Good Practice Guide on Making Rectangular Waveguide Connections at Frequencies above 100 GHz

Good Practice Guide on Making Rectangular Waveguide Connections at Frequencies above 100 GHz Thorsten Probst, Karsten Kuhlmann, Nick Ridler, James Watts July, 2019 This document and all parts contained therein are protected by copyright and are subject to the Creative Commons user license CC BY-ND 4.0 (https://creativecommons.org/licenses/by-nd/4.0/deed.en). ; Suggestion for the quotation of the references: Probst, Thorsten ; Kuhlmann, Karsten ; Ridler, Nick ; Watts, James. Good Practice Guide on Making Rectangular Waveguide Connections at Frequencies above 100 GHz, 2019. Physikalisch-Technische Bundesanstalt. DOI: https://doi.org/10.7795/530.20190805 Rectangular Waveguide Connections at Frequencies above 100 GHz Acknowledgement The authors acknowledge support by the European Metrology Programme for Innovation and Research (EMPIR) Project 17SIP08 “New Waveguide Interfaces for Terahertz Technologies”. The EMPIR program is co-financed by the participating countries and from the European Union’s Horizon 2020 research and innovation program. Contents 1 Introduction 1 1.1 Overview of rectangular waveguide standard documents . . . 1 1.2 Waveguide designation and compatibility . . . . . . . . . . . 3 1.3 Rectangular waveguide basics . . . . . . . . . . . . . . . . . . 3 2 The IEEE 1785 series of standards 5 2.1 IEEE Std 1785.1-2012 . . . . . . . . . . . . . . . . . . . . . . 6 2.2 IEEE Std 1785.2-2016 . . . . . . . . . . . . . . . . . . . . . . 6 2.3 IEEE Std 1785.3-2016 . . . . . . . . . . . . . . . . . . . . . . 7 3 Performing precise measurements 9 3.1 Preparations ........................... 9 3.2 Measurement set-up . . . . . . . . . . . . . . . . . . . . . . . 11 3.3 Somepracticaltips........................ 13 3.3.1 Use of a precision waveguide section as a test interface 14 3.3.2 Different waveguide apertures . . . . . . . . . . . . . . 14 3.3.3 Waveguide to coaxial adapters . . . . . . . . . . . . . 14 3.4 Example Measurements . . . . . . . . . . . . . . . . . . . . . 15 3.4.1 Measurements for 75 to 110 GHz: WR10 / WM-2540 / R 900 / WG 27 . . . . . . . . . . 16 3.4.2 Measurements for 220 to 330 GHz: WR3 / WM-864 / R 2.6k / WG 32 . . . . . . . . . . . 21 4 Software 23 4.1 Software for Uncertainty Calculation of Waveguide Connections 23 References 24 https://doi.org/10.7795/530.20190805 i Rectangular Waveguide Connections at Frequencies above 100 GHz Purpose At frequencies of approximately 100 GHz and above, rectangular metallic waveguide is often the preferred transmission medium for making reliable, precision, measurements. A prerequisite for these types of measurements is the use of the correct types of waveguide interface, and, the need to follow the correct operating procedures. This document is aimed at people interested in making reproducible measurements with relatively low measurement uncertainty, in rectangular waveguide, at these frequencies. It can be considered as a supporting document for existing standards that are available for waveguide and waveguide interfaces used at these frequencies (e.g. as published by IEEE, IEC, MIL, EIA, etc). This document concentrates on rectangular waveguides. However, many aspects that are presented might also be applicable to circular waveguides. https://doi.org/10.7795/530.20190805 ii Rectangular Waveguide Connections at Frequencies above 100 GHz 1 Introduction 1.1 Overview of rectangular waveguide standard documents Today the IEC and IEEE are the two main international standardisation bodies for rectangular waveguides for use at frequencies above 100 GHz. Other standardisation bodies, such as United States Military (MIL), the Electronic Industries Alliance (EIA) or the United Kingdom Ministry of Defence (MoD), are also referenced by some manufacturers, although some of these documents are no longer maintained. Table 1 lists the latest editions of these documents. Standard Dimension Interface Publication year fmax / GHz MIL-DTL-85/3D X X 2012 325 Std1785.1-2012 [1] X X 2012 3300 IEC 60153-2:2016 [2] X X 2016 3300 EIA RS-261-B:1979 X X 1979 325 MIL-DTL-3922/67E w/Amendment 1[3] X2014 110 IEEE Std1785.2-2016 [4] X2016 3300 IEC 60154-2:2016 [5] X2016 3300 Table 1. Selection of existing standards for rectangular