对于开放式校准标准，可以有负偏移和电容吗？

我最初写道：“我真的没有看到开路标准在最初开始跟踪时看起来像是电容性物质。但有可能的是，你的开放标准物理长度相对于你的工作频率是如此之长
这可能会超过180度标记。这是可能的。除此之外，我看不出开放标准（设计为“开放标准”）将如何看似归纳。“
但我重新阅读了你的帖子，看到你在校准后正在查看你的数据。
现在，我可以看到为什么你看到你看到的痕迹。
您的NA可能已损坏，或输入的校准输入不正确，或者测试系统中的连接存在物理错误，或者您的开放式标准连接器损坏。
或许你喝了太多啤酒！
大声笑！
顺便说一下，圣诞快乐，新年快乐！
编辑：SOLT_guy于2012年12月30日上午7:17编辑：SOLT_guy于2012年12月30日上午7:21编辑：SOLT_guy于2012年12月30日上午7:45编辑：SOLT_guy于2012年12月30日7：
47 AMEdited：SOLT_guy于2012年12月30日上午7:54



     以下为原文

  I initially wrote this:

"I don't really see how physically an open standard can appear to be anything but capacitive when the trace initially commences. 
  It may be possible, that your open standard is so long in physical length relative to your working frequency that it might cross the 180 degree marker.   This is possible.   Other than that, I can't see how an open standard (which is designed to be an "open standard") will appear to be inductive."

But I re-read your post and saw that you were looking at your data after calibration.      Now, I can see why you saw the trace that you saw.   
    
It may be possible that your NA is broken, or incorrect calibration inputs were input, or there is something physically wrong with the connections in your test system, or your open standard connector is broken.   Or maybe you had one too many beers! LOL! 

By the way, MERRY CHRISTMAS and have a HAPPY NEW YEAR!!!

Edited by: SOLT_guy on Dec 30, 2012 7:17 AM

Edited by: SOLT_guy on Dec 30, 2012 7:21 AM

Edited by: SOLT_guy on Dec 30, 2012 7:45 AM

Edited by: SOLT_guy on Dec 30, 2012 7:47 AM

Edited by: SOLT_guy on Dec 30, 2012 7:54 AM

这个问题很有意思：“2）中心导体位于参考平面下方约1 mm。假设真空中的光速为299792458 m / s，则表明输入的偏移应为-0.001 / 299792458
= -3.335 ps。这有意义吗？“
有人可以给我们一些反馈吗？



     以下为原文

  This question is of interest:

"2) The centre conductor sits below the reference plane by about 1 mm. Given the velocity of light in a vacuum is 299792458 m/s, that would suggest the offset to be entered should be -0.001/299792458 = -3.335 ps. Does that make sense?"

Can someone offer us a little feedback?

在实践中，使用端口扩展，并将端口扩展-3.335 ps仍然显示归纳行为。
-6.5 ps的端口扩展似乎使N插头看起来像一个点，参数随频率没有变化。
鉴于有肯定会被边缘电容，并采取猜测这将是大约120法郎，然后用端口扩展播放显示有关-12.8 PS的偏移需要得到一个开放的N型连接器是某处的电容值
接近现实。
我认为N连接器的另一个问题是，除非公引脚的直径增加，否则最后几毫米不是50欧姆的传输线，而是更高的阻抗，因为未配合的公引脚非常薄。
这可能是我的VNA的一个问题。
如果你有一个N校准套件，我有兴趣看看你发现的是一个开放的公N连接器的属性。
也许你也会看到它似乎是归纳的。
顺便说一下，根据我的VNA，一个开放的N型连接器显示出相当大的电容（几百fF），不像它的男性对应物，它看起来是感应式的。
Dave编辑：drkirkby于2012年12月30日8:16 PM编辑：drkirkby于2012年12月30日下午8:34



     以下为原文

  In practice, using port extensions, and extending the ports by  -3.335 ps still shows inductive behaviour. A port extension of -6.5 ps appears to make an N plug look like a spot, with no change in parameters with frequency. 

Given there is certainly going to be fringing capacitance, and taking a guess it will be around 120 fF, then playing with the port extensions shows an offset of about  -12.8 ps is needed to get a capacitance value of an open N connector which is somewhere near realistic. 

I think the other issue with the N connector is that unless the diameter of the male pin is increased, the last few mm are not a 50 Ohm transmission line, but a somewhat higher impedance, as the unmated male pin is quite thin. 

It could be an issue with my VNA. If you have an N calibration kit, I'd be interested to see what you find are the properties of an open male N connector. Perhaps you too will see that it appears inductive. 

BTW, according to my VNA, an open female N connector shows a fair large capacitance (several hundred fF), unlike it its male counterpart, which appears inductive. 

Dave

Edited by: drkirkby on Dec 30, 2012 8:16 PM

Edited by: drkirkby on Dec 30, 2012 8:34 PM

亲爱的柯比博士;
我无法评论您的具体情况，因为我无法测量您的特定校准套件的参数。
我只能假设你为什么看到你所看到的。
我不知道你是如何进行测量的。
我认为最好等一位安捷伦工程师向我们提供一些反馈意见。
让我们拭目以待安捷伦工程师说些什么。
顺便说一句，新年快乐！
编辑：SOLT_guy于2013年1月1日上午8:22编辑：SOLT_guy于2013年1月1日上午8:29



     以下为原文

  Dear Dr. Kirby;

      I can't comment on your particular situation because I can't measure the parameters of your particular cal kit.    I can only hypothesize why you see what you see.   

     I have no idea how you made your measurements.   

     I think it would be best to wait for an Agilent engineer to offer us some feedback.    

     Let's wait and see what an Agilent engineer has to say.

