动量SMD端口校准

安捷伦以限制实际使用的方式实施了* SMD *端口校准：它仅适用于差分端口（两个引脚分组为一个差分端口），这些差分端口设计无法处理任何分流寄生效应。
因此，如果您使用Momentum SMD端口并连接S2P数据，则可能会得到错误的结果，因为S2P数据中的分流路径未在Momentum模型中正确接地。
应用于您的情况，Momentum端口配置将等效于S2P数据块上未连接（浮动）接地参考的原理图仿真。



     以下为原文

  Agilent have implemented the *SMD* port calibration in a way that limits the practical use: it only works for differential ports (two pins grouped into one differential port), and these differential ports by design can't handle any shunt parasitics. So if you use Momentum SMD ports and connect S2P data, you will likely get wrong results because the shunt path in the S2P data isn't properly grounded in the Momentum model.

Applied to your case, the Momentum port configuration would be equivalent to the schematic simulation with unconnected (floating) ground reference at the S2P data blocks.

嗨Volker，谢谢你的快速回复。
我仍然不清楚你的答案。
在图中，我没有接通连接到EM模型的S2P数据块。
这是你的意思吗？
如果是这样，不幸的是它不起作用。
这是相同的。
我也尝试在理想的原理图模拟中删除。
而且，我的另外一个问题是：作为您的解释，它仅适用于差分端口的引脚组，因此如果我不使用SMD校准端口，如何模拟集中元素在动量中的影响？
谢谢先进的RinC



     以下为原文

  Hi Volker,
thank you for your quick reply.
I still get unclear about your answer. 
In the picture, i ungrounded the S2P data blocks which are connected to EM model. Was that your point? If so, unfortunately it doesnt work. It is the same. I also try to removed in the ideal schematic simulation.
And, my additional question is: as your explanation, it only works for groups of pin at differential ports, so how can i simulate the effect of Lumped elements in Momentum if i don't use SMD calibration port?
Thanks in advanced,
RinC

附件

• Ungrounded.PNG14.2 KB
  

> {quote：title = RinC写道：} {quote}>在图片中，我没有接触连接到EM模型的S2P数据块。
这是你的意思吗？
如果是这样，不幸的是它不起作用。
这是相同的。
我也尝试在理想的原理图模拟中删除。
我的观点是Momentum SMD端口无法连接任意S2P块。
它们仅为没有分流路径的纯系列元素提供正确的结果。
不确定安捷伦在介绍这种SMD端口实现时的想法......当然，一般来说，S2P模块的接地路径必须正确连接（当然不是浮动的）。
我只建议删除它作为测试，看看这是否导致你看到的差异。
>而且，我的另外一个问题是：作为您的解释，它仅适用于差分端口的引脚组，因此如果我不使用SMD校准端口，如何模拟集中元素在动量中的影响？
我建议远离SMD和Delta Gap端口。
通过强制用户将引脚分组到一个差分端口=>全局接地访问丢失=>这有效地断开（浮动）您连接的S2P数据中的任何分流路径，两者都有缺陷。
您可以禁用端口校准，或使用零长度TML校准，而不是将这些差分端口与SMD校准一起使用。
然后你的端口应该引用全局接地（不要将引脚分组到差分端口！）现在从你的回答中，我知道差异有其他原因，在你的情况下。
显然，这可能是布局寄生效应造成的差异（焊盘电容接地）。
要进一步解决这个问题，如果你可以上传你的项目（使用文件>存档项目压缩它）将有所帮助编辑：volker_muehlhaus于2014年5月20日上午9:12



     以下为原文

  > {quote:title=RinC wrote:}{quote}
> In the picture, i ungrounded the S2P data blocks which are connected to EM model. Was that your point? If so, unfortunately it doesnt work. It is the same. I also try to removed in the ideal schematic simulation.

My point was that Momentum SMD ports are not valid to connect arbitrary S2P blocks. They only give correct results for pure series elements that have no shunt path. Not sure what Agilent was thinking when they introduced this SMD port implementation ...

In general, of course, the ground path of S2P blocks must be properly connected (not floating, of course). I only suggested to remove it as a test, to see if this is causing the difference that you see.

> And, my additional question is: as your explanation, it only works for groups of pin at differential ports, so how can i simulate the effect of Lumped elements in Momentum if i don't use SMD calibration port?

I would suggest to stay away from SMD and Delta Gap ports. Both are flawed by forcing the user to group pins to one differential port => loss of global ground access => this effectively disconnects (floats) any shunt path in the S2P data that you connect. Instead of using these differential ports with SMD cal, you could either disable port calibration, or use the zero length TML calibration. Your ports should then reference the global ground (do not group pins to differential ports!)