waveguides. Usually, one standard provides information concerning the frequency bands and the associated waveguide aperture sizes, and another standard describes the waveguide interfaces (also referred to as flanges). In the 5 th column of Table 1, fmax indicates the highest frequency the standard supports. In 2012 and 2016 IEEE and IEC published new standards to cover frequency bands up to the terahertz frequency region [1], [2], [4], [5]. Figure 1 shows an example of a waveguide flange used at millimetre(mm)- wave frequencies and above. Usually, the waveguide opening (aperture) is centred, while the alignment and connection mechanisms (i.e. dowels, holes, and threaded holes) are placed around the aperture. A more complete overview about existing waveguide sizes and flanges can be found in [6]–[9]. https://doi.org/10.7795/530.20190805 1 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz anticocking ring alignment pin threaded hole WG apperture precision alignment holes boss alignment hole ab Fig. 1. Example flange for rectangular waveguide. Waveguide Designation IEEE MoD IEC EIA f/ GHz WMWG R WR 2540 27 900 10 75-110 2032 28 1.2k 8 90-140 1651 29 1.4k 7 110-170 1295 30 1.8k 5 140-220 1092 31 2.2k 4 170-260 864 32 2.6k 3 220-330 710 - 3.2k - 260-400 570 - 4k - 330-500 470 - 5k - 400-600 380 - 6.2k - 500-750 310 - 7.4k - 600-900 250 - 9k - 750-1100 200 - 12k - 900-1400 164 - 14k - 1100-1700 130 - 18k - 1400-2200 106 - 22k - 1700-2600 86 - 26k - 2200-3300 Table 2. Overview of rectangular waveguide designations for selected frequencies[6]. https://doi.org/10.7795/530.20190805 2 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz 1.2 Waveguide designation and compatibility Table 2 gives an overview of some common waveguide designations. The recommended frequency ranges specified in the different standards might vary slightly, thus rounded values are used here. Most standardised flanges from IEEE and IEC are mechanically compatible with older designs (e.g. from MIL, EIA, MoD, etc.) of the same type of flange and compatible with each other, though the electrical performance might be significantly degraded [10]. One exception is IEEE 1785.2c, which is a plug and jack design. Only the plug type design is compatible with the other flanges. In practice, many manufacturers of waveguide components for use at frequencies above 100 GHz have chosen to mitigate this performance degradation by deviating from the standard designs. This is typically achieved by increasing the diameter of the alignment pins or reducing the diameter of the alignment holes. Whilst these non-standard flanges can still be mated with each other and with standardised flange designs, some combinations of manufacturers’ flanges will not mate at all due to the impact of tolerances on the critical dimensions. If it is not known which standards a pair of flanges follows, one must make the connection very carefully. Most waveguide interfaces follow a similar connection strategy. There is basically one flange design for several frequency bands, and the alignment is realised via holes and dowels with a certain precision. However, there are some exceptions, e.g. flange type G (IEC 60154-2:2016 [5]). In IEEE 1785.2-2016, there are actually three different flanges designs, and the alignment mechanisms are somewhat different when compared to each other, or to flange designs given in other standards. This will be described in detail in Section 2.2. A more complete overview of flange compatibility can also be found in [6], [9]. 1.3 Rectangular waveguide basics In this section, some equations are given relating to the propagation characteristics of these waveguides. This kind of information is usually provided by waveguide manufacturers. The cut-off frequency fcfor the rectangular waveguide [1] is fc=c √r· 1 2·a(1) where c : is speed of electromagnetic waves in vacuum, defined as 299 792 458 m/s r: is relative permittivity https://doi.org/10.7795/530.20190805 3 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz a: is waveguide broad wall dimension The attenuation constant αcan be approximated [1] with α= 0.023273 ·rρ ρ0· 1 b√a·f fc2 +2·b a qf fc·rf fc2 −1 dB/cm (2) where ρ: is resistivity of the waveguide conductor ρ0: is reference resistivity = 17.241 nΩ·m a: is waveguide broad wall dimension