     By the way, HAPPY NEW YEAR!!!

Edited by: SOLT_guy on Jan 1, 2013 8:22 AM

Edited by: SOLT_guy on Jan 1, 2013 8:29 AM

> {quote：title = SOLT_guy写道：} {quote}>亲爱的柯比博士;
>>我无法评论您的具体情况，因为我无法测量您的特定校准套件的参数。
我只能假设你为什么看到你所看到的。
同意。
我的'N'校准套件是85032B，看起来状况良好。
VNA是HP 8720D，具有有效的校准证书。
>我不知道你是如何进行测量的。
好的，这是显示我的意思的简单方法。
1）在VNA的端口或连接到端口的电缆上获取公N连接器。
2）使用体面校准套件的母N标准，校准公测试端口 - 一个开口，短路和校准套件负载的单端口校准就足够了。
3）从公N连接器上卸下所有东西。
4）在史密斯圆图上查看S11 - 当然假设你的男性N在1号港口！
你看到了什么？
我观察到的是：a）在低频率下，该点位于史密斯圆图的右侧，正如您所期望的那样，因为它是一个开路。
b）随着频率的增加，该点旋转*逆时针*。
因此，在参考平面上，我得出一个公N连接器，没有连接到它，充当电感器。
当然，你必须在开口N处有边缘电容。但是，这可能会增加电感的效果。
这些影响是否会使它归纳我不知道，但他们可能会这样做。
*公N插头的中心导体不会延伸到参考平面。
凹陷了一些。
我猜这是大约1毫米，虽然我没有检查N连接器的规格，看看这是什么限制。
*插入母头的公针的位较薄，因此形成阻抗大于50欧姆的传输线的内导体。
>我认为最好等一位安捷伦工程师向我们提供一些反馈意见。
>>让我们拭目以待安捷伦工程师说些什么。
是的，如果安捷伦工程师做出回应，那将会很有趣。
如果您有N卡套件，并且需要10分钟，请自行试一试。
我希望你能和我一样。
Tommorow我会尝试使用3.5毫米连接器，但我怀疑这可能与N完全不同。>顺便说一句，快乐的新年！
也祝你新年快乐。
戴夫



     以下为原文

  > {quote:title=SOLT_guy wrote:}{quote}
> Dear Dr. Kirby;
> 
>       I can't comment on your particular situation because I can't measure the parameters of your particular cal kit.    I can only hypothesize why you see what you see.   

Agreed. The 'N' cal kit I have is the 85032B, which looks in good condition. The VNA is an HP 8720D, with a valid cal certificate. 

>      I have no idea how you made your measurements.   

OK, this is the simplist method to show what I mean. 

1) Get a male N connector on a port of a VNA or cable connected to the port. 

2) Using the female N standards of a decent calibration kit, calibrate the male test port - a one port calibration with an open, short and load from a calibration kit will be sufficient. 

3) Remove everything from the male N connector. 

4) View S11 on a Smith Chart - assuming of course your male N is on port 1! 

What do you see? 

What I observe is that:

a) At low frequencies, the point is on the right of the Smith Chart, as you would expect, since it is an open-circuit. 

b) As the frequency is increased, the point rotates *anti-clockwise*. Hence at the reference plane, I conclude a male N connector, with nothing connected to it, acts as an inductor. 
 
Of course, you are right there must be fringing capacitance at the open N. However, that might be dominated by effects which add inductance. Whether these effects would make it inductive I don't know, but they might do. 

* The centre conductor of the male N plug does not extend up to the reference plane. It is recessed some amount. I'm guessing that is about 1 mm, though I have not checked the specs of the N connector to see what the limits on this are. 

* The bit of the male pin which inserts into the female is thin, and so forms the inner conductor of a transmission line with an impedance of greater than 50 Ohms. 

>      I think it would be best to wait for an Agilent engineer to offer us some feedback.    
> 
>      Let's wait and see what an Agilent engineer has to say.

Yes, it will be interesting if an Agilent engineer responds.

If you have an N cal kit, and 10 minutes to spare, give it a try yourself. I expect you will see the same as me. 

Tommorow I will try this with 3.5 mm connectors, but I suspect that might be quite different to N. 

>      By the way, HAPPY NEW YEAR!!!

Happy new year to you too. 

Dave

> {quote：title = drkirkby写道：} {quote} >>是的，如果安捷伦工程师做出回应，那将会很有趣。
>>如果>>顺便说一句，快乐新的一年！
>戴夫和新年快乐。
不需要它成为安捷伦人;
我们欢迎任何知识渊博的RF工程师参与其中。到目前为止，您已经做得非常好，所以我们不需要介入您的谈话。
但是，只要你问......偏移延迟是一个任意结构，旨在允许标准在开放（或短）元素之前具有一些短线长度。
对于较小的值，延迟的相位与频率或C0值（在打开的情况下）都是近似线性的。
因此，具有略微负延迟和较大C0的模型将得到与具有0延迟和较小C0的模型几乎相同的结果。
通常，延迟被设置为表示连接器的参考点，其通常是外部导体的配合表面。
对于男性和女性，3.5毫米和小的具有相同的参考长度。
N型有很大差异。
打开的原始工作将开路电容建模为bessel函数（由于圆柱形性质），并且由三阶多项式（4项，包括固定项）近似。
如果你设置标准（或测试连接器）有一些衰退，并参考测试引脚，那么你将合理地有一些负偏移。



     以下为原文

  > {quote:title=drkirkby wrote:}{quote}
> > Yes, it will be interesting if an Agilent engineer responds.
> 
> If > >      By the way, HAPPY NEW YEAR!!!
> Dave
And happy new year as well.

