微波EDA网,见证研发工程师的成长!
首页 > 研发问答 > 微电子和IC设计 > IC后端设计交流 > cell delay疑问

cell delay疑问

时间:10-02 整理:3721RD 点击:

cell delay由output load和input transition决定。
the delay increases with increasing input transition time?这句怎么理解,谢谢

单元的输出延迟是通过NDLM 来特征化的,工具通过查二维表来获得cell 的delay 信息,其中cell输入信号的transition 和cell
输出load 共同来确定cell 的延迟。输入信号的transition越大,cell的延迟越大。

谢谢回复。NLDM查找表这个我明白,但是我对lib中的具体来由有疑惑。cell的输出load越大,output的slew越大,所以cell的延迟越大。这个能理解
我的疑问是:input transition越大,cell的延迟越大?原因是什么

很形象的描述一下吧,在数字电路中,对MOS 管进行建模,我们认为他是一个带有开关的电流源,一个cell的速度的快慢与这个电流源对其负载电容的充放电的时间来决定,因此我们就要求这个电流源的充放电的电流越大越好,我们知道MOS管是一个压控电流源,其电流的大小取决于其栅端的电压的大小来决定。栅端电压即输入电压有一个很大的slew/transition会是这个电压在很长的时间内都在一个比较小的之内徘徊,因而电流源在很长的时间段内都在一个比较小的值内,充放电的电流在很长的时间内自然不是很大,cell的速度自然也就变慢了。最理想的情况就是在输入电压端加上一个阶跃电压的情况。
这样的描述不是很严谨,但是比较形象,建议看一看一些理论书,例如 COMS VLSI....

请教一下,您说通过查二维表获得cell 的delay值,但二维表不能穷举所有的值,如果cell的input transition和output load的值不在二维表列举的值当中,这个时候怎么得到cell的delay啊?
谢谢指教!

有一个公式,利用内插或者是外插法来计算。数学的理论知识在数值分析中有介绍。EDA工具会根据这个公式计算出来。

请问这个公式是什么样的啊?谢谢!

见书:static timing analysis for nanometer designP53

谢谢大侠!

This section illustrates the lookup of the table models through an example.

If the input transition time and the output capacitance correspond to a table

entry, the table lookup is trivial since the timing value corresponds directly

to the value in the table. The example below corresponds to a general

case where the lookup does not correspond to any of the entries available

in the table. In such cases, two-dimensional interpolation is utilized to provide

the resulting timing value. The two nearest table indices in each dimension

are chosen for the table interpolation. Consider the table lookup

for fall transition (example table specified above) for the input transition

time of 0.15ns and an output capacitance of 1.16pF. The corresponding section

of the fall transition table relevant for two-dimensional interpolation is

reproduced below.

fall_transition(delay_template_3x3) {

index_1 ("0.1, 0.3 . . .");

index_2 (". . . 0.35, 1.43");

values ( \

". . . 0.1937, 0.7280", \

". . . 0.2327, 0.7676"

. . .

In the formulation below, the two index_1 values are denoted as x1 and x2;

the two index_2 values are denoted as y1 and y2 and the corresponding table

values are denoted as T11, T12, T21 and T22 respectively.

If the table lookup is required for (x0, y0), the lookup value T00 is obtained

by interpolation and is given by:

T00 = x20 * y20 * T11 + x20 * y01 * T12 +

x01 * y20 * T21 + x01 * y01 * T22

where

x01 = (x0 - x1) / (x2 - x1)

x20 = (x2 - x0) / (x2 - x1)

y01 = (y0 - y1) / (y2 - y1)

y20 = (y2 - y0) / (y2 - y1)

Substituting 0.15 for index_1 and 1.16 for index_2 results in the

fall_transition value of:

T00 = 0.75 * 0.25 * 0.1937 + 0.75 * 0.75 * 0.7280 +

0.25 * 0.25 * 0.2327 + 0.25 * 0.75 * 0.7676 = 0.6043

Note that the equations above are valid for interpolation as well as extrapolation

- that is when the indices (x0, y0) lie outside the characterized range

of indices. As an example, for the table lookup with 0.05 for index_1 and 1.7

for index_2, the fall transition value is obtained as:

T00 = 1.25 * (-0.25) * 0.1937 + 1.25 * 1.25 * 0.7280 +

(-0.25) * (-0.25) * 0.2327 + (-0.25) * 1.25 * 0.7676

= 0.8516


mark

请问cell delay和Vth有关吗?

学习学习

Copyright © 2017-2020 微波EDA网 版权所有

网站地图

Top