in mm b: is waveguide narrow wall dimension in mm fc: is cut-off frequency in GHz f: is frequency at which attenuation constant is calculated in GHz Equation (2) does not hold for thinly plated surfaces (e.g. where the plating thickness is less than two times the skindepth) and also neglects any effects due the surface roughness of the waveguide or other defect (e.g. cracks or gaps in the internal corners of the waveguide). The skindepth [11] is δ=rρ πfµ =r2ρ ωµ =1ρ RS (3) where µ: is permittivity RS: is surface resistance Great care should be taken in using this equation for waveguides where the waveguide cross section is not formed from a single conductor (e.g. waveguides formed from a machined channel and a flat cover). Imperfect conduction at the joins of these parts can give rise to significantly greater attenuation. The information given in this section is only accurate for single mode operation and for the first T ransverse E lectric mode TE 10 . This means, that the electric field is transverse to the direction of propagation, while the magnetic field has components in the direction of propagation. Field illustrations can be found e.g. in [11]. The field orientation of the TE 10 mode is also the reason for the naming of the two reference planes E-plane and H-plane in rectangular waveguides (see Figure 2). The terms E-plane and H-plane are also used in antenna measurements and for other microwave devices to indicate the orientation of the polarisation of of the electromagnetic radio waves. https://doi.org/10.7795/530.20190805 4 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz E-plane H-plane a b ; Fig. 2. Reference planes in rectangular waveguides. 2 The IEEE 1785 series of standards At around the beginning of this century, there was a significant increase in use of the millimetre-wave part of the spectrum for commercial applications in electronics, materials science and communications. In response to this increase in usage, commercial test equipment gradually became available. Companies such as OML [12] or RPG [13] started developing accessories for vector network analysers that enabled the operating frequency to be extended above 110 GHz [14]. These measurement capabilities were extended subsequently to at least 500 GHz [15]. Other companies, such as VDI [16], continued this trend, developing systems that operated to beyond 1 THz [17]. During the first decade of this century, there was a realisation that knowledge and standardisation of waveguide at these frequencies was rather limited [18]. Schemes were subsequently proposed to extend waveguide sizes to cover these millimetre and submillimetre-wave frequencies [19]–[21]. However, none of these proposed waveguide sizes had been standardised or completely accepted by the end-user community. This led to the initiation of an IEEE standardisation activity, started in 2008, to provide standardised waveguide sizes for use at these frequencies. At the same time, if was widely recognised that the electrical and mechanical performance of existing waveguide flanges was inadequate for use at these frequencies. The IEEE standardisation activity aimed to address this issue as well. The IEEE Standards Association set up and launched the Working Group P1785. Beginning in 2008, three standards covering (i) the waveguide sizes, (ii) waveguide interface, and (iii) typical uncertainty specifications for measurements at these frequencies were developed. This resulted in the publication of the following three standards: (i) 1785.1-2012: “IEEE Standard for Rectangular Metallic Waveguides and Their Interfaces for Frequencies of 110 GHz and Above—Part 1: Frequency Bands and Waveguide Dimensions” (ii) 1785.2-2016: “IEEE Standard for Rectangular Metallic Waveguides and Their Interfaces for Frequencies of 110 GHz and Above—Part 2: https://doi.org/10.7795/530.20190805 5 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz RCVR A IN RCVR R1 IN RCVR R2 IN RCVR B IN P1 P2P3 P4 1 2 4 2 5 66 7 8 8 10 7 6 6 9 9 10 RF IN MEAS OUT MEAS OUT REF OUT REF OUT RF INLO IN LO IN 3 Fig. 5. Alternative schematic measurement set-up with 4-port VNA and frequency extenders. 