻辑值为0、1、2、3、4、……、L-2、L-1、L时,读出电路的输入Doutj逻辑值仍依次为0、1、2、3、4、……、L-2、L-1、L;Grdij,Dinj和Doutj逻辑值对应逻辑电平依次为VGrdij(n),VDinj(n)和VDoutj(n)(n=0~L),其中输入Grdij是非规范的K值信号,要求输出Doutj是规范的等阶梯的K值信号,即VDoutj(0)=VDinj(0)=0V、VDoutj(1)=VDinj(1)=VDon,VDoutj(2)=VDinj(2)=2VDon,VDoutj(3)=VDinj(3)=3VDon,……VDoutj(L-2)=VDinj(L-2)=(L-2)VDon,VDoutj(L-1)=VDinj(L-1)=(L-1)VDon,VDoutj(L)=VDinj(L)=LVDon;tbk为输入Grdij的非规范的K值信号的相邻逻辑电平VGrdij(k)和VGrdij(k-1)的中间值,满足VGrdij(k-1)<tbk<VGrdij(k),k=1、2、3、4、……、L-2、L-1、L,即满足不等式0<tb1VGrdij(1)<tb2<VGrdij(2)<tb3<VGrdij(3)<tb4<……<tbL-2<VGrdij(L-2)<tbL-1<VGrdij(L-1)<tbL<VGrdij(L);考虑管Qbk导通电压为0V(或近0V),管Qbk导通电压就是管Qbk导通时源极和漏极间的压降,管Qbk导通电流取较小的值(如30μA),即恒流源Ij电流取较小的值(如30μA),记VGrdij和VDoutj各自为读出电路输入Grdij和输出Doutj的电压(瞬时值),二极管导通电压VDon等于输出Doutj的规范的K值信号的阶梯电压,也即VDon等于输出Doutj各相邻逻辑电平的差VDoutj(m)-VDoutj(m-1),所以VDoutj(m)=mVDon,电源电压Vdc等写入电路的输入和读出电路的输出的最大逻辑电平VDinj(L)和VDoutj(L),Vdc=VDoutj(L)=LVDonThe readout circuit of any K-value DRAM is shown in Figure 5, which needs to be proved to meet the design requirements: when the logic value of the input G rdij of the readout circuit is 0, 1, 2, 3, 4, ..., L-2, L-1 , L, the logic value of the input D outj of the readout circuit is still sequentially 0, 1, 2, 3, 4, ..., L-2, L-1, L; G rdij , D inj and D outj logic values correspond The logic levels are V Grdij (n), V Dinj (n) and V Doutj (n) (n=0~L), where the input G rdij is a non-standard K value signal, and the output D outj is required to be a standard equal ladder K value signal of V Doutj (0)=V Dinj (0)=0V, V Doutj (1)=V Dinj (1)=V Don , V Doutj (2)=V Dinj (2)=2V Don , V Doutj (3)=V Dinj (3)=3V Don , ... V Doutj (L-2)=V Dinj (L-2)=(L-2)V Don , V Doutj (L-1)=V Dinj (L-1)=(L-1)V Don , V Doutj (L)=V Dinj (L)=LV Don ; tb k is the adjacent logic level V of the non-standard K value signal of input G rdij The intermediate value of Grdij (k) and V Grdij (k-1), satisfying V Grdij (k-1)<tb k <V Grdij (k), k=1, 2, 3, 4,..., L-2 . _ _ _ _ _ _ _ Grdij (L-2)<tb L-1 <V Grdij (L-1)<tb L <V Grdij (L); consider that the turn-on voltage of tube Qb k is 0V (or nearly 0V), and the turn-on voltage of tube Qb k It is the voltage drop between the source and the drain when the tube Qb k is turned on, the conduction current of the tube Qb k takes a smaller value (such as 30μA), that is, the current of the constant current source I j takes a smaller value (such as 30μA), Note that V Grdij and V Doutj are the voltages (instantaneous values) of the input G rdij and output D outj of the readout circuit, and the diode conduction voltage V Don is equal to the step voltage of the standard K value signal of the output D outj , that is, V Don is equal to Output the difference V Doutj (m)-V Doutj (m-1) of each adjacent logic level of D outj , so V Doutj (m)=mV Don , power supply voltage V dc , etc. are written into the input of the circuit and read out the circuit output maximum logic level V Dinj ( L) and V Doutj (L), V dc = V Doutj (L) = LV Don .

读出电路的输入输出关系证明如下:根据上述不等式0<tb1<VGrdij(1)<tb2<VGrdij(2)<tb3<VGrdij(3)<tb4<……tbL-2<VGrdij(L-2)<tbL-1<VGrdij(L-1)<tbL<VGrdij(L),由图5看出:①当Grdij输入0电平时,VGrdij(0)=0V,VGrdij(0)<tb1,所有管Qb1~QbL都截止,读出电路的输出电压VDoutj=0V,②当Grdij输入1电平时,tb1<VGrdij(1)<tb2,管Qb1导通,管Qb2~QbL截止,(L-1)个二极管Db2~DbL导通,VDoutj=Vdc-(L-1)VDon=LVDon-(L-1)VDon=VDon=VDoutj(1),③当Grdij输入2电平时,tb2<VGrdij(2)<tb3,管Qb1和Qb2导通,管Qb3~QbL截止,(L-2)个二极管Db3~DbL导通,VDoutj=Vdc-(L-2)VDon=LVDon-(L-2)VDon=2VDon=VDoutj(2),④当Grdij输入3电平时,满足tb3<VGrdij(3)<tb4,则管Qb1~Qb3导通,管Qb4~QbL截止,有(L-3)个二极管Db4~DbL导通,VDoutj=Vdc-(L-3)VDon=LVDon+Δ-(L-3)VDon=3VDon=VDoutj(3),………⑤当Grdij输入L-2电平时,tbL-2<VGrdij(L-2)<tbL-1,管Qb1~QbL-2导通,管QbL-1和QbL截止,2个二极管DbL-1和DbL导通,VDoutj=Vdc-2VDon=LVDon-2VDon=(L-2)VDon=VDoutj(L-2),⑥当Grdij输入L-1电平时,tbL-1<VGrdij(L-1)<tbL,管Qb1~QbL-1导通,管QbL截止,1个二极管DbL导通,VDoutj=Vdc-VDon=LVDon-VDon=(L-1)VDon=VDoutj(L-1),⑦当Grdij输入L电平时,tbL<VGrdij(L),管Qb1~QbL导通,VDoutj=Vdc=LVDon=VDoutj(L)。由此得出:尽管读出电路输入Gwrij是非规范的K值信号,而读出电路输出Doutj却是规范的等阶梯K值信号,即读出电路将非规范的K值信号Grdij转换为规范的的等阶梯的K值信号Doutj。附图中的Pspice计算机模拟波形也证实它的正确性,且能将连续波形(图18)转换为等阶梯K值信号Doutj。所用二极管为硅二极管,导通电流取较小值。The input-output relationship of the readout circuit is proved as follows: according to the above inequality 0<tb 1 <V Grdij (1)<tb 2 <V Grdij (2)<tb 3 <V Grdij (3)<tb 4 <...tb L- 2 <V Grdij (L-2)<tb L-1 <V Grdij (L-1)<tb L <V Grdij (L), it can be seen from Figure 5: ①When G rdij input 0 level, V Grdij ( 0)=0V, V Grdij (0)<tb 1 , all tubes Qb 1 to Qb L are cut off, and the output voltage V Doutj of the readout circuit is 0V, ②When G rdij inputs 1 level, tb 1 <V Grdij ( 1)<tb 2 , tube Qb 1 is turned on, tubes Qb 2 ~Qb L are off, (L-1) diodes Db 2 ~Db L are turned on, V Doutj =V dc -(L-1)V Don =LV Don -(L-1)V Don =V Don =V Doutj (1), ③When G rdij input 2 levels, tb 2 <V Grdij (2)<tb 3 , tubes Qb 1 and Qb 2 are turned on, tube Qb 3 ~Qb L is off, (L-2) diodes Db 3 ~Db L are on, V Doutj =V dc -(L-2)V Don =LV Don -(L-2)V Don =2V Don = V Doutj (2), ④ When G rdij is input with 3 levels, tb 3 <V Grdij (3)<tb 4 is satisfied, then tubes Qb 1 ~ Qb 3 are turned on, tubes Qb 4 ~ Qb L are cut off, there is (L- 3) Diodes Db 4 ˜Db L are turned on, V Doutj =V dc -(L-3)V Don =LV Don +Δ-(L-3)V Don =3V Don =V Doutj (3),... …⑤When G rdij input L-2 level, tb L-2 <V Grdij (L-2)<tb L-1 , tubes Qb 1 ~ Qb L-2 are turned on, and tubes Qb L-1 and Qb L are cut off , 2 diodes Db L-1 and Db L conduct, V Doutj =V dc -2V Don =LV Don -2V Don =(L-2)V Don =V Doutj (L-2), ⑥When G rdij input At L-1 level, tb L-1 <V Grdij (L-1)<tb L , tubes Qb 1 to Qb L-1 are on, tube Qb L is off, one diode Db L is on, V Doutj =V dc -V Don =LV Don -V Don =(L-1)V Don =V Doutj (L-1), ⑦When G rdij inputs L level, tb L <V Grdij (L), tubes Qb 1 ~Qb L are turned on, V Doutj =V dc =LV Don =V Doutj (L) . It can be concluded that although the input G wrij of the readout circuit is a non-standard K-value signal, the output D outj of the read-out circuit is a standard equal-step K-value signal, that is, the readout circuit converts the non-standard K-value signal G rdij is the standard equal-step K value signal D outj . The Pspice computer simulation waveform in the attached figure also proves its correctness, and the continuous waveform (Fig. 18) can be converted into the equal-step K value signal D outj . The diode used is a silicon diode, and the conduction current takes a smaller value.

