Microwave Journal
www.microwavejournal.com/articles/4152-question-of-the-month-answers-january-2007

Question of the Month Answers, January 2007

January 8, 2007

Published January 19, 2007

You be the expert: The responses to the December Question of the Month are below

Shanthi B. from Commercial Cellphone Makers has submitted this month’s question:

Dear Harlan,
Why is the open/short/thru method the most preferred calibration method for RF measurements?

From: Heinrich Zweifel, Fela Management AG

Dear Shanthi,
It is relatively easy to do and you can get rid of the unknown parameter of the measurement equipment or measurement setup so that you can really measure the parameters of your DUT.

From: Gopalan (Sam) Sampath, DoD Australia

Dear Shanthi,
1.) Open and short test conditions represent the extremes of terminations. Along with the thru termination, the impedance effects due to the cable length in the set up are assessed and the inbuilt algorithm computes the necessary compensation, which is applied in the calibration process. Thus any error in measurement due to the electrical length of the cables is cancelled out. 2.) Open and short conditions are standardized conditions that can be easily realized in practice cost-effectively thru specific termination hardware supplied as calibration aids.

From: Yueyan Shan, SPRING Singapore

Dear Shanthi,
One of the reasons is its simplicity. Open and short are two full reflection cases with phase opposite. The thru is just connecting two ports together. A more accurate saying is that this is the easiest calibration method, but may not provide the highest accuracy.

From: Andy Street, M/A-COM Inc.

Dear Shanthi,
The short/open/load/thru (SOLT) calibration has proved very popular when performing coaxial-based measurements, but is by no means the 'most preferred' VNA calibration method for all frequencies and applications. Ignoring the isolation terms, a VNA can be considered to have five systematic error terms in the forward direction, comprising of three reflection terms (directivity, source match and reflection tracking), transmission tracking and load match. A similar model exists for the reverse direction. In SOLT, three known reflection terms (for example, SOL) are used to solve for the three systematic reflection terms whilst the through is used to assess the transmission tracking and load match terms. SOLT-based coaxial calibration is popular since it is relatively straightforward to produce robust and repeatable coaxial calibration standards that operate over a wide range of frequencies. Further, the calibration standards offer good impedance coverage (two around the edge of a Smith Chart and one in the centre). Coaxial calibration kits are readily available to 50 GHz using SOLT-based standards, some of which are augmented with sliding loads to improve the directivity correction of the VNA at higher frequencies. However, for waveguide applications, it is difficult to produce a well-defined open circuit. In this case, it is common to use a flush short, a short offset by quarter of a wavelength and a load for the SOL part of the calibration, when using SOLT. For waveguide and on-wafer VNA calibration, TRL (thru-reflect-line) and its variants have proved popular and effective. In this instance, the error model is re-formulated into an error box on either side of the DUT, which in its simplest form is a two-port cascade of the LHS error box, the DUT and the RHS error box. The beauty of this approach is that very little needs to be known about the calibration standards. The thru and the line standards need to have the same Zo and the line must be standard and must provide an additional 20-160 degrees insertion phase when compared with the thru, over the frequency range of interest. The phase response of the reflect standard only needs to be known to within +/-90 deg. Frequency range limitations can be overcome using multiple line standards for different ranges. At low frequencies, however, line lengths become impractically long and the line standards can be replaced with high quality fixed loads. Excellent results can be obtained in waveguide, where it is easy to produce high quality lengths of line. However, TRL-based algorithms require four sampler test sets (to allow the 'ratioing out' of the transfer switch), as opposed to the lower cost three sampler test sets. A good discussion of general VNA calibration techniques and TRL calibration can be found in Agilent App Notes AN1287-3 and AN8510-8A, respectively.