1: 4-port VNA 2: Adapter 2.4 mm jack →3.5 mm jack 3: 1.85 mm termination 4: Adapter NMD2.4 mm plug →NMD3.5 mm jack 5: Power Splitter 3.5 mm jack 6: Cable 2.4 mm plug →2.92 mm plug 7: Cable 2.92 mm plug →2.92 mm plug 8: Cable 3.5 mm plug →3.5 mm plug 9: Frequency Extender 10 : Adapter WM-2540 →WM-2540 how all components and cables were arranged and whether the waveguide connections were done in horizontal or vertical manner. If only 1-port measurements are required, it might be advantageous to use a vertical set-up. Thus, gravity effects should not introduce any systematic influence on the waveguide connections. A vertical set-up can be achieved by using a waveguide bend or by vertically positioning the extender head. For the latter, a customized fixture might be required. For both horizontal and vertical arrangements, one should aim to have a mechanically stable set-up: no unnecessary movements or vibrations. Schematic drawings for two different set-ups are given in Figure 7. Depending on the manufacturer, the internal VNA settings e.g. AGC (automatic gain control) can influence the overall measurement uncertainty. Therefore, one should check these setting beforehand. This is even more important if the VNA firmware is not used for the calibration. If the uncalibrated (raw) data is measured, e.g. for use with custom calibration routines, the operator should check the following: 1. Receiver settings (AGC, …) 2. Possible pre-calibrations https://doi.org/10.7795/530.20190805 12 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz MEAS IN REF IN REF IN MEAS IN P1 P2 1 RFM M RRRF LOLO 10 MHz REF OUT LAN 1 LAN REF IN OUT 3 2 3 5 4 5 4 3 3 6 6 7 7 8 8 910 11 12 Fig. 6. Schematic measurement set-up with 2-port VNA, synthesizer and frequency extenders. 1: 2-port VNA 2: Signal generator 3: Cable 2.4 mm plug →2,92 mm plug 4: Cable 2,92 mm plug →2,92 mm plug 5: Adapter 2.4 mm jack →3.5 mm jack 6: Frequency Extender 7: Adapter R900 →R900 8: Cable 3.5 mm plug →3.5 mm plug 9: Power Splitter 3.5 mm jack 10 : Adapter 3.5 mm plug →3.5 mm plug 11 : BNC Cable 12 : Ethernet Cable 3. Correct recording of switch terms 3.3 Some practical tips Much of the information given in [23] (chapter 8: Best measurement practice and practical advice) can also be applied to waveguide measurements above 100 GHz. This section focuses on good measurement practice for typical waveguide applications. https://doi.org/10.7795/530.20190805 13 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz (a) Typical horizontal set-up. (b) Typical vertical set-up. Fig. 7. Extender head positions: (a) horizontal, and (b) vertical. 3.3.1 Use of a precision waveguide section as a test interface Over time and with use, the waveguide interfaces of VNA frequency extenders can become damaged or worn, and this impacts directly on measurement repeatability. It is good practice to fit short precision waveguide straight sections to the frequency extender test ports since these can easily be removed for inspection and calibration without invalidating manufacturers’ warranties. This procedure is well known from coaxial measurements. The additional coaxial adapter is often called a ”connector saver” or ”interface saver”. 3.3.2 Different waveguide apertures If the same flange is used, it is possible to connect waveguides with different aperture sizes. This may even be intended in some applications, like e.g. for the VDI power meter PM5 [16]. The PM5 is equipped with a WR-10 waveguide, but it is possible to measure power levels at frequencies much higher than 110 GHz (even up to 3 THz). Above a certain frequency however, more than one waveguide mode can propagate in the larger aperture, and it becomes difficult to predict how much energy of the actual signal is distributed to which mode. This may even change for each new flange connection (different excitation conditions), and the repeatability of such measurements should be checked very carefully. In general, it is recommended to use WG apertures of the same size or to introduce a WG taper. This should minimise coupling into unwanted modes, but does not completely eliminate it. 3.3.3 Waveguide to coaxial adapters For ’normal’ coaxial or waveguide devices it should not matter, at which orientation they are connected to the test ports. This is different for coaxial to waveguide adapters. The TEM (transverse electromagnetic) mode of the coaxial line is symmetrical to the inner conductor, the TE (transverse electric) mode