任意K值DRAM的读出电路中取K=8,则得出8值DRAM的读出电路,示如图3,同样方法证明8值DRAM的读出电路满足设计要求。图10为已有的一种多输出精密镜像电流源(恒流源)电路图和符号图,为降低功耗和提高性能等,其恒流源Ij电流取较小值。Take K=8 in the readout circuit of any K-value DRAM, then the readout circuit of the 8-value DRAM is obtained, as shown in Figure 3, and the same method proves that the readout circuit of the 8-value DRAM meets the design requirements. Figure 10 is a circuit diagram and symbol diagram of an existing multi-output precision mirror current source (constant current source). In order to reduce power consumption and improve performance, etc., the current of the constant current source I j takes a smaller value.

实施例4:PMOS管变阈电路(简称变阈电路)的说明。Embodiment 4: Description of PMOS transistor variable threshold circuit (referred to as variable threshold circuit).

第一种PMOS管变阈电路(简称第一种变阈电路)示如图6左虚框,它由NMOS管Q3,PMOS管Q4和电阻R3组成,管Q3的栅极接输入电压Vx,管Q4的栅极接参考电压Vref,管Q3的漏极为该电路输出Vout1,输出Vout1接受控PMOS管QT1;改变参考电压Vref,使QT1的新阈值改变(放大、缩小、改变开启性质和提高开启分辨率);接PMOS管变阈电路的管QT1称为变阈型PMOS管。设Vdd>Vd≥Vtn+|Vtp|,Vdd-Vd≥|Vtp|+Vtn,记Vextn1=Vref+Vtn+|Vtp|,Vref为参考电压,NMOS和PMOS管阈值电压分别为Vtn>0和Vtp<0。管Q3和Q4的栅极对源极电位差分别为Vgs3和Vgs4,因为Q3和Q4的二源极相接,Q3的漏极经电阻R3接电源Vdd,Q4的漏极接地,仅当Q3和Q4的二栅压的差Vg3-Vg4≥Vtn+|Vtp|时,管Q3和Q4才同时导通,否则同时截止。因Vg3=Vx,Vg4=Vref,由此得出:①当Vx-Vref=Vg3-Vg4≥Vtn+|Vtp|,即输入电压Vx≥Vref+Vtn+|Vtp|=Vextn1时,管Q3和Q4导通,电阻R3上的电压Vout1为很低,使QT1导通;②当Vx<Vextn1时,Q3和Q4截止,Vout1=Vdd,使QT1截止;表明经该变阈电路电路后,使QT1变成Vx≥Vextn1时导通,或变阈型PMOS管QT1的新阈值t大小变为Vextn1,即t=Vextn1,改变参考电压Vref,使t改变,开启性质改变(QT1变成Vx≥t时导通)。因Vdd≥Vref≥0,t=Vextn1最小值为Vtn+|Vtp|,第一种PMOS管变阈电路不能实现t小于Vtn+|Vtp|的新阈值,较小的t还需用第二种PMOS管变阈电路实现。The first PMOS transistor variable threshold circuit (referred to as the first variable threshold circuit) is shown in the left dotted box in Figure 6, which is composed of NMOS transistor Q3 , PMOS transistor Q4 and resistor R3 , and the gate of transistor Q3 is connected to the input voltage V x , the gate of tube Q 4 is connected to the reference voltage V ref , the drain of tube Q 3 is the output V out1 of the circuit, and the output V out1 is controlled by the PMOS tube Q T1 ; changing the reference voltage V ref makes the new threshold of Q T1 Change (zoom in, zoom out, change turn-on properties and improve turn-on resolution); the tube Q T1 connected to the PMOS tube variable threshold circuit is called variable threshold PMOS tube. Suppose V dd >V d ≥V tn +|V tp |, V dd -V d ≥|V tp |+V tn , record V extn1 =V ref +V tn +|V tp |, V ref is the reference voltage, The threshold voltages of the NMOS and PMOS transistors are V tn >0 and V tp <0 respectively. The gate-to-source potential differences of tubes Q3 and Q4 are V gs3 and V gs4 respectively, because the two sources of Q3 and Q4 are connected, and the drain of Q3 is connected to the power supply V dd through resistor R3 , Q The drain of 4 is grounded, and only when the difference between the two gate voltages of Q3 and Q4 V g3 -V g4V tn + |V tp |, transistors Q3 and Q4 are turned on at the same time, otherwise they are turned off at the same time. Since V g3 =V x , V g4 =V ref , it follows that: ①When V x -V ref =V g3 -V g4 ≥V tn +|V tp |, that is, the input voltage V x ≥V ref +V When tn +|V tp |=V extn1 , tubes Q3 and Q4 are turned on, and the voltage V out1 on resistor R3 is very low, making Q T1 turn on; ②When V x <V extn1 , Q3 and Q4 Q 4 is cut off, V out1 = V dd , so that Q T1 is cut off; it shows that after the variable threshold circuit circuit, Q T1 is turned on when V x ≥ V extn1 , or the new threshold t of the variable threshold PMOS transistor Q T1 The magnitude becomes V extn1 , that is, t=V extn1 , and the reference voltage V ref is changed to change t and turn-on properties (Q T1 turns on when V x ≥ t). Because V dd ≥ V ref ≥ 0, the minimum value of t=V extn1 is V tn + |V tp |, the first type of PMOS transistor variable threshold circuit cannot realize the new threshold value of t less than V tn + |V tp |, the smaller t needs to be realized with the second PMOS tube variable threshold circuit.

