相位噪声问题

嗨，我会尽力回答你的两个问题。
第一个是关于HP产品说明。
您会注意到，在文本中，您所引用的图表被称为相位波动的频谱，如频谱分析仪所示。
与该图并置的是边带之一，在y轴上显示sigma ^ 2 RMS。
这些图的目的是为您提供关于理想频谱分析仪上振荡器的相位噪声的显示效果的定性概念。
目的不是准确指定y轴上的单位。
一般来说，可以安全地假设“A”指的是幅度;
频谱分析仪通常在y轴上显示功率。
由于功率通常与电压的平方成正比，并且由于您可以将2.1b中的图与2.1a中的数据的子集相关联，因此您可以假设您在图2.1a中的功率范围内
。
但是，我想再次强调，这些图是出于定性目的，y轴的确切单位对于产品说明而言并不重要。
关于你关于SSB相位噪声定义的问题，我会尽力解释它。
我相信如何以图形方式解释NIST定义可能会有些混乱。
我们来看看更多资源。
在Agilent相位噪声选择指南（http://cp.literature.agilent.com/litweb/pdf/5990-5729EN.pdf）的第3页上，它说明了NIST对SSB相位噪声的定义：“功率密度的比率
在从载波到载波信号总功率的偏移频率处。“
功率密度可以图形化地计算为在某个频带内的曲线下面积（我们通常将标准化为1 Hz带宽）在距载波的某个偏移频率处。
在我看来，真正的困惑来自于“载波信号的总功率”。
理论上，载波信号的总功率将通过对所有频率（负无穷大与无穷大）的功率进行积分来图形化地计算，如您所建议的那样。
但是，如您所知，由于分析仪中的失真和噪声，输入信号的可能失真和其他噪声（包括热噪声）以及干扰信号，存在一些限制。
如果我们将所有这些整合在一起，我们最终会得到一个远大于载波信号实际功率的读数。
在大多数情况下，我们可以通过使用默认RBW并在载波信号上放置标记来估计频谱分析仪上载波信号的功率;
然而，更准确的方法是执行频带/间隔功率计算。
问题再次成为，整合的限制是什么（我们的频段/区间计算的带宽）？
这可能是一个困难的问题，取决于我提到的所有现实世界的限制。
这就是为什么我们通常坚持测量载波信号功率的方法，并允许它代表总信号功率。
实际上，这是一个很好的假设，错误通常并不重要。
有关更多信息，请参阅Kay Gheen演示文稿的幻灯片10。
此外，还有一篇由安捷伦的Bob Nelson撰写的伟大文章讨论了相位噪声测量，并以非常容易理解的格式解释了细节。
强烈建议将其作为参考资料：http：//mwrf.com/test-amp-measurement-analyzers/demystify-integrated-phase-deviation-results-phase-noise-measurements如果我没有正确解释您的问题，请让
我知道。
最诚挚的问候，斯科特



     以下为原文

  Hi,

I will try to answer your two questions.

The first one was about the HP product note. You will notice that in the text, the plot to which you are referring is referenced as the spectrum of the phase fluctuations, as seen on a spectrum analyzer. Juxtaposed to that plot is one of the sidebands, showing sigma^2 RMS on the y-axis. The purpose of these plots are to give you a qualitative idea of what the display would look like for the phase noise of an oscillator on an ideal spectrum analyzer. The purpose is not to specify exactly what units are on the y-axis. In general, it is safe to assume that the "A" refers to amplitude; it is common for spectrum analyzers to display power on the y-axis. Since power is generally proportional to the square of the voltage, and since you could associate the plot in 2.1b to be a subset of the data in 2.1a, you could then make the assumption that you are on a power scale in plot 2.1a. However, I want to emphasize again that these plots are for qualitative purposes and the exact units of the y-axis are not important for the purposes of the product note.

