E8364C PNA的不确定性和跟踪是什么？

我打电话给技术支持，他们回答了我的问题。
不确定性只是错误，与定义的标准相比。
跟踪不是可重复性，例如网络分析仪中的两个接收器通道在一切事物上彼此匹配的程度。
我的问题的原因是人们声称PNA可以测量分贝的百分之一到千分之一，这似乎不太可能。
这是在那个范围内的跟踪错误，愚弄了人们。
技术支持还向我指出了一些关于此问题的有用的Keysight文档。
他们的答复如下。
--------------------------------------------------
--------------------------------------------------
---------------这是该帮助文件在线版本中“测量错误”主题的直接链接：[http://na.support.keysight.com/pna/
help / WebHelp9_42 / S3_Cals / Errors.htm]此页面上有链接，可以讨论有关反射和传输跟踪以及其他残留错误的信息。
另请参阅“用于生成不确定性曲线的方程式”主题：[http://na.support.keysight.com/pna/help/WebHelp9_42/Specs/u_curve_equations.htm]每个模型分析仪的数据表都有测量不确定度
有限数量的设置 - 电缆，校准套件，校准类型和分析仪设置。
有一个基于Excel电子表格的计算器，它将为您提供更多的配置灵活性，数据表：[http://www.keysight.com/find/na_calculator] -----------------
--------------------------------------------------
----------这些回答了我的问题。



     以下为原文

  I called Tech Support, and they answered my question.

Uncertainty is just that, error compared to a defined standard.

Tracking is not repeatability, it's how closely for instance two receiver channels in the network analyzer match one another over everything.  The reason for my question was that people were claiming that the PNA could measure hundredths to thousandths of a decibel, which seemed unlikely.  It was the tracking error, which are in that range, that fooled people.

Tech Support also pointed me to some useful Keysight documents on the matter.  

Their reply follows.
-------------------------------------------------------------------------------------------------------------------

This is a direct link to “Measurement Errors” topic in the online version of that Help file:

[http://na.support.keysight.com/pna/help/WebHelp9_42/S3_Cals/Errors.htm]

There are links on this page that lead to discussions about reflection and transmission tracking, as well as other residual errors.

And also see the “Equations Used to Generate Uncertainty Curves” topic:

[http://na.support.keysight.com/pna/help/WebHelp9_42/Specs/u_curve_equations.htm]

The datasheet for each model analyzer has measurement uncertainty for a limited number of setups – cables, calibration kits, cal type and analyzers settings. There is a Excel spreadsheet based calculator that will give you more configuration flexibility that the datasheet:

[http://www.keysight.com/find/na_calculator]

-----------------------------------------------------------------------------

These answered my question.

不确定性是测量参数（例如S21）时的总不确定性（而不是误差）。
也就是说，当你看到测量值，比如-10 dB，不确定度为0.04 dB时，意思是误差可以达到那个水平，但不是必须的。
不确定性是从几个组件计算出来的，最常见的是跟踪，匹配和隔离原始和残差错误项。
例如，S21不确定性主要与跟踪误差（对于匹配良好的设备）有关，但可能受匹配误差的影响。
隔离误差是信号泄漏的一般术语，不应该是，而主要的是测试端口耦合器中的指向性误差。
对于大多数现代网络分析仪，端口之间的隔离度远高于本底噪声，因此通常会被忽略。



     以下为原文

  Uncerrtainty is the total uncertainty (not the error) in the measurement of a parameter such as S21.  That is, when you look at the measuerment, say -10 dB, with an uncertainty of 0.04 dB, the meaning is that error could be up to that level, but not necessarlly.  The uncertainty is computed from several components, the most common are Tracking, Match, and Isolation raw and residual error terms.  For example, the S21 uncertainty is primarily related to Tracking Error (for a well matched device), but can be affected by the Match error.  Isolation error is the general term of leakage of signals where the shouldn't be, and the primary one is Directivity error in the test port coupler.  For most modern network analyzers, isolation between the ports is much higher than the noise floor, so usually ignored.