使用Z分析仪E4991A进行反射测量

您正在测试的设备是主动还是被动？
如果它是活动的，那么| S11 |
任何无源设备超过1？
有多高| S11 |？
你说它是> 1，但没有校准是完美的。
我从未使用过其中一种阻抗分析仪，所以根据我使用LCR仪表和VNA的经验，这只是一种预感。
我希望你应该在电缆的末端进行校准 - 而不是在测试头上进行校准。
我非常怀疑你应该添加端口扩展来纠正电缆。
在电缆末端校准更有意义。
至少那就是人们如何在VNA上做到这一点。
我假设阻抗分析仪允许您选择校准套件。
APC7套件是您选择的相同型号的套件吗？
我知道HP / Agilent生产了至少4种不同的APC7校准套件。
开放标准的边缘电容因工具包而异，以及短路的延迟。
如果你有错误的校准套件，那么它将无法正确校准，并且肯定会解释| S11 |
> 1.戴夫



     以下为原文

  Is the device you are testing active or passive? If It is active, does |S11| exceed 1 with any passive devices? 

How high is |S11|? You say it is >1, but no calibration is perfect. 

I have never used one of those impedance analyzers, so this is just a hunch based on my experience of using LCR meters and VNAs. I would expect you should be calibrating on the end of the cable - not on the test head. I very much doubt you should be adding port extension to correct for the cable. It makes far more sense to calibrate on the end of the cable. At least that is how one would do it on a VNA. 

I assume the impedance analyzer allows you to select a calibration kit. Is the APC7 kit you are selecting the same model of the kit you have? I know HP/Agilent have produced at least 4 different APC7 cal kits. The fringing capacitance of the open standard will vary among kits, as well as the delay of the short. If you have the wrong cal kit, then it will not calibrate properly, and would certainly explain having |S11| > 1. 

Dave

端口扩展是阻抗分析仪的补偿功能，因此在测试头进行O / S / L校准后完成。
（列表中的选择1是正确的方法。



     以下为原文

  Port extension is a compensation function on the impedance analyzer and so it is done after the O/S/L calibration at the test head. (Selection 1 in your list is the right method.

HekenPT，为什么＃1优于＃2？
我原本认为在电缆末端进行校准远比在测试端口校准好，然后使用端口扩展来补偿电缆的延迟。
只是补偿电缆的延迟似乎给我带来了很多不必要的错误1）它忽略了电缆的损失。
2）为了能够计算电缆末端的阻抗，需要知道电缆的阻抗，这不会精确到50欧姆。
实际上，实际电缆的阻抗是复数，并且与频率有关。
3）您需要确定电缆的电气长度。
该测量中会出现一些错误。
我认为最好在电缆末端校准。所以上述所有错误来源都被删除了。
当然，人们不会选择以你建议的方式校准VNA，尽管可能存在与阻抗分析仪根本不同的东西。
戴夫



     以下为原文

  HekenPT,
Why is #1 preferable to #2? I would have thought calibrating on the end of the cable far better than calibrating on the test port then using a port extension to compensate for the delay of the cable. Just compensating for the delay of the cable seems to suffer a lot of unnecessary errors to me

1) It ignores the loss of the cable.

2) For one to be able to calculate the impedance ob the end of the cable, one needs to know the impedance of the cable, which will not be exactly 50 Ohms. In fact, the impedance of a practical cable is a complex number, and frequency dependant. 
3) You need to determine the electrical length of the cable. There will be some error in that measurement. 

I would have thought it better to calibrate.on the end of the cable, so all the above sources of error are removed. 

Certainly one would not chose to calibrate a VNA the way you suggest, although maybe there's something fundamentally different with an impedance analyzer. 

Dave

戴夫，关于你的问题：DUT是被动的。
S11幅度振荡高于1且高达1.3。
我使用为该仪器提出的原始calkit。
根据VNA的经验，我试图将参考平面移动到电缆的末端。
但是，HelenPT是对的。
仪器手册不包括此过程。
仅指端口延长时通过延迟时间的电缆延长。
我只是觉得这并没有纠正所有错误。



     以下为原文

  Dave,

about your question:
the DUT is passive. 
the S11 magnitude oscillates above 1 and up to 1.3.
I use the original calkit proposed for that instrument.

Based on the VNA experience I tried to move the reference plane to the cable's end.

However, HelenPT is right. The instrument's manual does not include this process.
Only refers to cable extension through delay time when the port is extended.
I just feel that this does not correct all errors.

