| Horns |
TransitionsMAAS can design and manufacture transitions according to customers' needs, including
Transitions between different forms of transmission line must be designed with care so that the performance of the RF system is not degraded by poor Return Loss in the transition. Waveguide TransitionsWaveguide transitions as designed by MAAS are generally between circular and rectangular waveguide. Rectangular to rectangular transitions are not usually needed but can be made on similar principles. Circular to rectangular transitions are needed to go from a circular horn to a standard rectangular waveguide output. The Return Loss of a horn can be poor at about -15 dB (small aperture smoothwalled) to very good <-45 dB for a narrowband corrugated horn. Even a wideband (octave) corrugated horn will be better than -25 dB across the band. The addition of a transition should not degrade the Horn Return Loss too much and so a Return Loss of better than -30 dB should be the design aim. Two forms of transition are possible, stepped and tapered. Stepped TransitionsStepped transitions made by MAAS are designed and optimised using software from Antenna Software Limited [1, 2]. The programs used are RTCC and its optimiser OPTRTCC. A number of sections with varying rectangular dimensions are used and the section dimensions are optimised. An example is shown in Figure 1 and Figure 2 which has two intermediate rectangular sections between the input circular waveguide and the output rectangular waveguide (WG12). Between 5.0 and 5.5 GHz, the Return Loss of the transition is less than -39 dB. The Return Loss of the horn plus transition was as designed. A stepped transition for a full waveguide band can be designed but more intermediate sections are required and the total length will be greater. A typical example from rectangular waveguide WG10 to circular waveguide diameter 80.0mm covered the full waveguide band. The geometry is shown in Figure 3 and the predicted Return Loss in Figure 4. A full tolerance analysis is needed for such transitions in order to ensure that performance is not degraded when the device is manufactured. For this particular device using WG10, the tolerances were 25 microns.
Tapered TransitionsWhen a wideband transition is required, say, of a full waveguide band, a stepped transition may not be economical because too many steps are needed and the total length increases. In particular, as the frequency increases, the tolerances may make manufacture prohibitively expensive In such cases, a better solution is a tapered transition which should be over as long a length as feasible. This introduces a slow transition from the circular cross-section to the rectangular cross-section. For ease in manufacture by wire-cutting, the diameter of the circular cross-section should enclose the rectangular cross-section. This is not necessary if the transition is to be electroformed. The length should be at least two wavelengths in free space at the lowest frequency and three or four will be better. MAAS has designed and manufactured many of these, using ANSOFT Corporation's Finite Element program, HFSS [3], for the computation of RF performance. The Return Loss of a short tapered transition (only one wavelength) is shown in Figure 5 and Figure 6. The results and the geometry for a longer transition (two wavelengths) in Figure 7 and Figure 8. There is a very clear improvement in Return Loss when the transition is doubled in length. The shorter version was manufactured and the measured Return Loss was as predicted.
