Agilent 8720 C网络分析仪内部问题

> {quote：title = jdiplexers写道：} {quote}>我对8720C网络分析仪有一个有趣的问题，想知道是否有人以前见过它，知道问题是什么，希望有办法解决它或装备
你会建议我帮我修复它。
你可以问安捷伦。
我知道他们提议在一年前对我的8720D做出免费估计。
>我正在为客户测试一些过滤器并注意到在通带区域发生了一些奇怪的波纹，所以我开始研究我的校准和分析仪，最后注意到了这个问题。
正如您在下面的第一张图片中所看到的，S21端口的本底噪声在接近8 Ghz时出现异常尖峰，并进行校准校正。
校准后噪声始终存在，并在校正关闭时消失（参见第二个附件）。
第二个附件似乎是空的 - 至少我看不到它。
但是没有S21端口这样的东西。
>我似乎在某个时刻在分析仪内部产生了一些影响，导致我的校准在8Ghz附近出现错误的尖峰。
我验证校准设备是好的，端口2没有我在端口1中看到的任何相同问题。在测量传输传输测量时，如何谈论端口1和端口2。
必须使用端口1和2。
您是如何验证校准设备是否良好的？
>我打开了我们在实验室一侧的另一台分析仪，看看它有什么问题，它有完全相同的问题。
另一个8720C ??
如果两个VNA具有相同的问题，这似乎很奇怪。
您是否尝试在VNA上更换电缆，以便端口的性别发生变化，从而移动校准标准？
我想知道你的一个校准标准是否有错。
'你用什么工具包？
如果它是6 GHz 85032B，你有扩展器用于公头中心引脚吗？
在任何情况下，这都超出了85032B的频率范围，但只要你有扩展器就不会太糟糕。
您可以尝试使用不同的校准套件进行类似的测试吗？
你能试试另一对电缆吗？
戴夫



     以下为原文

  > {quote:title=jdiplexers wrote:}{quote}
> I have an interesting issue with an 8720C Network Analyzer and was wondering if anyone had seen it before, know what the issue is, and hopefully a way to fix it or an outfit that you would recommend to fix it for me.

You could ask Agilent. I know they offered to give a free estimate on my 8720D about a year ago.

> I was testing some filters for a customer and noticed some strange rippling occurring at the pas***and region so I began to investigate my calibration and the analyzer and finally noticed the issue. As you can see in the first picture below, the noise floor for the S21 port has an abnormal spike at close to 8 Ghz with a calibration correction on. The noise is always there after the calibration and vanishes when the correction is turned off (see second attachment). 

The second attachment seems to be empty - at least I can't see it. But there is no such thing as an S21 port. 

> I seem to be getting an some em influence inside the analyzer at some point, causing my calibration to give an erroneous spike around 8Ghz.  I verified the calibration equipment was good and port 2 does not have any of the same issues I saw in port 1.

How can you talk of port 1 and port 2, when measuring a transmission transmission measurement. Both ports 1 and 2 must be used. 

How did you verify the calibration equipment was good? 

> I turned on another analyzer that we had in the side of the lab to see what issue it had and it has the exact same problem. 

Another 8720C ?? It seems odd if two VNAs have the same problem.

Have you tried swapping the cables over on the VNA so the sex of the ports changes and so the cal standard is moved? I wonder if one of your cal standards has a fault. 

'what kit are you using? If by chance it is the 6 GHz 85032B, do you have the extender for the male centre pin? In any case, that is outside the frequency range of an 85032B, but it wont be too bad there as long as you have the extender. 

Could you try a similar test with a different cal kit? Could you try another pair of cables? 

Dave

发布通过连接的不正确测量（S21和S11）和隔离测量（S21在两个端口上都有负载）。
然后我们可以看到什么是什么。



     以下为原文

  post an UNCORRECT measurement of a thru connection (S21 and S11) and an isolation measurement (S21 with loads on both ports).   Then we can see what's what.

