通过PNA 5242A校准涉及50 db衰减器的路径

作为一般规则，无论使用何种calkits，您都无法通过50 dB衰减器执行完整的2端口校准。
如果您必须保留50 dB衰减器，那么您的选项仅限于简单的响应校正或增强响应校正，该校正使用全矢量校正来校正输入失配，但使用响应校正来校正正向传输。
它不提供经过校正的S22或S12测量值。
具有53 dBm峰值功率输出且PNA-X的损坏水平约为+30 dBm（取决于您的选项），您需要的最小外部填充为30 dB，即使如此，也不可能实现良好的2端口校准
。
如果要进行完整的2端口校准，则必须在端口2上设置高功率外部耦合配置。有关使用VNA进行高功率测量的详细信息，您可以阅读此AppNote，您也可以参考第6.4节。
Joel博士的书“微波元件测量手册”2



     以下为原文

  as a general rule, you cannot perform a full 2-port calibration through a 50 dB attenuator, regardless of the calkits being used.  if you have to keep the 50 dB attenuator, then your options are limited either to a simple Response correction or an Enhanced Response correction that corrects for the input mismatch using full vector correction, but corrects forward transmission using a response correction.  it does not provide a corrected S22 or S12 measurements.

with 53 dBm peak power output and the damage level of the PNA-X being around +30 dBm (depending on your options), the minimum external padding you need is 30 dB and even with that, a good 2-port calibration is not possible.  If you want to do a full 2-port calibration, you have to setup a high power external couple configuration on port 2.  

for detailed information on high power measurements with a VNA, you can read this AppNote 

you can also refer to section 6.4.2 of Dr. Joel's book Handbook of Microwave Component Measurements

嗨Daras，谢谢你的回复。
你说的是ERC，但我的校准套件是N4433A，它的所有四个端口都有3.5毫米连接器，我的DUT中有一个N型（F）和一个SMA（F）连接器。
这意味着我无法使用（定义通过）选项进行ERC。但是我的N型端口校准有一个N85032F机械校准套件。
我卡住了不知道该怎么办。
因为我必须测量我的DUT的相位特性，这意味着我将不得不通过制作高功率电路进行2完全端口校准，或者是否有办法解决。
注意：我的DUT没有直接连接到仪器，其间有一个开关矩阵，迫使我考虑一个2满端口校准。:(另外一个问题Daras我做了2个全端口校准之后看到的是
当两个端口的LED开始发光时，端口开始产生RF信号。如果我想仅测量S21（相位，增益），这会损害我的读数/ DUT /开关。理想情况下，如果我想进行S21测量，只有1
端口应该发光，2端口应该是接收器，不应该发光..正确???



     以下为原文

  Hi Daras,

Thanks for your reply. You are talking about ERC but the cal kit which I have is N4433A which has 3.5 mm connectors on all of its four ports and I have a N-type(F) and a SMA(F) connector in my DUT. So that means I wont be able to do a ERC with (defined thru) option.But I do have a N85032F mechanical cal kit for my N type port calibration. I am stuck don't know how to proceed. As I have to measure phase characteristics also of my DUT does that mean that I will have to do a 2-full port calibration by making a high power circuit or is there a way around.
Note: My DUT is not directly connected to instrument, there is a switch matrix in between which is forcing me to think about a 2-full port calibration.:(  
    One more question Daras what i saw after doing a 2 full port calibration is both port starts generating RF signal as the LED for both port starts glowing. Will this harm my Readings/DUT/Switches if suppose I want to measure only S21 (both phase,gain). Ideally if I want to make a S21 measurement, only 1st port should glow and 2 port should be the receiver and should not glow..right???

