Feature: RF and microwave


The same problem can show up between two production stations built from the same bill of materials


valid, but the resulting measurands are not identical. At millimetre-wave frequencies, even


a good one-metre test assembly may add several decibels of loss, depending on its construction and frequency. If station A calibrates to the cable end, while station B applies a nominal loss table for a similar cable, the residual difference is the difference between the actual assembly and that table. Cable amplitude can also change with routing and flexure, which is why test cable manufacturers specify stability under defined mechanical conditions; see Figure 1. A fixture compounds the problem


because common correction approaches are not automatically interchangeable. Leaving fixture loss inside the result, using port extension, de-embedding from measured S-parameters, and performing TRL or another on-fixture calibration can agree only when they implement the same measurand and their residual errors are understood. Port extension can correct electrical delay and model loss but doesn’t in general remove the full mismatch behaviour of a transition. S-parameter de-embedding is only as representative as


the file used. An on-fixture calibration can move the reference plane to the device but requires suitable standards and additional calibration time. Two stations can therefore be internally


consistent and still disagree systematically because one corrects a section that the other leaves inside the result. Recalibration will not remove that difference if the stations continue to implement different measurands. Te reference plane should be stated explicitly in the test specification rather than leſt as an implementation assumption.


“Identical” RF paths? Assume the reference planes agree. Te physical paths can still differ. Mismatch. Mismatch can be a leading contributor because it belongs to the interaction between the device and the station rather than to either one alone. Two reflection coefficients, Γ1


and Γ2 Γ2 , produce


a magnitude term bounded approximately by 20 log10(1 ± |Γ1


12dB return loss (|Γ| = 0.251) into a path presenting 16dB (|Γ| = 0.158) gives about +0.34dB to -0.35dB for one interaction.


|). A device presenting


Two stations at opposite mismatch phases could therefore differ by as much as about 0.69dB from this mechanism alone. Repeated readings without changing the electrical state will not average it away. Connector interfaces and torque. Precision interfaces require the torque specified for the actual connector and hardware. Keysight’s connector-care guidance gives 0.9N·m for 2.92mm and 2.4mm connectors, but 0.57N·m for 1.85mm connectors; the applicable equipment documentation remains authoritative. Te same guide specifies that SMA, 3.5mm and 2.92mm interfaces are mechanically inter-mateable but warns that an SMA male pin can degrade a precision 2.92mm or 3.5mm interface. Connection repeatability should be established for the connector stack and mating procedure in use, rather than imported as one universal number. Te IEEE 287 series on precision coaxial connectors sets out the framework for connector effects and their specification. Cable routing. Phase-stable assemblies are characterised under specified flexure and routing conditions, and their amplitude can change as well. A cable dressed into a tight loop inside a rack is not in the same mechanical state as the same cable lying straight on a bench. Routing that is neither fixed nor documented remains a measurement variable. Switch matrices. Every switched route has its own insertion loss and return loss.


Figure 1: Three candidate reference planes in one measurement chain, with representative losses for an illustrative coaxial millimetre-wave path www.electronicsworld.co.uk November 2026 17


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44