From: John Curtis, Tandberg Television

Dear Shanthi,
The short/open/load/thru (SOLT) method is not the most preferred calibration method in all situations, but does have many advantages, especially in coaxial systems. The main advantage is that it can be wideband - the calibration standards (Short/Open/Load/Thru) can be designed to work from DC up to just below the frequency where waveguide modes occur in the coaxial system being used, with good accuracy throughout the entire range. A single calibration kit can cover this entire range, and with an appropriate vector network analyzer (VNA), full two-port calibrated measurements can be made over a very wide bandwidth. Other calibration methods (TRL, for example) tend to be relatively narrowband, and their accuracy goes down at the upper and lower frequency extremes of their range. A calibration kit does not need to have standards that are as close to their respective ideal impedances as possible; instead the impedance of the standards needs to be accurately known so that their imperfections can be compensated for by the 'models' of the standards used by a VNA. These models incorporate the parasitics, such as the fringing capacitance of the open and its frequency dependence. For coax, these models can be relatively simple with few parameters. Many years ago when they needed to be entered manually, having few parameters resulted in fast and less error-prone entry into the VNA - another point in favor of SOLT calibration. Cal kits in different coax standards (for example, N-type, 7 mm, 3.5 mm) can use the same set of parameters, but with different parameter values that are provided by the manufacturer. Some other calibration methods use standards and transmission lines that are not fully shielded, so the effects of radiation, pickup and the proximity of nearby objects affect accuracy. The standards for a SOLT coaxial cal kit can be fully shielded (even the open) and are not affected by these problems. Coaxial SOLT standards are very stable mechanically and can be made to very tight tolerance, so their parameters vary little after disconnecting and re-connecting. The use of air dielectric removes much uncertainty and temperature effects of the dielectric, and makes the loss of the transmission lines within the standards very predictable. All of this contributes to accuracy.

From: Saul Hernandez, Delphi

Dear Shanthi,
This method gives us reference points with known phases and 180 degrees between open-short, allowing ease of adjustment. The standards used are low cost.

From: Arun Kumar, Sameer Kolkata Centre

Dear Shanthi,
Imperfections exist in even the finest test equipment and, if uncorrected, these imperfections (termed as systematic errors, random errors and drift errors) will cause the equipment to yield less than ideal measurements. The basis of network analyzer error correction is the measurement of known electrical standards, such as a thru, open circuit, short circuit and precision load impedance. By calibrating a network analyzer with these standards, one can compensate for its inherent errors (mainly systematic errors). Calibration procedures include the popular Short-Open-Load (SOL) calibration technique for one-port calibration, or Short-Open-Load-Thru (SOLT) for two-port calibration, which is the most accurate, Short-Short-Load-Thru (SSLT) for waveguide, and Thru-Reflect-Line (TRL). In non-coaxial applications such as waveguide, TRL usually achieves better source match and load match corrections than SOLT. Coaxial TRL can also provide more accuracy than SOLT, but only if very high quality coaxial transmission lines (such as beadless airlines) are used. But the SOLT calibration is most widely used because it is broadband and can accommodate fixed probes. The dispersion effect in the case of TRL calibration is also removed here. The standards are comparatively easy to design and model. For more details, visit http://cp.literature.agilent.com/litweb/pdf/5965-7709E.pdf.

From: Heinrich Zweifel, Fela Management AG

Dear Shanthi,
It is relatively easy to do and you can get rid of the unknown parameter of the measurement equipment or measurement setup. This will allow you to be able to measure the parameters of the DUT.

From: Paul Sameal, DeVry

Dear Shanthi,
The reason why the open/short/thru method is the most preferred calibration method for RF measurements is as follows: open has capacitance (which, by the way, is often negative); short and load have inductance; it uses off-wafer standards; it is available on multiple vector network analyzers; it is sensitive to probe placement; and all standards must be common and known.

From: Walter Herrington, H&H Construction

Dear Shanthi,
This method is ideal for general purpose signal investigations for signals from 100 kHz to 6 GHz. Also, this method seems to be cost effective and can be used in a handheld portable device, which makes it easy to take on the job.

From: Michael Myers, Myers RF Engineering

Dear Shanthi,
Using the open-short-load method calibrates the measurement of VSWR to show just the reflection from the antenna. If this is not used, the loss from the cable shows up in the VSWR reading and values must be adjusted. This is a common problem seen in the cellular phone industry. The more loss the lower the VSWR.

From: Roger Chandra, Aero Inc.

Dear Shanthi,
There are mainly five reasons why it is preferred: 1.) This is the simplest method known for making RF measurements; 2.) The time needed for making a measurement is much less than any other method known; 3.) The results are very accurate and all bench marking studies have proven that; 4.) The measurement setup and calibration methods of the instruments are relatively easy; 5.) Last but not the least, the cost involved is also less. It is the most preferred because it is simple, fast, accurate and the calibration fixtures are relatively inexpensive to make and stable over time.

From: Timothy Howard

Dear Shanthi,
From what I understand, the most perferred method is an RF meter that is used in commercial and construction, for use of calibration communication systems. These meters are portable, rechargable and probably out of the price range of the casual user. However, searching the web and asking others in your area, you may find the right meter for you (check out Rohde & Schwarz on the web and look at their meters). This is all I can provide, because I do not know what your use of calibrating RF is for. Good luck.