With regards to your question about the definition of SSB phase noise, I will try to interpret it as best I can. I believe that there may be some confusion as to how to interpret the NIST definition graphically. Let's take a look at some more resources. On page 3 of Agilent's phase noise selection guide (http://cp.literature.agilent.com/litweb/pdf/5990-5729EN.pdf), it states the NIST definition of SSB phase noise: "the ratio of the power density at an offset frequency from the carrier to the total power of the carrier signal." The power density can be calculated graphically as the area under the curve within some band (we would normally normalize to 1 Hz bandwidth) at some offset frequency from the carrier. It seems to me that the real confusion comes from the "total power of the carrier signal." In theory, the total power of the carrier signal would be calculated graphically by integrating the power across all frequencies (negative infinity to infinity), as you suggest. However, as you know, there are limitations due to distortions and noise in the analyzer, possible distortion and other noise (including thermal noise) of the input signal, and interfering signals. If we were to integrate all of this, we would end up with a reading that was much larger than the real power of the carrier signal. In most cases, we can estimate the power of the carrier signal on a spectrum analyzer by using the default RBW and placing a marker on the carrier signal; however, a more accurate method would be to perform a band/interval power calculation. The question once again becomes, what are the limits of integration (bandwidth of our band/interval calculation)? This can be a difficult question and vary depending upon all of the real world limitations that I mentioned. That is why we normally stick to the method of measuring the carrier signal power and allowing that to represent the total signal power. In fact, this is a good assumption and the error is usually not significant. Refer to slide 10 of Kay Gheen's presentation for more information.

Also, there is a GREAT article written by Agilent's Bob Nelson which discusses phase noise measurements and explains the details in a format that is very easy to understand. It is highly recommended as reference material: http://mwrf.com/test-amp-measurement-analyzers/demystify-integrated-phase-deviation-results-phase-noise-measurements

If I did not interpret your questions correctly, please let me know.

Best Regards,
Scott

亲爱的斯科特：我想感谢你的反馈。
在我发现Gheen先生的网络研讨会的存在之后，我看了他的网络研讨会，并且Gheen先生确实参考了他的图表，我发现这些图表是“信息性的”。
然而，对于用于定义如何测量相位噪声的图表，仍然存在很小的解释空间。
我要求的只是验证“单边带相位噪声与载波比”的图形定义。
这只需要确认Gheen先生提出的“单边带相位噪声与载波比”的图形插值。
我将参考我上传的* .pdf文件中的图2.1（a），名为“HP Phase Noise”（注意：该文档的官方标题是HP产品说明11729C-2），（第5页），因为您已经制作
参考它。
我相信你不理解我在完成原帖后所要求的答案类型。
我在原帖中写道：“我不想对”A“表示什么做任何假设，请你明确告诉我”A“是什么意思表示”你回答：“然后你可以做
假设您在图2.1a中的功率范围内。“
我很欣赏你试图回答这个问题，但我需要得到学术上支持的答复 - 在这种情况下，解释或“假设”是不够的。
我之所以寻求明确定义的学术支持答案，是因为我遇到了公司发布的应用说明“MINI-CIRCUITS”，其中“s***相位噪声”的定义与安捷伦对“s***”的定义有所不同
相位噪声。“
Mini Circuit的应用笔记题为“VCO相位噪声”，如果您查看本应用笔记的图3，您将看到安捷伦对“s***相位噪声与载波比”的定义与Mini-Circuits定义的不同之处。
我已经上传了这份应用笔记的副本，我留下了一个链接，供您直接下载文档。
请使用我在下面提供的链接获取本应用笔记的副本。
我希望您确定已从Mini-Circuit自己的网站下载此文档。
http://www.minicircuits.com/app/VCO15-6.pdf正如您可以从本网站上的其他帖子中读到的，有很多不同的，专业的，意见，关于如何使“相位噪声”相关
计算。
因此，我只是寻求一个简单的图形化参考，我知道这是真实的，我可以依赖它作为参考目的。
我知道正确标记的图形是“明确的”。
我需要的只是一个正确标记的图形。
让我重申一下我的问题：让我们从图2.1（a）中的“单边带相位噪声与载波比”的“理论”定义开始。
国家标准局将“单边带相位噪声”定义为每赫兹的一个相位调制边带中的功率与偏离载波的偏移f f赫兹的“总信号功率”之比。
关于我们之前讨论的“总信号功率”的定义：如果在图2.1（a）的图的垂直轴上标记“功率”，则是NIST术语，“总信号功率”，由
从波形的中心频率fo到曲线上最右边的曲线下面积，边带功率变为零。
请注意，我指的是中心频率点作为下限，我没有参考中心频率左边的任何点。
我所做的只是“标记”曲线的上限和下限。
该区域是NIST术语“总信号功率”的正确图形表示吗？
如果这不是对术语“总信号功率”的正确图形解释，那么请明确说明曲线下面积的上下界，并用适当的物理单位标记垂直轴。
感谢您的反馈意见。
编辑：SOLT_guy于2013年2月1日7:31 PM编辑：SOLT_guy于2013年2月1日下午7:33



