嗨Angelo，在过去两个小时左右处理你的问题后，我没有意识到你已经更新了线程。
但无论如何，你的问题和工作空间中埋藏了多个有趣的主题.... ~~~~~~我看了你的工作区，发现有很多值得一提的话题。
在原理图中，您已为电阻器定义了100欧姆/平方。
在Momentum模型中，您已将片状电阻值定义为100欧姆/平方，但将其应用于3D物体。
对我来说，这没有意义，我很惊讶Momentum接受了这一输入。
薄板电阻（欧姆/平方）定义的损耗是针对薄板定义的，如果将堆叠中的电阻从3D更改为薄金属板，您会发现Momentum给出了预期的串联电阻。
请参见附件截图sheetres.png。
*解决了* ~~~~~~~~~~~~~~~~~这是答案可以结束的地方，但..... ~~~~~~~~~~~~~~~~
~~~~~~~~所以简单的答案就是你可以把你的薄膜电阻器换成钣金模型。
鉴于电阻器与水平尺寸相比非常薄，并且鉴于趋肤效应与此处无关，因为趋肤深度远大于电阻器的厚度，因此100欧姆/平方的薄片建模非常精确和准确。
但无论如何，你把它建模为3D物体，所以也许你有充分的理由让我不知道。
所以我开始研究建模为3D对象的电阻，因为如果我们提供正确的材料定义，它应该可以工作。
在基板厚度为1μm的情况下，我将100欧姆/平方转换为电导率sigma = 1 /（薄片*厚度），并获得sigma = 10kS / m的电阻电导率。
我将其应用于基板中的3D通道......并发现了意外的模拟结果。
电阻为70欧姆，而不是50欧姆。
从电流密度来看，与顶部相比，我看到电阻器底部的电流很小。
但对于低频电阻，我们真的希望双方的电流密度相同，所以有些不对劲。
*要做*：我认为安捷伦EM人员应该看一下 - 夹层导体在这里的表现可能与我们预期的不同。
我怀疑问题来自铜线和薄膜电阻器之间的接口/过渡（当前电流可以在解算器网格的限制范围内）。
在原始布局中，您已经在整条线下方建模了薄膜电阻，沿着铜线重叠。
为了测试这一点，我创建了一种替代布局，其中薄膜电阻限于铜馈线之间的区域，铜和电阻之间没有重叠。
这现在工作正常，我通过= 10kS / m 3D体积获得预期结果。
*使用其他布局解决了Momentum *（没有堆叠导体）。
所以最后一个问题是，在我们修复了材料定义之后，* FEM *现在是否给出了正确的结果。
原始材料定义失败并不奇怪 - 您要求使用“网格内部”解决3D过孔/导体，但材料仅定义为2D薄层电阻。
所以我们需要改变它。
我确实得到了预期的50欧姆串联电阻用于FEM的3D通孔和网状内部用于我的替代布局，其中电阻器仅跨越两条铜线之间的间隙。
对于我的替代布局，我还获得了预期的50欧姆系列电阻，用于薄板模型。
*解决了*替代布局，没有堆叠导体那么我们得到的原始布局有FEM然后，电阻和铜重叠？
首先，我尝试了薄板模型，它给出了预期的结果。
接下来，我尝试了3D通孔模型，导体设置为“网状内部”，这也给出了低频时的预期50欧姆。
*解决了*有限元的原始布局，带有堆叠导体。
我的修改后的工作区已附加（删除了“simulations”文件夹以限制存档大小）。
使用File> Unarchive导入ADS。
请注意两种基板定义（3D通孔，导电片型号为100欧姆/平方）。
最好的问候Volker编辑：volker_muehlhaus于2013年10月31日上午8:21
test_thin_film_resistors_wrk.7zads148.9 KB
以上来自于谷歌翻译
以下为原文
Hi Angelo,
after working on your question for the last two hours or so, I didn't realize that you have already updated the thread. But anyway, there are multiple interesting topics buried in your question and workspace ....
~~~~~~
I looked at your workspace and noticed that there are multiple topics worth mentioning.
In your schematic, you have defined 100 Ohm/square for the resistor. In the Momentum model, you have defined a sheet resistance value of 100 Ohm/square, but applied that to a 3D via object. To me, this doesn't make sense and I am surprised that Momentum accepts that input. Loss defined by sheet resistance (ohm/sqare) is defined for thin sheets, and you will find that Momentum gives the expected series resistance if you change the resistor in the stackup from 3D via to thin metal sheet. See attached screenshot sheetres.png.
*solved*
~~~~~~~~~~~~~~~~~ this is where the answer could end, but ..... ~~~~~~~~~~~~~~~~~~~~~~~~
So the trivial answer is that you could just change your sheet resistor to sheet metal model. Given that the resistor is very thin compared to the horizontal dimensions, and given that skin effect is not relevant here because the skin depth is much larger than the resistor's thickness, thin sheet modelling with 100 ohm/square is perfectly fine and accurate. But anyway, you modelled it as a 3D object, so maybe you had good reasons that I don't know.