第二种PMOS管变阈电路(简称第二种变阈电路)示如图7左虚框,它的结构由第一种PMOS管变阈电路(包括NMOS管Q3,PMOS管Q4和电阻R3)加一个CMOS反相器(包括PMOS管Q5和NMOS管Q6)组成,其中管Q4的栅极接输入电压Vx,管Q3的栅极接参考电压Vref,管Q3的漏极接CMOS反相器输入(管Q5和管Q6的栅极),CMOS反相器输出(管Q5和管Q6的漏极)为该电路输出Vout0,输出Vout0接受控PMOS管QT0;改变参考电压Vref,使QT0的新阈值改变(放大、缩小、改变开启性质和提高开启分辨率);接PMOS管变阈电路的管QT0称为变阈型PMOS管。设Vdd>Vd≥Vtn+|Vtp|,Vdd-Vd≥|Vtp|+Vtn,记Vextn0=Vref-Vtn-|Vtp|,NMOS和PMOS管变阈值电压分别为Vtn>0和Vtp<0。管Q3和Q4的栅极对源极电位差分别为Vgs3和Vgs4,同上理由,仅当Q3和Q4的二栅压的差Vg3-Vg4≥Vtn+|Vtp|时,管Q3和Q4才同时导通,否则Vg3-Vg4<Vtn+|Vtp|,管Q3和Q4同时截止。因Vg3=Vref,Vg4=Vx,由此得出:①当Vref-Vx=Vg3-Vg4<Vtn+|Vtp|,管Q3和Q4截止,即输入电压Vx>Vref-Vtn-|Vtp|=Vextn0时,管Q3和Q4截止,管Q3的漏极(即CMOS反相器输入)为Vdd,于是管Q5截止和管Q6导通,CMOS反相器输出Vout0=Vd,使QT0导通;②当Vx≥Vextn0时,Q3和Q4导通,管Q3的漏极(即CMOS反相器输入)为很低,于是管Q6截止和管Q5导通,CMOS反相器输出Vout0=Vdd,使QT0截止。表明经变阈电路后,使QT0变成Vx≥Vextn0时导通,即t=Vextn0。其中t=Vextn0=Vref-Vtn-|Vtp|可以小于Vtn+|Vtp|,表明变阈型PMOS管QT0的新阈值t大小变为Vextn0,即t=Vextn0。改变参考电压Vref,使t改变,开启性质改变(QT0变成Vx≥t时导通)。因Vdd≥Vref≥0,新阈值t最小值为0,最大值为Vdd-Vtn-|Vtp|。The second PMOS transistor variable threshold circuit (abbreviated as the second variable threshold circuit) is shown in the left dashed box in Figure 7, and its structure is composed of the first PMOS transistor variable threshold circuit (including NMOS transistor Q3 , PMOS transistor Q4 and resistor R 3 ) plus a CMOS inverter (including PMOS transistor Q 5 and NMOS transistor Q 6 ), wherein the gate of transistor Q 4 is connected to the input voltage V x , the gate of transistor Q 3 is connected to the reference voltage V ref , and the gate of transistor Q 3 is connected to the reference voltage V ref , and the gate of transistor Q The drain of 3 is connected to the input of the CMOS inverter (the gates of the tube Q5 and the tube Q6 ), and the output of the CMOS inverter (the drains of the tube Q5 and the tube Q6 ) is the output V out0 of the circuit, and the output V out0 Receive control PMOS tube Q T0 ; change the reference voltage V ref to change the new threshold of Q T0 (enlarge, shrink, change the opening property and improve the opening resolution); the tube Q T0 connected to the PMOS tube variable threshold circuit is called variable threshold type PMOS tube. Let V dd >V d ≥V tn +|V tp |, V dd -V d ≥|V tp |+V tn , record V extn0 =V ref -V tn -|V tp |, NMOS and PMOS tube change threshold The voltages are V tn >0 and V tp <0, respectively. The gate-to-source potential differences of transistors Q3 and Q4 are V gs3 and V gs4 respectively. For the same reasons as above, only when the difference between the two gate voltages of Q3 and Q4 is V g3 -V g4 ≥V tn + |V tp When |, tubes Q3 and Q4 are turned on at the same time, otherwise V g3 -V g4 <V tn + |V tp |, tubes Q3 and Q4 are turned off at the same time. Since V g3 =V ref , V g4 =V x , it can be concluded that: ① When V ref -V x =V g3 -V g4 <V tn +|V tp |, tubes Q 3 and Q 4 are cut off, that is, the input When the voltage V x >V ref -V tn -|V tp |=V extn0 , the transistors Q3 and Q4 are cut off, the drain of the transistor Q3 (that is, the input of the CMOS inverter) is Vdd , so the transistor Q5 is cut off And the tube Q 6 is turned on, and the CMOS inverter outputs V out0 =V d , so that Q T0 is turned on; ②When V x ≥ V extn0 , Q 3 and Q 4 are turned on, and the drain of the tube Q 3 (that is, CMOS Inverter input) is very low, so the tube Q6 is turned off and the tube Q5 is turned on, and the CMOS inverter outputs V out0 =V dd , so that Q T0 is turned off. It shows that after the variable threshold circuit, Q T0 is turned on when V x ≥ V extn0 , that is, t=V extn0 . Where t=V extn0 =V ref -V tn -|V tp | can be less than V tn + |V tp |, indicating that the new threshold t of the variable threshold PMOS transistor Q T0 becomes V extn0 , that is, t=V extn0 . Change the reference voltage V ref to change t and turn-on properties (Q T0 turns on when V x ≥ t). Because V dd ≥ V ref ≥ 0, the minimum value of the new threshold t is 0, and the maximum value is V dd -V tn - |V tp |.

改变参考电压Vref,使t改变,第一种PMOS管变阈电路图6(t=Vextn1=Vref+Vtn+|Vtp|)不能实现小于Vtn+|Vtp|的新阈值t,第二种PMOS管变阈电路图7(t=Vextn0=Vref-Vtn-|Vtp|)不能实现大于Vdd-Vtn-|Vtp|的新阈值t,常需用二种PMOS管变阈电路配合使用。将图6和图7中的Vdd改为Vdc,则分别得出图8和图9,图8不能实现小于Vtn+|Vtp|的新阈值t,图9不能实现大于Vdc-Vtn-|Vtp|的新阈值t,也常需用此二种PMOS管变阈电路配合使用。图6和图7(包括图8和图9)中的R3可用恒流源I3代替(电流流向Q3漏极)。Change the reference voltage V ref to change t, the first PMOS transistor variable threshold circuit Figure 6 (t=V extn1 =V ref +V tn +|V tp |) cannot achieve a new threshold t smaller than V tn +|V tp | , the second PMOS transistor variable threshold circuit in Figure 7 (t=V extn0 =V ref -V tn -|V tp |) cannot achieve a new threshold t greater than V dd -V tn -|V tp |, and two types of PMOS tube variable threshold circuit is used together. Change V dd in Figure 6 and Figure 7 to V dc , then Figure 8 and Figure 9 can be obtained respectively. Figure 8 cannot achieve a new threshold t smaller than V tn + |V tp |, and Figure 9 cannot achieve a new threshold t greater than V dc - The new threshold t of V tn -|V tp | is also often used in conjunction with these two PMOS transistor variable threshold circuits. R3 in Figure 6 and Figure 7 (including Figure 8 and Figure 9) can be replaced by a constant current source I3 (the current flows to the drain of Q3 ).

为获得一序列不同参考电压Vref(因为序列变阈型PMOS管有各自的新阈值tak或tbk,需用不同参考电压Vref按二种PMOS管变阈电路来获得所需新阈值),可用在直流电源和地间(按常用方法)接多个电阻串联的分压电路来实现,也可用在直流电源和地间接多个二极管(或场效应二极管)串联的分压电路来实现(其中根据情况需要还可串联一电阻),多个二极管正极和负极接法和常用电池串联的接法一样,如k个二极管D1~Dk,D1正极接直流电源,D1负极接D2正极,D2负极接D3正极,……,Dk-2负极接Dk-1正极,,Dk-1负极接Dk正极,Dk负极接地(或通过R接地)实现,因为序列不同参考电压Vref全都是输出到MOS管栅极,输出直流电流几乎为0,所以实现起来很方便。本发明中所述的电流源I示如图10,是一种常用的接地的多输出精密镜像电流源。In order to obtain a sequence of different reference voltages V ref (because the sequence variable threshold PMOS transistors have their own new thresholds ta k or tb k , it is necessary to use different reference voltages V ref to obtain the required new thresholds according to two PMOS transistor variable threshold circuits) It can be realized by connecting a voltage divider circuit with multiple resistors connected in series between the DC power supply and the ground (according to the usual method), or it can be realized by connecting a voltage divider circuit in series with multiple diodes (or field effect diodes) connected in series between the DC power supply and the ground ( Among them, a resistor can be connected in series according to the needs of the situation), and the connection method of the anode and cathode of multiple diodes is the same as that of common batteries in series, such as k diodes D 1 ~ D k , the anode of D 1 is connected to the DC power supply, and the cathode of D 1 is connected to D 2 positive pole, D 2 negative pole connected to D 3 positive pole, ..., D k-2 negative pole connected to D k-1 positive pole, D k-1 negative pole connected to D k positive pole, D k negative pole connected to ground (or grounded through R) to achieve, because The reference voltage V ref of different sequences is all output to the gate of the MOS transistor, and the output DC current is almost 0, so it is very convenient to implement. The current source I described in the present invention is shown in FIG. 10, which is a commonly used grounded multi-output precision mirror current source.

实施例5:对图1~3的Pspice计算机模拟波形图11~16的说明。Embodiment 5: Description of the Pspice computer simulation waveform diagrams 11-16 of FIGS. 1-3.

