如何使用公端口和公标准进行8510B分析仪校准

> {quote：title = igouzouasis写道：} {quote}>大家好，>>我目前正在使用HP 8510B网络分析仪，配备85052B 3.5mm校准套件。
>不幸的是，女性（F）负载，F开放和F短缺，所以我留下了男性版本，以及3.5毫米m-m，m-f，f-f适配器，滑动负载和量具。
>> 1）测试端口延伸了~4cm 90度弯角刚性电缆，公头2.92mm（不确定）连接器。
在损耗，校准和相位延迟/噪声方面，“好”是多少？
校准会减少缺点吗？
离开制造版本更好吗？
>> 2）如果我拆下这根4cm电缆，测试端口是3.5mm公头制造。
如何使用男性标准进行溶质校准？
我不是VNA专家，但这是我从一个相对新手到VNA的采取。
你使用的最高频率是多少？
这将决定40毫米电缆的重要性。
请记住，您需要更正电缆的速度系数。
如果你有2.92毫米连接器的设备，它告诉我你使用很多GHz，在这种情况下任何黑客都可能*非常*愚蠢。
考虑到校准套件的成本，我很想购买，乞讨，恐怖或窃取女性开口和短裤。
这是一个昂贵的工具包，由于缺乏一些cal标准而被破坏。
负载不是问题，因为原则上好的适配器不会改变事物。
我不知道该套件中男性和女性的偏移是否相同，就像它们在3.5 mm 85033E套件中一样。
如果是这样，在这种情况下进行校准时，可能不会要求性别，*可能*可能使用“交换等量适配器”方法的变体。
http://na.tm.agilent.com/pna/help/latest/S3_Cals/Swap_Adapters_Method.htm使用它，您可以在公端口或母端口进行校准，该端口位于适配器的末端。
就个人而言，我只是购买标准。
戴夫



     以下为原文

  > {quote:title=igouzouasis wrote:}{quote}
> Hi all,
> 
> I'm currently using an HP 8510B network analyzer, with the 85052B 3.5mm calibration kit.
> Unfortunately, Female (F) Load, F Open and F Short are missing, so i'm left with the male versions, along with the 3.5mm m-m, m-f, f-f adapters, sliding loads and gages.
> 
> 1) Test ports have been extended by a ~4cm 90degrees angled rigid cable, with male 2.92mm (not sure about that) connector. How "good" is that in terms of loss, calibration and phase delay/noise? A calibration will diminish the drawbacks? Is it better to leave the manufacture version?
> 
> 2) If i remove this 4cm cable, the test ports are 3.5mm male by manufacture. How can i perform solt calibration using the male standards?

I'm no VNA expert, but here is my take from a relatively newbie to VNAs. 

What's the maximum frequency you are using? That will dictate how significant a 40 mm section of cable is. Remember you will need to correct for the velocity factor of the cable. If you have equipment with 2.92 mm connectors, it suggests to me you are using many GHz, in which case any hacks are likelly to be *very* silly. 

Given the cost of that calibration kit, I'd be tempted to buy, beg, borror or steal the female opens and shorts. It's an expensive kit to be ruined by the lack of a few cal standards. The load is less of an issue, as in principle a good adapter will not change things. 

I don't know whether the offsets to the male and female are the same in that kit, like they are in the 3.5 mm 85033E kit. If they are, in which case when doing a calibration you will probably not be asked for the sex, it *might* be possible to use a variation of the "swap equal adapters" method. 

http://na.tm.agilent.com/pna/help/latest/S3_Cals/Swap_Adapters_Method.htm

Using that, you could calibrate at a male or female port, where that port is on the end of the adapters. 

Personally though, I'd just buy the standards. 