No need for it to be an Agilent guy; any knowledgeable RF engineer is welcome to chime in.  And you have beed doing very well so far so no need for us to intrude on your conversation.  But, as long as you ask...

The Offset Delay is an arbitrary construct intended to allow for standards having some short line length before the open (or short) element.

For small values, the phase vs. freq for a delay, or a C0 value (in the case of an open) are both nearly linear.  Thus, a model with slightly negative delay and a larger C0 will give nearly identical results as one with 0 delay and smaller C0.

In general, the delay is set to represent the reference point of the connector, which is typically the mating surface of the outter conductor.

3.5 mm and small have the same reference length for male and female.  Type N has a large difference.

The orginal work on opens modeled the open circuit capacitance as a bessel function (due to the cylindrical nature) and that was approximated by a 3rd order polynomial (4 terms, including the fixed term).  

If you setup you standards (or test connector) to have some recession, and reference to the test pin, then you will reasonably have some negative offset.

亲爱的Dunsmore博士;
许多年前，我试图准确地推理出网络分析仪中“开始”相角计数的工作原理。
但是，我没有得到工程师的完整答复，经过多年的追问，但没有得到足够的答复，我放弃了。
我非常高兴Kirby博士重新审视了这个主题。
这是我的问题：假设端口1已经通过校准套件校准（我将称之为校准套件1）。
如果开放式标准从不同的校准套件（我将称之为校准套件2）连接到PORT 1，并且我将端口延伸拨回零度，则现在已经测量了标准的物理长度和边缘电容。
现在我将cal kit 2中的开放标准数据手动输入到我自己的个人校准套件中。
我记录了偏移延迟和边缘电容时间延迟值。
根据频率，在网络分析仪内，是开始相位角计数，用于确定通过此方法计算的网络分析仪的零度初始参考点：[第二天为了清晰起见]减去（添加）偏移延迟
从任意初始时间数据输入的时间数据（“偏移延迟”定义为使用端口扩展表示拨号值= =物理长度+以时间为单位表示的边缘电容）然后加上（减去）多项式的适当和
边缘电容的术语。
我添加边缘电容的原因是因为我知道物理长度不存在但延迟是由于开放引脚尖端的电效应。
在此之后计算该时间值，参考零度点（“零度点”参考物理线长度）。
括号（）中的术语可以忽略，我只将它们插入到这个解释中，因为计算方法的确切“形式”可能会反转这些符号以用于特定的设计考虑。
相关的是符号反转 - 减去偏移延迟并添加边缘电容（反之亦然）。
过去，我还没有得到足够的解释，说明边缘电容如何纳入网络分析仪的计算中。
我确实在IEEE会议上对此事进行了调查。
我想借此机会感谢一位现在为Anritsu公司工作的前安静工程师，因为当我在IEEE会议上以数学形式向他提出一般的边缘电容问题时，他给了我一个明确的答复。
但是，我并没有向他提出这个特别的问题，因为这是一个多雨和沉闷的夜晚，我不想让他长久。
我写这篇文章的原因是因为我对Anritsu公司所写的内容感到不满。
我没有得到Anritsu产品的良好技术支持，我在这个论坛上写了关于这个例子的技术支持的质量。
然而，当我写这篇回复时，我回忆说，当我参加一个免费的Anritsu会议时，我问了一些关于校准套件输入的基本问题，我收到了Anritsu工程师的一些非常好的反馈，为此我永远感激不尽。
当我写下我与公司的经历时，我真的很想有良心，我想把所有细节都包括在内。
这是我清除良心的方式。
我想感谢Dunsmore博士，感谢您过去的技术支持。
我永远感谢您过去为我提供的反馈。
这非常有帮助。
这可能是我在这个论坛上的最后一个问题。
我终于绕过了NA校准的所有松散的末端。
我期待着阅读你的回复。
编辑：SOLT_guy于2013年1月2日上午11:10编辑：SOLT_guy于2013年1月2日上午11:15编辑：SOLT_guy于2013年1月3日上午6:40



     以下为原文

  Dear Dr. Dunsmore;

     Many years ago, I tried to theorize exactly how the "start" phase angle count works in a network analyzer.   However, I did not receive a complete answer from an engineer and after years of chasing the question but not getting an adequate reply I gave up.

     I am very happy that Dr. Kirby has revisited this subject.   

    This is my question:  Assume Port 1 has been calibrated by a cal kit (which I will call cal kit 1).

     If an open standard is connected to PORT 1 from a different cal kit (which I will call cal kit 2) and I dial back port extensions to zero degrees,  the physical length of the standard and the fringing capacacitance have now been measured.   

     Now I go to manually input the data of the open standard from cal kit 2 into my own personal cal kit.   I record the offset delay and the fringing capacitance time delay values.

     As per frequency, within the network analyzer, is the start phase angle count to determne the zero degree initial reference point of the network analyzer calculated by this method:

[Edited for clarity the following day]

     Subtract (add) the offset delay in terms of the time data inputs from an arbitrary initial time data value ("offset delay" is defined to mean the dialed value using port extensions = = physical length + fringing capacitance expressed in units of time) and then add (subtract) the appropriate sum of polynomial terms of the fringing capacitance.    The reason why I add the fringing capacitance is because I know that a physical length does not exist but the delay is due to an electrical effect at the tip of the open pin.   

     After this this time value has been calculated, the zero degree point is referenced (the "zero degree point" is referenced to the physical tline length).

     The terms in parenthesis () can be ignored, I have only inserted them into this explanation because the exact "form" of the calculation method may reverse these signs for specific design considerations.   What is relevant is the sign reversal - subtract offset delay and add fringing capacitance (or vice versa).