写入脉冲wri和读出脉冲rdi来自DRAM的控制电路(考虑地址译码器的字线输出Wi,读/写控制,片选,刷新等),在rdi和wri的作用下,对图1~3进行Pspice计算机模拟,得出各种模拟波形示如图11~16,注:图中wri,rdi,Dinj,Gwrij,Cmij,Folij,Grdij和Doutj的8个波形在每个图横坐标下边各自写为V(wri),V(rdi),V(Dinj),V(Gwrij),V(Cmij),V(Folij),V(Grdij),V(Doutj)的8个带有V的形式,即V后面的括号内分别写为wri,rdi,Dinj,Gwrij,Cmij,Folij,Grdij和Doutj(其中下标改为非下标的正常字体,即wri,rdi,Dinj,Gwrij,Cmij,Folij,Grdij和Doutj,这是Pspice模拟图表示方式),以下所有波形图横坐标下边都按类似表示方式写出,不再一一描述。图11为本发明8值DRAM的存储单元电路,写入电路和读出电路在wri和rdi依次作用下的Dinj、Gwrij、Cmij、Folij、Grdij和Doutj的先后上下分立的波形图,按图11从上到下的次序依次为wri,rdi,Dinj,Gwrij,Cmij,Folij,Grdij和Doutj的8个波形,图12是在wri和rdi的作用下,Dinj和Doutj的2个波形,图中将wri和rdi的高度缩小十倍,并放在图的最下部(rdi在图的最底部,wri在rdi上边),从图12看出,DRAM的写入电路的输入Dinj曲线和读出电路的输出Doutj曲线是等阶梯的多值信号,同逻辑值的Dinj和Doutj的逻辑电平是相等的,满足所述要求,Doutj是在rdi(见图的最底部)来到时变化的。图13是在wri(见图的最底部)的作用下,8值DRAM写入电路的输入Dinj和输出Gwrij的2个波形,从图13看出,输入Dinj曲线是等阶梯的多值信号(相对为下面的曲线),输出Gwrij曲线是增高的多值信号(相对为上面的非等阶梯的曲线),同逻辑值的Dinj和Gwrij的逻辑电平(除0电平相等外)是不相等的,由此克服存储单元电路中电压跟随器F的输入输出间有向下的直流电平偏移Δ的不足。图14是在rdi(见图的最底部)的作用下,8值DRAM读出电路的输入Grdij和输出Doutj的2个波形,从图14看出,输入Grdij曲线是阶梯较小(相对较低的曲线,与Dinj阶梯不相同)的多值信号,同逻辑值的Dinj和Grdij的逻辑电平是不相等的,输出Doutj曲线(相对较高的曲线)是与Dinj阶梯相同的等阶梯的多值信号,即读出电路将非规范的多值信号Grdij转换为规范的等阶梯的多值信号Doutj,由此克服存储单元电路中电压跟随器F电压放大倍数小于1的不足。图15是在wri(见图的最底部)的作用下,8值DRAM存储单元电路的电压跟随器F的输入Cmij和输出Folij的2个波形,从图15看出,输入Cmij曲线是增高的多值信号(相对为上面曲线,非等阶梯),而输出Folij曲线(相对为下面曲线)对输入Cmij曲线有向下的电平位移,且幅度缩小,表明存储单元电路中电压跟随器F存在有向下的电平位移和电压放大倍数小于1。图16为本发明8值DRAM的存储单元电路,写入电路和读出电路在wri和rdi(rdi在图的最底部,wri在rdi上边)依次作用下的Dinj、Gwrij、Cmij、Folij、Grdij和Doutj的上下不分立的波形图,即相当于图13,图14和图15中6个曲线的合并。The write pulse w ri and the read pulse r di come from the control circuit of DRAM (considering the word line output W i of the address decoder, read/write control, chip selection, refresh, etc.), under the action of r di and w ri , carry out Pspice computer simulation on Figures 1~3, and get various analog waveforms as shown in Figures 11~16, Note: w ri , r di , D inj , G wrij , C mij , F olij , G rdij and D The 8 waveforms of outj are respectively written as V(wri), V(rdi), V(Dinj), V(Gwrij), V(Cmij), V(Folij), V(Grdij) under the abscissa of each figure, The 8 forms of V(Doutj) with V, that is, the brackets behind V are respectively written as w ri , r di , D inj , G wrij , C mij , F olij , G rdij and D outj (the subscripts are changed to It is a normal font that is not subscripted, namely wri, rdi, Dinj, Gwrij, Cmij, Folij, Grdij and Doutj, this is the representation of the Pspice simulation diagram), all the following waveforms are written in a similar manner below the abscissa, and no longer Describe them one by one. Fig. 11 is the storage unit circuit of the 8-value DRAM of the present invention, the sequence of D inj , G wrij , C mij , F olij , G rdij and D outj of the write circuit and the read circuit under the sequential action of w ri and r di Discrete waveform diagrams, according to the order from top to bottom in Figure 11 are w ri , r di , D inj , G wrij , C mij , F olij , G rdij and D outj 8 waveforms, and Figure 12 is in w ri and r di , the two waveforms of D inj and D outj , the height of w ri and r di is reduced by ten times in the figure, and placed at the bottom of the figure (r di is at the bottom of the figure, w ri is at the bottom of the figure r di above), it can be seen from Figure 12 that the input D inj curve of the DRAM write circuit and the output D outj curve of the read circuit are multi-valued signals with equal steps, and the logic voltage of D inj and D outj of the same logic value Ping is equal to meet the requirements, D outj is changed when r di (see the bottom of the figure) comes. Figure 13 shows the two waveforms of the input D inj and output G wrij of the 8-value DRAM write circuit under the action of w ri (see the bottom of the figure). It can be seen from Figure 13 that the input D inj curve is equal to the ladder Multi-valued signal (relative to the lower curve), the output G wrij curve is an increased multi-valued signal (relative to the upper non-equal ladder curve), the logic level of the same logic value D inj and G wrij (except 0 voltage (except equal) are unequal, thereby overcoming the disadvantage of a downward DC level shift Δ between the input and output of the voltage follower F in the storage unit circuit. Figure 14 is the two waveforms of the input G rdij and output D outj of the 8-value DRAM readout circuit under the action of r di (see the bottom of the figure). It can be seen from Figure 14 that the input G rdij curve has a smaller step (relatively lower curve, different from the D inj ladder) multi-valued signal, the logic levels of D inj and G rdij of the same logic value are not equal, and the output D outj curve (relatively higher curve) is the same as The multi-valued signal with the same step in D inj , that is, the readout circuit converts the non-standard multi-valued signal G rdij into a multi-valued signal D outj with a standard equal step, thereby overcoming the voltage of the voltage follower F in the storage unit circuit Insufficient magnification of less than 1. Figure 15 is the two waveforms of the input C mij and output F olij of the voltage follower F of the 8-value DRAM storage unit circuit under the action of w ri (see the bottom of the figure). It can be seen from Figure 15 that the input C mij The curve is an increased multi-valued signal (relative to the upper curve, non-equal ladder), and the output F olij curve (relative to the lower curve) has a downward level shift to the input C mij curve, and the amplitude is reduced, indicating that the memory cell circuit The medium voltage follower F has a downward level shift and a voltage amplification factor less than 1. Fig. 16 is the storage unit circuit of the 8-value DRAM of the present invention, the writing circuit and the reading circuit under the actions of w ri and r di (r di is at the bottom of the figure, and w ri is on the top of r di ) sequentially act on D inj and G The waveform diagrams of wrij , C mij , F olij , G rdij and D outj that are not separated up and down are equivalent to the combination of the six curves in Fig. 13, Fig. 14 and Fig. 15.

实施例6:图17和图18的Pspice计算机模拟波形的说明。Example 6: Illustration of the Pspice computer simulated waveforms of Figures 17 and 18.