> {quote：title = Rohit写道：} {quote}>你说的是ERC，但我的校准套件是N4433A，它的所有四个端口都有3.5 mm连接器，我有一个> N型（F）
和我的DUT中的SMA（F）连接器。
这意味着我无法使用（定义通过）选项进行ERC。但是我的N型端口校准有一个> N85032F机械校准套件。
我卡住了不知道该怎么办。
您对ERC需要定义的直通是正确的，但您仍然可以使用您拥有的校准硬件来管理它。
您可以将3.5 mm至N型适配器连接到ECal上的其中一个端口，并在ECal中创建“3.5 mm f-N-type f”用户特性，然后使用它来执行ERC，就像您有一个ERC一样
3.5-N ECal。
> {quote：title = Rohit写道：} {quote}> ...因为我必须测量我的DUT的相位特性，这意味着我必须通过制作>高功率电路来进行2全端口校准
或者有办法解决。
如果您无法使用我上面提到的用户表征方法，那么您唯一的另一种选择是设置高功率外部测试集。
对于较小的损耗，去嵌入高功率焊盘是一种选择，但即使该技术对于50 dB焊盘也是不可行的（良好的去嵌入候选的限制是大约10-15 dB的损耗）。
>还有一个问题Daras在进行2次全端口校准后看到的是端口开始产生RF信号，因为两个端口的LED开始发光。
如果我想仅测量S21（相位，增益），这会损害我的读数/ DUT /开关。
理想情况下>如果我想进行S21测量，只有第一个端口应该发光，2个端口应该是接收器，不应该发光。
测量可能足够快，您可以感知两个端口同时打开（或者您可能打开点扫描模式），但除非您已将端口1和2的端口状态专门更改为ON（而不是在Auto上）
），这两个来源不会同时开启。
事实上，在N5242A盒中，由于散热原因，我们甚至不允许您同时手动打开端口1和2。



     以下为原文

  > {quote:title=Rohit wrote:}{quote}
> You are talking about ERC but the cal kit which I have is N4433A which has 3.5 mm connectors on all of its four ports and I have a >N-type(F) and a SMA(F) connector in my DUT. So that means I wont be able to do a ERC with (defined thru) option.But I do have a >N85032F mechanical cal kit for my N type port calibration. I am stuck don't know how to proceed.

You are correct about ERC needing a defined thru, but you can still manage that with the calibration hardware that you have.  you can attach an 3.5 mm to N-Type adapter to one of the ports on your ECal and create a "3.5 mm f- N-type f " user characterization in your ECal and then use that to perform an ERC as if you had a 3.5-N ECal.

> {quote:title=Rohit wrote:}{quote}
>...As I have to measure phase characteristics also of my DUT does that mean that I will have to do a 2-full port calibration by making >a high power circuit or is there a way around.

If you are unable to use the user characterization method I mentioned above, then your only other alternative will be to setup a high power external testset.  with smaller losses, de-embedding the high power pad would be an option but even that technique is not feasible for a 50 dB pad (the limit for a good de-embedding candidate is about 10-15 dB of loss).


>     One more question Daras what i saw after doing a 2 full port calibration is both port starts generating RF signal as the LED for >both port starts glowing. Will this harm my Readings/DUT/Switches if suppose I want to measure only S21 (both phase,gain). Ideally >if I want to make a S21 measurement, only 1st port should glow and 2 port should be the receiver and should not glow..right???

The measurements might be fast enough that you perceive both ports being on at the same time (or you might have point sweep mode turned on), but unless you have specifically changed the port status for ports 1 and 2 to be ON (instead on Auto), the two sources will not be on at the same time.  In fact in a N5242A box we don't even allow you to manually turn both ports 1 and 2 on at the same time due to thermal reasons.

感谢Daras，我将检查所有这些设置并告诉您结果。
最诚挚的问候，Rohit。



     以下为原文

  Thanks Daras,


I will check all this settings and let you know about the result.


Best Regards,
Rohit.

嗨Daras，我按照你说的那样通过表征适配器来做ERC，这个过程在没有任何衰减器的路径上工作正常，但是在我有50dB衰减器的路径上开始给我一个错误信息。
我正在附上PNA校准步骤的屏幕截图。如果您可以节省一些时间并浏览文档并指导我在校准步骤中做错了，那将会很棒。
谢谢＆amp;
此致，罗希特。
4Port_SOLT_Cal.docx6.4 MBECal_UserCharacterization.docx3.3校准时MBError.docx265.1 KB



     以下为原文

  Hi Daras,

I did the ERC by characterizing the Adapter as you told, the process works fine in path without any attenuators but starts giving me a error message on the path in which I have the 50dB attenuator.
               I am attaching the screen shots of the PNA calibration steps.It will be great if you can spare some time and go through the documents and guide me what's wrong that I am doing in my calibration step.


Thanks & Regards,
Rohit.