So I started looking at the resistor modelled as a 3D object, because that should work, if we supply the proper material definition. With the 1µm thickness from your substrate, I converted the 100ohm/square into conductivity
sigma = 1 / (sheetres * thickness)
and obtained sigma = 10kS/m for the resistor conductivity. I applied that to the 3D via in your substrate ... and found unexpected simulation results. The resistance is 70 Ohm, instead of 50 Ohm. Looking at the current density, I see little current on the bottomside of the resistor, compared to the top side. But for the resistor at low frequency, we would really expect same current density on both sides, so something is wrong.
*TO DO*: I think the Agilent EM guys should have a look at that - sandwiched conductors might behave different here than we expect.
I am suspecting that the trouble is from the interface/transition between the copper line and the sheet resistor (where currebt can go, within the limitations of the solver mesh). In your original layout, you had modelled the sheet resistor underneath the entire line, with overlap along the copper lines. To test this, I create an alternative layout, where the sheet resistor is limited to the area between the copper feedlines, with no overlap between copper and resistor. This works fine now, and I get the expected results with = 10kS/m 3D volume via.
*Solved* for Momentum with alternative layout (no stacked conductors).
So the final question was if *FEM* gives correct results now, after we have fixed the material definition. It is no surprise that the original material definition failed - you requested to solve 3D vias/conductors with "meshed interior", but the material was only defined as 2D sheet resistance. So we need to change that.
I do get the expected 50 ohm series resistance for FEM with 3D via and meshed interior for my alternative layout, where the resistor only spans the gap between both copper lines. I also get the expected 50 ohm series resistance for FEM with thin sheet model for my alternative layout.
*Solved* for the alternative layout, with no stacked conductors
So what do we get for the original layout with FEM then, where resistor and copper overlap? First, I tried the thin sheet model, which gives the expected result. Next, I tried the 3D via model with conductor set to "meshed interior", which also gives the expected 50 ohm at low frequency.
*Solved* for the original layout in FEM, with stacked conductors.
My modified workspace is attached (with "simulations" folder deleted to limit the archive size).
Import into ADS with File > Unarchive.
Note the two substrate definitions (3D via with conductivity sheet model with 100 ohm/square) .
Best regards
Volker
Edited by: volker_muehlhaus on Oct 31, 2013 8:21 AM
附件
嗨Angelo，在过去两个小时左右处理你的问题后，我没有意识到你已经更新了线程。
但无论如何，你的问题和工作空间中埋藏了多个有趣的主题.... ~~~~~~我看了你的工作区，发现有很多值得一提的话题。
在原理图中，您已为电阻器定义了100欧姆/平方。
在Momentum模型中，您已将片状电阻值定义为100欧姆/平方，但将其应用于3D物体。
对我来说，这没有意义，我很惊讶Momentum接受了这一输入。
薄板电阻（欧姆/平方）定义的损耗是针对薄板定义的，如果将堆叠中的电阻从3D更改为薄金属板，您会发现Momentum给出了预期的串联电阻。
请参见附件截图sheetres.png。
*解决了* ~~~~~~~~~~~~~~~~~这是答案可以结束的地方，但..... ~~~~~~~~~~~~~~~~
~~~~~~~~所以简单的答案就是你可以把你的薄膜电阻器换成钣金模型。
鉴于电阻器与水平尺寸相比非常薄，并且鉴于趋肤效应与此处无关，因为趋肤深度远大于电阻器的厚度，因此100欧姆/平方的薄片建模非常精确和准确。
但无论如何，你把它建模为3D物体，所以也许你有充分的理由让我不知道。
所以我开始研究建模为3D对象的电阻，因为如果我们提供正确的材料定义，它应该可以工作。
在基板厚度为1μm的情况下，我将100欧姆/平方转换为电导率sigma = 1 /（薄片*厚度），并获得sigma = 10kS / m的电阻电导率。
我将其应用于基板中的3D通道......并发现了意外的模拟结果。
电阻为70欧姆，而不是50欧姆。
从电流密度来看，与顶部相比，我看到电阻器底部的电流很小。
但对于低频电阻，我们真的希望双方的电流密度相同，所以有些不对劲。
*要做*：我认为安捷伦EM人员应该看一下 - 夹层导体在这里的表现可能与我们预期的不同。
我怀疑问题来自铜线和薄膜电阻器之间的接口/过渡（当前电流可以在解算器网格的限制范围内）。
在原始布局中，您已经在整条线下方建模了薄膜电阻，沿着铜线重叠。
为了测试这一点，我创建了一种替代布局，其中薄膜电阻限于铜馈线之间的区域，铜和电阻之间没有重叠。
这现在工作正常，我通过= 10kS / m 3D体积获得预期结果。
*使用其他布局解决了Momentum *（没有堆叠导体）。
所以最后一个问题是，在我们修复了材料定义之后，* FEM *现在是否给出了正确的结果。
原始材料定义失败并不奇怪 - 您要求使用“网格内部”解决3D过孔/导体，但材料仅定义为2D薄层电阻。
所以我们需要改变它。
我确实得到了预期的50欧姆串联电阻用于FEM的3D通孔和网状内部用于我的替代布局，其中电阻器仅跨越两条铜线之间的间隙。
对于我的替代布局，我还获得了预期的50欧姆系列电阻，用于薄板模型。
*解决了*替代布局，没有堆叠导体那么我们得到的原始布局有FEM然后，电阻和铜重叠？
首先，我尝试了薄板模型，它给出了预期的结果。
接下来，我尝试了3D通孔模型，导体设置为“网状内部”，这也给出了低频时的预期50欧姆。
*解决了*有限元的原始布局，带有堆叠导体。
我的修改后的工作区已附加（删除了“simulations”文件夹以限制存档大小）。
使用File> Unarchive导入ADS。
请注意两种基板定义（3D通孔，导电片型号为100欧姆/平方）。
最好的问候Volker编辑：volker_muehlhaus于2013年10月31日上午8:21