> {quote：title = chstergiou写道：} {quote}> Dave，>>关于你的问题：> DUT是被动的。
> S11幅度振荡高于1且高达1.3。
好的，在这种情况下，由于校准错误，远远超出了预期的范围。
>我使用为该仪器提出的原始calkit。
安捷伦是否会发布任何数据 - 例如开路和短路的偏移延迟，或开路的边缘电容？
>根据VNA的经验，我试图将参考平面移动到电缆的末端。
>>但是，HelenPT是对的。
仪器手册不包括此过程。
>仅指端口延长时通过延迟时间的电缆延长。
我只能假设VNA与阻抗分析仪的不同架构意味着在电缆末端进行校准，这是人们用VNA做的，不起作用。
我知道我的4284A LCR仪表使用自动平衡桥，有必要指定电缆的长度，应该是0,1,2或4米，尽管需要一个软件选项来获得2和4
我去上班。
我真的不明白。
>我只觉得这不能纠正所有错误。
它肯定无法纠正VNA可以纠正的所有错误，但当然VNA无法使用非常高的反射系数。
如果您有VNA，是否可以不使用它而不是E4991A？
戴夫



     以下为原文

  > {quote:title=chstergiou wrote:}{quote}
> Dave,
> 
> about your question:
> the DUT is passive. 
> the S11 magnitude oscillates above 1 and up to 1.3.

OK, in which case that is far outside the range one would expect due to calibration errors. 

> I use the original calkit proposed for that instrument.

Do Agilent publish any data on that - such as the offset delay of open and short, or fringing capacitance of the open? 

> Based on the VNA experience I tried to move the reference plane to the cable's end.
> 
> However, HelenPT is right. The instrument's manual does not include this process.
> Only refers to cable extension through delay time when the port is extended.

I can only assume the different architecture of a VNA vs an impedance analyzer means calibrating on the end of the cable, which is what one would do with a VNA, does not work. 

I know with my 4284A LCR meter, which uses an auto-balancing bridge, it is necessary to specify the length of the cables, which should be 0, 1, 2 or 4 m, although one needs a software option to get 2 and 4 m to work. I don't really understand that. 

> I just feel that this does not correct all errors.

It sure can't correct all the errors a VNA could correct, but then of course VNAs can't work with very high reflection coefficients. 

If you have a VNA, could you not use that instead of the E4991A? 

Dave

你好Dr. Kirby，E4991A使用“RF-IV”方法来测量阻抗，而不是S参数。
因此，虽然它具有传统的50欧姆同轴接口，如VNA，但校准算法和测量结果不同。
建议在7mm测试头校准，然后使用端口扩展。
端口扩展确实采用无损耗电缆，50欧姆电缆，如果保持短路，则会产生良好的结果。
话虽这么说，您可以将校准平面移动到延长电缆的末端。
如果它的末端是7mm连接器，那么您可以使用相同的内部校准套件。
否则，您必须在仪器中编辑用户校准套件。
但是，在User Cal套件中，每个元素都被建模为2元素阻抗模型。
此外，该算法期望标准是齐平的，即它们都具有相同的偏移延迟。
（这就是使用7mm的原因。）亲切的问候，海伦



     以下为原文

  Hello Dr. Kirby,
The E4991A uses a "RF-IV" method to measure impedance, rather than S-parameters.  So while it has the traditional 50ohm coax interface like a VNA, the calibration algorithms and the measurements are different. 

The recommendation is to calibrate at the 7mm test head and then use port extension.  Port extension does assume a loss-less cable, 50 ohm cable, which if kept short has produced good results. 

That being said, you can move the calibration plane to the end of the extension cable.  If it's a 7mm connector at the end, then you could use the same internal cal kit.  Otherwise, you would have to edit the User cal kit in the instrument. However, in the User Cal kit, each element is modeled as a 2-element impedance model.  Also, the algorithm expects that the standards are flush, i.e. they all have the same offset delay.  (That's why 7mm is used.)

Kind regards,

Helen

谢谢你们这次富有成果的讨论。
Kirkby博士你是对的。
我使用Z分析仪而不是VNA的唯一原因是它现在更容易使用。
HelenPT，我更喜欢在电缆末端执行校准，以减轻任何可能的错误。
所有使用的连接器都是7mm型，我使用的校准套件与7mm测试头（短路，开路，负载，电容器）相同。
从这个讨论我明白，理论上接受将参考平面移动到电缆的末端。
由于我不能责怪S11的校准过程高于1，可能是任何坏的连接器或电缆？



     以下为原文

  Thank you both for this fruitful discussion.

Dr. Kirkby

You're right. The only reason I'm using a Z analyzer, instead of a VNA, is that it's more readily available at the moment.

HelenPT,

I prefer to perform the calibration at the cable's end to mitigate any possible errors. All the employed connectors are 7mm type and the cal kit I use is the same as that for the 7mm test head (short, open, load, capacitor).

From this discussion I understand that, shifting the reference plane to the cable's end is theoretically accepted.
Since I can't blame the calibration process for the S11 being above 1, could that be any bad connectors or cable?

我看到不好的校准（松散的连接器......）会导致非实际结果。
请记住，有时我们会在阻抗分析仪测量中看到负阻抗。
例如，当测量低 - 低分量的损耗时，尤其是在电抗至少比电阻大100倍的频率下，就会发生这种情况。
亲切的问候，海伦



     以下为原文

  I've seen bad calibrations (loose connectors...) contribute to non-real results.  

Keep in mind, that sometimes we see negative impedance in impedance analyzer measurements.  This happens for example, when measuring the loss of a low-low component, especially at frequencies where the reactance is at least 100x larger than the resistance.  

kind regards,

Helen