      In the past, I have not received an adequate explanation of how fringing capacitance is incorporated into the network analyzer calculations.    I did make an inquiry into this matter at an IEEE conference.

      I would  like to take this opportunity to thank a former agilent engineer, who now works for Anritsu corporation, because when I posed a general fringing capacitance question to him in mathematical form at an IEEE conference he gave me a definitive reply.  However, I did not pose this particular question to him because it was a rainy and dreary night and I did not want to keep him long.  

      The reason why I am writing this is because I feel bad about something I wrote with regard to Anritsu corporation.   I didn't get good technical support for an Anritsu product and I wrote in this forum about the quality of technical support with regard to this example.    However, as I was writing this reply, I recalled, when I went to a free Anritsu conference, I asked some fundamental questions about cal kit inputs and I received some really good feedback from some Anritsu engineers and for this I am eternally grateful.     I really like to have clear conscience when I write about my experiences with corporations and I like to include all the details the good along with the bad.     This is my way of clearing my conscience.    

        I would like to thank you, Dr. Dunsmore, for your past technical support.   I am eternally grateful for the feedback that you provided for me in the past.    It was very helpful.

        This will probably be one of my last questions in this forum.   I am finally getting around to wrapping up all the loose ends with regard to NA calibration.     I look forward to reading your reply.

Edited by: SOLT_guy on Jan 2, 2013 11:10 AM

Edited by: SOLT_guy on Jan 2, 2013 11:15 AM

Edited by: SOLT_guy on Jan 3, 2013 6:40 AM

Dave，中心导体在参考平面下方1 mm（.039“）处于N型连接器的公差范围内（例如：.016”至.060“），因为母连接器设计为至少用另一个连接器封闭它
交配时〜.150“。
我认为它看似归纳的原因在附录B的这个应用笔记中有所涉及：[http://cp.literature.agilent.com/litweb/pdf/5989-4840EN.pdf]如果我是对的，那么问题就是
：在开球之前，公开赛的中锋会有多长时间？
所以，没有负延迟......在电容之前只有更多的延迟。
Spacecase



     以下为原文

  Dave,

The center conductor being 1 mm (.039") below the reference plane is within the tolerance for Type N connectors (something like: .016" to .060") since the female connector is designed to enclose it by at least another ~.150" when mated. 

I believe the reason it appears inductive is the covered in this app note in Appendix B:

[http://cp.literature.agilent.com/litweb/pdf/5989-4840EN.pdf]

If I am correct, the question would be: How long is the center condutor in the Open, before fringing? 

So, no negative delay... just more delay before capacitance.

Spacecase

谢谢Joel，SOLT_guy和Spacecase。
我的目的是试着看看为无线电火腿等人制作校准套件是多么实际，这不足以证明在校准套件上花费很多。
如果你看看eBay，那么有很多真正糟糕的校准套件正在销售中。
它们通常由随机的适配器组成！
但当然我需要找到合理的偏移和电容值。
Joel的书中有很多关于制作自己的校准套件的文章，但是针对那些制作它们用于PCB的人。
我认为原则上一个同轴电缆应该更容易，但我相信N可能是任何常见连接器中最难的，而APC-7可能是最简单的。
在Joel的书之后，他广泛使用了TDR功能。
幸运的是我的8720D有TDR选项。
我的85032B校准套件在母头开口处的偏移量恰好为0.000 ps - 即OPEN（M）。
如果我在该开放标准上使用TDR功能查看反射，则反射为0 ps，如果标准偏移为零，这并不是太令人惊讶。
但是，在开放的N型插头上，TDR上的反射为-12.5 ps。
作为反射，是单向时间的两倍，但单向时间相当于3.974毫米的距离。
很明显，距离与小距离无关（当然，一旦我确定了偏移量，将观察到的电容拟合到多项式以得到C0，C1，C2和C3是一个相对简单的问题。我注意到
图中至少我的N校准套件（85032B）中的开放标准的边缘电容图，电容随频率变化不大。我附上了几个图。你可以看到只有几个fF
两种标准都改变了0到6 GHz的范围。这个数据直接基于两个标准的85032B手册中的C0，C1，C2和C3的值。这几乎让我想找到电容变化的偏移量
频率只有几个fF，但我知道这不是解决这个问题的真正方法。虽然我个人没有HFSS许可证，但我知道有人这样做，所以我可以尝试使用它来模拟
男N插头。我确实知道如何使用HFSS，但不能为我的生活中任何方式解决问题
为此我可以使用任何有用的东西，所以我没有试过HFSS。
如果有人提到那篇论文，Joel谈到了边缘电容和贝塞尔函数，我会对副本感兴趣。
我一直在寻找它，从来没有真正找到一个很好的参考。
从我所看到的，我怀疑它是基于与同轴电缆成直角的无限平面，这对APC-7来说是一个很好的近似，但不是N连接器。
但我仍然想找到这篇论文。
戴夫
雌性 - 开放 - 来自85032B-calibration-kit.PNG27.9 KB



     以下为原文

  Thank you Joel,  SOLT_guy and Spacecase. 

My aim is to try to see how practical it is to make a calibration kit for people like radio hams, that can't justify spending a lot on a cal kit. If you look on eBay, there are so many real poor calibration kits being sold. They often consist of a random collection of adapters! 

But of course I need to find a reasonable value for offset and capacitance. There's a fair bit in Joel's book on making your own cal kits, but is aimed at those making them for use on a PCB. I think in principle a coax one should be easier, but I believe N is probably the most difficult of any common connector, and APC-7 is probably the easiest. 