of the waveguide is not. If the transmission coefficient of a https://doi.org/10.7795/530.20190805 14 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz waveguide to coaxial adapter is measured, the phase will change by 180 °, if the waveguide connection to the test port is rotated by 180 °. The reflection coefficient remains unchanged. This effect poses usually no problem, because in data sheets of such adapters often only the magnitude of the insertions loss is given. If, however, a calibration certificate (or similar document) includes all complex S-Parameters of a coaxial to waveguide adapter, the waveguide flange orientation during measurement must also be stated in such a document. 3.4 Example Measurements In this section, some WG measurements are presented and discussed. The focus lies on the repeatability of flange connections and the drift influences of the measurement equipment. The flanges that are used are: • UG-387, i.e. the type of flange that most measurement devices are equipped with. • IEEE 1785.2a, precision flange for frequencies up to THz. The original UG-387 flange design does not have any inner precision alignment holes. But, to achieve better repeatability, some manufacturers added added inner alignment holes before IEEE or IEC published the new standards. The dimension and tolerance of these holes and associated dowels are often not the same, between different manufacturers, thus an operator must be very careful about connecting flanges that look similar, at first glance, to avoid damage due to these dimensional differences. To characterise a flange in terms of repeatability, one can do an analysis of repeated connections of stable devices under test (DUT), e.g. a matched load or a short [23]. But it is also possible to estimate the repeatability based on the tolerance of the alignment mechanisms of the flange. The maximum misalignment for the UG-387 flange is approximately 150 µm in both the Hand E-plane, and the maximum angular misalignment is 1.2 °. For the IEEE 1785.2a flange, the maximum misalignment is 25 µm in the Hplane, 20 µm in the E-plane, and the maximum angular misalignment is 0.4 °. With this information and modern simulation tools, the worst case return loss can be calculated [4], [27]. Table 4 shows these worst case return loss UG-387 IEEE.2a WR 10 / WM-2540 ≈-30 dB -60 dB WR 03 / WM-864 ≈-12 dB -39 dB WM-570 ≈-8 dB -32 dB Table 4. Theoretical flange performance for selected WG dimensions of UG-387 (not modified, without inner precision holes) and IEEE Std. 1785.2:2016. https://doi.org/10.7795/530.20190805 15 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz values for standardised flange connections. This table shows that the UG-387 flange is not really suitable for use at frequencies above approximately 110 GHz. The repeatability should usually be better than the worst case return loss. In the following, it is shown, that this is not always the case. 3.4.1 Measurements for 75 to 110 GHz: WR10 / WM-2540 / R 900 / WG 27 The frequency band 75-110 GHz is the lowest band supported by the IEEE standard, but the UG-387 flange is usually precise enough to get a reasonable good repeatability and reproducible measurement results for these frequencies and is therefore still widely used. In Figure 8 the repeatability is shown for a matched DUT for a horizontal and a vertical measurement set-up (to reduce gravity offsets) and with and without additional inner alignment dowels used. 75 80 85 90 95 100 105 110 0 2 4 ·10−4 Frequency / GHz Repeatability with dowel (vertical) no dowel (vertical) with dowel (horiontal) no dowel (horiontal) Fig. 8. Repeatability evaluation with four measurements for vertical and horizontal position with and without alignment dowels. The repeatability was calculated as describe in [23]. It is obtained from repeated measurements of a stable DUT and calculating the maximum differences, taking both the real and imaginary components into account. This is a very conservative approach and outlier sensitive, but it acknowledges the fact that the repeatability can strongly vary from one flange pair to another. As can be seen, the different cases yield very similar repeatability results and are always better than 0.0316, (≈-30 dB, see Table 4). For a connection with the UG-387 flange four