As an example, four coaxial lines were used in an mTRL calibration (with and without Super-Weighting) and the same four lines were used (in a banded sense) in an LRL calibration (labeled 'normal TRL/LRL' in the plot). These various calibrations were used to measure yet another airline and the return loss results are shown in Figure: Measurements of Return Loss of an Airline After Various mTRL and TRL/LRL Calibrations.
Measurements of Return Loss of an Airline After Various mTRL and TRL/LRL Calibrations
The same set of four lines was used for each calibration.
While all of the results are quite good (return loss was expected to be in the area of 60 dB at low frequency and 50 dB near 40 GHz based on dimensional measurements of the DUT and materials knowledge), the mTRL calibration is capable of handling minor differences in characteristic impedances of the calibration lines and differences in connection parasitics. Thus the mTRL calibrations were able to produce results closer to those expected for the DUT. The Optimal Length Super-Weighting did not make a large difference in this calibration since the line length delta distribution was fairly uniform but it did improve the measured return loss values slightly below 20 GHz (making the result even closer to what was expected for the DUT).
As another example, consider a three-line mTRL calibration where match use is enabled with the threshold electrical length delta set to 10 degrees. The calibration was used to measure an additional coaxial airline and the results are shown in Figure: A Three Line-Plus-Match mTRL Calibration Used to Measure an Airline; the Results Show the Match Use Transition Near 1.68 GHz.. When the available electrical length difference falls below 10 degrees at low frequency, one can see a transition in return loss (near 1.68 GHz in this example). The return loss is still extremely high but since the match standard and the lines create different reference impedances, seeing a transition should not be unexpected. The match standard used in this measurement was not characterized (except for DC resistance being 50 ohms +/- 0.1 ohms). A smaller transition may be possible with a well-characterized match standard.
A Three Line-Plus-Match mTRL Calibration Used to Measure an Airline; the Results Show the Match Use Transition Near 1.68 GHz.
Other observations:
• As with regular LRM, when a match is used, the agreement between the standard and the model/.s1p file will directly affect residual directivity and the measurement of high return losses. The match impedance sets the calibration reference impedance when it is used (and the 'mean' line characteristic impedance sets the reference impedance when it is not; here 'mean' is in the context of the least-squares solution process).
• As with regular LRL, a central assumption is that the lines are ideal (no launch-related admittances, transmission acts like exp(-γ L) where γ is the complex propagation constant, …).
• Reflect offset lengths are referenced to the end of line 1 and are air-equivalent lengths (=2.9978 x108 (m/s) * Delay (s)). Asymmetry of the reflect standard (between ports) is much more important than the absolute reflection coefficient (which is solved for during the calibration anyway). Still, the sensitivities to reflect standards problems are much weaker than those for the lines (and match, if used).
• Generally, the more repeatability problems there are, the more lines that will be needed for an equal-quality calibration. If the repeatability is extremely poor (~<20-30 dB), the mTRL/LRL/LRM families may not be the best choice for the planes in question. Doing some full calibration at a stable plane and then using one of the partial information network extraction techniques (see Adapter Removal Calibrations and Network Extraction of this guide for more information) to get to the final reference plane may be a better choice.