Following Joel's book, he makes extensive use of the TDR function. Luckily my 8720D has the TDR option. My 85032B cal kit has an offset of exactly 0.000 ps on the female open - i.e OPEN(M). If I look at the reflection using the TDR function on that open standard, the reflection is at 0 ps, which is not too much of a surprise given the offset of the standard is zero. 

However, on an open N male plug, the reflection on the TDR is at -12.5 ps. Being  a reflection, that is of twice the one-way time, but that one-way time equates to a distance of 3.974 mm. Clearly that distance has nothing to do with the small (< 1 mm) distance by which the male pin is recessed behind the reference plane. I suspect the reflection is from where the male pin is reduced in diameter, and so makes a transmissoin line of an impedance signifiicantly higher than 50 Ohms. I can't see a second reflection - there is only one, but I suspect that it just an issue with the resolution of my VNA in the time domain mode. It does not help the fact I only have a 6 GHz cal kit, though the VNA works to 20 GHz.  

Of course, once I have settled on an offset, it is a relatively simple matter of fitting the observed capacitance to a polynomial to get C0, C1, C2 and C3. 

I do notice if one plots graphs of the fringing capacitance of the open standards in at least my N calibration kit (85032B), there is not much change of capacitance with frequency. I've attached a couple of plots. You can see there's only a couple of fF change over the range 0 to 6 GHz in both standards. That data is based directly on the values of C0, C1, C2 and C3 in the 85032B manual for the two standards. It almost makes me want to find an offset where the capacitance change with frequency is only a couple of fF, but I know that is not the real way to crack this problem. 

Although I personally don't have an HFSS license, I know someone who does, so I could try to use that to model the male N plug. I do know how to use HFSS, but can't for the life of me work out any way of getting anything useful out of it for this, so I've not tried HFSS. 

If anyone has a reference to that paper Joel spoke of about fringing capacitance and Bessel functions, I'd be interested in a copy. I've hunted for it, and never actually found a good reference to this. From what I have seen, I suspect it is based on an infinite plane at right angles to the coax, which would be a good approximation on APC-7, but not N connectors. But I'd still like to find the paper.  

Dave

附件

• male-open-from-85032B-calibratin-kit.PNG25.0 KB
  
• female-open-from-85032B-calibration-kit.PNG27.9 KB
  

> {quote：title = drkirkby写道：} {quote}>谢谢Joel，SOLT_guy和Spacecase。
> Snip>但是，在打开的N插头上，TDR上的反射为-12.5 ps。
作为反射，是单向时间的两倍，但单向时间相当于3.974毫米的距离。
很明显，距离与小距离无关（>当然，一旦我确定了偏移量，将观察到的电容拟合到多项式以获得C0，C1，C2和C3是一个相对简单的问题。我看到了
现在混淆......你在校准后指的是开放式公N连接器，而不是Open N Cal标准。是的，它在参考平面后面。这是因为中心导体直径减小（所以它不是
50欧姆，也是凹陷的尖端，减少到一点）。Spacecase



     以下为原文

  > {quote:title=drkirkby wrote:}{quote}
> Thank you Joel,  SOLT_guy and Spacecase. 
> 
Snip

> However, on an open N male plug, the reflection on the TDR is at -12.5 ps. Being  a reflection, that is of twice the one-way time, but that one-way time equates to a distance of 3.974 mm. Clearly that distance has nothing to do with the small ( 
> Of course, once I have settled on an offset, it is a relatively simple matter of fitting the observed capacitance to a polynomial to get C0, C1, C2 and C3. 

I see the confusion now... you are refering to the open ended male N Connector after calibration, not the Open N Cal Standard. Yes, it is behind the reference plane. This is because of the reduced diameter of the center conductor (so it is not 50 ohms, also the recessed tip and reduction to a point).

Spacecase

> {quote：title = Spacecase写道：} {quote}>我现在看到了混乱......你在校准后指的是开放式公N连接器，而不是Open N Cal Standard。
是的，它在参考平面后面。
这是因为中心导体的直径减小（因此它不是50欧姆，也是凹陷的尖端并且减少到一点）。
>> Spacecase是的，我说的是开放的公N连接器，而不是校准标准。
我很想假设男性N对我在8720D的TDR功能上所测量的任何东西都有负偏移，并看看事情是如何发生的。
但我真的需要更好地掌握乔尔的书中的相关部分。
有关TDR的整章，以及本书另一部分关于创建自己的校准套件的部分，该部分广泛使用了TDR。
戴夫



     以下为原文

  > {quote:title=Spacecase wrote:}{quote}

> I see the confusion now... you are refering to the open ended male N Connector after calibration, not the Open N Cal Standard. Yes, it is behind the reference plane. This is because of the reduced diameter of the center conductor (so it is not 50 ohms, also the recessed tip and reduction to a point).
> 
> Spacecase

Yes, I'm talking of the open male N connector, not a calibration standard. 

I'm tempted to assume the male N has a negative offset of whatever I measured on the TDR function of the 8720D, and see how things work out from there. But I really need to get a better grip of the relevant sections in Joel's book. There's a whole chapter on the TDR, and a section in another part of the book on creating your own cal kits, which makes extensive use of the TDR. 