screws must be tightened. Because in some set-ups one or more screws might be difficult to reach (or due to time reasons) some operators do not use all four screws. This is likely to have a significant influence not only on the repeatability, but also on the measured value, as is shown in Figure 9. One should always use all four screws and also a suitable torque to tighten the screws. The latter can be a difficult task, because the torque is not standardised, and manufacturers specify different values [24], [28]. The https://doi.org/10.7795/530.20190805 16 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz 75 80 85 90 95 100 105 110 0.98 0.99 1 Frequency / GHz |S11| loose screws; fixed screws Fig. 9. Calibrated return loss measurements of a flush short-circuit. torque needed to produce a precise and reproducible waveguide connection is at least dependent on: • base material of the flanges (e.g. copper, beryllium copper, brass, aluminium, etc.), • thickness of the flange (i.e. the flange could be formed using a block) • quality of the threaded holes and screws (different friction loss), • surface material (often several metal layers, e.g. nickel, gold, silver, etc.), • surface flatness, • surface roughness. For a high quality pair of flanges (i.e. with flat surface and low surface roughness, etc.) a very small torque is often sufficient to achieve a good repeatability for waveguide connections. But, of equal importance as the torque is the weight of a DUT for horizontal set-ups or the alignment of a second extender head for horizontal set-ups (needed for 2-port measurements). Both might reduce the effective torque on the interface itself very much and give the wrong impression, e.g. that a high torque must always be used. In Figure 10 the repeatability is shown for two torque values (0.06 N · m and 0.58 N · m) as well as for the torque achieved by an experienced operator. The latter is estimated to be approximately 0.3 N · m (verified with a calibrated torque meter). Figure 10 shows that the repeatability achieved by the experienced operator can be better than that achieved using the specified torque values. It also shows that more (or less) torque does not automatically mean a better connection. This is one reason why values of torque for making waveguide connections are not currently specified in the standards. For example, during connection, an experienced operator can consider other aspects that might impact the quality of the connection – e.g. the weight of the component being connected when tightening the flange screws. However, it is still usually recommended to use torque drivers to achieve reproducible results. https://doi.org/10.7795/530.20190805 17 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz 75 80 85 90 95 100 105 110 0.5 1 1.5 2 ·10−2 Frequency / GHz Repeatability 0.06 N·m; exp. operator; 0.58 N·m Fig. 10. Repeatability measurements [23] of an offset-short for different torques. When performing repeatability tests, one should not blindly do the statistical analysis but also investigate the single measurements to identify any possible outliers caused by poor connections. Figure 11 shows the difference of the single reflection coefficient measurements Γn to the mean value Γall,mean of all repeatability measurements for two experiments with the same DUT (short-circuit). In Figure 11(a), two outliers (measurement 3 and 4) can be identified. These should not be taken into account for the repeatability calculation. If the number of remaining measurements is too few (at least 4 or more should be used), one should repeat the experiment completely. In Figure 11(b), the repeated experiment (performed with more care) without outliers is shown. All repeatability measurements might also be distorted due to: • Noise, • Drift, • Cable movements. Depending on the measurement scenario, 1-port measurements usually don’t require cable movement, the latter one can often be avoided completely. Even if not a complete uncertainty budget is required, the procedure to get this are given in [22], [23], one should still do some simple tests to quantify these effects. For example to get an idea how large the instrument drift is, one can do a series of measurements (e.g. one every 60 sec.) of a stable DUT (without reconnecting). In Figure 12 the drift