主要Pspice模拟波形在实施例5中已完成,现在考虑写入电路和读出电路各自分别输入为正弦波时的情况,作为进一步了解写入和读出电路优点的参考,注:本Pspice模拟时为了与上述模拟区分开,输入正弦波时的写入电路输入Dinj和读出电路输出Doutj各自改用符号INDi和OUTDi(各自仍然是在写入电路输入Dinj和读出电路输出Doutj二点上,即符号不同,各表示的输入和输出点相同,说明时仍称为写入电路输入Dinj和读出电路输出Doutj),于是17中的写入电路输入Dinj和图18中的读出电路输出Doutj的Pspice模拟波形图横坐标下边各自写为V(INDi)和V(OUTDi),而Gwrij、Grdij的Pspice模拟波形图横坐标下边和前述相同仍然依次写为V(Gwrij)、V(Grdij)。图17为本发明8值DRAM的写入电路在wri=0传输门G1截止时且输入Dinj为正弦波时的输入Dinj和输出Gwrij的波形图,从图17看出,正弦波连续信号输入Dinj曲线经写入电路后得出的输出Gwrij曲线为不连续的多值信号(但有增高作用),表明写入电路具有良好的量化整形作用(类似4舍5入的量化作用),当输入Dinj电压上升或下降不越过上下二新阈值时,输出Gwrij仍可恢复原多值信息,即具有恢复原多值信息能力,此能力还可用于刷新。图18为本发明8值DRAM的读出电路在rdi=0传输门G2截止时且输入Grdij为正弦波时的输入Grdij和输出Doutj的波形图。从图18看出,正弦波连续信号输入Grdij曲线经读出电路后得出的输出Doutj曲线为不连续的(校正为等阶梯)多值信号,表明读出电路具有良好的量化整形作用(类似4舍5入的量化作用),当输入Grdij电压上升或下降不越过上下二新阈值时,输出Doutj仍可恢复原多值信息,即具有恢复原多值信息能力,此能力还可用于刷新。为提高抗干扰性能,增加恢复原多值信息能力,减轻刷新任务,可适当加大存储单元的存储电容的数值。另外,读出电路输入也是MOS管的栅极,同样具有信息存储作用,为提高抗干扰性能,增加恢复原多值信息能力,减轻刷新任务,其栅极输入电容也以稍大为有利。The main Pspice simulation waveform has been completed in Embodiment 5. Now consider the situation when the write circuit and the readout circuit are respectively input as sine waves, as a reference for further understanding the advantages of the write and readout circuits. Note: This Pspice simulation In order to distinguish from the above-mentioned simulation, the writing circuit input D inj and the reading circuit output D outj when inputting a sine wave are respectively replaced with symbols IND i and OUTD i (respectively are still in the writing circuit input D inj and the reading circuit output D outj two points, that is, the symbols are different, and the input and output points of each representation are the same, and they are still called the input circuit input D inj and the output circuit output D outj when explaining), so the input circuit input D inj and the output circuit in 17 The lower side of the Pspice analog waveform diagram abscissa of the readout circuit output D outj in Fig. 18 is respectively written as V (INDi) and V (OUTDi), and the lower side of the Pspice analog waveform diagram abscissa of G wrij and G rdij is the same as the foregoing and still sequentially Written as V(Gwrij), V(Grdij). Fig. 17 is the oscillogram of the input D inj and the output G wrij when w ri = 0 transmission gate G 1 is cut off and the input D inj is a sine wave in the writing circuit of the 8-value DRAM of the present invention, as seen from Fig. 17 , the sine wave The output G wrij curve obtained after the wave continuous signal input D inj curve is written into the circuit is a discontinuous multi-valued signal (but has the effect of increasing), indicating that the write circuit has a good quantization and shaping effect (similar to the rounding of 4 and 5 Quantization function), when the input D inj voltage rises or falls without crossing the upper and lower thresholds, the output G wrij can still restore the original multi-valued information, that is, it has the ability to restore the original multi-valued information, and this ability can also be used for refreshing. Fig. 18 is a waveform diagram of the input G rdij and the output D outj of the readout circuit of the 8-value DRAM of the present invention when r di = 0 and the transmission gate G 2 is cut off and the input G rdij is a sine wave . It can be seen from Figure 18 that the output D outj curve obtained after the sine wave continuous signal input G rdij curve passes through the readout circuit is a discontinuous (corrected to equal steps) multi-valued signal, indicating that the readout circuit has a good quantization and shaping effect (similar to the quantization effect of rounding to 5), when the input G rdij voltage rises or falls without crossing the upper and lower thresholds, the output D outj can still restore the original multi-valued information, that is, it has the ability to restore the original multi-valued information, and this ability is also Available for refresh. In order to improve the anti-jamming performance, increase the ability to restore the original multi-valued information, and reduce the refresh task, the value of the storage capacitor of the storage unit can be appropriately increased. In addition, the input of the readout circuit is also the gate of the MOS transistor, which also has the function of information storage. In order to improve the anti-interference performance, increase the ability to restore the original multi-valued information, and reduce the refresh task, the gate input capacitance is also slightly larger.

实施例7:选取tak和tbk数值的说明。Embodiment 7: Description of selecting values of ta k and tb k .

理论证明上述(1)和(2)只要求满足下述不等式即可:证明(1)只要求满足0<ta1<VDinj(1)<ta2<VDinj(2)<ta3<VDinj(3)<ta4<……<taL-2<VDinj(L-2)<taL-1<VDinj(L-1)<taL<VDinj(L),证明(2)只要求满足0<tb1<VGrdij(1)<tb2<VGrdij(2)<tb3<VGrdij(3)<tb4<……<tbL-2<VGrdij(L-2)<tbL-1<VGrdij(L-1)<tbL<VGrdij(L);从证明过程看出,理论上新阈值可取相邻逻辑电平之间的任意值,实际上,一.按抗干扰性要求,最好取接近或等于相邻逻辑电平的平均值,本发明就是基于抗干扰性要求设计的;二.按刷新性能要求,最好取略高于相邻逻辑电平的平均值,使允许放电电荷较多,但兼顾抗干扰性要求,只能适当偏离平均值,不能偏离平均值很大;三.当输入电压等于新阈值时,输出处于相邻逻辑电平间的过渡区,过渡区的大小取决于新阈值的开启分辨率(数学上‘当输入≥新阈值,则管导通’,仅是理想的,实际不含=),该开启分辨率是较好的,但达不到数学上的理想要求。It is theoretically proved that the above (1) and (2) only need to satisfy the following inequalities: prove that (1) only needs to satisfy 0<ta 1 <V Dinj (1)<ta 2 <V Dinj (2)<ta 3 <V Dinj (3)<ta 4 <...<ta L-2 <V Dinj (L-2)<ta L-1 <V Dinj (L-1)<ta L <V Dinj (L), prove (2) It is only required to satisfy 0<tb 1 <V Grdij (1)<tb 2 <V Grdij (2)<tb 3 <V Grdij (3)<tb 4 <...<tb L-2 <V Grdij (L-2) <tb L-1 <V Grdij (L-1)<tb L <V Grdij (L); It can be seen from the proof process that theoretically the new threshold can take any value between adjacent logic levels, in fact, one. According to the requirements of anti-interference, it is better to get close to or equal to the average value of adjacent logic levels, the present invention is based on the design of anti-interference requirements; The average value of the allowable discharge charge is more, but taking into account the anti-interference requirements, it can only deviate from the average value appropriately, and cannot deviate greatly from the average value; 3. When the input voltage is equal to the new threshold value, the output is between adjacent logic levels The transition zone, the size of the transition zone depends on the opening resolution of the new threshold (mathematically 'when the input ≥ the new threshold, the tube is turned on', it is only ideal, and actually does not contain =), the opening resolution is better Yes, but not up to the ideal requirements in mathematics.

实施例8:用电阻Rj代替恒流源Ij的说明。Embodiment 8: Explanation of using resistor R j instead of constant current source I j .

‘恒流源Ij取为电阻Rj’即“(4)所述的K值DRAM存储单元电路中,恒流源Ij取为电阻Rj”,指的是在K值DRAM存储单元电路中‘用电阻Rj代替恒流源Ij’,也就是电阻Rj一端接射极Folij,电阻Rj另一端接地,这样以来管Qm2和电阻Rj构成射极跟随器。因为射极Folij经恒流源Ij接地也就是恒流源Ij一端接射极Folij,恒流源Ij另一端接地,此时管Qm2和恒流源Ij构成射极跟随器;换言之,就是改‘管Qm2的射极Folij经恒流源Ij接地’为‘管Qm2的射极Folij经电阻Rj接地’,前者用恒流源Ij则Qm2的射极电流恒定,后者用电阻Rj则Qm2的射极电流不恒定,但此二者(接Ij和接Rj)都是常用的射极跟随器结构。注:管Qm1的源极接电阻Rm1(Rm1的另一端接地),即构成源极跟随器(共漏极放大电路)。源极跟随器和射极跟随器是常用的电路,与射极跟随器(三极管共集电极放大电路)类似,场效应管源极跟随器(共漏极放大电路)没有电压放大作用,其电压增益小于1,输出电压与输入电压相位相同,输入电阻高,输出电阻低,可作阻抗变换用。源极跟随器的输入电阻极高,适合接存储电容Cj(漏电极小),射极跟随器的输入电阻低,不适合接存储电容Cj(漏电大),但源极跟随器的输出电阻比射极跟随器的输出电阻高,负载能力比较差,为增加负载能力,可将源极跟随器的输出接射极跟随器的输入,将射极的输出作为电压跟随器的输出,其输出负载能力就大大加强。'Constant current source I j is taken as resistance R j ', that is, "in the K-value DRAM storage unit circuit described in (4), constant current source I j is taken as resistance R j ", referring to the K-value DRAM storage unit circuit In 'Replace constant current source Ij with resistor Rj ', that is, one end of resistor Rj is connected to emitter F olij , and the other end of resistor Rj is grounded, so that tube Qm2 and resistor Rj form an emitter follower. Because the emitter F olij is grounded through the constant current source Ij , that is, one end of the constant current source Ij is connected to the emitter F olij , and the other end of the constant current source Ij is grounded. At this time, the tube Q m2 and the constant current source Ij form an emitter follower In other words, it is to change "the emitter F olij of the tube Q m2 is grounded through the constant current source I j " to "the emitter F olij of the tube Q m2 is grounded through the resistor R j ", the former uses the constant current source I j and Q m2 The emitter current of Q m2 is constant, and the emitter current of Q m2 is not constant if the latter uses resistor R j , but both (connected to I j and connected to R j ) are commonly used emitter follower structures. Note: The source of tube Q m1 is connected to resistor R m1 (the other end of R m1 is grounded), which constitutes a source follower (common drain amplifier circuit). Source follower and emitter follower are commonly used circuits. Similar to emitter follower (transistor common collector amplifier circuit), field effect tube source follower (common drain amplifier circuit) has no voltage amplification function, and its voltage The gain is less than 1, the output voltage is in the same phase as the input voltage, the input resistance is high, and the output resistance is low, which can be used for impedance transformation. The input resistance of the source follower is extremely high, suitable for connecting to the storage capacitor C j (small drain electrode), the input resistance of the emitter follower is low, not suitable for connecting to the storage capacitor C j (large leakage), but the output of the source follower The resistance is higher than the output resistance of the emitter follower, and the load capacity is relatively poor. To increase the load capacity, the output of the source follower can be connected to the input of the emitter follower, and the output of the emitter can be used as the output of the voltage follower. The output load capacity is greatly enhanced.