Now from your answer, I understand that the difference has other reasons, in your case. Obviously, it could be the layout parasitics that are causing the difference here (pad capacitance to ground). To troubleshoot this further, it would help if you could upload your project (zip it by using File > Archive Project)

Edited by: volker_muehlhaus on May 20, 2014 9:12 AM

嗨沃尔克，我想我明白你的意思了。
但是，在我看来，当我们使用TML校准端口（零长度）进行集总元件时，它只是因为S2P文件中的地面丢失而不知何故。
只要我知道，S2P文件（至少来自公司），他们总是单独考虑寄生效应。
例如，您可能会分别为分流器和串联电容器获得2个不同的S2P文件。
对于S2P数据项目，作为我的实验，也与我的同事讨论过，Ref端口对于“正常”组件（如电感器，电容器或电阻器）来说是“无用的”。
此端口可能会影响某些复杂的电路，您需要指定Ref点。
在附件中，我将工作区发送给您。
实例“Testprog”是我目前的工作。
此外，ELJQF1N0是用于电感器的S2P文件，其余用于电阻器我的新想法是：从ADS的定义来看，零长度的TML端口与集总元件一起使用，因此我们应该在集总元件的2个端口使用2个引脚
，并定义它们的极性。
例如，端口1：TML（零长度），Pin1：（+），Pin2 :( - ）。
只要我定义了正常的TML零长度端口（1个引脚，极性为+，另一个为接地）。
它给出了与理想原理图相同的结果。
最好的问候，RinC
ELJQF1N0.s2p60.4 KBERJ1GEJ560.s2p60.4 KB



     以下为原文

  Hi Volker,
i think i got your point. However, in my opinion, it somehow makes no senses when we use TML calibration port (zero length) for lumped elements just because the lost of ground in S2P files. As long as i know, the S2P file (at least from company), they always consider the parasitic effects individually. For example, you will likely get 2 different S2P files for shunt, and series capacitor respectively. For S2P Data items, as my experiment, also having discussions with my colleagues, Ref port is somehow "useless" for "normal" component like inductors, capacitors or resistors. This port might be impact in some complicated circuit, which you need to specify the Ref point.
In the attachment, i send my workspace to you. The instance "Testprog" is my current work. 
Also, ELJQF1N0 is S2P files for inductors, and the rest is for resistor
My new idea is: as from definition from ADS, TML port with zero length is used with lumped elements, so should we use 2 pins at 2 ports of a lumped component, and define the polarities of them. For example, Port 1: TML (zero length), with Pin1: (+), Pin2: (-).
As long as i define normal TML zero length port (1 pin with polarity +, another is ground). it gives the same results as ideal schematic.

Best regards,
RinC

附件

• scalar_detect.7zap1.2 MB
  
• ELJQF1N0.s2p60.4 KB
  
• ERJ1GEJ560.s2p60.4 KB
  

嗨，>但是，在我看来，当我们使用TML校准端口（零长度）进行集总元件时，它只是因为S2P文件中的地面丢失而不知何故。
TML（零长度）和常规TML都可以，因为它们可以在没有*将引脚分组到差分端口的情况下使用*。
这就是SMD cal的根本问题：需要将引脚分组到差分端口，从而有效地将其从2端口减少到1端口。
但如上所述，TML（零长度）和常规TML不需要这样，因此这些可以安全地与S2P数据一起使用。
>只要我知道，S2P文件（至少来自公司），他们总是单独考虑寄生效应。
例如，您可能会分别为分流器和串联电容器获得2个不同的S2P文件。
对于S2P数据项目，作为我的实验，也与我的同事讨论过，Ref端口对于“正常”组件（如电感器，电容器或电阻器）来说是“无用的”。
此端口可能会影响某些复杂的电路，您需要指定Ref点。
这真的取决于你拥有的数据。
通常，我们在测量数据中都有串联和分流元件。
对于不包括寄生效应的理想RLC，我们有纯系列路径。
只有这些才能与差分端口准确连接（1端口连接）。
>我的新想法是：从ADS的定义来看，零长度的TML端口与集总元件一起使用，因此我们应该在集总元件的2个端口使用2个引脚，并定义它们的极性。
例如，端口1：TML（零长度），Pin1：（+），Pin2 :( - ）。
如上所述......对于您设备的2个打击垫，我们希望有2个端口。
我们*不想*创建一个端口，其中P1位于左侧焊盘上，P2位于右侧焊盘上，因为这只是一个端口连接。
相反，我们希望拥有完整的2个端口。
如果您的基板定义在基板下方具有无限接地，则无需为端口定义显式（ - ）端子。
动量将自动将（ - ）置于最近的无限地面。
只有当您的基板中没有无限的接地时，您才需要自己为接地定义额外的（ - ）端子。
这是我之前写过的一个应用笔记，用于显示S2P数据的未接地差分端口的问题：http：//muehlhaus.com/support/ads-application-notes/momentum-port-global-ground-or-differential>
在附件中，我将工作区发送给您。
实例“Testprog”是我目前的工作。
谢谢，我会在今天晚些时候调查一下！



