THE IMPORTANCE OF USING CALIBRATION STANDARD DEFINITIONS FOR A VECTOR NETWORK ANALYSER

Introduction

This article was triggered by a very well presented web page on the Internet about home made BNC calibration standards and at the end somebody had asked a very pertinent question "What capacitance do you use for the OPEN standard definition?". Without knowing the capacitance and other characteristics of the standard, it is not possible to make a known measurement and the accuracy of the measurement will always be in doubt. It is clear from the connectors selected in the article and the way that they are modified that the author is knowledgeable about making these calibration devices and he does give several clues as to the specific requirements. Making your own calibration kit from connectors his way is not too difficult  but characterising them is not for beginners. It is not until you are able to define the characteristics of your connector that it is possible to judge if the quality as a standard is adequate. There are a lot of connectors manufactured that are of dubious quality, or have large reactive characteristics and are quite unsuitable for making into calibration standards.  

Aim

The aim of this article is to give information, that has not previously been readily available, for engineers to make a more informed decision as to when it is necessary to enter offset date into a Vector Network Analyser definition table. It is also intended to give a basic understanding of the parameters required in the definition table and to give a guide to the likely performance of a home made standard.

Calibration Principle

There are several basic methods used to calibrate a VNA and this article covers the standards required for the SOLT (Short, Open, Load, Thru') method but a lot of this information is also applicable to TRL and other methods of calibration.

 In order to calibrate a VNA, standards are fitted to the open end of the measurement cables and calibration data collected. The analyser then mathematically removes the OFFSET data in the standard definition tables from the calibration data to give a 'zero' reference point at the connector measurement reference plane. The measurement reference plane of a connector is regarded as the contact point of the outer conductors and an article on connectors that gives drawings showing the measurement reference plane of SMA and N type connectors can be found at http://www.maurymw.com/pdf/datasheets/5A-021.pdf  (Maury are probably the most well known manufacturer of VNA calibration kits and give a lot of useful data on their website.) The OFFSET data entered into the VNA tables includes delay in ps, capacitance in the form of a fourth order polynomial, inductance as a fourth order polynomial, loss as resistance per second, characteristic impedance, plus minimum and maximum frequencies.

LOAD

The LOAD is easiest standard to define because it just has Zo plus  minimum and maximum frequencies to enter. These frequency values can normally be left as the VNA default and the only major concern is that the LOAD should have a good return loss over the required frequency range. In the article on home made standards by Andrey E. Stove, which can be found at http://blog.kotarak.net/2009/10/n2pk-vna-calibration-standards.html  Andrey uses a high frequency precision 0.1% 0603 surface mount resistor as the load. The resistor is soldered to the back of the BNC connector and this makes a mechanically matching standard to the SHORT and OPEN which looks nice, but there is no need to use a physically identical (length) connector as the physical length of the LOAD is unimportant. A perfect LOAD would have no reflection and therefore it would not be possible to measure its electrical length.

SHORT

The SHORT standard is more difficult to characterise than the LOAD standard because the OFFSET DELAY and OFFSET LOSS are required. At high frequencies the inductance coefficients may be required but this is only on some very high frequency analysers and these are unlikely to be used with  home made calibration kits. In fact, some lower frequency VNA do not have the facility to enter inductance coefficients  but the SHORT OFFSET DELAY can often be adjusted slightly to compensate for the inductance variations over frequency by using the electrical delay at  a frequency somewhere near the centre of the 'delay curve'. One method to achieve this would be to read the inductance of the SHORT standard directly from the VNA Smith chart with a calibrated analyser and then make a linear graph showing the 'inductance curve' over frequency so that a mid point on this curve can then be selected.

OPEN

The OPEN is the most difficult standard to define because it requires the same offsets as the SHORT plus the capacitor coefficients. Although information from Agilent can be found stating that the capacitance is not needed for frequencies of less than 3 GHz, a home made standard made from a BNC connector is very unlikely to have a low capacitance and correction for this will  normally be necessary. The capacitance is defined in terms of 

C0 = 10-15 Farads

C1 = 10-27 Farads/Hz

C2 = 10-36 Farads/Hz2

C3 = 10-45 Farads/Hz3

 During the calibration process, the analyser requires the SHORT and OPEN standards to have 180° phase difference. As the frequency applied to an OPEN increases, the fringing and hence electrical length also increase thus making the OPEN look longer at high frequencies than at low ones. To compensate for this, the OPEN is made slightly shorter in electrical length than the SHORT at low frequencies. If the OPEN was made longer than the SHORT, then the Smith chart could run anti-clockwise with an OPEN or SHORT standard connected. The difference in electrical length for the SHORT and OPEN standards in Agilent calibration kits vary and can be from 2.05 ps for the N 75 Ohm female standards to 5.4 ps for the N 50 Ohm male standards. The specified difference for some kits can be found in the VNA definition table but may require some calculation of C0 and OFFSET DELAY to find the total delay. This is because DELAY and C0 do the same thing. If you have experimented with polynomials, you will probably know that the first coefficient (C0) is just an overall offset for all values of  X (frequency) which is just what DELAY does.

WHY MAKE STANDARDS?

There are several reasons to make your own standards. For instance there may not be one available for the type of connector that you need to use. Agilent supply only a limited range but there are more available from Maury including the BNC which can be found at http://www.maurymw.com/pdf/datasheets/2Z-029B.pdf  It may be that you have an analyser that does not demand a very high frequency calibration kit such as the DG8SAQ Vector Network Analyser which can be found at  http://www.sdr-kits.net/VNWA/VNWA_Description.html (this analyser should work well with our SMA connectors). Probably one of the biggest reasons to make your own is cost because the kits made by Agilent and Maury are very expensive. However, it is almost certainly a false economy to make your own kits if they are to be properly verified for traceability reasons.  Even if they do not require verification but do need characterising in order to complete the definition table, the cost of the time involved in doing this from first principals could easily more than justify the cost of buying a genuine kit.  If the quality of a home made standard is poor it may not even be possible to obtain verification as the capacitance or inductance coefficient values would fall out of the acceptable range of the VNA. The input range of each capacitance coefficient is set at a maximum of +/- 10,000 by Agilent VNA software and all BNC connectors tried while researching this article fell out of the input range.

Further information on calibration standards can be found in Specifying Calibration Standards for the Agilent 8510 Network Analyser which is available from Agilent at http://cp.literature.agilent.com/litweb/pdf/5956-4352.pdf 

Technical Article Index    <1  2  3>  Page Number