Feature: RF and microwave


Figure 2: Illustrative magnitudes of mechanisms that may appear in a station comparison at 28GHz


A single nominal correction removes the average loss but not necessarily the residual path-to-path difference. Temperature behaviour, power level and repeatability must be taken from the selected matrix and verified in the assembled station. Cycle- count degradation is relevant to electro- mechanical switching; it should not be assumed to describe a solid-state matrix. Waveguide. At millimetre-wave frequencies, alignment, flange condition, screw pattern and tightening sequence can become measurable variables. Alignment- pin interfaces improve repeatability, but the work instruction still has to control how the connection is made. Temperature. Cable loss, fixtures and the DUT can all move with temperature, but the coefficients are product-specific. For illustration, a DUT coefficient of 0.02dB/°C across an 8°C laboratory-to-floor difference would produce 0.16dB. Tat value is an example for explaining scale, not a generic coefficient for millimetre-wave devices. A correlation study should measure the coefficient of the actual artefact and record both station and DUT temperature.


Software is part of the RF path Hardware receives most of the attention, but configuration can produce station differences of a similar scale. Several result- critical settings are controlled entirely in soſtware. Detector and averaging domain. For


Gaussian noise, instantaneous power is exponentially distributed. Averaging its logarithm rather than averaging power produces the well-known 2.51dB under- response:


10γ / ln(10) = 2.5068dB


where γ is the Euler-Mascheroni constant. Instrument manufacturers explicitly document this effect for log-averaged noise measurements. It is not a universal 2.51dB correction for every 5G NR or Wi-Fi power measurement. Whether it appears depends on the signal statistics, detector, averaging domain, resolution bandwidth, trace processing and the analyser implementation. A sample detector used where an RMS or power-average result is required can also create bias, but that is not automatically the same fixed 2.51dB mechanism. A correlation study should therefore exercise the actual modulated waveform and measurement mode, not only a CW source. Integration bandwidth. For an ideal flat power spectral density, changing the integration width from 100-101MHz changes integrated power by about 10 log10(101/100) = 0.043dB. A real modulated spectrum is not perfectly flat, so the actual change depends on its spectral shape, filtering and whether noise correction is enabled. Correction tables. Tese are easy to


18 November 2026 www.electronicsworld.co.uk


overlook because they persist as ordinary files. A table generated for an earlier cable set can remain active on one station and create a permanent offset that an internal instrument calibration will not remove. Residual state. A sequence that assumes a known starting state instead of establishing it inherits what the previous test leſt behind: attenuation, reference level, trigger mode, trace mode or alignment status. A station that has just run a high-power test and one that has just been powered on may not begin from the same condition. Trigger and timing. For burst or TDD signals, different trigger delays or settling intervals capture different parts of the waveform. On a device with a power- control transient, that difference can become a repeatable station offset. For correlation purposes, the


measurement architecture includes the RF hardware, configuration, soſtware state and physical reference plane. A change in any one of them can change the reported result. Figure 2 compares the scale of several


mechanisms using the earlier examples. It is not a decomposition of measured data and it is not an uncertainty budget. Te bars mix a calculated mismatch magnitude with illustrative deterministic offsets, so they should not be summed or combined in quadrature. Te figure is intended only to show that ordinary, non-fault mechanisms can be of the same order as a 0.7dB station difference.


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