Dave

如果你想写一堆数学，解决方案很简单：将电缆连接到VNA的端口2，确保它有一个母端口。
使用您的男性标准校准端口2上的母端口。将端口2连接到端口1并使用QSOLT（由都灵大学的Andre Ferrero和我的朋友推导）创建一个2端口校准。
断开电缆，现在您有一个带校准的公测试端口。
使用PNA，它是微不足道的。
使用8510，你可以算一算：A。Ferrero和U. Pisani，“QSOLT用于双端口S参数测量的新型快速校准算法”，第8届ARFTG Cant Dig。，圣地亚哥，加利福尼亚州，12月5日至6日，
1991年



     以下为原文

  If you want to write a bunch of math, the solution is straightforward:

Connect a cable to port 2 of the VNA, ensure it has a female port.
Use your male standards to calibrate the female port on port 2.
Connect port 2 to port 1 and use the QSOLT (derived by Andre Ferrero of Univ of Turin, and a friend of mine) to create a 2-port calibraiton.
Disconnect the cable and you now have a male test port with a calibration.

With a PNA, it's trivial.  With an 8510, you do the math:

  A. Ferrero and U. Pisani, “QSOLT A new fast calibration algorithm for two port S parameter measurements,” in 8th ARFTG Cant Dig., San Diego, CA, Dec. 5-6, 1991.

嗨，我一直在研究Ferrero和Pisani的论文，我有几个问题。
在论文中没有关于校准程序的参考。
这是你帖子中的那个吗？
为什么我必须在最后断开电缆？
通过观察方程，我发现该方法假设源和负载不匹配，如TRL *。
这不会在测量中引入某种错误吗？



     以下为原文

  Hi

I've been working around the paper of Ferrero and Pisani and i have a couple of questions.
In the paper there is no reference about the procedure of calibration. Is this the one in your post? Why do i have to disconnect the cable at the end?
By observing the equations, i see that the method assumes equal source and load mismatch, as in TRL*. Doesn't this introduce some kind of error in the measurements?

QSOLT基于8项错误模型。
与TRL校准一样，需要进行开关误差校正以消除开关误差，以使负载匹配和源匹配相同。
另一种可行的方法是使用适配器交换威廉希尔官方网站 。
85052B校准套件应该有一组匹配的适配器。
：您可以使用带有测试柱的适配器进行校准，以便您可以使用男性标准。
然后用匹配的t t替换适配器和一个公测试端口连接器。
附加校准误差取决于这些适配器在传输和反射特性方面的匹配程度。
在3 GHz，可能是可以接受的。
肯



     以下为原文

  QSOLT is based on the 8-term error model.  Like TRL calibrations, a swith error correction is required to remove the switch error so that load match and source match are the same.

Another way that may work is to use adapter swap technique.  The 85052B Cal kit should have a set of matched adapters.  :You can calibrate with an adapter with a female test post so that you can use your male standards.  Then you swap out the adapter with a match one t with a male test port connector.  The addional calibration error  depends how well these adapters match in transmission and reflection characteristics. At 3 GHz, it may be acceptable.

Ken

Ken的注释是正确的......源和负载匹配的处理并不是必需的，但是需要知道差异（我们称之为delta-match）并补偿它的响应（它被视为
在端口上有一个额外的反射信号。至于你的具体问题：如果适配器是惠普校准套件中常见的适配器，它们是匹配长度的适配器。如果是，你有MF和FF，那么你可以做
以下：制作一个女性测试端口，使用男性标准校准，添加短路并测量相位，保存到存储器。添加MF adpater，添加公头短接，使用端口扩展调整相位以匹配存储器跟踪
（或者只是测量短路的延迟和适配器的延迟加上短路。）现在你有了适配器的长度.FF适配器的长度是相同的。符合女性标准（实际上是3.5毫米，女性和男性标准
具有相同的型号），并将FF adpater的延迟添加到标准定义中
你可以用标准的女性 - 女性 - adpater到osl来校准一个男性测试端口校准将参考正确的平面。
但！
当您使用F-F进行直通测量时，请务必修改直通标准，以使其具有适配器的长度。
这将为你提供30岁以下至40岁以下的剩余比赛，即使你没有付出很多努力就能获得最好的效果，当然比你需要的更好，除非你是计量学家......如果你是一个计量学家，你呢？
d有一个带有所有部件的校准套件！



     以下为原文

  Ken's notes are spot on... The treatement of source and load match as equal isn't required, but what is required is to know the difference (we call it delta-match) and compensate the response for it (it is treated as an additional reflection signal at the ports.