     以下为原文

  Dear Scott:

      I would like to thank you for your feedback.    After I discovered the existence of Mr. Gheen's webinar,  I watched his webinar and Mr.Gheen did make some references to his graphs which I found to be "informative."  However, there does remain a little room for interpretation with regard to the graphs he used to define how phase noise is measured.   

       All I require is verification of the graphical definition of "the single sideband phase noise to carrier ratio."   This would only entail confirmation of the graphical interpration of "the single sideband phase noise to carrier ratio" as set forth by Mr. Gheen.  I will refer to graph 2.1(a) from the *.pdf file that I uploaded, named "HP Phase Noise" (Note:  the document's official title is HP Product Note 11729C-2), (page 5), since you have made reference to it.   

      I do believe that you failed to understand the type of answer that I had required when I had made my original post.   I had written in my original post:

     "I don't want to make any assumptions about what the "A" denotes, can you please explicitly tell me what the "A" is intended to denote"

You replied:

"you could then make the assumption that you are on a power scale in plot 2.1a."

I do appreciate that you tried to answer the question but I needed an academically supported reply - an interpretation or an "assumption" in this case will not suffice for good reason.    The reason why I am seeking an academically supported answer that is clearly defined is because I have come across an application note published by the company, "MINI-CIRCUITS", whose definition of "s*** phase noise" substancially differs from Agilent's definition of "s*** phase noise."

The Mini Circuit's application note is entitled, "VCO Phase Noise", and if you look at figure 3 of this application note, you will see where Agilent's definition of "s*** phase noise to carrier ratio" differs from Mini-Circuits definition.

I have uploaded a copy of this application note and I have left a link, for you to download the document directly from.   
Please use the link that I have provided for you below to obtain your copy of this application note.  I want you to be certain that you have downloaded this document from Mini-Circuit's own website.
  
http://www.minicircuits.com/app/VCO15-6.pdf

       As you can read from other posts on this website there are a lot of differing, professional, opinions, with regard to, how to make "phase noise" related calculations.  Therefore, I am seeking only a simple, graphical, reference that I know to be true and that I can rely upon as true for reference purposes.  I know that a properly labelled graph is "unambiguous."  All I require is a properly labelled graph.   

Let me restate my question:

Let us begin with the "theoretical" definition of the "the single sideband phase noise to carrier ratio" using graph 2.1(a).

The National Bureau of Standards defines "Single Side Band Phase Noise" as the ratio of power in one phase modulation sideband per Hertz, at an offset, f, Hertz away from the carrier, to the "total signal power".

With regard to the definition of "total signal power" that we have discussed before:

If "power" is labelled on the vertical axis of the graph of figure 2.1(a), is the NIST term, "total signal power", represented by the area under the curve from center frequency, fo, of the waveform to the rightmost point on the graph where the sideband power goes to zero.  Please take notice that I am referring to the center frequency point as the lower limit, I am not referencing any points to the left of the center frequency.

All I did was "mark" the upper and lower limits of the curve.  

Is this area the correct graphical representation of the NIST term, "total signal power"?  
 
If this is not the correct graphical interpretation of the term, "total signal power," then please explicitly set forth the upper and lower bounds of the area under the curve and label the vertical axis with the appropriate physical unit.

Thank you for your feedback.