Dave

> {quote：title = drkirkby写道：} {quote} >> {quote：title = Spacecase写道：} {quote} >>>我现在看到了混乱......你在校准后指的是开放式公N连接器
，而不是Open N Cal标准。
是的，它在参考平面后面。
这是因为中心导体的直径减小（因此它不是50欧姆，也是凹陷的尖端并且减少到一点）。
>>>> Spacecase >>是的，我说的是开放式公N连接器，而不是校准标准。
>>我很想假设男性N对我在8720D的TDR功能上测得的任何东西都有负偏移，并看看事情是如何发生的。
但我真的需要更好地掌握乔尔的书中的相关部分。
有关TDR的整章，以及本书另一部分关于创建自己的校准套件的部分，该部分广泛使用了TDR。
>> Dave No，它希望在参考平面上连接另一个连接器。
由于之前我说过的原因，它没有校准后与参考/校准平面的负偏移。
如果在电缆末端执行校准并添加N适配器（母对公或母对母）并将另一端打开并进行测量，则它将具有正偏移。
这将允许您将*适配器*建模为“打开”。
Spacecase



     以下为原文

  > {quote:title=drkirkby wrote:}{quote}
> > {quote:title=Spacecase wrote:}{quote}
> 
> > I see the confusion now... you are refering to the open ended male N Connector after calibration, not the Open N Cal Standard. Yes, it is behind the reference plane. This is because of the reduced diameter of the center conductor (so it is not 50 ohms, also the recessed tip and reduction to a point).
> > 
> > Spacecase
> 
> Yes, I'm talking of the open male N connector, not a calibration standard. 
> 
> I'm tempted to assume the male N has a negative offset of whatever I measured on the TDR function of the 8720D, and see how things work out from there. But I really need to get a better grip of the relevant sections in Joel's book. There's a whole chapter on the TDR, and a section in another part of the book on creating your own cal kits, which makes extensive use of the TDR. 
> 
> Dave

No, it expects another connector to be connected at the reference plane. Since there is not, for the reasons I said before, it has a negative offset from the reference/calibration plane after calibration. If you perform the calibration at the end of the cable and add an N adapter, either Female to Male or Female to Female and leave the other end open and measure it, then it will have a positive offset. This will allow you to model the *adapter* as an "Open."

Spacecase

> {quote：title = Spacecase写道：} {quote}>不，它希望在参考平面上连接另一个连接器。
由于之前我说过的原因，它没有校准后与参考/校准平面的负偏移。
我明白那个。
当然可以使用开放式N作为校准装置，尽管精度有点差。
如果您支付QuickCal选项，安捷伦FieldFox系列便携式VNA可以执行此操作。
这真是令人印象深刻。
您只需要一个开放式连接器，如果预计DUT的回波损耗很好，那么您需要一个接近50欧姆的负载，但它不一定是计量级负载。
不需要短路。
我将QuickCal与我的85032B机械校准套件进行了比较，两者非常相似，尽管安捷伦称机械校准套件更准确。
>如果您在电缆末端执行校准并添加N适配器，无论是母对公还是母对母，并将另一端打开并进行测量，那么它将具有正偏移。
同意。
>这将允许您将*适配器*建模为“打开”。
>> Spacecase据我所知，开放的N公测试端口比开放的N母测试端口更令人痛苦。
如果看一下85032B校准套件，它包括一个带有母标准的小型扩展器，以增加公头中心销的直径。
女性不需要这样的装置。
当然其他校准套件可能有扩展器作为开放的一部分，但在85032B的开放时，必须坚持使用这个小适配器。
由于男性N没有伸展，我试图用两个女性到女性的桶形连接器制作一个开放的雌性和短雌性。
我把一个短路放在一个，然后让另一个打开。
然后抵消将大致相似。
我买了一些不同的女性对女性桶，并在回报损失方面找到了最好的。
然后我将一个短片连接到一个，然后将另一个打开。
我认为这样可以让我对女性N进行建模，就像安捷伦女性N一样，只需要很少的东西 - 只需稍微屏蔽一下，不需要增加公针的直径。
但是，经过一段时间的尝试，我得出的结论是，不可能像这样建立一个cal标准。
很简单，如果我用这个校准，然后放上一个真正的校准标准，史密斯圆图就到处都是。
实际标准的阻抗进出史密斯圆图，显示出负阻力和正阻力。
我认为问题可能是没有正确考虑到桶的损失。
PTFE电介质可能意味着损失不能仅仅归因于趋肤效应，因此输入的损耗偏差不能充分代表性能。
至少是开路公N，阻抗保持无功，并且不会阻抗阻抗实部的奇怪变化。
如果你愿意，我可以张贴我所制作的图片，以及它在VNA上的样子。
它不漂亮！
DaveEdited：drkirkby于2013年1月3日上午2:11



     以下为原文

  > {quote:title=Spacecase wrote:}{quote}

> No, it expects another connector to be connected at the reference plane. Since there is not, for the reasons I said before, it has a negative offset from the reference/calibration plane after calibration. 

I understand that.

It is certainly possible to use an open N as a calibration device, although the accuracy suffers a bit. 

Agilent's FieldFox range of portable VNAs do this if you pay for the QuickCal option. That is pretty damm impressive. All you need is an open connector, and if the return loss of the DUT is expected to be good, then you need a load which is somewhere near 50 Ohms, but it does not have to be a metrology grade load. No short is needed. 

I compared that QuickCal to my 85032B mechanicanical cal kit, and the two were very similar, although Agilent state a mechanical cal kit will be more accurate.  

> If you perform the calibration at the end of the cable and add an N adapter, either Female to Male or Female to Female and leave the other end open and measure it, then it will have a positive offset. 

Agreed. 

> This will allow you to model the *adapter* as an "Open."
> 
> Spacecase

As far as I can determine, the open N male test port is a lot more of a pain to make than an open N female test port. If one looks at the 85032B cal kit, it includes a small extender with the female standard to increase the diameter of the male centre pin. 

No such device is necessary on the female. Of course other cal kits may have the extender as part of the open, but on the open in the 85032B one has to stick on this little adapter. 

Because of the fact there is no extension on the male N, I tried to make an open female and short female from two female-to-female barrel connectors. I put a short on one, and left the other open. Then the offsets would be roughly similar. 