As to your specific problem: if the adapters are the one's often found in HP cal kits, they are matched length adapters.  If so, and you have the M-F and the F-F, then you can do the following:

Make a female test port, calibrate with male standards, add the short back on and measure the phase, save to memory. Add the M-F adpater, add the male short on, use port extension to adjust the phase to match the memory trace (or just measure the delay of the short and the delay of the adapter plus short).  Now you have the adpater length.  The F-F adapter is the same length. 

Go to the female standards (in 3.5 mm actually, the female and male standards have the same model), and add the delay of the F-F adpater to the standard definition.  Now you can calibrate a male test port wtih the standards which are Female-Female-adpater-to-osl

The calibration will be referenced to the proper plane.  BUT! when you do the thru measurement, using the F-F, be sure to modify the thru stnadard so that it also has the lenght of the adapter. 

This will give you mid-30's to low 40's for residual match, about as good as you can get without a lot of effort and certainly better than you need unless you are metrology guy...and if you were a metrology guy, you'd have a cal kit with all the parts!

> {quote：title = Dr_joel写道：} {quote}>校准将参考正确的平面。
但！
当您使用F-F进行直通测量时，请务必修改直通标准，以使其具有适配器的长度。
我是否正确假设他需要输入26.6毫米的长度 - （或ps中的任何东西）？
http://www.home.agilent.com/upload/cmc_upload/All/EPSG086193.html各种HP连接器的参考平面之间的尺寸，包括85052-60012 3.5（f）至3.5（f）85052-60013 3.5
（f）至3.5（m）85052-60014 3.5（m）至3.5（m），这是85052B校准套件中的那些。
我假设列出的26.6毫米是电气长度，而不是机械长度，尽管没有说明。
这可能是一个相当古老的文档，因为它的名称是惠普，而不是安捷伦。
我还假设“M”等级表示计量等级，也许“I”表示工业，但也没有说明。
当然，如果缺少3位的校准套件，如果适配器不是正确的部件，我绝不会感到惊讶。
也许'igouzouasis'应该在Agilent的Find-a-parts上查找部件号并检查适配器看起来是正确的。
戴夫



     以下为原文

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

> The calibration will be referenced to the proper plane.  BUT! when you do the thru measurement, using the F-F, be sure to modify the thru stnadard so that it also has the lenght of the adapter. 

Would I be right in assuming he needs to enter a length of 26.6 mm - (or whatever that is in ps)? 

http://www.home.agilent.com/upload/cmc_upload/All/EPSG086193.html

has dimensions between the reference planes of various HP connectors, including the 

85052-60012  3.5 (f) to 3.5 (f) 
85052-60013  3.5 (f) to 3.5 (m) 
85052-60014  3.5 (m) to 3.5 (m) 

which are the ones in the 85052B cal kit. I assume the 26.6 mm listed are electrical lengths, not mechanical ones, although that is not stated. It is probably quite an old document too, as it has the name HP and not Agilent on it. 

I also assume the grade of "M" indicates metrology grade, and perhaps "I" indicates industrial, but again that is not stated. 

Of course, with a cal kit with 3 bits missing, it would never surprise me if the adapters are not the correct pieces. Perhaps ' igouzouasis' should look up the part number on Agilent's Find-a-part and check the adapter looks to be the right one. 