> {quote：title = jdiplexers写道：} {quote} >> {quote：title = drkirbywrote：} {quote} >>你是如何验证校准设备的好？
>>我们的高级射频工程师告诉我校准设备很好。
你知道他如何验证校准设备是否良好？
对于你有两个具有相同问题的VNA，它告诉我，校准套件可能有问题。
我不是专家，但未修正的结果对我来说没问题。
应用纠错时会出现问题。
顺便说一下，如果您使用默认的3.5毫米校准套件，它会从您的第一张照片中看出来。
在我的8720D上，这是85052B而不是85033E。
但幸运的是85052B和85033E具有相同的常数，因此没有区别。
但你的8720C可能比我的8720D还要老。
您的8720C有没有机会假设一个较旧的校准套件，例如85052A，可能（我没有检查过）与您使用的85033E有不同的系数？
我会检查偏移延迟C0，C1，C2和C3的假设，看看它们是否与85033E套件相同。
我肯定乔尔会回来找到比我更有用的答案，但他们是我的一些想法。
戴夫*更新。
我刚刚检查了85052A手册* * http://cp.literature.agilent.com/litweb/pdf/85052-90011.pdf* *它与85033E的系数不同。
因此，如果你的VNA假设一个较旧的校准套件，可能会出现问题，尽管我希望它能表现为涟漪而不是你的涟漪。
但值得一试。*编辑：drkirkby于2013年8月1日下午6:05



     以下为原文

  > {quote:title=jdiplexers wrote:}{quote}
> > {quote:title=drkirbywrote:}{quote}
> > How did you verify the calibration equipment was good? 
> 
> Our senior RF engineer informed me the calibration equipment was good.

Do you know how he verified the calibration equipment was good? 

For you to have two VNAs with the same problem, it suggests to me the cal kit might have a problem. I'm no expert, but the uncorrected results look ok to me. It is when the error correction is applied the problem appears. 

BTW, it looked from your very first picture if you were using the default 3.5 mm cal kit. On my 8720D, that is the 85052B and not the 85033E. But luckily the 85052B and 85033E have the same constants, so it makes no difference. But your 8720C is probably older than my 8720D. Is there any chance your 8720C is assuming an older cal kit such as the 85052A, which might (I have not checked) have different coefficients to the 85033E you are using? I would check what it is assuming for the offset delays, C0, C1, C2 and C3 and see if they are the same as the 85033E kit. 

I'm sure Joel wil come back with much more useful answers than me, but they are some thoughts I had. 

Dave

*Update. I just checked the 85052A manual*
*http://cp.literature.agilent.com/litweb/pdf/85052-90011.pdf*
*and it does have different coefficients to the 85033E. So if your VNA assumed an older cal kit, it is possible there might be a problem, though I would have expected it to manifest itself as ripples rather than what you have. But worth checking anyway.*

Edited by: drkirkby on Aug 1, 2013 6:05 PM

前瞻性术语看似干净
在你的第一篇文章中，显示了峰值，测量的是什么？
你是否终止了两个端口或让它们打开？
您可以在相同的设置下发布S21和S11的图，并关闭校正。
另外，我们也检查反向术语。
S12，S22直通，纠正和未纠正。
同样重新发布S21未校正，因为迹线远离屏幕，重新缩放以保持屏幕上的迹线。
如果未修正参数好看，并纠正看起来很糟糕，那么它通常是校准套件是有过错。



     以下为原文

  Forward terms seem clean.

In your first post, that shows the spike, what was being measured?  Did you terminate both ports or leave them open?  Can you post a plot of S21 and S11 for the same setup with correction off.

Also, let's check reverse terms too. S12, S22 of a thru, corrected and uncorrected.  Also repost S21 uncorrected as the trace is mosly off screen, rescale to keep the traces on screen.

If the uncorrected parameters look good, and corrected look bad, then it's usually the cal kit that is at fault.

你做隔离卡吗？
或者你省略隔离？
我建议使用省略隔离重复校准。
并且......如果你使用隔离：隔离卡连接了什么标准？
看起来你有一个糟糕的串扰错误术语（如果我不得不猜测，你打开隔离端口？）。



     以下为原文

  Do you do an isolation cal?  or do you omit isolation?  I recommend repeating the calibration using omit isolation.