result of such a measurement is shown for the linear reflection coefficient of a flush short-circuit. The values are obtained by simply calculating the difference to the first measurement. The drift for 4 measurements (corresponds to 4 minutes) at the beginning and at the end of a experiment is nearly identical and approximately 1 ·10−3 . The drift for 15 measurements (corresponds to 15 minutes) is somewhat larger and of the same order of magnitude as the expected repeatability. In Figure 13 the cable stability is shown after performing the test prohttps://doi.org/10.7795/530.20190805 18 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz 75 80 85 90 95 100 105 110 0 0.5 1 1.5 2·10−3 Frequency / GHz ||Γn|−|Γall,mean|| 1; 2; 3; 4 (a) Difference between reflection coefficient magnitudes and their mean value, with outliers. 75 80 85 90 95 100 105 110 0 0.5 1 1.5 2·10−3 Frequency / GHz ||Γn|−|Γall,mean|| 1; 2; 3; 4 (b) Differences between reflection coefficient magnitudes and their mean value, without outliers. Fig. 11. Reflection measurements of a stable DUT normalised to the mean of n=4 measurements: (a) port 1 and (b) port 2. 75 80 85 90 95 100 105 110 0.5 1 1.5 2 ·10−3 Frequency / GHz Stability drift n=4; drift n=15; driftend n=4 Fig. 12. Stability evaluation (drift) with n=15 measurements, one measurement each minute. cedure given in [23]. This consists of repeated movements and reflection measurements of a cable (here several cables including frequency extender) terminated with a matched load and a short-circuit and calculating the maximum differences, taking both magnitude and phase into account. https://doi.org/10.7795/530.20190805 19 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz 75 80 85 90 95 100 105 110 1 1.5 2 2.5 ·10−2 Frequency / GHz Transmission stability / dB 10 cm movement; 25 cm movement (a) Transmission stability in dB (cg-12) 75 80 85 90 95 100 105 110 0.1 0.15 0.2 Frequency / GHz Transmission stability / ° 10 cm movement; 25 cm movement (b) Transmission stability in deg (cg-12) 75 80 85 90 95 100 105 110 −90 −85 −80 −75 Frequency / GHz Reflection stability / dB 10 cm movement; 25 cm movement (c) Reflection stability in dB (cg-12) Fig. 13. Cable stability [23] for 10 cm and 25 cm cable movements. In this case, the results are in the order of magnitude one would expect at these frequencies and don’t change much for cable movements of 10 cm or 20 cm. This indicates, that all cables connected to the extenders are quite stable. If the results are not as good as expected, one should test all cables separately. Usually this involves coaxial measurements up to only 20 GHz or less and following the same procedure. https://doi.org/10.7795/530.20190805 20 of 25 Rectangular Waveguide Connections at Frequencies above 100 GHz 3.4.2 Measurements for 220 to 330 GHz: WR3 / WM-864 / R 2.6k / WG 32 The waveguide band 220 GHz to 330 GHz is supported by the flanges described in IEEE Std. 1785.2-2016 and the UG-387 flange. Although the UG-387 flange can be used at these frequencies, it is not recommended, unless a modified version incorporating improved alignment features is used. The IEEE Std. 1785.2-2016 flange designs provide better electrical performance at these frequencies. 220 230 240 250 260 270 280 290 300 310 320 0 5 10 ·10−3 Frequency / GHz Stability drift n=4; drift n=15 driftend n=4 Fig. 14. Stability evaluation (drift) with n flush short-circuit measurements. In Figure 14 the stability based on repeated measurements on a fixed timescale (one every 60 sec.) of a stable DUT (without reconnecting) is evaluated. Compared to the measurements up to 110 GHz, see Figure 12, the system drift is larger. In 4 minutes it is approximately 2·10−3 and for 15 minutes it reaches 10·10−3 at the highest frequency. The main reason for this is the lower stability of the frequency extenders, thus any repeatability measurements are strongly superimposed by the instrument drift. 220 230 240 250 260 270 280 290 300 310 320 0 5 10 ·10−3 Frequency / GHz Repeatability UG-387 mod. IEEE 1785.2a Fig. 15. Repeatability measurements [23] of a stable DUT for different flanges. In Figure 15 repeatability measurements are shown for different flange configurations: standard flange of the frequency extender (modified UGhttps://doi.org/10.7795/530.20190805 21 of 25