Dave

通常，校准套件中的适配器是相位匹配的。
85052B套件中的那些是计量等级。
一种说法是查看女性联系人是否是无槽版本。
使用匹配适配器的适配器交换威廉希尔官方网站 ，无需修改延迟等。我可能没有明确它是如何工作的。
您从两个端口上的女性测试端口开始。
其中一个测试端口连接器必须是相位匹配版本。
1）使用公校准标准校准两个端口。
2）将相位匹配版本换成将测试端口转换为公测试端口的版本。
3）通过将端口1连接到端口2来完成2端口校准。



     以下为原文

  Usually, the adapters in the cal kits are phase matched.  The ones in the 85052B kit are metrology grade.  One way to tell is to see if the female contacts are of the slotless version.  

Using the adapter swpping technique with matched adapters, there is no need to modify delays etc.  I may not have made it clear on how it works.  You start out with female test ports on both ports.  One of the test port connector must be the phase matched version. 
1)  Calibrate both ports with male calibration standards.
2)  Swap out the phase match version with one that converts the test port to a male test port.
3)   Finish the 2-port calibration by connecting port 1 to port 2.

谢谢您的回答。
我现在的工作频率是3ghz。
我希望我能找到一种方法来获取pdf，因为我无法访问ieee。
如果我必须永久地将这些4cm电缆留在端口上，我是否必须进行“端口扩展”？
或校准是否补偿延迟/损失？



     以下为原文

  Thank you for your answers.
My operating frequencies will be 3ghz for now.
I hope i'll find a way to get the pdf, cause i have no access to ieee.

In case I have to leave these 4cm cables on the ports for ever, do i have to do a 'port extension' ? Or does the calibration compensate for the delays/losses?

如果你不喜欢数学，我会把你不完整的85052B 26.5 GHz校准套件换成我的完整85033E 9 GHz校准套件 - 假设你的校准套件状况良好。那就是说，如果有人从一个卡上丢失3位
套件，它表明它可能没有得到照顾。
您可以以约30美元的价格购买IEEE论文。
但是，如果您正在使用VNA，那么值得加入IEEE并成为可以访问您可能需要的论文的社团成员。
戴夫



     以下为原文

  If you don't fancy the maths, I'll swap you your incomplete 85052B 26.5 GHz cal kit for my complete 85033E 9 GHz cal kit - assuming your cal kit is in good condition.That said, if someone loses 3 bits from a cal kit, it suggests it has probably not been cared for. 

You can buy IEEE papers for about $30. But if you are working with a VNA, it is worth joining the IEEE and becoming a member of the societies that give you access to the papers you might need. 

Dave

很抱歉有时候回复，我有很多工作人员要做。
但我正在研究这个问题。
好的，所以我看到有两种方法可以做到这一点。
使用Dr_joel的方法，我可以将f-f适配器+男性开放式短载荷组合在一起，并将整体视为男性端口的女性开放式短载荷。
有一点是测量适配器的长度，以便将其添加到开放 - 短载标准的定义中。
85052B套件的定义表（第A-9页）给出了94.75ps的标称长度值。
你可以在这里找到套件手册：http：//engineering.mq.edu.au/research/groups/cnerf/resources/85052-90077.pdf我使用了Joel的方法，vna的值为~95.18 ps。
假设这是自由空间中的值，我可以通过乘以0.99来补偿以获得电气长度。
（我的测量中是否有错误？我猜这取决于我想要的频率）顺便说一句，适配器是无槽的。
第二件事是我应该赔偿损失。
我应该估计适配器+短路，适配器+开路的损耗并且每秒进行一次归一化（假设损耗是电阻性的并且可以简单地添加）？
作为第三件事......我应该测量短路和开路的偏差，还是应该根据定义表默认设置？
kenwong的交换适配器方法可以如下工作。
请注意，我的DUT是一个带端口的双端口设备。
1）我将vna的端口1设置为公，vna的端口2为女性2）使用ff适配器测量osl到端口1. 3）使用mf适配器测量osl到端口2. 4）保留两个适配器
放置，使用mm适配器进行适配器测量。
5）断开所有适配器，将f-m适配器插入端口1，将m-m适配器插入端口2.编辑：igouzouasis于2013年3月13日下午2:37



     以下为原文

  Sorry for replying now and then, I have a lot of staff to do. But i'm working on this.