Claims (4)

1.一种任意K值DRAM的写入电路,其特征在于:所述的K值DRAM的写入电路中,设K=3,4,5,……;采用K-1=L个变阈型PMOS管Qak,k=1,2,3,……,L,管Qak的栅极经变阈电路连接到写入电路的输入Dinj,变阈型PMOS管Qak的新阈值为tak,管Qak导通时源极漏极间压降为0;管Qak的源极接电源Vdd,选取Vdd的电压比写入电路输入和读出电路输出的最大逻辑电平VDinj(L)和VDouj(L)高Δ,Δ是电压跟随器F输入输出间向下的直流电平偏移;采用L-1个二极管Dan,n=2,3,……,L,二极管Dan的导通电压是VDon;Dan的正极和负极分别连接到变阈型PMOS管Qan-1的漏极和管Qan的漏极;管QaL的漏极经过恒流源Ij接地,管QaL的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管QaL的漏极形成写入电路的输出Gwrij,写入电路输出Gwrij接到存储单元电路的写位线输入;选取tak为写入电路输入Dinj的K值信号的相邻逻辑电平VDinj(k)和VDinj(k-1)的平均值(VDinj(k)+VDinj(k-1))/2,即tak为VDinj(k)和VDinj(k-1)的中间值,VDinj(k)>VDinj(k-1);写入电路输入Dinj的K值信号和读出电路输出Doutj及DRAM输入输出的规范的等阶梯的K值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,也即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,L,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;写入电路的输出Gwrij除0电平外比写入电路输入的K值信号高Δ,0电平仍为0,该K值写入电路又称为K值写入增高电路。1. a write circuit of arbitrary K value DRAM is characterized in that: in the write circuit of described K value DRAM, establish K=3,4,5,...; Adopt K-1=L variable threshold Type PMOS transistor Qa k , k=1, 2, 3,..., L, the gate of transistor Qak k is connected to the input D inj of the writing circuit through the variable threshold circuit, and the new threshold value of the variable threshold type PMOS transistor Qak is ta k , when the tube Qak is turned on, the voltage drop between the source and the drain is 0; the source of the tube Qak is connected to the power supply V dd , and the voltage of V dd is selected to be higher than the maximum logic level of the write circuit input and the read circuit output V Dinj (L) and V Douj (L) are high Δ, Δ is the downward DC level offset between the input and output of the voltage follower F; L-1 diodes Da n are used, n=2, 3,..., L , the turn-on voltage of the diode Dan is V Don ; the anode and cathode of Dan are respectively connected to the drain of the threshold-variable PMOS transistor Qa n-1 and the drain of the transistor Qa n ; the drain of the transistor Qa L is subjected to constant current The source I j is grounded, the drain of the tube Qa L is connected to the constant current source I j so that the current flowing through the conduction diode remains at the same fixed value, and the output G wrij of the writing circuit is formed on the drain of the tube Qa L , and the writing circuit outputs G wrij is connected to the write bit line input of the memory cell circuit; choosing ta k is the average value ( V Dinj (k)+V Dinj (k-1))/2, that is, ta k is the intermediate value of V Dinj (k) and V Dinj (k-1), V Dinj (k)>V Dinj (k-1 ); the characteristics of the K value signal of the K value signal of the input circuit D inj and the standard K value signal of the read circuit output D outj and DRAM input and output are the same: the difference of each adjacent logic level of the input D inj is equal , the difference between adjacent logic levels of the output D outj is equal, and the step voltages of the input D inj and the output D outj are the same, and the step voltage is V Don , that is, V Dinj (m)-V Dinj (m-1)= V Doutj (m)-V Doutj (m-1)=V Don , m=1, 2, 3,..., L, V Dinj (m) and V Doutj (m) are the input and readout of the write circuit respectively The output logic value of the circuit is a logic level of m; the output G wrij of the write circuit is higher than the K value signal input by the write circuit by Δ except for the 0 level, and the 0 level is still 0. The K value write circuit is also called Write the boost circuit for the K value. 2.一种按照权利要求1所述的任意K值DRAM的写入电路的相同结构特征而形成的任意K值DRAM的读出电路,其特征在于:所述的K值DRAM的读出电路中,设K=3,4,5,……;采用K-1=L个变阈型PMOS管Qbk,k=1,2,3,……,L,管Qbk的栅极经变阈电路连接到读出电路的输入Grdij,变阈型PMOS管Qbk的新阈值为tbk,管Qbk导通时源极漏极间压降为0;Grdij接到存储单元电路的读位线输出,管Qbk的源极接电源Vdc,选取Vdc的电压等于写入电路输入和读出电路输出的最大逻辑电平VDinj(L)和VDouj(L);采用L-1个二极管Dbn,n=2,3,……,L,二极管Dbn的导通电压是VDon;Dbn的正极和负极分别连接变阈型PMOS管Qbn-1的漏极和管Qbn的漏极;变阈型PMOS管QbL的漏极经过恒流源Ij接地,管QbL的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管QbL的漏极形成读出电路的输出Doutj;选取tbk为读出电路输入Grdij的K值信号的相邻逻辑电平VGrdij(k)和VGrdij(k-1)的平均值(VGrdij(k)+VGrdij(k-1))/2,即tbk为VGrdij(k)和VGrdij(k-1)的中间值,VGrdij(k)>VGrdij(k-1);读出电路输出Doutj的信号和写入电路输入Dinj及DRAM输入输出的规范的等阶梯的K值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,也即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,L,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;读出电路输入Grdij是来自存储单元电路输出的不规范的K值信号,所述不规范的K值信号就是对比DRAM输入输出和写入电路输入及读出电路输出的规范的等阶梯的K值信号为逻辑电平幅度不一致;读出电路输出Doutj为规范的等阶梯的K值信号,即读出电路将不规范的K值信号输入Grdij转换为规范的等阶梯的K值信号输出Doutj,该K值读出电路又称为K值读出校正电路。2. A readout circuit of any K-value DRAM formed according to the same structural feature of the write-in circuit of any K-value DRAM according to claim 1, is characterized in that: in the readout circuit of described K-value DRAM , assuming K=3, 4, 5, ...; K-1 = L variable-threshold PMOS transistors Qb k , k = 1, 2, 3, ..., L, the gate of the transistor Qb k is variable threshold The circuit is connected to the input G rdij of the readout circuit, the new threshold value of the variable threshold PMOS transistor Qb k is tb k , and the voltage drop between the source and the drain is 0 when the transistor Qb k is turned on; G rdij is connected to the readout of the memory cell circuit Bit line output, the source of the tube Qb k is connected to the power supply V dc , and the voltage of V dc is selected to be equal to the maximum logic level V Dinj (L) and V Douj (L) of the write circuit input and read circuit output; adopt L- 1 diode Db n , n=2, 3, ..., L, the turn-on voltage of diode Db n is V Don ; the anode and cathode of Db n are respectively connected to the drain and the drain of threshold-variable PMOS transistor Qb n-1 The drain of Qb n ; the drain of variable-threshold PMOS transistor Qb L is grounded through constant current source I j , and the drain of tube Qb L is connected to constant current source I j so that the current flowing through the conduction diode remains at the same fixed value. The drain of the tube Qb L forms the output D outj of the readout circuit; select tb k as the average of the adjacent logic levels V Grdij (k) and V Grdij (k-1) of the K-value signal input to the readout circuit Grdij Value (V Grdij (k)+V Grdij (k-1))/2, that is, tb k is the intermediate value of V Grdij (k) and V Grdij (k-1), V Grdij (k)>V Grdij (k -1); The characteristics of the signal of the readout circuit output D outj and the K value signal of the standard equal ladder of the write circuit input D inj and DRAM input and output are the same: the difference of each adjacent logic level of the input D inj is equal , the difference between adjacent logic levels of the output D outj is equal, and the step voltages of the input D inj and the output D outj are the same, and the step voltage is V Don , that is, V Dinj (m)-V Dinj (m-1)= V Doutj (m)-V Doutj (m-1)=V Don , m=1, 2, 3,..., L, V Dinj (m) and V Doutj (m) are the input and readout of the write circuit respectively The circuit output logic value is the logic level of m; the readout circuit input G rdij is the non-standard K value signal from the storage unit circuit output, and the non-standard K value signal is exactly to compare the DRAM input and output with the write circuit input and The standard equal-step K value signal output by the readout circuit is inconsistent with the logic level and amplitude; the output D outj of the readout circuit is a standardized equal-step K value signal, that is, the readout circuit inputs the non-standard K value signal into G r dij is converted into a standard equal-step K value signal output D outj , and the K value readout circuit is also called a K value readout correction circuit. 3.