     以下为原文

  Hi,

> However, in my opinion, it somehow makes no senses when we use TML calibration port (zero length) for lumped elements just because the lost of ground in S2P files. 

TML (zero length) and regular TML are ok, because they can be used *without* grouping pins into differential ports. And that is what the underlying problem with SMD cal is: the need to group pins into a differential port, so effectively reducing it from 2-port to 1-port. But as said, that is not required for TML (zero length) and regular TML, so these are safe to use with S2P data.

> As long as i know, the S2P file (at least from company), they always consider the parasitic effects individually. For example, you will likely get 2 different S2P files for shunt, and series capacitor respectively. For S2P Data items, as my experiment, also having discussions with my colleagues, Ref port is somehow "useless" for "normal" component like inductors, capacitors or resistors. This port might be impact in some complicated circuit, which you need to specify the Ref point.

It really depends what data you have. In general, we have both series and shunt elements in measued data. For ideal RLC that do not include parasitics to ground, we have pure series path. Only these can accurately be connected with differential ports (1-port connection).

> My new idea is: as from definition from ADS, TML port with zero length is used with lumped elements, so should we use 2 pins at 2 ports of a lumped component, and define the polarities of them. For example, Port 1: TML (zero length), with Pin1: (+), Pin2: (-).

As said ... for the 2 pads of your devices, we want to have 2 ports. We do *not* want to create one port where P1 is on the left pad and P2 is on the right pad, because that would be only a one port connection. Instead, we want to have the full 2 ports. If your substrate defintion has an infinite ground below the substrate, there is no need to define explicit (-) terminals for the ports. Momentum will automatically place the (-) at the nearest infinite ground. Only if you have no infinite ground in your substrate, then you need to define additional (-) terminals for the grounds yourself. 

Here's an appnote that I had written a while ago to show the issue with ungrounded differential ports for S2P data:
http://muehlhaus.com/support/ads-application-notes/momentum-port-global-ground-or-differential


> In the attachment, i send my workspace to you. The instance "Testprog" is my current work. 

Thanks, I will look into that later today!

> {quote：title = RinC写道：} {quote}>另外，ELJQF1N0是电感器的S2P文件，剩下的是电阻器Ok，我现在看了你的S2P数据。
这确实是纯系列元素，没有对ref节点的分流元素。
在这种情况下，使用Momentum中的差分端口不会导致问题。
我现在来看看你的工作区。



     以下为原文

  > {quote:title=RinC wrote:}{quote}
> Also, ELJQF1N0 is S2P files for inductors, and the rest is for resistor

Ok, I looked at your S2P data now. This is indeed pure series elements, with no shunt elelements to the ref node. In this case, using the differential ports in Momentum doesn't cause problems.

I will now look at your workspace.

我查看了您的工作区，只是将测试平台移动到与emModel不同的单元格中，以便更好地概述模型和结果。
模拟结果看起来不错，EM设置看起来不错，而且带有/不带EM布局效果的S11非常接近。
两者都很匹配。
一切看起来都很好。
scalar_detect_volker_wrk.7zads741.2 KB



     以下为原文

  I looked at your workspace and just moved the testbench into a different cell than the emModel, for a better overview of the models and results.

Simulation results look ok, EM setup look ok, and S11 with/without EM layout effects are very close. Both are well matched. Everything looks fine to me.

附件

• match.PNG49.1 KB
  
• scalar_detect_volker_wrk.7zads741.2 KB
  

嗨Volker，对我迟到的回复很抱歉。
我在等你的意见，我没有收到你已回答的通知。
我已经阅读了有关端口校准的文章。
现在让我更清楚了。
来自柏林的干杯:)最好的问候，RinC



     以下为原文

  Hi Volker,
Sorry about my late reply to you. I was waiting for your opinion, and i didn't receive notification that u already answered. 
I already read your article about the port calibration. It makes more clear  to me now. 
Cheers from Berlin :)
Best regards,
RinC