N4373C：了解测量不确定度的规范和程序

嗨曼努埃尔，谢谢你的兴趣。
实际上，除了简单地将光学转发器添加到网络分析仪之外，规范的可追踪校准和不确定性分析是LCA系统的关键值。
您的问题的一些细节确实会进入我们的内部生产流程（并且也有点复杂，就像您在此处写的那样。）用户指南确实有一些额外的细节，特别是有关定义和一般测量方法的
术语定义部分。
无论如何，这是一个部分响应，我希望能帮到你：相对频率响应是指响应频率的变化。
其偏差对应于频率响应中绝对误差频率的变化。
例如，如果测量结果在所有频率上偏离设备的真实响应（仅为迹线的刚性偏移），那么相对频率响应误差将为零，而绝对误差为1 dB
。
对于确定信号的带宽和可能的失真，设备的相对频率响应通常是令人感兴趣的。
绝对频率响应也决定了通过器件的信号强度，并以E / O器件的单位W / A或O / E器件的A / W表示。
相位测量是相对的。
此外，E / O和O / E测量在长度或群延迟中具有任意偏移，其通过线性斜率与频率改变相对相位响应。
在O / O测量中，用户校准可以提供零长度的参考，从而测量可以确定绝对GD。
校准和测试基于具有校准响应的参考接收器的测量，该参考接收器在此处开发并且可以追溯到NIST。
LCA发射器和LCA接收器都通过该过程校准其响应。
接收器在外部条件方面稍微更加稳定，因此通常通过在当前条件下对LCA接收器进行用户校准LCA发射器然后使用新校准的LCA发射器来完成光电设备的测量。
用于O / E测量。
“用户指南”描述了无需特殊设备即可完成的验证程序。
GUM的原理用于确定不确定性。
使用覆盖因子2。
A类和B类输入都有贡献。
例如，使用人工制品的规格将是类型B.系统测量结果分布是类型A.还需要注意的是，规范与不确定性结果不同。
规格还有额外的余量，以便老化和漂移。
（并且应该给出合理的产量。）随货物附带的校准认证数据给出了实际的校准日结果。
这特别指的是保证的规格。
还有一些典型的规范可以提供有关通常预期的性能水平的更多信息，这些信息可用于测量其他参数或超出保证的其他条件。
其中一些在生产过程中进行检查，但在定期重新校准期间通常不会重新检查。
此致，Mike P.S.
我们可能希望将此主题移至LCA主题聚集的论坛。
光子 - 电光元件的表征



     以下为原文

  Hi Manuel,

Thanks for your interest. Indeed the tracable calibration and uncertainty analysis for the specifications are a key value for the LCA system, beyond simply adding optical transponders to the network analyzer. Some of the details to your questions really get into our internal production processes (and are also kind of complex, like you note, to write out here.) The User Guides does have some additional detail, especially with the definitions and general measurement method in the Definition of Terms section.

Anyway here is a partial response that I hope helps you:

The relative frequency response refers to the variation with respect to frequency of the response. A deviation in this corresponds to a variation with frequency of the absolute error in the frequency response. For example, if the measurement result  were offset by 1dB  at all frequencies from the true response of the device (that is just a rigid offset of the trace), then the relative frequency response error would be zero, while the absolute error is 1 dB. The relative frequency response of a device is often of interest, as for determining the bandwidth and possible distortion of the signal. The absolute frequency response also determines the strength of the signal through the device and is expressed in terms of the units W/A for E/O devices or A/W for O/E devices. 

The phase measurement is relative. In addition, the E/O and O/E measurements have an arbitrary offset in the length or group delay, which changes the relative phase response by a linear slope vs. frequency.
In the O/O measurements, the user calibration can provide the reference for zero-length, so that the measurement can determine absolute GD.

The calibration and testing is based on measurements of a reference receiver with calibrated response that was developed here and is tracable to NIST. Both the LCA transmitter and the LCA receiver get calibration of their responses through this process. The receiver is somewhat more constant with respect to external conditions, so often measurement of optical-to-electrical devices is first done by making a user calibration of the LCA transmitter in the current conditions against the LCA receiver and then using the freshly calibrated LCA transmitter for the O/E measurement. The User Guide does describe verification procedures that you can make without special equipment.

The principles of the GUM are used in the determinations of uncertainty. A coverage factor of 2 is used. There are contributions from both Type A and Type B inputs. For example, using the specifications of an artefact would be Type B. The system measurement result distribution is Type A.

One more thing to note is that the specifications are not the same as the uncertainty result. The specifications have additional margins to allow for aging and drift as well. (And should give reasonable production yield.)  Your calibration certification data that come with the shipment give the actual day-of-calibration results. This refers  particularly to the guaranteed specifications. There are also the typical specifications that give further information about the usually expected performance level in measuring other parameters or under other conditions beyond those that are guaranteed. Some of these are checked in the production process but not usually rechecked during periodic recalibration.

Best regards,
Mike

P.S. We may want to move this thread to the forum where the LCA topics are clustered.
Photonic - Characterization of Electro Optical Components

嗨，迈克，是的，请将此主题移至LCA论坛。
我没有看到那个论坛，抱歉。
当你解释相对频率响应和绝对频率响应时，我理解你的意思，但我不理解与GUM标准不确定性定义的关系。
此外，在规范中，您将这两个术语定义为“相对频率响应不确定性”和“绝对频率响应不确定性”以及如何定义它们，它们不是不确定性测量。
从这个意义上说，在我看来，“频率响应可重复性”可能是GUM中解释的标准不确定性。
可能吗？
非常感谢你的支持，迈克。
我想准确理解规范术语的含义，以便了解我的测量结果。
亲切的问候曼努埃尔



     以下为原文

  Hi Mike,

Yes please, move this thread to the LCA forum. I didn't see that forum, sorry about that.

