电容器质量测量设置是否错误

我认为你需要一个透视帮助你 - 我怀疑任何工程师都可以。
也许如果你1）拍下你的灯具，并解释它是如何工作的。
2）说你有什么VNA - 模型和选项。
3）上传Touchstone文件。
4）写出电容值是什么5）描述你如何校准VNA - 什么校准套件等.6）说明你如何确定正确的端口扩展。
也许工程师可以帮助你，你也不需要透视！
戴夫



     以下为原文

  I think you need a clairvoyant to help you - I doubt any engineer can. 

Maybe if you 

1) Photographed your fixture, and explained how it worked. 

2) Said what VNA you had - model and options. 

3) Uploaded the Touchstone file. 

4) Wrote what the capacitor value was

5) Described how you calibrated the VNA - what cal kit etc. 

6) State how you determined the right port extension. 

then maybe an engineer could help you, and you would not need the clairvoyant! 

Dave

嗨，我希望这次我能说清楚。
夹具是一个简单的PCB评估板，其中只有我想要测量的电容器通过一个大约15mm长的50欧姆轨道连接到SMA连接器。
夹具中还有短而开放的端口，长度与其他轨道相同。
没有50欧姆的端口，这就是我使用端口扩展的原因。
我正在使用的VNA是hp8714，它是一个非常古老的模型，它没有太多选项。
它有两个端口，反射和传输，基本上你可以测量s11和s21。
在连接评估板之前，我在连接到VNA的电缆末端校准它。
我在VNA中选择反射校准，这需要开路，短路和负载标准。
我有一个3.5mm的3GHz校准套件。
对于端口扩展，我只需输入数据表中指定的值。
要检查值是否正确，请连接电路板中提供的开放端口，以查看相位响应是否确实是平坦的。
电容值为2.5pF。
我得出质量的方法是首先从s11参数获得阻抗Zc = 50 *（1 + s11）/（1-s11）然后，Q = | X | / R你会说我的程序是正确的吗？
你看到一些strabge吗？
我上传了touchstone文件，如果有帮助的话。
谢谢你的时间。
乔恩



     以下为原文

  Hi,

I hope this time I make it clear.

The fixture is a simple PCB evaluation board where there is just the capacitor I want to measure connected to a SMA connector through a 50ohm track that is more or less 15mm long. There are also short and open ports in the fixture, same length as the other track. There is not 50ohm port, so that's why I use port extension.

The VNA I am using is hp8714, it is a pretty old model and it doesn't have too many options. It has two ports, reflection and transmission and basically you can just measure s11 and s21.
I calibrate it at the end of the cable that is connected to the VNA, before connecting the evaluation board. I select reflection calibration in the VNA, which requires open, short and load standards. I have a 3.5mm calibration kit for 3GHz. 

For port extension I just put the value specified in the datasheet. To check out if the value is correct I connect the open port provided in the board to see if phase response is actually flat.  

The capacitor value is 2.5pF.

The way I derive the quality is obtaining first the impedance from s11 parameters Zc = 50*(1+s11)/(1-s11)
and then, Q = |X|/R

Would you say my procedure is correct? do you see something strabge maybe?
I upload the touchstone file, if that is helpful.

Thanks for your time.
Jon

附件

• trace10.s1p9.6 KB
  

是的，它可能是长度为1.5米的坏电缆，但更可能是电缆上的连接器。
在最坏的情况下，它是校准套件中50欧姆负载上的破坏连接器。
点击这里查看如何不丢弃校准套件！http：//i.imgur.com/A5xSr06.jpg！
由于纹波主要是电阻性的，它可能与开路或短路校准插头无关。
每75MHz重复一次纹波，这就是我假设1.5米电缆长度的原因。
点击此处查看有关如何执行测量设置以及如何添加自动端口扩展器以及为8714获取更多测量功能的更多信息。编辑：E Kafeman于2015年2月26日上午2:29



     以下为原文

  Yes it can be a bad cable with a length of 1.5 meter but more likely is it the connectors on that cable.
In worst case is it a destroyed connector on 50 Ohm load in calibration kit. Check here how to not trash a calibration kit
!http://i.imgur.com/A5xSr06.jpg!
As ripple mainly is resistive is it probably not related to open or short calibration plugs.
Ripple is repeated each 75MHz, so that is why I assume 1.5 meter cable length. Check here for more information about how to perform measurement setup as well as how to add auto port extender and get many more measurement functions for your 8714.

Edited by: E Kafeman on Feb 26, 2015 2:29 AM

在较低频率下，Xc很大。
如果预期损耗（R）很小，则很难得到R的精确测量，因此Q.



     以下为原文

  At lower frequencies, Xc is large.  If losses (R) are expected to be small, it's very difficult to get an accurate measurement for R, and therefore Q.

我猜你有一个TR（不是全2端口）分析仪，所以你的校准不能正确校正源和负载匹配。
如果不匹配，你会看到波纹。
当纹波接近史密斯图表的外部时，数学“爆炸”，随着损失变得越来越小，Q因子的误差增加（并不是绝对误差必然更糟，只是相对于它的损失而言
更重要的）。
可能的是，如果没有完整的2端口校准，就无法实现测量所需的质量（这就是为什么几乎没有人再销售TR系统的原因）。
尝试在PCB固定装置前面的端口2电缆上放置一个非常高质量的衰减器（回波损耗至少为25 dB），看看它是否能减少纹波，或者如果你刚刚使用的话，只需在固定装置末端使用负载
用于计算上限值的S11。



     以下为原文

  And I guess you have a TR (not full 2 port) analyzer, and so your calibration doesn't correct properly for source and load match.  The ripple you see if from mismatch.  The math "explodes" when the ripple reaches close to the outside of the smith chart, and as the loss gets less, the error increases in Q factor (it's not that the absolute error is necessarily worse, just relative to the loss it is much more significant).  Likely, the quality that you want for your measurement can't be achieved without a full 2port calibraiton (which is why almost no one sells TR systems any more).

Try putting a very high quality attenuator (with at least 25 dB return loss) on your port 2 cable just before the PCB fixture and see if it reduces the ripple, or just use a load on the end of you fixture if you are just using the S11 for computing your cap value.