SOLT原始数据看起来不对

使用E5071C VNA和相关的SOLT校准时遇到问题。
我所看到的是奇怪的，我相信并迫使我派遣一个单位进行维修无济于事。
在端口1和端口上使用相同的标准执行标准的2端口校准。
2.当在端口2上测量开路或短路时，我看到S22曲线上0dB的急剧偏差;
视频中的标准纹波和过去的8GHz至20GHz显示在端口2上测量的负载也将显示频率较高的响应，在频谱的高端超过10dB。
它是如此糟糕，负载看起来比短/开放更糟糕。
我知道显示的数据是原始的，校准似乎正在纠正它。
我比较了测量一个有损电缆之间的租金（没有显示这个现象）VNA和“坏”VNA和相关性非常好。
我的直觉告诉我，2号港口的匹配非常差，这可以解释围绕史密斯圆图的“远远”阻抗。
使用租赁VNA测量每个港口并未在比赛中显示出任何明显的差异。
知道发生了什么事吗？
我被告知VNA的维修是换掉端口2上的NMD连接器。我不认为维修设施已经意识到这个问题，只是发回了一个符合规格的装置。
提前致谢



     以下为原文

  Having trouble with an E5071C VNA and associated SOLT calibration.

What I am seeing is odd I believe and has forced me to send a unit out for repair to no avail.

Performing a standard 2 port calibration, using the same standards on ports 1 & 2.  When measuring on port 2 with either the open or short, I see a drastic deviation from 0dB on the S22 plot; with the standard ripple in view and past 8GHz up to 20GHz showing <-20dB.  Port 1 shows a nearly flat line at 0dB as expected.

A load measured on port 2 will also show poor response higher in frequency, exceeding 10dB on the high end of the spectrum.  It is so bad, that the load looks worse RAW than the short/ open.

I understand the data being shown is raw, and the calibration appears to be correcting for it.  I have compared measuring a lossy cable between a rental (does not show this phenomenon) VNA and the "bad" VNA and the correlation is excellent.

My gut told me that port 2 would have a very poor match which would explain this "far out" impedance wrapping around the Smith Chart.  Measuring each port with a rental VNA did not show any appreciable difference in the match.

Any idea what is going on?  I was told the repair to the VNA was to swap out the NMD connector on port 2.  I don't think the repair facility was made aware of the problem and just sent back a unit that was in spec.

Thanks in advance