Добавил:
Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Скачиваний:
0
Добавлен:
01.04.2024
Размер:
416.1 Кб
Скачать

Synthesis Approach for Compact Ku-Band Waveguide Lowpass Filters with Wide Rejection Bandwidth

Metehan Cetin

Gokhan Boyacioglu

Bulent Alicioglu

Radar and Electronic Warfare Systems

Radar and Electronic Warfare Systems

Former Aselsan Inc. Microwave Design

Business Sector

Business Sector

Engineer

Aselsan Inc.

Aselsan Inc.

Ankara, Turkey

Ankara, Turkey

Ankara, Turkey

balicioglu37@gmail.com

metehanc@aselsan.com.tr

gboyaci@aselsan.com.tr

 

Nevzat Yildirim

 

 

Department of Electrical and

 

 

Electronics Engineering

 

 

Middle East Technical University

 

 

Ankara, Turkey

 

 

nyil@metu.edu.tr

 

 

Abstract—Classical corrugated waveguide (WG) LP filters with wide rejection bandwidths rely on designing a band stop block with reduced height waveguides augmented by impedance matching blocks at terminations. In this paper direct synthesis approach is used to synthesize the whole filter as a single unit. The initial filter is designed as a TEM mode commensurate length LPF with very high quarter wavelength frequency for size reduction. Filter is made up of series SC stubs interlaced by contributing Unit Elements (UE). Then the filter is converted into its WG equivalent. The finite transmission zeros formed by the SC stubs are spread into the stopband properly to get both steep skirt and high rejection level. Return loss and rejection level is then adjusted by manual tuning of stub lengths and heights of UE’s. The resulting filter is then optimized on electromagnetic simulators. A Ku-band filter is designed and implemented as example.

Keywords—E-plane corrugated filters, low-pass filters, microwave filters, waveguide filters.

I. INTRODUCTION

Recent publications on miniaturized corrugated waveguide (WG) LP filters with wide rejection bandwidths for use in satellite applications rely on designing a band stop (BS) block with reduced height waveguides augmented by impedance matching blocks at input and output [1]-[4]. In [1], Levy uses TEM mode commensurate length distributed element LPF synthesis approach leading to series arm SC stubs interleaved by contributing Unit Elements (UE). Then the SC stubs and UE’s are converted into equivalent reduced height WG SC stubs and WG UE’s. The SC stubs form finite transmission zeros (FTZ’s), all at the same (quarter wavelength) frequency, creating the stopband. This band stop block is then augmented by adding quarter wavelength WG sections to match the BS block to the actual WG terminations. In [2], the architecture is similar to that of Levy, based on direct connection of corrugated composite resonators formed by a symmetric E-plane cavity (corresponding to the SC Stubs of Levy) in between WG pieces. These resonators create FTZ’s and are spread into the targeted stopband for achieving the required rejection level. The circuit models of these resonators are used for direct optimization with multimode variational formulation. In [3] both BS block and matching sections are also

designed as BS elements (series SC stubs) separated by short WG sections (WG UE’s) and optimization routine is applied to get the targeted response. In both [2] and [3], lengths of UE’s and widths of all stubs are set to be the same and fixed at a minimum value to minimize overall length of the filter and gaps of the BS block are constrained to avoid multipaction. Thus, optimization variables are limited to stub lengths and matching section WG UE heights only. This approach leads to very compact devices. In [4] and [5] more sophisticated optimization approach is adopted to decrease the effects of multimode interaction between closely spaced adjacent resonators.

In this paper advantageous sides of both synthesis approach of Levy [1] and miniaturization features of [2] and [3] are brought together to get a circuit model which form a good starting point for a EM optimization. The approach of Levy is extended to synthesize the whole filter (band rejection block + matching sections) as a single unit by using transformed frequency domain technique which can synthesize very high degree filters [6]. The WG sections of [2]-[3] are treated as contributing UE’s and obtained by selecting very high quarter wavelength frequency (fq) in synthesis stage to minimize overall length. The number of UE’s and series arm SC Stubs are selected to get both minimum size for the filter and also to set the required stopband rejection level by design trials.

II. DESIGN STAGES

The targeted specifications and constraints of the filter are as follows:

Passband edges are fp1=11.6 GHz, fp2=12.7 GHz with a return loss level better than 30 dB. The stopband edges are fs1=13.5 GHz to fs2=40 GHz with a minimum of 60 dB rejection.

At terminations WR75 WG will be used with width a=19.05 mm and height b=9.525 mm. In the filter block reduced height WR75 WG will be used with a=19.05 mm and height b varying from 3 mm to about 8 mm.

The initial circuit is synthesized in the filter synthesis program FILPRO [7] as a TEM mode commensurate length

978-1-5386-7132-0/18/$31.00 ©2018 IEEE

221

distributed element LPF with formed by series SC stubs interlaced by UE’s. Then the UE’s and SC stubs will be converted into WG UE’s and WG SC stubs respectively. The commensurate lengths of the UE’s and SC stubs are set by their fq. Since the overall length of the filter is desired to be as small as possible, it is set as fq=95.45 GHz to give 1 mm length for WG UE’s after conversion of UE’s and stubs into waveguide forms.