Edited by: SOLT_guy on Feb 1, 2013 7:31 PM

Edited by: SOLT_guy on Feb 1, 2013 7:33 PM

附件

• Mini Circuits VCO Phase Noise.pdf31.8 KB
  

您好，感谢您提出后续问题和意见。
你是对的 - 当你在产品说明中询问y轴标签的含义时，我确实在响应中使用了词假设。
更确切地说，图2.1a中的“A”代表“幅度”。
请注意，不提供进一步的信息（例如用于了解比例，电压，功率等的单位）。
但是，正如我所提到的，该图的目的是为您提供频谱而不是定量的定性概念，因此不需要诸如单位之类的细节。
我恭敬地不同意您在Mini Circuit的应用笔记中对单边带相位噪声的定义存在差异。
如果您参考HP产品说明的图2.2，您会注意到它与Mini Circuit应用笔记的图3基本相同。
这两个音符都以图形方式表示总信号功率为Ps或载波信号的功率电平。
此外，它们都参考了文本中关于“总信号功率”的单边带相位噪声的等式。
HP产品说明图2.2的标题表明，这是在频谱分析仪显示屏上获得单边带相位噪声的图形方式。
通常，我们可以互换地使用术语“载波功率”和“总功率”。
实际上，如果曲线图代表Ps的理论真值（载波在中心频率Fo处的功率），则这是正确的总功率。
另一方面，在频谱分析仪显示器上，所示的总功率限于通过分辨率带宽滤波器到达检测器的功率。
正如我在上一次回复中提到的，总积分功率与通过将标记放置在中心频率测量的功率之间的差异（因此误差）是最小的。
在现代分析仪中，载波的“总功率”来自仪器分析带宽内载波信号的功率。
要回答您的其他问题，您需要测量整个频率范围以获得总信号功率。
因此，在两种情况下，您的积分极限将从曲线的最左侧点（不是中心频率）到最右侧点。
这是功率基本上为零的地方。
从理论上讲，真正的总功率将从负无穷大变为正无穷大。
当然，在频谱分析仪上，您受到测量带宽，杂散和其他干扰信号以及本底噪声的限制。
我希望这更清楚一点。
我感谢任何进一步的讨论或反馈。
最诚挚的问候，斯科特



     以下为原文

  Hi,

Thank you for your follow up questions and comments. You are correct- I did use the word assumption in the response when you had asked for the meaning of the y-axis label in the product note.
To be more definite, the "A" in figure 2.1a stands for "Amplitude." Note that no further information is provided (such as units to give you an idea of the scale, voltage, power, or otherwise). However, as I mentioned, the purpose of that figure is to give you a qualitative idea of the frequency spectrum rather than quantitative, and so details such as units are unnecessary.

I would respectfully disagree with you that there is a difference in definition of single sideband phase noise in the app note from Mini Circuit. If you refer to Figure 2.2 of the HP product note, you will notice that it is essentially identical to Figure 3 of the Mini Circuit app note. The two notes both graphically represent the total signal power as Ps, or the power level of the carrier signal. Also, they both reference the equation for single sideband phase noise in the text with respect to "total signal power."

The caption for Figure 2.2 of the HP product note states that this is the graphical way to derive the single sideband phase noise on a spectrum analyzer display. In general, we can use the term "carrier power" and "total power" interchangeably. In fact, if the plots represent the theoretical true value of Ps (the power of the carrier at the center frequency Fo), then this is the correct total power. On the other hand, on the spectrum analyzer display, the total power shown is limited to the power that reaches the detector through the resolution bandwidth filter. As I mentioned in my last response, the difference (and therefore the error) between the total integrated power and the power measured by placing a marker at the center frequency is minimal. In modern analyzers, the "total power" of the carrier is derived from the power of the carrier signal within the instrument's analysis bandwidth.

To answer your other question, you would need to measure the entire frequency range to get the total signal power. Therefore, your limits of integration would be from the leftmost point (not the center frequency) on the curve to the rightmost point, in both cases. This is where the power is essentially going to zero. 

In theory, the true total power would go from negative infinity to positive infinity. On a spectrum analyzer, of course, you are limited by your measurement bandwidth, spurs and other interfering signals, and the noise floor.

I hope that is a bit more clear. I appreciate any further discussion or feedback.

Best Regards,
Scott

Dear Sir:

        I would like to thank you for your reply.