附件

• Create_User_Characterize_ECal.docx5.4 MB
  
• 4Port_SOLT_Cal.docx6.4 MB
  
• ECal_UserCharacterization.docx3.3 MB
  
• Error while calibration.docx265.1 KB
  

你必须关闭ECal自动定位。
使用端口2上的50 dB焊盘，端口2没有足够的电源来正确运行ECal的自动定向程序。
在高级设置页面（ERC校准文档的第3页）中，您将找到自动方向复选框。
将其关闭然后在后续页面中，告诉cal向导ecal的哪个端口连接到PNA的哪个端口。
应解决错误情况。



     以下为原文

  you have to turn off ECal Auto Orientation.  with the 50 dB pad on port 2, there will not be enough power from port 2 to properly run the Auto Orientation routine for the ECal.  in the advanced settings page (page 3 of your ERC calibration document), you'll find the auto orientation check box.  turn that off and then in the subsequent page, tell the cal wizard which port of the ecal is connected to which port of the PNA.  that should resolve the error condition.

嗨Daras，我尝试了你所指的选项，但有了这个选项，我没有得到我的DUT的增益和相位的所需输出，我的读数失真，也失去了适配器，如0.1 - 0.2
dB显示为1.1.-1.2 dB，我相信这种失真是因为50dB衰减器导致的低功耗，你对此有何看法。
这是ERC最好的读数吗？
我在0dBm校准我的系统，如果我将功率增加到10或15 dBm。应该有帮助吗？



     以下为原文

  Hi Daras,

I tried the option which you are referring, but with that option I am not getting the desired output of Gain and Phase of my DUT, I am getting distortions in my reading and also the loss of adapter which is like 0.1 - 0.2 dB is showing 1.1.-1.2 dB,I believe this distortions are because of low power which I am getting due to the 50dB attenuator,what's your take on this. Is this the best reading which we can get by doing this ERC???


I am calibrating my system at 0dBm should I increase the power to 10 or 15 dBm.Should that help??

在0 dBm时，您将实际压缩ECal。
大多数ECal型号压缩到-3 dBm以上。
尝试将通道功率设置为-5 dBm以进行校准。



     以下为原文

  at 0 dBm, you will actually compress the ECal.  most ECal models will compress above -3 dBm.  try setting the channel power to -5 dBm for your calibration.

在0 dBm时，您将实际压缩ECal。
大多数ECal型号压缩到-3 dBm以上。
尝试将通道功率设置为-5 dBm以进行校准。



     以下为原文

  at 0 dBm, you will actually compress the ECal.  most ECal models will compress above -3 dBm.  try setting the channel power to -5 dBm for your calibration.

在0 dBm时，您最有可能压缩ECal。
大多数ECal模型压缩功率高于-3 dBm。
尝试将通道功率设置为t0 -5 dBm进行校准，并可能发布一些结果（S11，S21），以便我们可以更好地了解您的测量结果。



     以下为原文

  at 0 dBm, you are most likely compressing the ECal.  most ECal models compress at powers above -3 dBm.  try setting the channel power t0 -5 dBm for calibration and perhaps post some results (S11, S21) so that we could have a better idea of what you are measuring.

在0 dBm时，您最有可能压缩ECal。
大多数ECal模型压缩功率高于-3 dBm。
尝试将通道功率设置为t0 -5 dBm进行校准，并可能发布一些结果（S11，S21），以便我们可以更好地了解您的测量结果。



     以下为原文

  at 0 dBm, you are most likely compressing the ECal.  most ECal models compress at powers above -3 dBm.  try setting the channel power t0 -5 dBm for calibration and perhaps post some results (S11, S21) so that we could have a better idea of what you are measuring.

Daras提到ecal压缩到-5 dBm以上，但是在你超过+5之前它是一个非常小的误差。
你能发布你的直通适配器结果的.png图吗？
如果它有噪音，那很可能是因为噪音！
（杜）;
如果它的纹波不随噪声而变化，那就是校准错误。
但是，让我们试着让事情变得非常简单，然后我们可以在出现错误时理解错误（人们总是首先尝试跳到最复杂的案例，但保持简单会让事情变得更容易理解）。
首先，连接你的直通适配器，做data-> mem和data / mem;
跟踪如何看待。
如果噪音太大则误差与噪声有关。
降低IF BW或添加平均值。
10倍的变化都会使痕量噪声发生3倍的变化（详细描述......是的，你猜对了，我的书）。
如果这看起来干净，那么进行1端口校准，然后看一下开路响应（它是否平滑且漂亮），然后连接到50 dB衰减器;
这是一场很好的比赛吗？
如果没有，那么ERC将是一个问题，因为它不能在测量期间纠正错误的负载匹配（尽管如此）。
大多数情况下，50 dB衰减器应该非常好（30 dB或更高）。
最后，通过良好的平均和紧密的连接器，ERC与Ecal;
但请注意，如果你的频率超过20 GHz，那么Ecal through不是那么稳定（可能是0.2 dB），所以最好使用一个物理定义的直通。
让我知道你的结果，我们可以从那里拿走它。
（是的，daras也给我重复了六次！）