And...if you were using isolation: what standards were connected for the isolation cal?  It looks like you have a bad crosstalk error term (if I had to guess, you did isolation with the ports open?).

我继续将分析仪上的子弹从端口1交换到端口2，以便切换每个端口的从属关系并运行完整的2端口校准，作为对实际校准设备的完整性检查。
由于全男性校准器和全女性校准器被反转，我预计每当端口2驱动信号时就会看到错误的响应，这与我第一次观察到的情况相反。
在校准之后，我发现当启用校准时，端口2确实给出了大约8Ghz的衰减尖峰，当校准被禁用时，它看起来像是正常的本底噪声。
看起来全男性校准器是整个时间的罪魁祸首，这比两个8720c在同一时间遇到完全相同的问题更有意义。
感谢您帮助我解决问题，以及您对此过程的见解和反馈。
我还是新手，非常感谢您的及时帮助！



     以下为原文

  I went ahead and swapped the bullet on the analyzer from port 1 to port 2 in order to switch each port's affiliation and ran a full 2 port calibration as a sanity check on the actual calibration equipment. Since the all-male calibrator and all-female calibrator were reversed, I expected to see the erroneous response whenever Port 2 was driving the signal, opposite of what I observed the first time.

After the calibration, I found that Port 2 was indeed giving me the attenuation spike at around 8Ghz when the calibration was enabled and looked like a normal noise floor when the calibration was disabled. It looks like the all-male calibrator was the culprit the entire time, which makes a whole lot more sense than having two 8720cs suffer the exact same issue at the same time.

Thanks for helping me troubleshoot the issue and for your insight and feedback on the process. I am still new to the field and appreciate your timely help!

虽然你似乎发现了你的calkit有问题，但Joel指出有很多串扰。
我以前从未注意到这一点，但在乔尔提到之后，它确实看起来很可疑。
我很快就在8720D上测量了未校正的隔离度。
我没有负载电缆，但即使没有负载，没有校准，最大信号为-98 dB。
你有大约-30分贝。
这显然是非常错误的。
我确实有propor测试电缆，而不是便宜的RG58，但似乎很难看出你如何只能获得30 dB的隔离。
即使负载应该在电缆上，我发现很难相信那里没有什么问题。
戴夫



     以下为原文

  Although you seem to have found a problem with your calkit, as Joel notes there is a lot of crosstalk. I never noticed that before, but after Joel mentioned it, it did look suspicious. 

I quickly measured the uncorrected isolation on my 8720D. I did not have loads on the cables, but even without loads, with no calibration, the maximum signal was at -98 dB. You have about -30 dB. That is clearly very wrong. I do have the propor test  cables, not bits of cheap RG58, but it seems hard to see how you would only get 30 dB isolation. Even though loads should be on the cables, I find it hard to believe there is not something very wrong there. 

Dave

虽然你似乎发现了你的calkit有问题，但Joel指出有很多串扰。
我以前从未注意到这一点，但在乔尔提到之后，它确实看起来很可疑。
我很快就在8720D上测量了未校正的隔离度。
我没有负载电缆，但即使没有负载，没有校准，最大信号为-98 dB。
你有大约-30分贝。
这显然是非常错误的。
我确实有propor测试电缆，而不是便宜的RG58，但似乎很难看出你如何只能获得30 dB的隔离。
即使负载应该在电缆上，我发现很难相信那里没有什么问题。
戴夫



     以下为原文

  Although you seem to have found a problem with your calkit, as Joel notes there is a lot of crosstalk. I never noticed that before, but after Joel mentioned it, it did look suspicious. 

I quickly measured the uncorrected isolation on my 8720D. I did not have loads on the cables, but even without loads, with no calibration, the maximum signal was at -98 dB. You have about -30 dB. That is clearly very wrong. I do have the propor test  cables, not bits of cheap RG58, but it seems hard to see how you would only get 30 dB isolation. Even though loads should be on the cables, I find it hard to believe there is not something very wrong there. 

Dave