        The reason why I pose fundamental questions of this type, where I require clear and exact definitions, is because the customer, who uses your test instruments to make measurements and design products, must clearly understand, in an exact physical sense, exactly what is being  measured.   Therefore, if the measurement is an "interpretation" of a theoretical value, or concept, the limits of the measurement's accuracy must then be considered (how much deviation from the theoretical value exists).      

        Definitions  which are not clearly specified are "useless" to a scientist or an engineer because the uncertainty of a measurement cannot be determined.   

        There is room for misinterpretation if one reads material which is not clearly specified (the case of the meaningless "A" for example).     

        There is also room for misinterpretation if one relies upon the "spoken word" for a definition.   I saw the phase noise webcast and the "spoken word" led me to hypothesize about what the "limits of the area under the curve" are.    However, I do not rely upon the "spoken word", therefore, I posed the question about the exact graphical definition of signal power.   I appreciate that you have answered my question to a degree.   

        I realize that it is possible to express a value in a simple manner which is easy for people to understand - but it is incorrect because it is not in "conformance" with the exact "formal" definition which fits into an established physical framework.  

        Let's say everyone took everyone else's word as true, and we all copied our academic data from the same source, and no one really knew exactly what that data physically meant, do you think there would be some confusion?

        When you wrote:

"  I would respectfully disagree with you that there is a difference in definition of single sideband phase noise in the app note from Mini Circuit. If you refer to Figure 2.2 of the HP product note, you will notice that it is essentially identical to Figure 3 of the Mini Circuit app note. The two notes both graphically represent the total signal power as Ps, or the power level of the carrier signal. Also, they both reference the equation for single sideband phase noise in the text with respect to "total signal power." ... "

       I will pose the question:

Do you think that the graph in the Mini-Circuits application note was copied?

      (What a coincidence!, Is it just random luck that the graphs are identical?  Notice the meaningless "A" which is labelled on vertical axis of Figure 3 - it kind of resembles the meaningless "A" labelled on the vertical axis of the HP product note, doesn't it).  However, not every term in the graph is identical, there exists one exception, and I will go into detail about this one exception.  But, first, I must pose a question for you:

       The question I will pose to you is, look above figure 3 of the Mini-Circuits's Application Note entitled, "VCO Phase Noise", there exists a definition:    Sc(f)

The Mini-Circuit App Note defined Sc(f)  [which I will refer to as "Equation 1"] as:

[Equation 1]:   " 
                          Sc(f) =  Ps / Ps*** 
                                                            "

Quoting the Mini-Circuits's application note:

"Ps is the carrier power and Ps*** is the sideband power in one Hz bandwidth at an offset frequency f from the center"


Page 5 of the HP product note 11729C references this ratio [that I will refer to as "Equation 2"] as:

[Equation 2]:  "

                           Ps*** / Ps
                                                "
         
         I would like to continue this discussion with you, but, first, can you reply and assert that I have correctly "transcribed" equation 1 from the Mini-Circuits application note and if you believe that I have not correctly "transcribed" this equation from the application note please say so in your reply.     (Please take notice that I have used the term "transcribed" in reference to the Mini Circuit's equation, this means that I have identically "transferred" what was written on the Mini Circuit's application note to this forum -  I am not indicating that the content of this equation is "correct" - only that the equation was transferred onto this forum as it was written in the application note).

        I would like to thank you for your past replies and please do not take my line of questioning in the wrong way.     There are other issues, not specifically related to science and engineering, that I must consider before I continue on with this discussion.   I must verify that we are both on the same page.   Do you see what I see?  If you want to check the authenticity of the Mini Circuit's application note that I have uploaded and linked to you, then please conduct an independent search of this application note using any web search engine.  In other words, do not click on the link, or look at the *.pdf file I have uploaded for you, conduct an independent search of this same application note on the internet for yourself to verify that "you see what I see."  I would prefer if you did this.

       I look forward to reading your reply and please do not take my response in the wrong way .   There is a reason why I have to communicate in this manner to you.    Again, I would like to thank you for your replies.