     以下为原文

  Daras mentioned that ecal compresses above -5 dBm, but it is a very small error until you exceed about +5.

Can you post a .png plot of your through adapter result.  If it is noisy, then it is likely due to noise! (duh); if it has ripple that does not change with noise, it is a calibration error.

But, let's try to make things really simple, then we can understand errors as they come up (people always try to jump to the most complicated case first, but keeping it simple makes things much easier to understand).

First, connect your thru adapter, do data->mem and data/mem; how does the trace look.  If it is too noisey then the error is noise related.  Lower the IF BW or add averaging.  10x change in either gives 3x change in trace noise (this is described in detail in...yes, you guessed it, my book).

If this looks clean, then do a 1 port calibration, and look a the open response (is it smooth and nice) and then connect through to the 50 dB attenuator; is it a good match?  If not, then ERC will be a problem as it does not correct for bad load  match during the measurement (it does during the cal, though).  Mostly, a 50 dB attenuator should be quite good (30 dB or more).
 
Finally, with good averaging and tight connectors, do the ERC with the Ecal; but note, if you are going over 20 GHz, the Ecal through is not so stable (maybe 0.2 dB) so it is better to use a phyisical defined thru.

Let me know your results and we can take it from there.

(and yes, daras repeats things to me half a dozen times too!)

> {quote：title = Dr_joel写道：} {quote}>（是的，daras也向我重复了六次！）我只做了绝对必要的事情:-)



     以下为原文

  > {quote:title=Dr_joel wrote:}{quote}
> (and yes, daras repeats things to me half a dozen times too!)

I only do what is absolutely necessary :-)

你好Daras＆amp;
Joel博士，感谢您的回应和耐心，帮助我理清这个问题。
如果您想要通过响应的屏幕短路，请在附件中找到它们。我已经在各种IF带宽中包含了有无衰减器（50 dB）通过响应的屏幕。
最诚挚的问候，Rohit。



     以下为原文

  Hello Daras & Dr. Joel,


Thanks for your response and patience in helping me sort out this issue. As you wanted some screen shorts of the thru response,find those in the attachment.I have included screens of thru response with and without attenuator.(50 dB) at various IF bandwidth.


Best Regards,
Rohit.

附件

• Path Calibration having 50 dB PAD in Tx_Mode.docx166.4 KB
  

嗨Daras和Joel博士，上周我已经附上了使用和不使用衰减器的直通响应的屏幕截图。如果您有时间通过​​这些屏幕截图并帮助我摆脱这个问题，那将会很棒。等待您的回复
。:) Regrds，Rohit。



     以下为原文

  Hi Daras and Dr. Joel,


Last week I have attached the screen shots of the thru response with and without attenuator.It will be great if you spare some time to go through those screen shots and help me out of this issue.Awaitin your response.:)


Regrds,
Rohit.

我看了一下你的数字并且它们对我来说看起来不那么好：1）没有50 dB打击垫的S21的直通响应看起来不正确。
我看到S21 = .6到1.2 dB，它应该是0 dB。
这是一个非常大的错误。
ecal用户表征不好或校准效果不佳。
对于图1，你应该得到平坦的回应。
我建议在很宽的频率范围（比如50 MHz到6 GHz）上重复测试，看看你是否有系统的纹波;
如果是这样，它验证了cal的基本问题。
2）当您在校准后更改IF BW时，50 dB衰减器的噪声响应具有相同的外观，因此我认为您在校准期间使用过宽的IF BW来“校准”噪声。
在校准期间使用1kHz IF BW和10个平均值;
并重复结果。
然后将平均值提高到100;
找到可以产生良好结果的价值。
3）对于您看到的错误，错误可能是由于错误的Ecal Characterization造成的。
我建议不要使用ecal特征描述，然后向PNA说谎，说明端口是可插入的（一个是f，另一个是男性），选择端口1型连接器。
然后执行ERC并查看结果的外观。
您将使用SMA到N适配器（如果这是您的两种连接器类型）将端口2转换为匹配端口1，并暂时忽略适配器。
看看这看起来如何。
4）如果这看起来不错，那么我们将通过这个来表征它，并将它用于已知的直通（只需知道它的延迟，也许是损失.5）同样，你的垫的回波损失看起来不太好。
对于像你这样的狭窄范围，它应该是平坦的和低的（如-30 dB）。
有涟漪的事实表明校准不好（可能来自ecal特征）。
在宽频带上尝试这个并回复结果。