From: Jeo Zimmer, Slidersnet

Dear Shanthi,
A method of calibrating a multi-port vector network analyzer compromises: (i) performing two-port calibrations on pairs of ports, to determine forward and reverse systematic error terms that can arise with each pair of ports, where the pairs of ports are selected such that each port's systematic error terms (directivity, source match, reflection tracking and load match) are determined; (ii) generating a switch error correction matrix using data from the two-port calibrations; and (iii) performing unknown thru calibration for at least one pair of ports that was not utilized in step (i), wherein the unknown thru calibration comprises applying the switch error correction matrix to measurement data and determining transmission tracking error terms using the corrected measurement data.

From: Mike Branum, Branum Enterprises

Dear Shanthi,
It is easy and there is a quick way to calibrate. The calkit standards are measured and then de-embedded. This method provides a high resistivity substrate, which is preferred.

From: Don Boby, CM

Dear Shanthi,
The transmission calibration required the use of a thru standard. A thru is supposed to have no length. The "open" and "short" dummy structures de-embed the probe-pad parasitics of a device-under-test (DUT). To accurately estimate the input/output interconnect parasitics, including the resistive, inductive, capacitive and conductive components, the "thru" dummy device has been characterized after probe-pad de-embedding. With the combination of transmission-line theory and cascade-configuration concept, this method can efficiently generate the scalable and repeatable interconnect parameters to completely eliminate the redundant parasitics of the active/passive DUTs of various device sizes and interconnect dimensions. Consequently, this method is very suitable for on-wafer automatic measurement.

From: John Hollis, Barrios Technology

Dear Shanthi,
The open/short/thru method is the most preferred method for RF measurements because it is quicker and does not take as long to calibrate, even though it is not the most accurate way.

From: Leonid Schalanovsky, UberBand Microwave

Dear Shanthi,
Aside from the fact that calibration fixtures are more or less not as expensive, the open/short/thru method is known for simplicity, accuracy and quickness. Stability is also reliable.

From: Arunabh Chattopadhydy, IITK

Dear Shanthi,
The open/short/thru method removes directivity, source match and reflection tracking errors. It defines the VSWR and return loss of the DUTs.The reasons of its popularity include: 1.) It requires fewer standard classes (three reflection classes) in comparison to more classes needed for full/TRL two-port calibration; 2.) Calibration techniques requiring less than three reflection classes (open short) do not always work in all situations.

From: Vaughn Maycock, FMF Cape Breton

Dear Shanthi,
Open or shorting the end of a cable allows its length and loss to be measured with a time domain reflectometer. A precision thru-line termination allows a very accurate power measurement to be made with an accurate RF voltmeter. The losses in connecting cable can be accounted for to get the true measurement at the output of a device.

From: Richard Pinkerton

Dear Shanthi,
These methods are the most commonly used in calibration methods and perform well if accurate models of calibration standards can in fact be determined. These methods of calibration are sensitive to probe placement, which also makes it the most preferred calibration method for RF measurements.

From: Tod Criswell, Surfit Center Machines

Dear Shanthi,
With the combination of transmission-line theory and cascade-configuration concept, this method can efficiently generate the scalable and repeatable interconnect parameters to completely eliminate the redundant parasitics of the active/passive DUTs of various device sizes and interconnect dimensions. Consequently, this method is very suitable for on-wafer automatic measurement.

From: Tammy Berg, Anderson

Dear Shanthi,
Through calibration, a simple technique is provided for the calibration of devices for which matched loads are difficult to obtain. The method only requires three standards: a through connection, a high reflectivity termination and a section of uniform line.

The winning response will win a free book from Artech House, along with an "I Asked Harlan!" t-shirt. In addition, everyone who submits a legitimate response will be sent an "I Asked Harlan!" t-shirt.

If your response is selected as the winner, you'll receive a free book of your choice from Artech House. Visit the Artech House on-line bookstore at www.artechhouse.com for details on hundreds of professional-level books in microwave engineering and related areas (maximum prize retail value $150).

Disclaimers: Microwave Journal reserves the right to determine the legitimacy of all responses. The determination of the winning response is not based on any set criteria and is based only on the opinions of the editors of Microwave Journal.

Questions and answers submitted to Ask Harlan are collected for sole use by Microwave Journal and www.MWJournal.com. Names and company affiliations will be published with questions and answers that qualify according to our editorial policies. Email addresses are collected in case we need to contact you.

Mailing address information submitted for the delivery of the T-Shirt is collected solely for this purpose.

Email addresses and mailing addresses are not published and are not shared with any third party.

This form is not a magazine subscription application nor a Website registration. For magazine registration services, please click here.

Entry into the T-Shirt give away is while supplies last.