PNA-X-FW A10.49.03使用端口扩展进行错误的K系数计算错误

使用端口扩展时打开固定装置，问题就会消失。
为什么......长长的st ..
端口扩展是“电延迟”和真正固定之间的混合功能。
与许多事情一样，父亲的罪过传给了儿子，在这种情况下，8510A中端口扩展的实现是以一种有罪的方式完成的，直到今天仍然引起了我们的问题：端口扩展应用于
8510中的PER TRACE基础（并且每个参数有一个跟踪），因此当您在端口2上放置端口扩展但测量端口1时，它不会修改S22的基础更正值（如果未显示S22）。
端口扩展功能只是改变S21（如果是端口2）或S11（如果端口1）的相位，并且不强制进行S22扫描，也不更新S22值。
因此，您无需进行2次扫描，并且可以在不进行校准的情况下应用端口扩展。
快进到PNA大约2002年：我们实现它是相同的兼容。
快进到PNA大约2007年（好吧，我真的不记得日期，你得到了图片）：添加了Fixturing但是对于端口匹配，你想在端口匹配之前添加端口扩展。
通过修改校准阵列来应用夹具，但是将端口扩展应用于迹线，因此为了支持正确的所需操作顺序，在Fixturing打开的情况下，在校准过程中应用端口扩展。
但是，要正确应用它，必须前进和后退扫描，因此如果不应用校准，我们必须应用虚拟或“单位”校准。
这使得可以在cal中应用夹具，并且更新所有4个s参数。
如果cal关闭，你仍然可以进行cal扫描。
但是现在您获得了应用端口扩展的所有参数，并且当公式编辑器将S参数从数据存储中拉出时，数据已经针对端口扩展进行了更正。



     以下为原文

  Turn on fixturing when using port extensions and the problem will go away.  Why...long storty.

Port extensions are kind of a hybrid function between "electrical delay" and true fixturing.  

As with many things, the sins of the father are passed down to the sons, and in this case, the implementation of port extensions in the 8510A was done in a sinful way that causes us problems to this day:

The port extensions are applied on a PER TRACE basis in the 8510 (and it had one trace per parameter) so when you put on port extensions on port 2, but measured port 1, it did not modify the underlying corrected value of S22 (if S22 was not displayed).  The port extension function just change the phase of S21 (if port 2) or S11 (if port 1) and did not force an S22 sweep nor update the S22 value.  Thus you don't need to take 2 sweeps, and port extension can be applied with out calibration.

Fast forward to PNA circa 2002: we implemented it the same to be compatible.
Fast forward to PNA circa 2007 (OK, I don't really remember the dates, you get the picture though): Fixturing was added but for port matching you want to add the port extension before you do port matching. Fixturing is applied by modifying the calibration arrays, but port extension is applied to the trace, so to support the proper desired order of operation, with Fixturing on, port extension is applied in the calibration process.  However, to apply it properly, you must take forward and reverse sweeps and so we must apply a dummy or "unity" cal, if no calibration is applied. This makes it possible to have fixturing applied in the cal, and all 4 s-parameters are updated.  If cal is off, you still get the cal sweeps. But now you get all the parameters getting port extension applied, and when equation editor yanks the S-parameters out of the data store, the data is already corrected for port extension.

Joel博士感谢你的回复。
一如既往的信息。
一个建议。
为什么不将Port Extension移动到Fixturing菜单中？
这将解决一些混乱。
问候。



     以下为原文

  Dr Joel

Thank You for the reply. As always welf of information.
One suggestion. 
Why not move Port Extension into Fixturing menu? That will solve some of the confusion.

Regards.