盖子和地面有什么区别？

对不起，这个数字的顺序错了;
第一个案例与最后一个数字相对应;
第二种情况与第二种情况相对应;
并且第一个数字对应于所有三个案例的眼图。



     以下为原文

  sorry , The figure's order is wrong ;
The first case correspondences to the last figure;
The second case correspondences  to second figure;
and the first figure correspondences to  the eye diagram of the all three cases.

> {quote：title = echo write：} {quote}>在第二种情况下，我添加了图层，端口在此图层开始，就像第一种情况一样，但是图层是我自己添加的。
你可以看到第二张图，也许会更清楚。
我认为在这种情况下，我自己添加的图层就是地面图层。
你可以做到这一点，但是模拟效率很低，因为地面必须由求解器进行网格划分。
根据您的设置，这可能不如使用基板底部的无限接地那么准确。
在大多数情况下，使用基板底部的无限接地更好（也更有效）。
不用担心：无限接地回路的电感/路径长度正确包含在结果中。
那是因为无限地被模拟为分布式金属，这正是我们想要的。
您甚至可以为其分配金属，以包括地平面损耗。
~~但是你的EM模型中肯定会产生一些不好的结果：传输线与默认设置啮合，在整个线宽上有恒定的电流。
这不仅会导致损失（趋肤效应！），还会导致线之间的耦合产生大的误差。
在emSetup中，转到选项>网格并启用边网格设置。
查看Momentum网格划分的在线帮助。
这是取得好成绩的关键。
~~此外，介质叠层在3D视图中看起来很意外。
您可以上传基板定义的屏幕截图，还有一些文字说明真正的硬件基板配置是什么？



     以下为原文

  > {quote:title=echo wrote:}{quote}
> In the second case, I add the layer, and the port started at this layer, just like the first case but the layer is add by myself. You can see the second Figure and maybe that will be more clear. I think in the case the layer add by myself acts as the ground layer.

You can do this, but it is inefficient for simulation, because the ground must be meshed by the solver. Depending on your settings, this can be less accurate than using the infinite ground at the bottom of the substrate. 

In most cases, it is better (and more efficient) to use the infinite ground at the bottom of the substrate. 

Don't worry: The inductance/path length of that infinite ground return is properly included in results. That is because the infinite ground is simulated as a distributed metal, which is what we want. You can even assign a metal to it, to include the ground plane losses.

~~

But there is something in your EM model that will for sure create bad results: the transmission lines are meshed with default settings, with a constant current across the entire line width. That will give large error not only in loss (skin effect!), but also in the coupling between the lines. In the emSetup, go to Options > Mesh and enable the Edge Mesh setting.

Have a look at the online help on Momentum meshing. This is the key to good results.

~~

Also, the dielectric stackup looks unexpected in the 3D view. Can you upload a screenshot of your substrate definition, and some text what the real hardware substrate configuration is?

> {quote：title = echo写道：} {quote}>感谢您的建议！
>你的意思是如果我自己添加一个无限的地面层是正确的。
在你的情况下，最好的叠加是第一个叠加，具有无限的基础。
这是因为这避免了地平面中的离散化误差。
但重要的是在层上的导体上有一个更精细的网格。
在网格设置中，打开边网格。
这将对您的情况下的线路阻抗产生重大影响。
这些近距离导体上的默认网格不够准确，你真的需要边缘网格！
>我添加基板的图片！
这看起来是错误的：在基板的底部，在地平面下方，你有一层无限厚的氧化铝，而不是空气。
这不应该改变地平面以上的结果，但肯定不是你想要的模型。
但无论如何：放下这个“绘制”的地平面，并采用无限地平面模型。
这很简单，也很安全，并避免了一些错误，你可以用“绘制”的网格划分和端口。



     以下为原文

  > {quote:title=echo wrote:}{quote}
> Thanks for your advices!
> You mean it  would be right if i add a  infinite ground layer by myself . 

The best stackup in your case is the first stackup, with the infinite ground. 
This is because that avoids discretization errors in the ground plane.

But it is important to have a finer mesh on the conductors on layer cond. In the mesh settings, switch on edge mesh. This will have a significant effect on line impedance in your case. The default mesh on these close space conductor is NOT accurate enough, and you really NEED edge mesh here!

> And I add the picture of the substrates!

That's what looked wrong: at the bottom side of the substrate, below the ground plane, you have an infinite thick layer of Alumina, instead of air. That should not change the results above the ground plane, but certainly it is not what you wanted to model.

But anyway: drop this "drawn" ground plane, and go with the infinite ground plane model. That's easy and safe, and avoid some errors that you can make with "drawn" grounds for meshing and ports.

我知道了。
非常感谢！



     以下为原文

  I get it.

Thanks a lot!