Ok, so I see there are two ways of doing this.
Using Dr_joel's method, I can have a combination of f-f adapter + the male open-short-load and treat the whole as a female open-short-load for a male port. 

One thing is to measure the length of the adapter, so as to add it to the definition of the open-short-load standard. The definitions table of the 85052B kit (page A-9) gives a value of 94.75ps as nominal length. You may find the kit manual here:
http://engineering.mq.edu.au/research/groups/cnerf/resources/85052-90077.pdf

I used Joel's method and the vna gave a value of ~95.18 ps. Giving the fact that this is the value in free space, i can compensate by multiplying by 0.99 to get the electrical length.  (is there a mistake in my measurement? I guess it depends on the frequencies i want)

By the way, the adapters are slotless.

A second thing is that i should compensate for the loss. Should I estimate the loss for adapter+short, adapter+open and make a normalization per sec (giving the fact that losses are resistive and can be simply added)?

As a third thing.. Should I measure the offset short and open, or should I take it by default according to the definitions table?


The swap adapter method of kenwong could work as follows. 
Note that my DUT is a two-port device with female ports.
1) I set port 1 of vna to be male, port 2 of vna to be female
2) use the f-f adapter to measure osl to port 1.
3) use the m-f adapter to measure osl to port 2.
4) keep both adapters placed, use the m-m adapter to make adapter-thru measurement.
5) disconnect all adapters, plug f-m adapter to port 1, plug m-m adapter to port 2.

Edited by: igouzouasis on Mar 13, 2013 2:37 PM

> {quote：title = igouzouasis写道：} {quote} >>>我使用了Joel的方法，vna的值为~95.18 ps。
假设这是自由空间中的值，我可以通过乘以0.99来补偿以获得电气长度。
（我的测量是否有误？我想这取决于我想要的频率）>>编辑者：igouzouasis于2013年3月13日下午2:37我建议的方法取决于适配器长度相等，但它们不是 -
完全相同的长度。
你看到.43 psec的差异。
这可能是由于两个适应器之间的长度差异，或者可能是这些加速器不是标称的。
让我们检查一些数字：1 ns = 1英尺，100 pse = 1英寸，1 psec = .010 inc，.43 = 4.3 mils。
3.5毫米的标称偏移大约为1-3密耳，因此通过几个适配器，看到4密耳的差异是完全合理的;
欢迎来到真实的世界。
很可能你的测量方法很好。
此外，交换相等的适配器可能与我提出的方法非常接近，因为它也依赖于适配器具有相同的长度。
但请随意尝试并报告;
具体，包括情节。
好问题。
事实上，在PNA上，使用新方法（QSOLT）这些问题甚至更有趣（更难以确定什么是最好的方法）。
`



     以下为原文

  > {quote:title=igouzouasis wrote:}{quote}
> > 
> I used Joel's method and the vna gave a value of ~95.18 ps. Giving the fact that this is the value in free space, i can compensate by multiplying by 0.99 to get the electrical length.  (is there a mistake in my measurement? I guess it depends on the frequencies i want)
> 
> Edited by: igouzouasis on Mar 13, 2013 2:37 PM
The method I propose depends upon adapters being equal length, but they are not -exactly- the same lenght. You see a .43 psec difference.  This could be due to a length difference between the two adapaters, or it could be that these adatpers are not nominal.  Let's check some numbers:

1 ns = 1 foot,  100 pse= 1inch, 1 psec =.010 inc, .43 = 4.3 mils.  The nominal offset in 3.5 mm is on the order of 1-3 mils, so through several adapters, it is entirely reasonable to see 4 mils of difference; welcome to the real world. Very likely your measuerement methods are just fine.

Further, swap equal adapters is probably very closely aligned with the method I proposed, because it also relies on the adapters being of equal lenght. But feel free to try it and report back; be specific, include plots.

Good question.  On the PNA, in fact, there using new methods (QSOLT) these issues are even more interesting to explore (and more difficult to determine what's the best method).  `