test_thin_film_resistors_wrk.7zads148.9 KB
以上来自于谷歌翻译
以下为原文
Hi Angelo,
after working on your question for the last two hours or so, I didn't realize that you have already updated the thread. But anyway, there are multiple interesting topics buried in your question and workspace ....
~~~~~~
I looked at your workspace and noticed that there are multiple topics worth mentioning.
In your schematic, you have defined 100 Ohm/square for the resistor. In the Momentum model, you have defined a sheet resistance value of 100 Ohm/square, but applied that to a 3D via object. To me, this doesn't make sense and I am surprised that Momentum accepts that input. Loss defined by sheet resistance (ohm/sqare) is defined for thin sheets, and you will find that Momentum gives the expected series resistance if you change the resistor in the stackup from 3D via to thin metal sheet. See attached screenshot sheetres.png.
*solved*
~~~~~~~~~~~~~~~~~ this is where the answer could end, but ..... ~~~~~~~~~~~~~~~~~~~~~~~~
So the trivial answer is that you could just change your sheet resistor to sheet metal model. Given that the resistor is very thin compared to the horizontal dimensions, and given that skin effect is not relevant here because the skin depth is much larger than the resistor's thickness, thin sheet modelling with 100 ohm/square is perfectly fine and accurate. But anyway, you modelled it as a 3D object, so maybe you had good reasons that I don't know.
So I started looking at the resistor modelled as a 3D object, because that should work, if we supply the proper material definition. With the 1µm thickness from your substrate, I converted the 100ohm/square into conductivity
sigma = 1 / (sheetres * thickness)
and obtained sigma = 10kS/m for the resistor conductivity. I applied that to the 3D via in your substrate ... and found unexpected simulation results. The resistance is 70 Ohm, instead of 50 Ohm. Looking at the current density, I see little current on the bottomside of the resistor, compared to the top side. But for the resistor at low frequency, we would really expect same current density on both sides, so something is wrong.
*TO DO*: I think the Agilent EM guys should have a look at that - sandwiched conductors might behave different here than we expect.
I am suspecting that the trouble is from the interface/transition between the copper line and the sheet resistor (where currebt can go, within the limitations of the solver mesh). In your original layout, you had modelled the sheet resistor underneath the entire line, with overlap along the copper lines. To test this, I create an alternative layout, where the sheet resistor is limited to the area between the copper feedlines, with no overlap between copper and resistor. This works fine now, and I get the expected results with = 10kS/m 3D volume via.
*Solved* for Momentum with alternative layout (no stacked conductors).
So the final question was if *FEM* gives correct results now, after we have fixed the material definition. It is no surprise that the original material definition failed - you requested to solve 3D vias/conductors with "meshed interior", but the material was only defined as 2D sheet resistance. So we need to change that.
I do get the expected 50 ohm series resistance for FEM with 3D via and meshed interior for my alternative layout, where the resistor only spans the gap between both copper lines. I also get the expected 50 ohm series resistance for FEM with thin sheet model for my alternative layout.
*Solved* for the alternative layout, with no stacked conductors
So what do we get for the original layout with FEM then, where resistor and copper overlap? First, I tried the thin sheet model, which gives the expected result. Next, I tried the 3D via model with conductor set to "meshed interior", which also gives the expected 50 ohm at low frequency.
*Solved* for the original layout in FEM, with stacked conductors.
My modified workspace is attached (with "simulations" folder deleted to limit the archive size).
Import into ADS with File > Unarchive.
Note the two substrate definitions (3D via with conductivity sheet model with 100 ohm/square) .
Best regards
Volker
Edited by: volker_muehlhaus on Oct 31, 2013 8:21 AM
附件
举报
