论文部分内容阅读
能用MOS电荷耦合技术完成的电荷运算的主要方法有存储、传输、分裂、组合、输入和提取。我们把这些运算方法理想化、通用化,以便定义一个用于离散时间线性滤波的通用型网络。P相电荷流程网络(CRN)由一群分成P个子群,其流程受P相钟控制的存储单元组成。在一特定钟相位期间,电荷从一特定子群存储单元流入另一子群。这样,电荷就逐次通过每个子群的各个单元,通过确定其加权值与流程有关的矩阵来确定周期时变线性离散时间网络。由P相CRN的分析得到一组简化的线性时不变动态方程,这就为研究信号处理提供了方便。本文给出了这组动态方程的充分和必要条件,以便实现P相CRN。由CRN获得的各种部件可用来实现无限脉冲响应滤波器的传递函数。但是,对用CRN实现的传递函数的种类有一些重要的限制。具体地说,可以看到在Z平面上的单位圆内存在禁区,这里不可能有CRN的极点(或自然模)。在经典的RC网络和电荷耦合网络之间作个比较,可以看到,两种网络都对能用它们本身实现的滤波器的种类有较大的限制。对于RC网络来说,我们知道通过外加另一种元件(电感或运算放大器)就能消除这些限制。对于CRN网络来说,可能存在同样的潜力。本文给出了一般的理论基础,它能提供用电荷耦合网络进行离散时间滤波的实现条件和综合方法。
The main methods of charge calculation that can be performed using MOS charge-coupled techniques are storage, transfer, splitting, combining, inputting, and extraction. We have idealized and generalized these algorithms in order to define a general purpose network for discrete-time linear filtering. The P phase charge flow network (CRN) consists of a group of memory cells divided into P subgroups and whose flow is controlled by the P phase. During a particular clock phase, charge flows from one particular subgroup memory cell to another subgroup. In this way, the charge passes through each element of each subgroup sequentially, and the periodic linear time-varying discrete-time network is determined by determining the matrix whose weight is related to the flow. By the analysis of P-phase CRN, a set of simplified linear time-invariant dynamic equations is obtained, which is convenient for the study of signal processing. In this paper, the sufficient and necessary conditions for this group of dynamic equations are given in order to realize P-phase CRN. The various components available from the CRN can be used to implement the transfer function of an infinite impulse response filter. However, there are some important limitations on the types of transfer functions implemented with CRNs. In particular, it can be seen that there is a forbidden zone within the unit circle in the Z plane, where there is no pole (or natural mode) of the CRN. In comparing classical RC networks with charge-coupled networks, it can be seen that both networks have a greater limitation on the types of filters that can be implemented with them. For the RC network, we know that these limitations can be eliminated by adding another component (inductor or op amp). The same potential for CRN networks may exist. This paper gives a general theoretical basis, which can provide the realization conditions and synthesis method of discrete time filtering by charge coupled network.