Edited by: SOLT_guy on Feb 1, 2013 8:36 PM

Edited by: SOLT_guy on Feb 2, 2013 10:23 AM

Edited by: SOLT_guy on Feb 3, 2013 5:22 AM

嗨，你说的没错。
您在Mini-Circuit应用笔记中引用的等式（在图中引用）显示了HP产品说明中类似等式的倒数。
我之所以说我对Mini Circuit与安捷伦的定义不一致的原因是Mini-Circuit应用笔记以下列方式说明了文本中的定义：“Sc（f）=（每个边带的功率密度）
在偏离载波的偏移频率f处的Hz带宽）/（总信号功率）“它们然后遵循NIST定义”，在每个赫兹带宽的一个相位调制边带中的功率比，在距离载波的偏移f赫兹处，
对于总信号功率，“用所讨论的等式。
因为符号方程与NIST定义中的单词或我之前引用的前一文本不匹配，所以这似乎是等式或图中的拼写错误。
我建议联系Mini-Circuits，在他们的应用笔记中询问此问题。
谢谢，最诚挚的问候，斯科特



     以下为原文

  Hi,

You are correct. The equation that you are referring to in the Mini-Circuit app note (and which is referenced in the figure) is showing the inverse of the similar equation in the HP product note. The reason that I said that I disagree with you about the definition from Mini Circuit versus Agilent is that the Mini-Circuit app note states the definition in the text in the following way:

"Sc(f) = (Power density in one sideband per Hz bandwidth at an offset frequency f away from the carrier) / (Total signal power)" 

They then follow the NIST definition, "the ratio of power in one phase modulation sideband per Hertz bandwidth, at an offset f Hertz away from the carrier, to the total signal power," with the equation in question. Because the symbolic equation does not match the words from the NIST definition or the previous text that I have quoted above, this appears to be a typo in the equation or figure. I would suggest contacting Mini-Circuits to ask about this issue in their app note.

Thanks, and Best Regards,
Scott

亲爱的斯科特;
我发现这篇文章写在杂志“微波和RF”中。
http://mwrf.com/active-components/managing-phase-noise-microwave-sources?NL=MWRF-06&Issue=MWRF-06_20130305_MWRF-06_525&YM_RID=go_er00@yahoo.com&YM_MID=1377019&sfvc4enews=
42本文展示了Mini-Circuits最初发布的错误相位噪声定义，它展示了错误如何拥有自己的生命，如同病毒一样失控。
很长一段时间后我回信了，因为我正在组装一个粗略的相位噪声测试装置，我想知道是否有人在这个论坛上曾经组装过“松散PLL”相位噪声测试电路，该电路采用双平衡混频器和频谱
分析仪进行相位噪声测量。
关于频谱分析仪和相位噪声测量的使用，我有一个计量测量问题，我希望有人可以插入并告诉我他们是否已将这个测试设置放在一起。
这个特定测试设置的细节最初是由NIST发布的，然而，“魔鬼在细节中”并且测试设置的一些细节没有公布。
我相信NIST的员工确实有责任回答有关“公益”的测试和测量方法的问题，或者至少将所有感兴趣的人发布和引导到包含详细问题答案的文件中
。编辑：SOLT_guy于2013年6月16日上午8:28



     以下为原文

  Dear Scott;

I found this article written in the magazine,  "Microwaves and RF".

http://mwrf.com/active-components/managing-phase-noise-microwave-sources?NL=MWRF-06&Issue=MWRF-06_20130305_MWRF-06_525&YM_RID=go_er00@yahoo.com&YM_MID=1377019&sfvc4enews=42

The article displays the erroneous phase noise definition that was originally published by Mini-Circuits and it demonstrates how errors can have a life of their own, growing like a virus out of control.

I wrote back after a long time because I am putting together a crude phase noise test set and I was wondering if anyone here at this forum had ever put together the "loose PLL" phase noise test circuit that utilizes a double balanced mixer and a spectrum analyzer to make phase noise measurements.

I have a metrological measurement question with regard to the use of spectrum analyzers and phase noise measurements and I was hoping someone could chime in and tell me if they had put together this test set up.   This details of this particular test set up was originally published by NIST, however, the "devil is in the details" and some of the details of the test set up were not published.    I do believe that the employees at NIST do have a public duty to answer questions about their test and measurement methods for the "public good" or, at least, publish and direct all interested persons to the documents which contain the answers to the detailed questions.

Edited by: SOLT_guy on Jun 16, 2013 8:28 AM