COAXIAL TO WAVEGUIDE TRANSITIONSWaveguide to coaxial transitions are commonly required for microwave components and may come in many forms. MAAS has designed and built
Standard rectangular waveguide to coaxial transitions covering a full waveguide band are available from many commercial suppliers and there has been no need for MAAS to design specials. The performance of commercial transitions is typically better than -20 dB. Where high quality is needed, for example, when APC-7 connectors are used for interface to a Vector Network Analysis, the Return Loss can be as low as -33 dB across the full waveguide band. Circular Waveguide Transitions - Side LaunchedThe side launched probe is the commonest form of coaxial connection to a waveguide. In this case, the geometrical parameters which can be varied are the length and diameter of the probe and the distance of the probe from the rear short-circuit. This needs to be as short as possible in order to avoid making the component narrowband. Typically a dual polarised transition is required to provide high isolation between the two output coaxial ports. If the two output ports are placed at the same distance from the rear short-circuit, then isolation is poor at about -10 dB. This can be improved by using two opposed outputs for each polarisation and adding the outputs together but this makes for a much more complex component. One solution is to separate the two ports by a quarter wavelength in which case the isolation will be better than -45 dB but the bandwidth of the port further from the rear short-circuit will be narrower (Figure 9). The measured Isolation is shown in Figure 10 and has a maximum value of -40 dB in the band shown but less than -45 dB over the band of interest which was 40 MHz centred on 4.97 GHz. This was used in a splashplate feed reflector [4]. The reduction in bandwidth of the upper port is not always acceptable and an alternative geometry is to use a vane. In this a thin vane is placed parallel to the probe of the upper output port and this acts as a short-circuit for the upper port. The lower (orthogonal) port can now be placed within the area of the vane and at the same distance away from the rear short-circuit. Thus the same bandwidth can be obtained for both ports. The design of each probe (length and diameter) will be identical but the effective short-circuit position is slightly to the rear of the forward edge of the vane and so some adjustment is needed here. It should be noted that the vane should be quite long but should not reach the rear short-circuit or the fields for the lower port will be incorrect. The vane need not be very thick, 1.0 mm at 10 GHz is quite adequate. This type of design has several advantages. It allows the device to be shortened because the Return Loss of each port is now independent of the position of the other port. The isolation is however dependent on the separation of the outputs and will control the minimum length. -25 dB is easily achieved and much better values result as the device is lengthened. Figure 11 and Figure 12 show the predicted and measured results for this device. Figure 13 and Figure 14 show a cross-section of the device.
Ridged Waveguide TransitionsWideband horns are often in dual or quad ridged waveguide. While the output can be in similar waveguide, more often a coaxial output is required. This can affect the overall Return Loss of the horn. Work by MAAS on such transitions has shown that the geometry for best match in dual and quad ridged transitions is quite different. Several such transitions have been optimised using HFSS. In optimising such a horn, the most time-effective method of modelling using HFSS is to treat the transition as a closed waveguide/coaxial device and apply a waveguide port to the output waveguide. Many examples have been designed and built, for example
In summary, by careful modelling of the transition alone, a dual ridged or quad ridged transition to coaxial line can be designed and built to give a good Return Loss over a wide band. Typical achievable figures are listed below.
Figure 15 Cutaway drawing of a small quad-ridged waveguide operating over 7 to 18 GHz
End-launched TransitionsThese are not commonly used because of the difficulty of obtaining good performance from them but occasionally, the geometry of the system is such that only an end launched transition can be fitted into the available space, for example, for waveguide elements in a phase array, or similar positions where space is very restricted. A few papers have been published on this topic. Ragan [5] mentions that end-launchers are possible using a loop and gives no further discussion. Deshpande [6] and Saad [7] consider analytical methods of dealing with such transitions and their optimisation. Wheeler [8] discusses the practicalities of designing such transitions and gives several examples. This is the most useful paper although Saad also gives some dimensions which appear to be based on those of Wheeler. Saad points out that it is possible to optimise performance over a reasonable bandwidth but that it is a case of minimising the Return Loss ripple. The optimisation of the probe and the matching steps was carried out using HFSS and the predicted Return Loss is shown in Figure 23. Figure 24 shows a SolidWorks model of the transition. The transition was constructed from Solidworks drawings. No extra parts were made to assist in measuring the Return Loss of the final transitions. They were therefore measured back-to-back. Since the Return Loss of the two devices will be identical, the Return Loss of the pair will be 6 dB higher at some frequencies and very low at other frequencies. In order to check on the results, HFSS was run with two identical units back-to-back and the measured results are compared to the measured results [Figure 25]. This indicates that the measured results are slightly worse around 4.5 GHz and slightly better around 6 GHz. The data can also be extracted from the measured peaks of Return Loss and this is plotted in Figure 23. The match is better than 1.2 above 4.6 GHz and better than 1.6 over the entire waveguide band.
Figure 25. Comparison of Measured and Predicted Return Loss for a pari of End Launchers Back-to-back REFERENCES
|
