Hello,

I read the following online document and have a question about the physical offset length from reference plane.

http://cp.literature.agilent.com/litweb/pdf/5989-4840EN.pdf

I want to engineering coaxial solt calibration standards and I don't have an idea how to define the physical offset length from reference plane of the short and open. My frequency span is from 10 MHz to about 20 GHz and I want an return loss of about 40 dB in average over the span. It is posibble to calculate my length with the following formula:

S11=-e^(-j*4*pi*l/lambda) with lambda= c/fcenter

If I calculate, I get a length of about 8.8 mm. Is it right how I am thinking? Or has anyone a hint to me how I can calculate the length l?

I would be grateful, if someone could help me.

I read the following online document and have a question about the physical offset length from reference plane.

http://cp.literature.agilent.com/litweb/pdf/5989-4840EN.pdf

I want to engineering coaxial solt calibration standards and I don't have an idea how to define the physical offset length from reference plane of the short and open. My frequency span is from 10 MHz to about 20 GHz and I want an return loss of about 40 dB in average over the span. It is posibble to calculate my length with the following formula:

S11=-e^(-j*4*pi*l/lambda) with lambda= c/fcenter

If I calculate, I get a length of about 8.8 mm. Is it right how I am thinking? Or has anyone a hint to me how I can calculate the length l?

I would be grateful, if someone could help me.

> Hello,

> I read the following online document and have a question about the physical offset length from reference plane.

>

> http://cp.literature.agilent.com/litweb/pdf/5989-4840EN.pdf

>

> I want to engineering coaxial solt calibration standards and I don't have an idea how to define the physical offset length from reference plane of the short and open. My frequency span is from 10 MHz to about 20 GHz and I want an return loss of about 40 dB in average over the span. It is posibble to calculate my length with the following formula:

The exact offset length is not critical. The main thing is that the offsets of the open and short should be close - ideally the short has a little longer offset distance than the open. But unless you can characterize them, then they are going to be useless to you. It is no trivial task to characterise them.

Despite all the web pages that tell you how to make your own calibration kits, the truth is that it is not easy, and especially at 20 GHz. My own company produces low-cost calibration kits

http://www.kirkbymicrowave.co.uk/

but we do nothing that would achieve the sort of specifications you are asking for. I don't know what use an "average" return loss is to you. Most people want a worst case in some frequency range, and you are not going to achieve 40 dB at 20 GHz with home-made standards. One would need sliding loads for that.

I just checked the manual for the Keysight 85052B 3.5 mm cal kit. The loads in there have a return loss of at least 36 dB between 8 and 20 GHz, falling to 34 dB between 20 and 26.5 GHz, so not even a Keysight kit will give you fixed loads with a 40 dB return loss.

Seriously, if you want to work at 20 GHz, but a new cal kit from Keysight, Rosenberger, Anritsu, Rohde & Schwarz or Maury Microwave. If you have a HP/Agilent/Keysight VNA, then it makes far more sense to buy a kit from Keysight.

Since you are working to 20 GHz, I assume you are using 3.5 mm. The 85052D is a reasonably priced kit for that frequency, and the 85052B is better, since it has sliding loads.

You will never be able to make your own kit that is useful to 20 GHz.

Dave

thanks for your detailed answer.

We should forget the old specifications of 20 GHz , I also read a manual of a kit from keysight. :D

Let`s say the max. frequency is about 6 GHz of the coaxial standard, this would be more realistic. I think it is very difficult to define the offsets, because you can't find any practical description how to do it.

For example how to define the offsets and fringing capacitances of the open?

Is it only possible with a lot of experience and a professional simulations software like ads or cst?

I don't have any idea how to begin with it. It would be so great, if someone could give me a approach or some hints.

Sorry I am a beginner in this topic, but everyone starts small. :)

regards

1. Characterization and verification of coaxial open-circuit primary standards for millimeter-wave vector network analyzer calibration

2. Characterizing Artefact Standards for Use with Coaxial VNAs at Millimeter-wave frequencies

3. Characterization of Calibration Standards by Physical Measurements

4. Characterizing Calibration Standards Using one airline as a transfer standard

5. Specifying Calibration Standards for the Agilent 8510 Network Analyzer (Keysight P/N 5956-4352)

I found the first one on that list (#1) to be the most thorough and informative. In attempting to characterize a coaxial open-circuit:

An Air Gauging Measurement System and a three-dimensional Coordinate Measuring Machine (CMM) are used to measure the diameter of the center conductor (a) and outer conductor (b). A Laser Displacement Measurement System (LDMS) is used to measure offset length of center conductor (l). A VNA calibrated using a TRL cal kit is used to produce S11 data on the open (Gamma). This is all the data you need (a, b, l, Gamma). From this data, you use formulas to find the attenuation constant (alpha), resistivity (p), characteristic impedance (Z), and propagation constant (y). You then use this data to modify the Gamma data to represent Gamma at the end of the offset instead of Gamma at the mating plane. With this new Gamma data, you have have the correct phase data for the coaxial open-circuit. You use a formula to convert frequency dependent phase data into frequency dependent capacitance data. Then you put this frequency dependent capacitance data into a program that can do a third order polynomial regression data fit and you get the coveted capacitance coefficients.

The second document listed (#2) has some more insight. Instead of making all these difficult measurements using equipment you don't have, the diameters (a, b), are simply assumed to be standard dimensions according to the IEEE standard. And the length of the open-circuit offset? They just take it apart and measured the length! At this point you have all the mechanical dimensions you need (a, b, l). Proceed as mentioned above.

The third document (#3), is the least useful, but it at least tells you one thing: with Agilent HFSS software, you can plug in the mechanical dimensions and it will simulate everything and do all the math for you, giving you the capacitance coefficients.

The fourth document (#4) vaguely describes a different method to determining the capacitance coefficients. You calibrate a VNA using a TRL cal kit. Then you place a long beadless airline on the VNA port, and then connect the coaxial open-circuit to the end of that. The resulting S11 linear magnutide data is a ripple instead of the straight line you might expect from an open-circuit. This is a known method of revealing residual source match error. At this point you are supposed to adjust/tweak the generic capacitance coefficients until you get the flattest trace you can get, or at least reduce the amplitude of the ripple. How this is done is not described in detail. but if you could do it, it would mean the residual source match error is being better corrected and therefore the capacitance coefficients are being optimized.

Lastly, document #5 is readily available on the Internet. Page 10 summarizes three methods to compute the capacitance coefficients. They are all fundamentally the same; you use a VNA to find the S11 data for the open-circuit. The trick is compensating for the offset length. This document brings up something none of the other documents do; using time-domain gating to compensate for the offset length. Of course, the document doesn't go into detail on how to do this. However, this technique is nothing new and you should be able to find information out there. Time domain is usually an option for HP\Agilent\Keysight VNAs, so don't be surprised if your VNA can't do it.

This is only from what I've read. I've only recently started researching this stuff, so i have no hands on experience.

Good luck.

Edited by: cmains on Apr 4, 2016 12:26 AM