I bought a number of different female-to-female barrels, and found the best of them in terms of return loss. Then I attached a short to one, and left the other open. I thought that would allow me to model the female N the same as the Agilent female N's, which require very little - just a bit of shielding around it, with no need to increase the diameter of the male pin. 

However, after trying this for some time, I came to the conclusion it was not possible to build a cal standard like this. Quite simply if I calibrated with this, and then put a real calibration standard on, the Smith chart was all over the place. The impedance of real standards went in and out the Smith Chart, showing both negative resistance and positive resistance. I believe the issue may be that the loss of the barrels was not taken into account properly. The PTFE dielectric may mean the loss can't be assumed to be just due to the skin effect, and so the loss offset entered can't adequitely represent the properties. 

At least wth an open male N, the impedance stay reactive, and does not exibit odd changes in real part of the impedance. If you want, I can post a picture of what I made, and what it looked like on the VNA. It was not pretty! 

Dave

Edited by: drkirkby on Jan 3, 2013 2:11 AM

亲爱的Dunsmore博士和Kirby博士;
我编辑了我的帖子，这是在spacecase开始与Kirby博士讨论之前写的。
该帖子的编辑是为了清晰起见。
Dunsmore博士可以告诉我，这个计算是在零度点的确定中随时进行的（“零度点”项是指可以观察到的传输线的物理端）。
感谢您过去的反馈。编辑：SOLT_guy于2013年1月3日上午6:58



     以下为原文

  Dear Dr. Dunsmore and Dr. Kirby;

     I edited my post which was written before spacecase began his discussion with Dr. Kirby.

     The post was edited for clarity.

      Dr. Dunsmore can you tell me if this calculation takes place at any time in the determination of the zero degree point ("zero degree point" term is referenced to what can be observed as a the physical end of a transmission line).

       Thank you for your past feedback.

Edited by: SOLT_guy on Jan 3, 2013 6:58 AM

> {quote：title = SOLT_guy写道：} {quote}> Dunsmore博士你能告诉我这个计算是否在零度点的确定中随时发生（“零度点”术语是指什么可以参考
观察到传输线的物理端）。
SOLT_guy，我不知道Joel，但是当你提到“这个计算”时，我不清楚你在描述什么计算。
戴夫



     以下为原文

  > {quote:title=SOLT_guy wrote:}{quote}

>       Dr. Dunsmore can you tell me if this calculation takes place at any time in the determination of the zero degree point ("zero degree point" term is referenced to what can be observed as a the physical end of a transmission line).

SOLT_guy,

I don't know about Joel, but is not clear to me what calculation you are describing, when you refer to "this calculation".

Dave

I decided to reprint the post because I knew it would be lost after your discussion with spacecase.  Can you please not comment after I make this post because I would like an answer to this question.  I also believe you will find the answer to this question to be of value.   The answer to this question may only require a brief explanation, that is, if we don't get too complicated into the details, such as how the short will play a role in the determination of the zero phase angle point (I am aware of the theoretical 180 degree phase angle difference between the short and the open).  I am only seeking to determine how the zero degree reference point is determined.  The equation is simple to understand and I have only posed this question for academic purposes - verification.    If you answer the question (although you may be right), I will not have achieved verification (even though you may have been right).  If Dr. Dunsmore replies to the question then verification will be achieved because he is a designer of Agilent network analyzers and calibration kits.   Please do not be offended by my request.    The reason why he has to answer is because if at calibration lab when a manager asks me how I know my knowledge of this particular facet of NA measurements I can show him this post.      

Here is the post:

""

Many years ago, I tried to theorize exactly how the "start" phase angle count works in a network analyzer. However, I did not receive a complete answer from an engineer and after years of chasing the question but not getting an adequate reply, I gave up.

I am very happy that Dr. Kirby has revisited this subject. 

This is my question: 

Assume Port 1 has been initially calibrated by a cal kit (which I will call cal kit 1).

If an open standard is connected to PORT 1 from a different cal kit (which I will call cal kit 2) and I dial back port extensions to zero degrees, the physical length of the standard and the fringing capacacitance have now been measured.   Afterward, another formal measurement is made to determine the fringing capacitance of the open.  

Now I go to manually input the data of the open standard from cal kit 2 into my own personal cal kit. I record the offset delay and the fringing capacitance time delay values.

As per frequency, within the network analyzer, is the start phase angle count to determne the zero degree initial reference point of the network analyzer calculated by this method:

Edited for clarity the following day

Subtract (add) the "total offset delay" in terms of the time data inputs from an arbitrary initial time data value ("total offset delay" is defined to mean the dialed value using port extensions = = physical length + fringing capacitance expressed in units of time) and then add (subtract) the appropriate sum of polynomial terms of the fringing capacitance expressed in unit of time. The reason why I add the fringing capacitance is because I know that a physical length does not exist but the delay is due to an electrical effect at the tip of the open pin. 

After this time value has been calculated, the zero degree point is referenced (the "zero degree point" is referenced to the physical tline length).

The terms in parenthesis () can be ignored, I have only inserted them into this explanation because the exact "form" of the calculation method may reverse these signs for specific design considerations. What is relevant is the sign reversal - subtract the "total offset delay" ("total offset delay" is defined to mean the dialed value using port extensions = = physical length + fringing capacitance expressed in units of time)  and add fringing capacitance (or vice versa).

In the past, I have not received an adequate explanation of how fringing capacitance is incorporated into the network analyzer calculations. I did make an inquiry into this matter at an IEEE conference.

I would like to take this opportunity to thank a former Agilent engineer, who now works for Anritsu corporation, because when I posed a general fringing capacitance question to him in mathematical form at an IEEE conference he gave me a definitive reply. However, I did not pose this particular question to him because it was a rainy and dreary night and I did not want to keep him long. 

