диафрагмированные волноводные фильтры / a5cf8ad1-b9f6-4cb7-ab29-5ff18b17015a

Keywords: E-plane, evanescent mode, transmission zero, waveguide filter, wideband filter.

Abstract

In this paper, practical implementation aspects of recently proposed ultra compact inline waveguide bandbass filters that utilise stoband performance of their pseudo-elliptic responses through positive sign bypass coupling are explored. Fundamental working principles are outlined from different perspectives and a number of X band filter samples are described in details, together with fabrication process using a pair of thin metal inserts positioned parallel along two different E-planes of a straight waveguide section. Developing from the core strengths of compactness, being 70% smaller than conventional E-plane filters, and ease of fabrication, discourse is continued by exposing the filters to different application challenges. It is demonstrated that the proposed structure is not only suitable for narrowband designs, giving a sample of a filter with 12% fractional bandwidth, still maintaining dimensions that allow fabrication with the same machining equipment. In addition, flexibility of producing the same filter response by modifying dimensions of different elements is shown. Sensitivity analysis is carried out and an approach to fine mechanical filter tuning is proposed. Furthermore, an example of higher degree filter of the fifth order is presented to illustrate the ability of the filter to meet demanding real-life specifications.

1 Introduction

Waveguide filters have been integral infrastructural part of numerous terrestrial, aeronautical and satellite radio communication and radar systems, both civil and military. Their first use was for processing of high power, namely radar, signals in shielded environment. However, as there is more demand over efficient use of power over time, their low loss attribute becomes more important. Likewise, more cramped packaging of devices signifies perfect electromagnetic isolation property, especially in emerging millimetre wave applications. Unlike filters implemented in other competing technologies, waveguide filters have least difficulties at upper microwave frequencies and drawbacks like larger size start to diminish. In addition, waveguide filters are good for satisfying strict stability criteria under different

uncontrollable influences in the working environment, especially in harsh space conditions.

Ultra compact pseudo-elliptic inline waveguide bandpass filters [1] mainly address two of the shortcomings of waveguide filters: ease of fabrication and large size. At the same time, they preserve enough of the mentioned waveguide filter advantages to have the edge in performance over other similar and smaller size filters. The first problem is tackled by applying E-plane building technology [2], inheriting its inexpensive and mass producible manufacturing characteristics. Unlike in the original paper [2] and most of the subsequent ones [3], there is not just one metal insert to be put between the two parts of the housing, but two of them, with additional pieces to keep the spacing between the inserts. Inline E-plane design also enables cascading other sections made in the same technology [4].

Although visually similar, compact E-plane filters presented in [3-5] are based on different working principles. There is full waveguide profile around discontinuities, hence no evanescent mode propagation in these sections. Subsequently, fin elements, modelled to form extracted pole sections, produce finite transmission zeros (TZs), and not cross couplings. This can be verified from the frequency responses as there is equal number of fins and TZs. Only offsetting metallic inserts from the centre E-plane to the side walls is not enough to form a TZ. In [6] are presented related singlets and doublets with capacitive cross couplings able to produce lower stopband TZs. It is demonstrated as well how to cascade compact E-plane bypass coupled and extracted pole filter sections.

Regarding the resonator size reduction, it is accomplished as the transmission pole (TP) frequencies are dominantly controlled by the lengths of the metal fins along the waveguide height, whereas septa between them reduced the interresonator mainline couplings, making the filters shorter lengthwise. As the filter operates below the TE10 mode cutoff frequency in the main body section between the rectangular waveguide feeds, it can be classified as evanescent mode filter. Thus, for practical operation, filter volume can be further significantly reduced by decreasing the waveguide width in this central section to the ground plane septum from the narrower side with no insert, as there is no port to port wave propagation in that back branch at the frequencies of interest. However, for prototyping, this hasn’t been adopted

because it makes housing fabrication as well as impedance matching between different waveguide profiles more difficult.

2 Filter structure

In Figure 1 is presented prototype assembling of an ultra compact pseudo-elliptic inline waveguide bandpass filter.

Figure 1: Exploded view of a waveguide bandpass filter with inductive bypass coupling constructed from two metal inserts: a singlet that produces a transmission pole and a transmission zero in the upper stopband.

In the sandwich structure, between the two sides of the aluminium housing there are three more layers stacked and aligned by two roll pins. They are altogether fastened by four bolts. The general front copper insert from Figure 1 contains fins and septa that form resonators and coupling in between them respectively. Since this is a singlet structure in Figure 1, the resonator is only coupled to the source and load – the two waveguide ports, and there are no septa to hold together the top and bottom contacts. The back copper insert contains a wide grounded septum. Top and bottom aluminium pieces determine the distance between the two metal inserts. All the metal elements inside the waveguide are connected to the housing contact parts, hence there is no need for a solid dielectric and practically only losses in the conductors exist. Layouts of the two of the metal inserts can be viewed in Figure 2.

In Figure 3 is diagrammed accompanying coupling schematic. Numbered resonant nodes are localised around fins and to the closest conductors around them, spurious mode is localised in the central region between the top insert and side wall, whereas non-resonant source and load nodes are in full width WR-90 rectangular waveguide region (a = 22.86 mm and b = 10.16 mm).

