THE IMPORTANCE OF USING CALIBRATION STANDARD DEFINITIONS FOR A VECTOR NETWORK ANALYSER page 3

SHORTS

The DELAY is required in the definition table for a SHORT standard in order to enable that the length of the actual standard being used to be removed from the measurement data during calibration. A good demonstration of this can be easily performed using the default data for a SHORT 3.5mm standard because this has an electrical length of 16.695 ps that can be observed after calibration. First the 3.5 mm calibration kit is selected on the VNA and then a  SHORT standard of any type (SMA, N, BNC etc) which is much greater than 16.695 ps length is connected and a RESPONSE - SHORT calibration made. The Smith Chart is then selected with the SHORT standard still connected and the response curve observed going clock-wise from a point at the left hand side. A port extension is then performed and a dot is achieved at the left hand side of the Smith chart when the port extension is equal to16.695 ps. This result indicates that the analyser is deducting a length equal to the default data in the definition table which is not related to the actual length of the standard used. Using the analyser with the wrong DELAY in the definition table for the SHORT standard will give an absolute phase measurement error and delay measurement error because the analyser will set the measurement reference plane in the wrong position. This could be a problem for some measurements, for instance delay lines, and it also makes a noticeable difference to measured power as will be shown later.

The OFFSET LOSS @ 1 GHz of a SHORT standard also has to be entered into the definition table and this causes great problems with BNC connectors. The maximum insertion loss of a military standard BNC connector should be less than 0.2 dB according to MIL-C-39012. When this connector is used as a standard, the return loss will be seen as twice the insertion loss because the signal has a 'round trip'. When a connector of lower quality is used the return loss can then exceed 0.4 dB and one of the many sample connectors measured during the research for this article gave a return loss of -1.02 dB. The loss in a connector is dependant upon several factors and one of these is the dielectric material. The dielectric in a BNC connector can be one of several different types but the most common seem to be DELRIN and PTFE/TEFLON. The TEFLON has good performance as a dielectric with an εr of 2.08 and a loss tangent of 0.0028 @ 3 GHz but the DELRIN has a poorer performance with an εr of 3.7 and a loss tangent of 0.0085 @ 0.5 MHz. The greater loss of a solid dielectric cannot be compensated correctly by the VNA which has a mathematical model for a gold plated standard with an air dielectric whereas the BNC may have for instance a nickel plated body, a silver-plated centre contact and a DELRIN dielectric. It is possible to enter the correct OFFSET LOSS at 1 GHz into the definition table and this will correct the LOSS at 1 GHz but it will probably not correct the loss at higher frequencies and the analyser phase also gets shifted thus also causing the DELAY to be incorrect. The solution to this is to select a connector with a good quality dielectric and only use BNC at the lower frequencies. A a Teflon dielectric appears to be the favourite material electrically and this will also has the advantage that it will not melt from the heat of a soldering iron.

OPENS

The OPEN standard definition requires all of the data used for the SHORT standard plus the capacitance coefficients. Again the dielectric of the connector becomes a major factor. Not only does the solid dielectric give high OFFSET LOSS but it also gives higher capacitance thus causing the capacitor coefficients to become too large to enter into the analyser. Selecting a connector with a TEFLON dielectric helps but it was still not possible during our research to find a BNC connector that will give calculated capacitance coefficients that are within the VNA input range. Although an approximate definition can be obtained by manually entering capacitor coefficients into a optimisation engine that displays both the calibration standards and input values capacitance curves.

It is possible to measure the capacitance of the standard at different frequencies using the VNA Smith Chart with a calibrated analyser and to use this data to find the polynomial coefficients using either Excel or MathCad. Excel can handle the calculation using the frequency in Hz and the capacitance in Farads using the Linest Function  An old version of MathCad had some difficulty with the calculation but was adequate when the the units were scaled to MHz and fF. When the coefficients are entered into the VNA, they have to be the opposite to the standards characteristics and so have to be entered as the negative of the numbers found in the polynomial calculation.

It was found during the characterisation of the OPEN standards that using the Termination Box with the lid fitted produced less fringing with an OPEN standard than the 'open tube' construction similar to that of an Agilent calibration standard.

 

RESULTS WITH DEFAULT DEFINITIONS

The OFFSET PHASE was calculated from the phase characteristics of the BNC(m) OPEN standard and the default values in the VNA 3.5 mm definition table.  Figure 6 shows the resulting Phase errors for OPEN standards and the BNC(m) connector is usable up to 10.9 MHz  for Phase within the previously set limit of 0.33°. The exercise was repeated for the BNC(f) OPEN connectors and then both the SMA OPEN connectors (all using the 3.5mm kit definitions in the VNA) and the results are shown in Figure 6 and Table 1.

[Graph showing calculated OFFSET phase of OPENS using default definition]

Similarly, the OFFSET PHASE was calculated from the phase characteristics of the BNC(m) SHORT standard and the default values in the VNA 3.5 mm definition table.  Figure 7 shows the resulting Phase errors for the BNC(m) SHORT standard connector which is usable up to 10.6 MHz  for Phase within the limit of 0.33°. The calculation was then repeated for the remaining SHORTS and the results are shown in Figure 7 and Table 1.

[Graph showing calculated OFFSET phase of SHORTS using default definition]

 The Termination Box was fitted with BNC(m) connectors and the OFFSET LOSS measured with the 3.5 mm cal kit definition  in the VNA. Figure 8 shows the resulting OFFSET LOSS errors for OPEN standards and the BNC(m) connector is usable up to 288 MHz  for OFFSET LOSS within the previously set limit of 0.033 dB. The exercise was repeated for the BNC(f) OPEN connectors and then both the SMA OPEN connectors (all using the 3.5mm kit definitions in the VNA) and the results are shown in Figure 8 and Table 1.

[Graph showing measured OFFSET LOSS of OPENS using DEFAULT definition]

Again the measurements were repeated for an SHORT  BNC(m) connector and the OFFSET LOSS measured with the 3.5 mm default definition table values in the VNA. Figure 9 shows the resulting OFFSET LOSS errors for OPEN standards and the BNC(m) connector is usable up to 159 MHz  for OFFSET LOSS within the previously set limit of 0.033 dB. The measurement was then repeated for the remaining SHORTS and the results are shown in Figure 9 and Table 1.

[Graph showing measured OFFSET LOSS of SHORTS using DEFAULT definition]

 

Technical Article Index   < 2  3  4 >  Page Number