VSWR数据在S2P文件中不可用

> {quote：title = nosjojo写道：} {quote}>这是我们已经有一段时间的问题了，我们还没弄清楚究竟是什么原因引起的。
>>保存S2P文件时，我们将获得具有spikey且无法使用的VSWR数据。
我现在打开的一个例子的范围是+438：1到-1073：1。
显然这些数字是无效的，但我们无法弄清楚是什么导致它在我们的数据中发生。
我把它附在这篇文章中以供参考。
>>我们最初的假设与标准化和窗口选择有关。
我们注意到，如果我们将S21标准化并选择带有S11或S22的窗口，则会出现此问题。
当选择S21时，它没有。
然而，我们昨天记录了一堆数据，只选择了S21窗口并且仍然发生这种情况。
所以现在我们很难过。
>>任何想法？
什么是DUT？
你做了什么样的校准？
你有什么校准套件？
（看过Touchstone文件，我可以看到你有一个N5242A VNA，但它也值得把它放在你的问题描述中，所以有人不必找到它。）在某些频率（例如500 MHz）
您的S11幅度> 0 dB，而在其他频率，您的S22幅度> 0 dB。
如果这是某种放大器，它可能只是不稳定。
值得测量一个衰减器，说明它的标称值（10 dB，20 dB或它应该是什么），并显示其数据。
戴夫



     以下为原文

  > {quote:title=nosjojo wrote:}{quote}
> This is an issue we've had for some time and we haven't figured out exactly what causes it.
> 
> When saving S2P files, we'll get VSWR data that is spikey and unusable. An example I have open right now has a range of +438:1 down to -1073:1. Clearly the numbers aren't valid, but we can't figure out what causes it to happen in our data. I attached it to this post for reference.
> 
> Our original assumption was related to normalization and window selection. We had noticed that if we normalized S21 and selected the window with S11 or S22, this issue occurred. When selecting S21, it didn't. However, we recorded a bunch of data yesterday that only had the S21 window selected and this still occurred. So now we're stumped.
> 
> Any ideas?

What is the DUT? What sort of calibration did you do? What calibration kit do you have? (Having looked at the Touchstone file I can see you have a N5242A VNA, but it would have been worth putting that in the description of your problem too, so someone did not have to find it.)

At some frequencies (e.g. 500 MHz) you have the magnitude of S11 being > 0 dB, and at other frequencies you have the magnitude of S22 being > 0 dB. If this is some sort of amplifier, it could just be unstable. 

It would be worth measuring an attenuator, stating it nominal value (10 dB, 20 dB or whatever it is supposed to be), and showing the data for that.  

Dave

是的，正常化是个问题。
以下是保存S2P数据时发生的情况：S2P数据被定义为S11，S21，S12，S22。
在分析仪上，您可以使用S11获得多个不同的迹线，一些具有数据/存储器，一些具有幅度偏移，一些具有相位偏移，等等。
但是我们只能保存一个S2P文件，因此我们必须选择一个。
为了简化实现，我们选择做的是：采取主动跟踪，将参数更改为S11，保存数据，更改为S21，保存，更改为S12，保存，更改为S22更改，将其返回到
原始参数。
如果原始参数上有数据/ mem，则所有参数都会选择它。
这是S2P文件中的限制，但是否则用户实现会复杂得多。
标准化应用于S11数据，这会使其变坏。
如果要保存4个数据跟踪并希望每个数据具有单独的偏移量，内存规范化等，则尝试将保存数据用作csv，保存显示的跟踪，并仅显示所需的4个s参数跟踪。
如果你必须有S2P文件，有标准化的痕迹，你可以做一些后门技巧的东西来做到这一点，这很乏味，但是回复后我会告诉你秘密公式。
你可能不会喜欢它。
编辑：p.s.
这是在帮助文件中:-)“重要 - 使用与活动测量相同的格式和设置（跟踪数学，偏移，延迟等）保存所有有效数据。这可能导致为此保存的数据
非活动测量与显示的数据显着不同。例如，保存S2P文件时，如果活动S11测量设置为数据/存储（数据除以存储器），则保存所有4个S参数
使用数据/存储器。数据/存储器操作中使用的存储器跟踪与活动（S11）测量中使用的存储器跟踪相同。“编辑者：Dr_joel于2015年2月6日上午12:06



     以下为原文

  Yup, normalization is the problem.  

Here's what happens when you save S2P data:  S2P data is defined to be exactly S11, S21, S12, S22.   On the analyzer, you can have several different traces wtih S11, some with data/mem, some with magnitude offsets, some with phase offsets, whatever.  But we can only save one S2P file, so we have to choose one. 

To make the implementation simple, what we chose to do is this:  Take the active trace, change the parameter to S11, save the data, change to S21, save, change to S12, save, chagne to S22 change, return it to the original parameter.  

If you have data/mem on the original parameter, all the parameters will pick it it.  It is a limitation in S2P file, but otherwise the user implemenation would be much more complicated.  The normalization is applied to your S11 data and that makes it go bad.

If you want to save 4 traces of data and want each one to have individual offsets, memory normalization, etc, then try using save data as csv, save displayed traces, and only show the 4 s-parameter traces that you want.

If you must have S2P files, with normalized traces, there is a back-door tricky bit of stuff you can do to get to it, tedious to do, but post back and I will tell you the secret formula.  you probably won't like it.

edit: p.s.  This is in the help file :-)
"IMPORTANT - ALL valid data is saved using the same format and settings (trace math, offset, delay, and so forth) as the active measurement. This can cause the data that is saved for the non-active measurements to be dramatically different from the data that is displayed. For example, when saving an S2P file, if the active S11 measurement is set to Data/Mem (data divided by memory), then ALL 4 S-parameters are saved using Data/Mem. The memory trace that is used in the Data/Mem operation is the same as that used in the active (S11) measurement."

Edited by: Dr_joel on Feb 6, 2015 12:06 AM

谢谢！
听起来我们与原始想法非常接近。
值得庆幸的是，它只发生在手动测试中，我们的自动测试清除了内存。
不断触发这种情况的技术人员说他们当时没有任何规范化，他们运行的测试不应该正常化。
虽然标准化测试确实在它之前，所以如果他们错过了允许它发生的事情我也不会感到惊讶。
可能需要做一些实验，看看是否有他们错过的边缘情况。
很高兴知道S2P数据是以它的方式保存的，我没有意识到它是这样做的。
它对我未来的测试设置肯定会派上用场。



     以下为原文

  Thanks! It sounds like we were pretty close with our original thoughts. Thankfully it only occurs on manual testing, our automated tests clear out memory.

The technicians who keep triggering this were saying that they didn't have anything normalized at the time, and the tests they were running aren't supposed to have normalization. A normalization test did come right before it though, so I wouldn't be surprised if they're missing something that allows it to occur. Might have to do some experimentation to see if there is an edge case they miss.

It's good to know that S2P data is saved the way it is though, I didn't realize it did that. It will definitely come in handy with my future test setups.