The cross coupling between the source and the load is used to create steep transition between the passband and the upper stopband with a TZ.

SP

Figure 3: Coupling diagram of a 3rd order ultra compact pseudo-elliptic inline waveguide bandpass filters using inductive bypass coupling.

Tables 1 and 2 give the exact geometric and electromagnetic description of the fabricated 3rd older filter with the centre frequency of 9.4 GHz and TZ frequency of 10.4 GHz, which measurement results are graphed in Figure 4.

Table 1: Physical dimension of the designed, fabricated and tested 3rd order filter.

Table 2: Coupling matrix of the designed, fabricated and tested 3rd order filter: S - Source; L - Load; 1, 2, 3 - Resonators; SP – Spurious Resonance.

Figure 4: S-parameters of the fabricated 3rd older filter (inset image), compared with responses calculated by full wave EM simulation in CST Microwave Studio and from the coupling matrix.

It is interesting to compare a bypass coupled resonator (Figure 1) and an EPS resonator [3]. Both singlets have been designed at 9.5 GHz centre frequencies, having TZs at 11.5 GHz, with fin cross sections of 0.1 mm x 1.6 mm and total structure lengths of 30 mm so as to have week source/load couplings. Numerically calculated unloaded Q factors obtained from transmission S-parameters are very similar, and around 2100. If the bypass coupled resonator is manufactured in a way that favours its compactness, removing the back path, it has clear advantage in the Q to volume ratio, though, since E-plane filters become longer with reduced coupling, it is hard to exactly compare their sizes. However, due to more complex fabrication of bypass coupled resonator, EPS resonator may have advantage in practical implementation owing to smaller signal leakage, especially in E-plane technology. Comparison with other resonators can be found in [7]. EPS resonators in [7] have larger quality factors because eigenmode solver was used, hence losses is sections with septa were excluded.

3 Wideband filter

Presented design can be used not just for narrowband filters. In Table 3 are given dimensions of a wideband ultra compact filter corresponding to the layouts in Figure 2.

The separation between the two metal inserts is increased to 7 mm. At the same time, the inserts still have equal offsets from the central E-plane. Together with narrower septa and fins, it increases the couplings that are needed for wider passband. This is similar to well-known microstrip line characteristics depending on the w/h ratio.

In addition, less pronounced way to increase the couplings is done by decreasing the distance between the fins of the resonators. This is similar to control of bandwidth in interdigital filters, between narrowband/moderate and wideband [8]. The topology of the ultra compact pseudo-elliptic inline waveguide bandpass filter using bypass coupling is analogue to interdigital filter topology with first and last lines open.

In either case, what limits the bandwidth is how physically small dimensions can be reliably manufactured. Different for the presented waveguide filters are the couplings of the first resonator to the source and the last resonator to the load. These couplings are made stronger by decreasing the length of the ground septum. This way, the end fins are put closer to the unperturbed waveguide feeds and sections of narrow waveguide with the evanescent mode propagation are made shorter.

Making filter wideband also included proportional shift of the TZ to a higher frequency. Otherwise, ripple factor in the passband is increased because of the vicinity of the TZ. There is a physical limitation as well of how much higher the TZ can be displaced by altering positions of the two metal strips in order to change the frequency at which the signals taking two different paths cancel each other.

In Figure 5 are presented transmission and reflection scattering parameters of the modelled wideband 3rd order filter. The filter has 12.0% fractional bandwidth, compared to 5.3% fractional bandwidth of the filter in Table 1 (Figure 4).

Figure 5: S-parameters of a wideband ultra compact bypass coupled filter.

4 Design flexibility

One of advantages of the discussed filters is variety of available ways how to obtain the desired filter response. In Table 4 are listed side by side three of such filters, and in Figure 6 are plotted their simulated S-parameters.

Table 4: Comparison of different dimensions between three same response 3rd order filters. Constant dimensions are

D=3.0 and WFin1=WFin2=WSep1=1.6. All dimensions are in millimetres.

Figure 6: Overlapping of S-parameters: filters with the same insert geometrical shapes, but different dimensions.

First obvious conclusion is that the fin lengths are the most sensitive dimensions of those listed in Table 4, as changing them in steps of 0.1 mm requires change of other parameters in several times larger steps. Furthermore, filter becomes shorter when inner fin is getting shorter and outer longer.

5 Sensitivity analysis

In Figures 7 and 8 are plotted S-parameters of the equal fin 3rd order filter from the preceding section together with new responses obtained when one at the time deviations of different filter dimensions are present in order to investigate

fabrication tolerances. Parameters a, b, LGr, L1, L2, WSep1, D, LFin1, LFin2, WFin1 and WFin2 have been changed by ±0.1 mm, since entire design had been made with 0.1 mm resolution

expecting to be feasible to have fabrication accuracy within that range. In addition, metallization thickness has been tested to be +0.1 mm, as well as the longitudinal misalignment of the two inserts.

Figure 7: Dependence of the filter transmission response on the dimension variations.