根据权利要求1所述的一种任意K值DRAM的写入电路,其特征在于:取K=8,得出8值DRAM的写入电路,其中采用7个变阈型PMOS管Qak,k=1,2,3,……,7,管Qak的栅极经变阈电路连接到写入电路的输入Dinj,变阈型PMOS管Qak的新阈值为tak,管Qak导通时源极漏极间压降为0;管Qak的源极接电源Vdd,选取Vdd的电压比写入电路输入和读出电路输出的最大逻辑电平VDinj(7)和VDouj(7)高Δ,Δ是电压跟随器F的输入输出间向下的直流电平偏移;采用6个二极管Dan,n=2,3,……,7,二极管Dan的导通电压是VDon;Dan的正极和负极分别连接到变阈型PMOS管Qan-1的漏极和管Qan的漏极;管Qa7的漏极经过恒流源Ij接地,管Qa7的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管Qa7的漏极形成写入电路的输出Gwrij,写入电路输出Gwrij接到存储单元电路的写位线输入;选取tak为写入电路输入Dinj的8值信号的相邻逻辑电平VDinj(k)和VDinj(k-1)的平均值(VDinj(k)+VDinj(k-1))/2,即选取tak为VDinj(k)和VDinj(k-1)的中间值,VDinj(k)>VDinj(k-1);写入电路输入Dinj的8值信号和读出电路输出Doutj及DRAM输入输出的规范的等阶梯的8值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,也即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,7,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;写入电路输出Gwrij除0电平外比写入电路输入的8值信号高Δ,0电平仍为0,该8值写入电路又称为8值写入增高电路。3. the write-in circuit of a kind of arbitrary K value DRAM according to claim 1, is characterized in that: get K=8, draws the write-in circuit of 8-value DRAM, wherein adopts 7 variable-threshold type PMOS transistors Qak , k=1, 2, 3,..., 7, the gate of the tube Qak k is connected to the input D inj of the writing circuit through the variable threshold circuit, the new threshold value of the variable threshold PMOS transistor Qa k is ta k , and the tube Qa k When k is turned on, the voltage drop between the source and the drain is 0; the source of the tube Qa k is connected to the power supply V dd , and the voltage of V dd is selected to be higher than the maximum logic level V Dinj of the write circuit input and the read circuit output (7) and V Douj (7) high Δ, Δ is the downward DC level offset between the input and output of the voltage follower F; using 6 diodes Dan , n=2, 3,..., 7, the conduction of the diode Dan The on-voltage is V Don ; the positive pole and the negative pole of Da n are respectively connected to the drain of the variable-threshold PMOS transistor Qa n-1 and the drain of the pipe Qa n ; the drain of the pipe Qa 7 is grounded through the constant current source Ij , and the pipe The drain of Qa 7 is connected to the constant current source Ij to keep the current flowing through the conduction diode at the same fixed value, and the output G wrij of the writing circuit is formed at the drain of Qa 7 , and the output G wrij of the writing circuit is connected to the storage unit The write bit line input of the circuit; select ta k as the average value (V Dinj ( k ) + V Dinj (k-1))/2, that is, select ta k as the intermediate value of V Dinj (k) and V Dinj (k-1), V Dinj (k)>V Dinj (k-1); write into the circuit The characteristics of the 8-value signal input D inj and the output D outj of the readout circuit and the 8-value signal of the standard equal ladder of the DRAM input and output are the same: the difference of each adjacent logic level of the input D inj is equal, and the output D outj each The difference between adjacent logic levels is equal, and the step voltages of input D inj and output D outj are the same, and the step voltage is V Don , that is, V Dinj (m)-V Dinj (m-1)=V Doutj (m) -V Doutj (m-1)=V Don , m=1, 2, 3,..., 7, V Dinj (m) and V Doutj (m) are respectively the logic values of the input of the write circuit and the output of the read circuit The logic level of m; the write circuit output G wrij is Δ higher than the 8-value signal input by the write circuit except for the 0 level, and the 0 level is still 0. The 8-value write circuit is also called 8-value write increase circuit. 4.根据权利要求2所述的一种按照权利要求1所述的任意K值DRAM的写入电路的相同结构特征而形成的任意K值DRAM的读出电路,其特征在于:取K=8,得出8值DRAM的读出电路,其中采用7个变阈型PMOS管Qbk,k=1,2,3,……,7,管Qbk的栅极经变阈电路连接到读出电路的输入Grdij,变阈型PMOS管Qbk的新阈值为tbk;管Qbk导通时源极漏极间压降为0;Grdij接到存储单元电路的读位线输出,管Qbk的源极接电源Vdc,选取Vdc的电压等于写入电路输入和读出电路输出的最大逻辑电平VDinj(7)和VDouj(7);采用6个二极管Dbn,n=2,3,……,7,二极管Dbn的导通电压是VDon;Dbn的正极和负极分别连接变阈型PMOS管Qbn-1的漏极和管Qbn的漏极;变阈型PMOS管Qb7的漏极经过恒流源Ij接地,管Qb7的漏极接恒流源Ij使流经导通二极管的电流保持同一固定值,在管Qb7的漏极形成读出电路的输出Doutj;选取tbk为读出电路输入Grdij的8值信号的相邻逻辑电平VGrdij(k)和VGrdij(k-1)的平均值(VGrdij(k)+VGrdij(k-1))/2,即选取tbk为VGrdij(k)和VGrdij(k-1)的中间值,VGrdij(k)>VGrdij(k-1);读出电路输出Doutj的信号和写入电路输入Dinj及DRAM输入输出的规范的等阶梯的8值信号的特性是相同的:输入Dinj各相邻逻辑电平的差相等,输出Doutj各相邻逻辑电平的差相等,且输入Dinj和输出Doutj的阶梯电压相同,阶梯电压为VDon,即满足VDinj(m)-VDinj(m-1)=VDoutj(m)-VDoutj(m-1)=VDon,m=1,2,3,……,7,VDinj(m)和VDoutj(m)分别是写入电路输入和读出电路输出逻辑值为m的逻辑电平;读出电路输入Grdij是来自存储单元电路输出的不规范的8值信号,所述不规范的8值信号就是对比DRAM输入输出和写入电路输入及读出电路输出的规范的等阶梯的8值信号为逻辑电平幅度不一致;读出电路输出Doutj为规范的等阶梯的8值信号,即读出电路将不规范的8值信号Grdij转换为规范的等阶梯的8值信号Doutj,该8值读出电路又称为8值读出校正电路。4. a kind of readout circuit of the arbitrary K value DRAM that a kind of according to the same structural feature of the write circuit of arbitrary K value DRAM described in claim 1 forms according to claim 2, is characterized in that: get K=8 , to obtain the readout circuit of 8-value DRAM, wherein seven variable-threshold PMOS transistors Qb k are used, k=1, 2, 3, ..., 7, and the gate of the transistor Qb k is connected to the readout circuit through the variable-threshold circuit The input G rdij of the circuit, the new threshold value of the threshold-variable PMOS transistor Qb k is tb k ; when the transistor Qb k is turned on, the voltage drop between the source and the drain is 0; G rdij is connected to the output of the read bit line of the memory cell circuit, and the transistor The source of Qb k is connected to the power supply V dc , and the voltage of V dc is selected to be equal to the maximum logic level V Dinj (7) and V Douj (7) of the write circuit input and read circuit output; use 6 diodes Db n , n =2, 3, ..., 7, the conduction voltage of the diode Db n is V Don ; the positive pole and the negative pole of Db n are respectively connected to the drain of the variable threshold PMOS transistor Qb n-1 and the drain of the tube Qb n ; The drain of the threshold-type PMOS transistor Qb 7 is connected to the ground through the constant current source Ij , and the drain of the transistor Qb 7 is connected to the constant current source Ij so that the current flowing through the conduction diode remains at the same fixed value, and the drain of the transistor Qb 7 forms The output D outj of readout circuit ; Choose tb k to be the average value (V Grdij ( k ) +V Grdij (k-1))/2, that is, select tb k as the intermediate value of V Grdij (k) and V Grdij (k-1), V Grdij (k)>V Grdij (k-1); read out The characteristics of the signal of the circuit output D outj and the 8-valued signal written into the circuit input D inj and DRAM input and output specifications are the same: the difference of each adjacent logic level of the input D inj is equal, and the output D outj of each phase The difference between adjacent logic levels is equal, and the step voltage of the input D inj and the output D outj are the same, and the step voltage is V Don , which satisfies V Dinj (m)-V Dinj (m-1)=V Doutj (m)-V Doutj (m-1)=V Don , m=1, 2, 3,..., 7, V Dinj (m) and V Doutj (m) are respectively the logic values of the input circuit input and the output logic value of the readout circuit m Logical level; the readout circuit input G rdij is an irregular 8-value signal from the memory unit circuit output, and the irregular 8-value signal is exactly compared to the standard of the DRAM input and output, the write circuit input and the readout circuit output The 8-valued signal of the equal ladder is inconsistent with the logic level amplitude; the output D outj of the readout circuit is a standardized 8-valued signal of the equal ladder, that is, the readout circuit converts the non-standard 8-valued signal G rdij to If it is replaced with a standard equal-level 8-value signal D outj , the 8-value readout circuit is also called an 8-value readout correction circuit.
CN201310211023.2A 2011-04-19 2011-04-19 Arbitrarily K value and 8 is worth write circuit and the reading circuit of DRAM Expired - Fee Related CN103345936B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310211023.2A CN103345936B (en) 2011-04-19 2011-04-19 Arbitrarily K value and 8 is worth write circuit and the reading circuit of DRAM