     以下为原文

  I looked and your figures and they don't look that great to me:

1) the thru response for S21 with no 50 dB pad does not look correct.  I see S21 =.6 to 1.2 dB, it should be 0 dB.  This is a very big error.   Either the ecal user characterization is not good or the calibration was not performed well.

For figure 1 you should get a flat response.  I recommend repeating the test over a wide freq range (say 50 MHz to 6 GHz) to look and see if you have a systematic ripple; if so it verifies a base problem with the cal.

2) The noisy response of the 50 dB attenuator has the same look when you change IF BW after cal, thus I think you are "calibrating in" the noise by using too wide an IF BW during cal.

Use a 1kHz IF BW and perhaps 10 averaging during the cal; and repeat the results.  Then up the averaging to 100; find the value that gives good results.

3) For the errors you see, the errors are probably due to a bad Ecal Characterization.  I would recommend not using an ecal characterization, and starting by lying to the PNA saying that the ports are insertable (one is f, the other male) choosing the port 1 type connector.  Then do the ERC and see how the results look.  You will use an SMA to N adapter (if those are your two connector types) to transform port 2 to match port 1, and ignore the adapter for now.  See how this looks.

4) If this looks good, then we will characterize this thru, and use it for the known thru (just need to know its delay, and maybe loss.

5) also, the return loss of your pad doesn't look good.  For a narrow span like you you have it should be flat and low (like -30 dB).  The fact that there are ripples indicates a bad cal (probably from the ecal characterization).

try this over a wide band and reply with results.

感谢Joel博士，我将通过更新截图回复您。感谢您的帮助和耐心。
问候。



     以下为原文

  Thanks Dr. Joel, I will get back to you with updated screen shots.Thanks for your help and patience. 


Reagrds.

你好Joel博士＆amp;
Daras，找到带有和不带50 dB衰减器的Ecal User特性的新附加屏幕截图，结果看起来好多了。:)请提供您宝贵的反馈。
如果这个校准足够继续我的测试（我必须读取我的DUT的增益和相位特征），或者我仍然应该尝试实现更好的校准。
等待你的宝贵回应:)最诚挚的问候，Rohit。



     以下为原文

  Hello Dr. Joel & Daras,


Find the new attached screen shots of the Ecal User characterization with and without 50 dB attenuator, the results look a lot better.:)  Please provide your valuable feedbacks. Should this calibration be sufficient in order to continue with my test (I have to read both Gain and Phase characterization of my DUT) or I should still try and achieve better calibration.
                    
                  Waiting for your valuable response:)

Best Regards,
Rohit.

附件

• THRU response with and without 50 dB PAD.docx152.7 KB
  

这确实看起来好多了，但你也必须意识到，在这种情况下，通过测量直通与使用T / R模块得到的结果不是很好。
在使用ERC和50 dB衰减器的实验中，您将源功率设置为0 dBm，这意味着在焊盘就位后，端口2测试端口的功率约为-50 dBm。
但是，当你的T / R模块输出功率为53 dBm时，测试端口的功率将在+3 dBm左右。
我认为使用T / R模块进行的测量比使用直通式实验的噪声要小得多。
所以你应该没事:-)



     以下为原文

  this does look a lot better, but you also have to realize that in this case measuring a thru is not a good comparison to what you will get with the T/R module in place.  in your experiment with ERC and the 50 dB pad, you are setting the source power to 0 dBm, which means with the pad in place, the power at port 2 testport is around -50 dBm.  however, when you have your T/R module with 53 dBm output power in place the power at the testport will be around +3 dBm.  I think your measurements with the T/R module in place will be a lot less noisy than your experiment with the thru.  so you should be fine :-)