The reason why I am writing this is because I feel bad about something I wrote with regard to Anritsu corporation. I didn't get good technical support for an Anritsu product I bought and I wrote in this forum about the quality of technical support with regard to this example. However, as I was writing this reply, I recalled, when I went to a free Anritsu conference and I had asked some fundamental questions about cal kit inputs, I received some really good feedback from some Anritsu engineers.  For this, I am eternally grateful. I really like to have clear conscience when I write about my experiences with corporations and I like to include all the details, the good along with the bad.  This is my way of clearing my conscience. 

I would also like to thank you, Dr. Dunsmore, for your past technical support. I am eternally grateful for the feedback that you provided for me in the past. It was very helpful.

This will probably be one of my last questions in this forum. I am finally getting around to wrapping up all the loose ends with regard to NA calibration. I look forward to reading your reply.

""

Edited by: SOLT_guy on Jan 3, 2013 10:13 AM

Edited by: SOLT_guy on Jan 3, 2013 4:37 PM

“零相位”一案的确定是有点问题和假定的标准的延迟是恒定的，并且校准这样的点间距那点之间发生小于180度的相位变化。
但考虑到condiiton（和几乎所有的标准满足这样的条件下），校正过程迫使开放的响应-exactly-匹配的开放模式，短的响应-exactcly-匹配很短的模型，并响应
负载与负载resposne完全匹配。
因此，没有相位的“零分配”，而是计算提取的误差项，使得上述陈述保持真实。
在创建模型时，它们基于开放和短路（非电气测量）的物理测量以及根据标准的直径和长度计算的模型。
这就是为什么关键是要增加“男性扩展”（我可以像很多垃圾邮件过滤器不会让这句话通）为男性测试端口，该模型假定不变outter直径。
此外，对于N型，参考平面是从阴测试端口触点的尖端（它是外导体的配合表面） - 并且因此，阳测试端口从参考平面凹进。
所以男性测试引脚 - 没有扩展器 - 将显示负延迟。
添加测试引脚延长器使得公引脚的直径恒定，并且允许非负偏移的长度。
事实上，它被设定为使得其偏移几乎阴相匹配，以使它们保持180在宽的频率度的相分离通过range.Edited：Dr_joel上2013年1月3日下午7点53分



     以下为原文

  The determination of the "zero-phase" case is a bit problematic and presumes that the delay of the standard is constant and the point spacing of the calibration such that less than 180 degrees phase change occurs between points.  

But given that  condiiton (and almost all standards meet such conditions), the correction process forces the response of the open to -exactly- match the open model, the response of the short to -exactcly- match the short model, and the response of the load to -exactly- match the load resposne.  

Thus, there is no "zero-settng" of the phase, but rather the extracted error terms are computed such that the above statement hold trues.

In creating the models, they are based soley on physical measurements of the open and shorts (not electrical measurements) and the models computed from the diameter and length of the standards.

That's why is is critical to add the "male-extender" (I can image a lot of spam filters won't let that phrase thru) for male test ports, as the model presumes constant outter diameter.

Further for type N, the reference plan is -recessed- from the tip of the female test port contact (it is the mating surface of the outer conductor)  and thus, the male test port is recessed from teh reference plane.  So male test pins -without an extender- will show negative delay. Adding the test pin extender makes the male pin constant diameter and of a length that allows non-negative offset.  In fact, it is set so that its offset nearly matches the female, so that they maintain 180 degrees phase separation over a wide frequency range.

Edited by: Dr_joel on Jan 3, 2013 7:53 PM

亲爱的Dunsmore博士;
感谢您的最佳答复。
我相信你可能已经回答了Kirby博士的原始问题，关于他为什么看到他在NA显示屏上看到的东西（只有他可以验证这一点）。
Kirby博士确实提出了一个关于在开放标准模型结束时研究“边缘电容”的学术问题，为了学术目的，我也希望能够引用这项研究的来源。
我相信你已经将“贝塞尔”功能与这项努力联系起来了。
您能否引用消息来源，以便我们更好地了解开放模型？
再次，我想感谢您的回复。
编辑：SOLT_guy于2013年1月3日下午4:29编辑：SOLT_guy于2013年1月3日下午4:34编辑：SOLT_guy于2013年1月3日4:43 PM编辑：SOLT_guy于2013年1月4日7:12
下午



     以下为原文

  Dear Dr. Dunsmore;

     Thank you for your most excellent reply.

     I believe that you may have answered Dr. Kirby's original question with regard to why he is seeing what he is seeing on the NA display screen (only he can verify this).   
 
     Dr. Kirby did raise an academic question with regard to researching the "fringing capacitance" at the end of the open standard model and for academic purposes  I would also like to be able to cite the source of this research also.    I believe that you had associated the "Bessel" function with this endeavour.   Can you please cite the source so that we may better understand the open model better?

     Again, I would like thank you for your informative reply.

Edited by: SOLT_guy on Jan 3, 2013 4:29 PM

Edited by: SOLT_guy on Jan 3, 2013 4:34 PM

Edited by: SOLT_guy on Jan 3, 2013 4:43 PM

Edited by: SOLT_guy on Jan 4, 2013 7:12 PM

Joel博士，你提到的论文可能不是这篇论文，但它可能是参考文献之一：[http://www.dtic.mil/dtic/tr/fulltext/u2/a243236.pdf] Spacecase



     以下为原文

  Dr Joel,

The paper you were refering to probably wasn't this paper, but it might have been one of the references:  [http://www.dtic.mil/dtic/tr/fulltext/u2/a243236.pdf]


Spacecase