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310211023.2A CN103345936B (en) 2011-04-19 2011-04-19 Arbitrarily K value and 8 is worth write circuit and the reading circuit of DRAM

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
CN 201110097206 Division CN102290095B (en) 2011-04-19 2011-04-19 Storage unit circuit for any K-valued and 8-valued DRAM (dynamic random access memory)

Publications (2)

Publication Number Publication Date
CN103345936A true CN103345936A (en) 2013-10-09
CN103345936B CN103345936B (en) 2016-08-03

Family

ID=49280726

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310211023.2A Expired - Fee Related CN103345936B (en) 2011-04-19 2011-04-19 Arbitrarily K value and 8 is worth write circuit and the reading circuit of DRAM

Country Status (1)

Country Link
CN (1) CN103345936B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113436660A (en) * 2020-03-23 2021-09-24 长鑫存储技术有限公司 Latch circuit
CN113571107A (en) * 2020-07-14 2021-10-29 台湾积体电路制造股份有限公司 Method and integrated circuit for operating an integrated circuit

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1231481A (en) * 1998-03-05 1999-10-13 日本电气株式会社 Multivalue semiconductor memory device and error erasing method thereof
CN1232272A (en) * 1998-04-14 1999-10-20 日本电气株式会社 semiconductor multi-valued memory device
CN1428867A (en) * 2001-12-27 2003-07-09 株式会社东芝 Nonvolatile semiconductor memory for storage multiple-valued data in single storage unit
US20100118597A1 (en) * 2007-01-16 2010-05-13 Chungbuk National University Industry-Academic Cooperation Foundation Multiple valued dynamic random access memory cell and thereof array using single electron transistor
CN101828233A (en) * 2007-10-15 2010-09-08 S.阿夸半导体有限公司 Multivalue memory storage with two gating transistors
US20100254178A1 (en) * 2007-12-12 2010-10-07 Sony Corporation Storage device and information re-recording method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1231481A (en) * 1998-03-05 1999-10-13 日本电气株式会社 Multivalue semiconductor memory device and error erasing method thereof
CN1232272A (en) * 1998-04-14 1999-10-20 日本电气株式会社 semiconductor multi-valued memory device
CN1428867A (en) * 2001-12-27 2003-07-09 株式会社东芝 Nonvolatile semiconductor memory for storage multiple-valued data in single storage unit
US20100118597A1 (en) * 2007-01-16 2010-05-13 Chungbuk National University Industry-Academic Cooperation Foundation Multiple valued dynamic random access memory cell and thereof array using single electron transistor
CN101828233A (en) * 2007-10-15 2010-09-08 S.阿夸半导体有限公司 Multivalue memory storage with two gating transistors
US20100254178A1 (en) * 2007-12-12 2010-10-07 Sony Corporation Storage device and information re-recording method

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113436660A (en) * 2020-03-23 2021-09-24 长鑫存储技术有限公司 Latch circuit
CN113436660B (en) * 2020-03-23 2022-05-24 长鑫存储技术有限公司 Latch circuit
US11705893B2 (en) 2020-03-23 2023-07-18 Changxin Memory Technologies, Inc. Latch circuit
CN113571107A (en) * 2020-07-14 2021-10-29 台湾积体电路制造股份有限公司 Method and integrated circuit for operating an integrated circuit

Also Published As

Publication number Publication date
CN103345936B (en) 2016-08-03

Similar Documents

Publication Publication Date Title
CN111046617B (en) Memristor-based three-value digital logic gate circuit
CN113467751B (en) Analog domain memory internal computing array structure based on magnetic random access memory
CN110942792B (en) Low-power-consumption low-leakage SRAM (static random Access memory) applied to storage and calculation integrated chip
CN109979503B (en) Static random access memory circuit structure for realizing Hamming distance calculation in memory
US20020071308A1 (en) Semiconductor memory device having memory cells each capable of storing three or more values
CN112133348B (en) Storage unit, storage array and memory computing device based on 6T unit
CN104966532A (en) One-time programmable memory unit and circuit
CN111816233B (en) In-memory computing unit and array
CN102081959A (en) A memory readout circuit and memory
CN112185447B (en) 8-pipe double-split control storage unit, storage array and in-memory computing device
CN114496010B (en) An analog domain near-memory computing array structure based on magnetic random access memory
CN110176264A (en) A kind of high-low-position consolidation circuit structure calculated interior based on memory
CN102426855B (en) 8-value memory cell embedded in DRAM storage matrix, and corresponding conversion circuit thereof
CN101986389A (en) Flash memory unit, flash memory device and programming method thereof
CN103345936B (en) Arbitrarily K value and 8 is worth write circuit and the reading circuit of DRAM
CN106158022B (en) Word line driving circuit and method for common source architecture embedded flash memory
CN204102573U (en) A kind of novel 12 pipe sram cell circuit improving read noise tolerance limit simultaneously and write nargin
CN102290095A (en) Storage unit circuit, write circuit and read circuit for any K-valued and 8-valued DRAM (dynamic random access memory)
CN102436847B (en) PMOS (P-channel Metal Oxide Semiconductor) pipe band-pass-band-stop changeable threshold circuit and PMOS pipe high-pass-low-pass changeable threshold circuit
CN103714853A (en) NAND content addressable memory
CN206461708U (en) ADC dynamic logics reverse circuit, word line voltage selection circuit and storage unit circuit
CN108717859A (en) A kind of three value SRAM cell circuits using carbon nano field-effect transistor
CN116631468A (en) Dynamic random access memory array circuit
Wei et al. Extending non-volatile operation to DRAM cells
CN115050406A (en) Bit line leakage current compensation circuit and module of SRAM (static random Access memory) and memory

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20160803

Termination date: 20200419

CF01 Termination of patent right due to non-payment of annual fee