差分阻抗

单线

        图1(a)演示了一个典型的单根走线。其特征阻抗是Z0,其上流经的电流为i。沿线任意一点的电压为V=Z0*i(根据欧姆定律)。

        一般情况,线对:图1(b)演示了一对走线。线1具有特征阻抗Z11,与上文中Z0一致,电流i1。线2具有类似的定义。当我们将线2向线1靠近时,线2上的电流开始以比例常数k耦合到线1上。类似地,线1的电流i1开始以同样的比例常数耦合到线2上。每根走线上任意一点的电压,还是根据欧姆定律,为:

V= Z11*i1 + Z11*k*i2 (1)
V2 = Z22*i2 + Z22*k*i1

        现在我们定义Z12 = k*Z11以及Z21 = k*Z22。这样,式(1)就可以写成:

V= Z11*i1 + Z12*i2  (2)
V= Z21*i1 + Z22*i2

        这是一对熟悉的联立方程组,我们可以经常在教科书中看到。这个方程组可以推广到任意数量的走线,并且可以用你们中大部分人都熟悉的矩阵形式来表示。

        特殊情况,差分对:图1(c)演示了一对差分走线。重写式1:

V= Z11*i1 + Z11*k*i(1)
V2 = Z22*i2 + Z21*k*i1

        现在注意在仔细设计并且是对称的情况下,Z11 = Z22 = Z0,且i= -i1这将导致(经过一些变换):

V1 = Z0*i1*(1-k)  (3)
V2 = -Z0*i1*(1-k)

        注意V1 = -V2,当然,这是我们已经知道的,因为这是一个差分对。

有效(差模)阻抗

        电压V1以地为参考。线1的有效阻抗(单独来看,在差分对中叫做“差模”阻抗,通常叫做“单线”阻抗)为电压除以电流,或:

Zodd = V1/i1 = Z0*(1-k)

        由上可知,因Z0 = Z11 且 k = Z12/Z11,上式可写成:Zodd = Z11 - Z12这也是一个在许多教科书中都可以看到的公式。

        为了防止反射,正确的端接方法是用一个值为Zodd的电阻。类似地,线2的差模阻抗与此相同(在对称差分对的特定情形下)。

差分阻抗

        假定在某一瞬间我们将两根走线用电阻端接到地。因为i1 = -i2,所以根本没有电流流经地。也就是说,没有真正的理由把电阻接地。事实上,有人认为,为了将差分信号和地噪声隔离,一定不能将它们接地。因此通常的连接形式如图1(c)中所示,用单个电阻连接线1与线2。电阻的值是线1和线2差模阻抗的和,或:

Zdiff  = 2*Z0*(1-k) 或
  2*(Z11 - Z12)

        这就是为什么你经常看到实际上一个差分对具有大约80Ω的差分阻抗,而每个单线阻抗是50Ω。

计算

        知道Zdiff是2*(Z11-Z12)不是很有用,因为Z12的值并不直观。但是,当我们看到Z12与耦合系数k有关,事情就变得清晰了。事实上,耦合系数与我在Brookspeak中关于串扰的专栏[1]中谈到的耦合系数是相同的。国家半导体发布的计算Zdiff的公式[2]已经被广泛接受:

Zdiff = 2*Z0(1-.48*e-.96*S/H)   微带线
Zdiff = 2*Z0(1-.347*e-2.9*S/H)  带状线

其中的术语在图2中定义。Z0为其传统定义[3] 。


 
图 2 差分阻抗计算中的术语定义

共模阻抗

        为了讨论完整起见,共模阻抗与上面略有不同。第一个差别是i1 = i2(没有负号),这样式3就变成:

V1 = Z0*i1*(1+k) (4)
V2 = Z0*i1*(1+k)

        并且正如所期望的,V1 = V2。因此单线阻抗是Z0*(1+k)。在共模情况下,两根线的端接电阻均接地,所以流经地的电流为i1+i2且这两个电阻对器件表现为并联。也就是说,共模阻抗是这些电阻的并联组合,或:
Zcommon = (1/2)*Z0*(1+k),或

Zcommon = (1/2)*(Z11 + Z12)

注意,这里差分对的共模阻抗大约为差模阻抗的1/4。


Just when you thought you had mastered Zo, the characteristic impedance of a PCB trace, along comes a data sheet that tells you to design for a specific differential impedance. And to make things tougher, it says things like: “… since the coupling of two traces can lower the effective impedance, use 50 Ohm design rules to achieve a differential impedance of approximately 80 Ohms!” Is that confusing or what!! This article shows you what differential impedance is. But more than that, it discusses why it is, and shows you how to make the correct calculations. Single Trace: Figure 1(a) illustrates a typical, individual trace. It has a characteristic impedance, Zo, and carries a current, i. The voltage along it, at any point, is (from Ohm’s law) V = Zo*i. General case, trace pair: Figure 1(b) illustrates a pair of traces. Trace 1 has a characteristic impedance Z11, which corresponds to Zo, above, and current i1. Trace 2 is similarly defined. As we bring Trace 2 closer to Trace 1, current from Trace 2 begins to couple into Trace 1 with a proportionality constant, k. Similarly, Trace 1’s current, i1, begins to couple into Trace 2 with the same proportionality constant. The voltage on each trace, at any point, again from Ohm’s law, is: V1 = Z11 * i1 + Z11 * k * i2 Eqs. 1 V2 = Z22 * i2 + Z22 * k * i1 Now let’s define Z12 = k*Z11 and Z21 = k*Z22. Then, Eqs. 1 can be written as: V1 = Z11 * i1 + Z12 * i2 Eqs. 2 V2 = Z21 * i1 + Z22 * i2 This is the familiar pair of simultaneous equations we often see in texts. The equations can be generalized into an arbitrary number of traces, and they can be expressed in a matrix form that is familiar to many of you. Special case, differential pair: Figure 1(c) illustrates a differential pair of traces. Repeating Equations 1: V1 = Z11 * i1 + Z11 * k * i2 Eqs. 1 V2 = Z22 * i2 + Z22 * k * i1 Now, note that in a carefully designed and balanced situation, Z11 = Z22 = Zo, and i2 = -i1 This leads (with a little manipulation) to: V1 = Zo * i1 * (1-k)
评论 2
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值