N9923A系统动态范围怎么验证

我认为这是规范：规格是90 dB系统动态范围，具有高功率和负载。
降低功率会降低信号源，使接收机的噪声保持不变，因此低功耗时动态范围会降低40-45 dB ...所有网络分析仪都以这种方式工作。
我原本以为会有更多的余量，但我认为噪声基底的工厂测试是看RMS本底噪声，这是跟踪中间的一种。
噪声很有趣：对于接收机噪声测量，由于噪声是统计的，理论上如果你采取足够的扫描，你将获得噪声功率的任何值（达到无限的噪声，但机会无限小）。
实际上，噪声基底附近的迹线上的噪声（以dB表示）约为11 dB峰 - 峰值。
因此，RMS本底噪声比您看到的峰值噪声低约5.5 dB。
平滑具有平滑峰和谷的效果，因此数据将显示更接近RMS噪声基底。
在工厂中，在大多数VNA上，通过采用固定频率201点扫描并计算迹线的均方根（平均值）来测量本底噪声。
如果您将平均值提高10倍，如果您不处于串扰限制，则会将有效本底噪声降低10 dB。
在大多数现代VNA中，串扰限制低于本底噪声，但不是很多。
因此，对于100 dB的典型本底噪声规格，您可能会在迹线中看到-95 dB“峰值”。
正如我所说，该单位的动态范围规格为-90。



     以下为原文

  I think that's in spec: The specification is 90 dB system dynamic range, with high power and loads.

Lowering the power lowers the source leaving the receiver noise the same, so you would expect about 40-45 dB lower dynamic range in low power... all network analyzers work this way.

I would have thought there would be a bit more margin, but I think the factory test for noise floor is to look at the RMS noise floor, which is kind-of in the middle of the trace.

Noise is funny: for receiver noise measurements, since noise is statistical, theoretically if you take enough sweeps, you will get any value for the noise power (up to infinite noise, but the chances are infinitely small).  In practice, the noise on a trace (expressed in dB) near the noise floor is around 11 dB peak-to-peak.  So the RMS noise floor will be about 5.5 dB below the peak noise you see.  

Smoothing has the effect of averagine peaks and valleys, so the data will show closer to the RMS noise floor.

In the factory, on most VNAs, the noise floor is measured by taking sevaral fixed frequency 201 point sweeps and computing the rms (mean) value of the trace.

If you kick up the average by 10 x, it will lower the effective noise floor by 10 dB, provided you are not at the cross-talk limit.  In most modern VNAs, the cross talk limit is below the noise floor, though not by much.

Thus, for a 100 dB typical noise floor spec, you might see -95 dB "peaks" in the trace.  And as I said, the dynamic range spec is -90 on that unit.

感谢您的全面分析。
虽然毫无疑问VNA符合其规格并且我得到的测试结果并未表明该设备已经从其出厂测试条件转移，但我仍然想知道是否有一种方法可以使用VNA本身来显示动态范围结果
非常接近工厂测试结果。
例如使用内置VNA的数学？
我有点坚持使用E4402B，安捷伦有一个明确的逐步方法来验证本底噪声。
我们也可以为N9923A提供一个。
对于接收机本底噪声的低电平输出效应，我可以解释它，因为接收器自动增加其增益，这也会使前端噪声增加40 dB？



     以下为原文

  Thanks for the thorough analysis.

While having no doubt the VNA meets its spec and the test result I am getting does not indicate the unit has shifted from its factory tested conditions, I am still wondering is there a method that we can use the VNA itself to show the dynamic range result very close to the factory test result. for example using the math built-in the VNA? 

I am kind of stuck to the mind set that with E4402B, there is a clear step by step method from Agilent to verify the noise floor. It would be good we can also have one for N9923A.

For low level output effect on the receiver noise floor, can I interpret it as the receiver automatically increases its gain which also boost the front end noise by 40 dB?

Tim，系统动态范围指定为90 dB保证，典型值为100 dB。
基于SysDynRg =最大水平功率-RcvrNsFlr，基于+5 dBm信号源，保证达到-85 dBm。
接收器动态范围典型值为104 dB。
基于RcvrDynRng = MaxRcvrpwr-RcvrNsFlr，其最大接收器功率为+9 dBm，典型本底噪声为-95 dBm。
大多数网络分析仪的噪声基底测试都是在线性模式下完成的，而不是作为频谱分析仪的对数。
在设置仪器时，请查看8753ES系列网络分析仪的操作步骤。
这是如何手动测试动态范围的最佳示例之一，因为没有为此模型编写的手动过程。
正如Joel博士所说，由于低端的A / D噪声，测试基于至少10个平均值。
如果您正在寻找@ High Q组件，这可能不是正确的产品。
戴夫



     以下为原文

  Tim,

The system dynamic range is specified at 90 dB guaranteed, 100 dB typical.   Based on SysDynRg= Max leveled power-RcvrNsFlr which would come out to - 85 dBm guaranteed based on a +5 dBm source.

The receiver dynamic range is 104 dB typical.  Based on RcvrDynRng= MaxRcvrpwr- RcvrNsFlr which would be +9 dBm maximum receiver power and - 95 dBm typical noise floor.

Most Network analyzers noise floor tests are done in the linear mode, not logarithmic as a spectrum analyzer.   Look at the procedure for doing the 8753ES series network analyzer in setting up the instrument.  This is one of the best examples of how to test dynamic range manually since there is not a manual procedure written for this model.

As Dr. Joel said, the test is based on at least 10 averages as due to A/D noise at the low end.    This may not be the right product to use if you are looking @ High Q components.

Dave

感谢Dave和Joel博士的大力支持和指导。
根据您提供的信息，我很肯定在即将到来的夜晚，我可以解决一些问题。



     以下为原文

  Thank you both Dave and Dr Joel for the great support and guidance. With the information you have provided, I am positive that  in the coming nights I can work something out.