是德科技
直播中

戚歆敦

7年用户 978经验值
私信 关注
[问答]

DSO80204B怎么测量振荡器的抖动

嗨,我有一个DSO80204B 40Gsa / s Infiniium,想要测量振荡器的抖动。
示波器没有特定于抖动的测量功能,因此可以尝试直方图。
考虑附加图像。
我将52MHz波均匀地置于零上下(去除了直流偏移)并使用了高存储深度。
抖动看起来像一个高斯形状的随机。
我有两件事要问。
1)通常说rms抖动是直方图的标准偏差。
但是,如果我没有弄错的话,RMS抖动是rms与理想转换的偏差,意味着零均值。
直方图的平均值不为零(参见附件),并且根据定义,标准偏差是从平均值(Xi-mean)^ 2的每个点的*差*的平方计算的。
换句话说,当均值为零时,rms抖动应该是标准偏差,因此直方图的stdev是*不是rms抖动。
我失去了什么吗?
2)直方图测量受到示波器的触发抖动的影响,这意味着可以非常快速地触发示波器并获取数据点的能力。
触发抖动在多大程度上影响测量 - 限制是什么?
我确定200fs测量准确吗?

以上来自于谷歌翻译


     以下为原文

  Hi

I have a DSO80204B 40Gsa/s Infiniium and want to measure jitter of an oscillator. The scope has no jitter-specific measurement functions, so thought to try an histogram. Consider attached image. I placed a 52MHz wave uniformly above and below zero (removed dc offset) and used high memory depth. Jitter looks random with a gaussian like shape. I have two things to ask.

1) It's commonly said that rms jitter is the standard deviation of the histogram. However, if I'm not mistaken, RMS Jitter is the rms deviation from the ideal transition, meaning having zero mean. Histogram's mean is not zero (see attached) and also by definition, standard deviation is computed from the squares of the *difference* of each point from the mean (Xi-mean)^2. In other words, it should be that rms jitter is the standard deviation when the mean is zero, and consequently stdev of the histogram is *not* the rms jitter. Am I loosing something?

2) The histogram measurement is subjected to the scope's trigger jitter, meaning the ability of the scope to be triggered and acquire data point very quickly. To what extent the trigger jitter affects the measurement-what are the limits? Am I assured that a 200fs measurement is accurate?   
附件

回帖(3)

周菊

2019-3-1 15:21:47
我要说的第一件事就是“获得一个抖动包”。
它适用于您的范围,并将使您的生活更轻松。
抖动包是示波器SW的一部分,因此您可以轻松打开短期许可证并进行试用。
我无法从你的照片中确切地知道这是什么。
它是正弦波下降沿的交叉点吗?
如果是这样,那就是测量抖动的最坏方法,即使它是抖动测量包变得可用且易于使用之前的唯一方法。
使用此方法会丢失许多抖动组件。
此外,存储器深度无关紧要,因为示波器仅查看相对于触发器的单个交叉点。
为了解决你的信息中的一些具体问题: - “但是,如果我没有弄错,RMS抖动是rms偏离理想转换,意味着零均值”平均值可能是也可能不是零。
系统中的抖动可能不是完全高斯的。
它可能有DJ组件,或者交叉点可能与理想位置有正偏移或负偏移。
如果抖动不是高斯,则RMS值没有意义。
抖动工具将高斯分量与非高斯分量分开,并分别使用它们来得出总抖动的良好估计。
- “触发抖动在多大程度上影响测量 - 限制是什么?我确信200fs测量是准确的吗?”
200fsec规格是准确的。
很多工作都能够将触发抖动降低到极少数。
您可以使用与用于进行“抖动”测量的相同技术自行测量。
在真正的抖动测量中,触发器没有任何意义或影响,并且不用于测量。
重要的是交叉点相对于理想的时间,无论是“明确的”还是计算的。
免责声明:为了获得更可靠的响应,您应该考虑致电当地的安捷伦技术呼叫中心。
安捷伦论坛在“可用”的基础上进行监控,并不一定是解决技术问题的最快方式。

以上来自于谷歌翻译


     以下为原文

  First thing I'll say is, "Get a Jitter Package". It's available for your scope, and will make your life so much easier. The jitter package is part of the scope SW, so you could easily turn on a short-term license and try it out.