I understand what you mean when you explain the relative frequency response and the absolute one but I don't understand the relationship with definition of standard uncertainty of the GUM. Moreover, in the specifications you define these two terms as "Relative frequency response uncertainty" and  "Absolute frequency response uncertainty" and how you are defining them, they aren't uncertainty measurements.  

In this sense, It seems to me that the "frequency response repeatability" could be the standard uncertainty that is explained in GUM. Could it be?

Thank you so much for your support, Mike. I would like to understand exactly the meaning of the terms of the specifications in order to understand results of my measurements.

Kind regards
Manuel

嗨曼努埃尔，也许这个描述有帮助。
该仪器用于例如测量设备（例如光电探测器）的绝对频率响应。
该测量的结果具有一些不确定性，这取决于几个因素，例如仪器中光学转发器的校准，噪声水平等。当我们开发仪器和规格时，我们确定我们可以保证的不确定性水平。
使用GUM中的评估和分析选择条件。
用于验证仪器性能的测试检查测量结果是否在指定的不确定度内是否正确。
因此，测量不是确定不确定性，而是检查测量结果，例如绝对频率响应，满足规范。
根据条件，可重复性可以是噪声水平的直接指示，因此它可以是总结一些统计变化的方式，因此可以是A类型对测量不确定度的贡献。
诸如参考伪像的校准不确定性之类的其他贡献通常会增加这一点，使得测量参数的不确定性高于重复性分布。
反过来，如果您需要根据这些测量得出结果的不确定性，您可以使用仪器测量的参数的指定不确定度作为B类贡献。
最好的问候，迈克



     以下为原文

  Hi Manuel,

Maybe this description helps.

The instrument is used for example to measure the absolute frequency response of a device, such as a photodetector. The result of that measurement has some uncertainty that depends on several factors, like the calibration of the optical transponder in the instrument, the noise level, etc.
When we develop our instruments and specifications, we determine what level of uncertainty we can guarantee under the chosen conditions using evaluations and analysis like in the GUM. The tests used to verify the performance of the instrument check that the measurement result is correct within the specified uncertainty. So that measurement is not determining the uncertainty, but checking that the measurement result, such as absolute frequency response, satisfies the specification.
The repeatability, depending on the conditions, can be a straight-forward indication of noise level, so it can be a way to summarize some statistical variations and thus Type A type contributions to the measurement uncertainty. Other contributions like the calibration uncertainty of the reference artefacts usually add to this so that the uncertainty of a measured parameter is higher than the repeatability distribution.

In turn, you can use the specified uncertainty of parameters measured by the instrument  as Type B contributions if you need to derive uncertainty for results based in part on these measurements.

Best regards,
Mike

你好Mike，感谢您的评论。
我想我明白你的意思。
另一方面，我在N4373C的数据表中看到了令我惊讶的一件事。
在规范的“测量条件”中说：“在反向耦合器配置中配置的网络分析器的端口2（4，用于选项314）（”RCVB B in“到”CPLR THRU“，”SOURCE OUT“到”CPLR ARM“）
”。
这些连接位于PNAX的前面板上，默认情况下，PNAX连接到“CPLR ARM”，“SOURCE OUT”连接到“CPLR THRU”。
我应该按照规范说明修改这种连接吗？
我们一直使用默认连接的仪器。
亲切的问候曼努埃尔



     以下为原文

  Hello Mike,

Thanks for your comments. I think I understand what you mean.

On the other hand, I've seen in datasheet of N4373C one thing that surprises me. In "Measurement conditions" of specifications says that:
"Port 2 (4, for option 314) of network analyzer configured in reverse coupler configuration (“RCVB B in” to “CPLR THRU”, “SOURCE 
OUT” to “CPLR ARM”)". 
These connections are in the front panel of the PNAX and by default PNAX came with "RCVR B IN" connected to "CPLR ARM" and "SOURCE OUT" connected to "CPLR THRU". Should I modify this connections as specifications says? We've always used the instrument with the default connections.

Kind regards
Manuel

曼努埃尔你好，我很抱歉反应缓慢，但一直在旅行。
事实上，如果您本周碰巧在OFC，您可以在我们的展位上停下来展示LCA，并直接与我们联系。
端口2（或4）PNA耦合器的标准配置如您所述配置，以增加LCA光接收器的信号电平（并且由于不需要此连接的S22反射）。
因此，它改善了测量的动态范围并降低了噪音。
您可能希望使用此功能，尤其是使用LCA接收器的EO测量。
但是，当校准也是如此时，这两种配置在噪声限制内都是有效且准确的。
最好的问候，迈克



     以下为原文

  Hello Manuel,
I'm sorry for the slow response but have been traveling. In fact, if you happen to be at OFC this week, you could stop by our booth where an LCA is being shown and talk to us directly about it.
The standard configuration of the Port 2 (or 4) PNA coupler is configured as you described to increase the signal level from the optical receiver of the LCA (and since the S22 reflection of this connection is not needed). So it improves the dynamic range of the measurement and reduces noise. You may want to use this, particularly for the EO measurements that use the LCA receiver. But both configurations are valid and accurate within the noise limits when the calibrations are made that way too.

Best regards,
Mike