Figure 8: Dependence of the filter reflection response on the dimension variations.

It is confirmed that fin lengths are the only dimensions which have alone given significantly different responses, and the only bigger concern regarding sensitivity. Apart from them, increasing metal strip thickness produces visible shift of the design frequency.

6 Mechanical tuning

As in practice the fabricated filters inevitably possess certain imperfections, filter tuning is implemented in most of real life applications to correct created differences and satisfy the specifications as a more economical approach if extremely accurate fabrication process is needed. In Figure 9 is given the model of the same 3rd order filter with equal fins, which was first detuned by altering the most sensitive dimensions, and then tuning elements were introduced to restore the filter response. The relevant dimensions are given in Table 5.

The tuning screws opposite the fins mainly affect resonant frequencies of the resonators they are part of, whereas tuning screws at the edges of the large grounded septum and between it and the narrow septa that separate fins mainly affect input/output and interresonator couplings respectively.

However, the former ones are more effective in the tuning process. All the used tuning screws have radiuses of 0.8 mm.

In order to control the TZ position, this design requires that the ground septum can be slid between the two smaller waveguide side walls. This assumes some compromise regarding the fabrication in E-plane technology. Alternatively, additional moveable septum can be used between the insert with signal lines and the nearer parallel waveguide side wall.

Figure 10: Undisturbed filter response compared with responses after changed dimensions and tuning to compensate for them.

7 A 5th order filter sample

In Figure 11 is given a longitudinal cross section of the same topology filter, but with increased order to the 5th degree.

Figure 9: 3rd order filter with added elements for mechanical tuning (dark).

Table 5: Dimension of the original filter elements that are changed, values of those changes, and dimensions of the correcting tuning elements.

In Figure 10 are shown all the three pairs of responses from the structure in Figure 9, as previously described. It was interesting to notice that the capacitances produced by the tuning screws opposite the fins move further away upper spurious passband in similar way as in combline filters.

Figure 11: 3D model of a 5rd order ultra compact pseudoelliptic inline waveguide bandpass filter using inductive bypass coupling.

In Figure 12 are displayed its frequency responses similar to those in Figure 4, having 9.45 GHz centre frequency and upper stopband transmission zero at 10.41 GHz, with dimensions given in Table 6. Parameters in Table 6 are analogue to those in Figure 2. Apart from increased selectivity, fractional bandwidth is increased as well to 6.5 %, whereas simulated insertion loss at the centre frequency is around 0.4 dB. It can be seen that increasing filter order improves its compactness per resonator.

Table 6: Physical dimension of the designed 5rd order filter.

Figure 12: S-parameters of the 5th order ultra compact bypass coupled filter.

8 Conclusion

Main improvements achieved with ultra compact pseudoelliptic inline waveguide bandpass filters are in compactness and fabrication simplicity. Here, these filters have been put against further challenges occurring in practical applications showing well rounded characteristics. They are able to satisfy wide range of passband bandwidths, have versatile ways to obtain the same network response, are not very sensitive, have tuning possibility and are suitable for making higher order filter units.

References

[1]N. Mohottige, U. Jankovic and D. Budimir, “Ultra compact pseudo-elliptic inline waveguide bandpass filters using bypass coupling,” 2014 IEEE MTT-S

International Microwave Symposium (IMS2014), Tampa, FL, 2014, pp. 1-4.

[2]Y. Konishi and K. Uenakada, “The Design of a Bandpass Filter with Inductive Strip - Planar Circuit Mounted in Waveguide,” IEEE Transactions on Microwave Theory and Techniques, vol. 22, no. 10, pp. 869-873, Oct. 1974.

[3]O. Glubokov and D. Budimir, “Extraction of Generalized Coupling Coefficients for Inline Extracted Pole Filters With Nonresonating Nodes,” IEEE Trans. Microw. Theory Tech., vol. 59, no. 12, pp. 3023–3029, Dec. 2011.

[4]O. Glubokov and D. Budimir, “Novel inline waveguide E-plane filters using dual-mode extracted pole section,”

41st European Microwave Conference (EuMC 2011),

2011, pp. 99–102.

[5]N. Mohottige, O. Glubokov and D. Budimir, “Ultra Compact Inline E-Plane Waveguide Extracted Pole Bandpass Filters,” IEEE Microwave and Wireless Components Letters, vol. 23, no. 8, pp. 409-411, Aug. 2013.

[6]N. Mohottige, O. Glubokov, U. Jankovic and D. Budimir, “Ultra Compact Inline E-Plane Waveguide Bandpass Filters Using Cross Coupling,” IEEE Transactions on Microwave Theory and Techniques, vol. 64, no. 8, pp. 2561-2571, Aug. 2016.

[7]U. Jankovic, N. Mohottige and D. Budimir, “E-plane resonators for compact inline waveguide filters,” Active and Passive RF Devices Seminar, London, 2016, pp. 1- 7.

[8]G. L. Matthaei, L. Young, and E. M. T Jones,

Microwave filters, impedance-matching networks, and coupling structures. Dedham, Mass: Artech House, 1980.