I can't tell from your picture exactly what this is. Is it the crossing point of the falling edge of the sine wave? If so, that's about the worst way of measuring jitter, even though it was the only way before jitter measurement packages became available and easy to use. There are many jitter components that will get missed using this method. Also, memory depth won't matter, since the scope is only looking at a single crossing point relative to the trigger.

To address some specific issues in your message:
- "However, if I'm not mistaken, RMS Jitter is the rms deviation from the ideal transition, meaning having zero mean"
The mean may or may not be zero. The jitter in your system may not be completely Gaussian. It may have DJ components, or the crossing may have a positive or negative offset from the ideal location. If the jitter is not Gaussian, then the RMS value has no meaning. The Jitter tools separate the Gaussian components from the non-Gaussian components, and use them separately to come up with a good estimate of the Total Jitter.
- "To what extent the trigger jitter affects the measurement-what are the limits? Am I assured that a 200fs measurement is accurate?"
The 200fsec specification is accurate. A lot of work went into being able to reduce the trigger jitter to a very small number. You could measure it yourself using the same technique you used to make your "jitter" measurement.

In a true jitter measurement, the trigger has no meaning or effect and is not used in the measurement. All that matters is the timing of the crossing points relative to the ideal, whether that's 'explicit' or calculated.

Al

Disclaimer: For more reliable response, you should consider calling your local Agilent Technical Call Center. The Agilent Forums are monitored on an "as available" basis, and aren't necessarily the fastest way to get technical questions answered.
举报

戚歆敦

2019-3-1 15:35:05
引用: nvywyerwer 发表于 2019-3-1 15:21
我要说的第一件事就是“获得一个抖动包”。
它适用于您的范围,并将使您的生活更轻松。
抖动包是示波器SW的一部分,因此您可以轻松打开短期许可证并进行试用。

谢谢您的回答。
如果我不能购买抖动包怎么办?
我应该依靠直方图方法吗?
任何建议的程序?
我还有一个89441A vsa,它可以通过相位噪声测量的时域变换计算抖动,但作为本论坛中提到的另一个人,vsa拍摄了一个非常小的信号快照(~1.5ms)。
不知道这是否是有效的测量。

以上来自于谷歌翻译


     以下为原文

  Thank you for your answer.
What if I can't buy the jitter package? Should I count on the histogram method?
any suggested procedure?
I also have an 89441A vsa, which can compute jitter via time domain transform from phase noise measurement, but as another fellow mentioned in this forum, the vsa tooks a really small snapshot of the signal (~1.5ms). don't know if this is valid measurement.
举报

周菊

2019-3-1 15:47:41
好的,还有一点我忘记了。
我不是一定要向你推销一个抖动包,只是推荐它。
如果您选择不购买抖动包,您可以自己做同样的事情,通过一些编程:1)卸载轨迹,更深更好2)找到所有交叉点3)计算理想交叉点。
这可以是具有SW PLL或理想时钟的圆顶,或者是从数据导出的恒定频率时钟。
4)计算理想交叉点与实际交叉点之间的差异。
现在你有一个抖动记录,真正的乐趣开始了。
5)将DJ(ISI,DCD,BUJ)与PJ和RJ分开6)将PJ与RJ分开7)现在将它们全部放在一起,确定什么是重要的以及每个组件对测量的影响程度。
我会警告你,这可能需要一段时间。
安捷伦(和我们的竞争对手)有多年来一直在考虑这个问题的人,改进算法,使它们更快,更可靠。


以上来自于谷歌翻译


     以下为原文

  OK, one more point I forgot.

I'm not trying to necessarily sell you a jitter package, just recommending it. If you choose to not buy a jitter package, you can do the same thing yourself, with some programming:

1) Unload the trace, deeper is better
2) Find all of the crossing points
3) Calculate the ideal crossing points. This could be dome with a SW PLL, or an ideal clock, or a constant frequency clock derived from the data.
4) Calculate the difference between the ideal crossings and the real crossings. Now you have a jitter record, and the real fun begins.
5) Separate the DJ (ISI, DCD, BUJ) from the PJ and RJ
6) Separate the PJ from the RJ
7) Now fit it all back together, deciding what's important and how much each component affects your measurement.

I'll warn you, this could take a while. Agilent (and our competitors) have people who have been thinking about this for years, refining the algorithms, making them faster, and more reliable.

Al
举报

更多回帖

发帖
×
20
完善资料,
赚取积分