After a few design trials the number of SC stubs is set as Nstub=17 with the number of UE’s being Nue=Nstub+1=18. So, a degree N=35 LPF will be synthesized. The passband ripple is selected as 0.01 dB temporarily as the return loss will change after the brute force circuit transformations that will be applied in the succeeding stages. The passband return loss will be set to desired level at the tuning/optimization stages.

The synthesized circuit is shown in Fig. 1.a. The 17 SC stubs form 17 finite transmission zeros (FTZ) all at the same (quarter wavelength) frequency fq=95450 MHz.

Then the circuit is scaled for realization in reduced height WR75 waveguide with width a=19.05 mm and height b=3 mm. Scaling is done by inserting inverters at terminations such that the middle UE’s (termed as reference UE’s) will have b=3 mm after conversion to WG UE form (Fig. 1.b). The termination resistors are the characteristic impedance of the reduced height waveguide at passband center of f0=12.5 GHz.

Next all UE’s are converted into WG UE form. In this transformation the reference WG UE will have b=3 mm while all others UE’s will be scaled to different heights, increasing towards the terminations (Fig. 1.c).

Then the TEM mode SC stubs are converted into WG SC stubs. In these transformations parameters of the stubs are adjusted to shape the stopband response as follows:

Width of the WG stubs are kept at a=19.05 mm.

Height (b) of stubs are set as 6 mm. In implementation, each stub will be split into two stubs for realization on top and bottom surfaces of the main waveguide in a symmetric manner with actual heights of b=3 mm (assumed to act as series connected stubs).

Lengths (Lng) of WG SC stubs are adjusted to spread

the FTZ’s into the range from fs1=13.5 GHz to fs2=40 GHz to shape stopband response to get more than 60 dB rejection level.

For very steep skirt slope, 9 of the FTZ’s are placed close to passband edge, at 14 GHz. The other 8 FTZ’s are spread into the stopband, as shown in Fig. 1.c. Since all three parameters, Lng, a and b of the SC stubs are set by brute force, their input impedances will be different from the original TEM mode stubs. Therefore return loss level degrades after these conversions.

Next, we delete inverters at terminations and change termination resistors as RS=RL=494.76 ohms which is the characteristic impedance of the actual WR75 waveguide terminations with a=19.05 mm and b=9.525 mm at

passband center f0=12.5 GHz, as seen in Fig. 1.d. This action degrades return loss further.

This circuit is then used to start a tuning/optimization procedure to shape both passband stopband responses.

Heights (b) and lengths (Lng) of the stubs and UE’s before optimization are shown underneath each element, named as initial (Fig. 1.d).

The first 9 FTZ’s at 14 GHz (steep skirt FTZ’s) are placed on the source side. It is noticed that the two FTZ’s nearest to source are most effective in degrading passband return loss. By tuning lengths (Lng) of all stubs and heights (b) of all WG UE’s both passband and stopband responses are improved to a great extent (Figs. 1.e-1.f). A 34 dB return loss with greater than 60 dB stopband rejection is obtained after optimization. The optimized heights and lengths are shown underneath elements, named as final in Fig. 1.d.

A conceptual picture is shown in Fig. 1.g.

Inspection of heights (b) of WG UE’s shows that inner heights are almost constant at b=3 mm while they increase towards the source and load ends to match the filter to the WR75 terminations.

The total length of the filter is 6.9 cm which is smaller than all the previous filters cited in the references.

The final FILPRO design is optimized by using the electromagnetic simulation software FEST3D. Scattering parameters obtained from the optimization is shown in Fig. 2.

Implementation trials are continuing.

III. CONCLUSIONS

The classical corrugated WG LPF’s are designed by cascading a BS block with two matching blocks. In this paper direct synthesis approach is used to synthesize the filter as a single unit. The initial filter is designed as a TEM mode commensurate length LPF with very high quarter wavelength frequency for size reduction. Filter is made up of series SC stubs interlaced by contributing Unit Elements (UE). Then the filter is converted into its WG equivalent. The FTZ’s formed by the SC stubs are spread into the stopband properly to get both steep skirt and high rejection level. Return loss and rejection level is then adjusted by manual tuning of stub lengths and heights of UE’s. The resulting filter is then optimized on EM simulators. A Kuband filter is designed and implemented as example.

In addition, patent application has been made for this novel work.

ACKNOWLEDGMENT

This work is supported by ASELSAN Inc. and all products related to this work will be produced with the facilities of ASELSAN Inc.

REFERENCES

[1]R. Levy, “Compact waveguide bandstop filters for wide stopbands”, IEEE MTT-S Digest, Boston, 2009, pp. 1245-1248.

[2]F. D. Paolis, R. Gouliyev, J. Zheng, M. Yu, “CAD procedure for high performance composite corrugated filters”, IEEE Trans. MTT, v. 61, No. 9, pp. 3216-3224, Sept. 2013.

[3]F. Teberio, I. Arregui, A. Gomez-Torrent, E. Menargues, I. Arnego, M. Chudzik, M. Zedler, F. J. Gortz, R. Jost, T. Lopetegi, M. A. G. Laso, “Low-Loss Compact Ku-Band Waveguide Low-Pass Filter”, IEEE MTT-S Digest, 2015, Phoenix, pp. 1-4.

222

[4]O. A. Peverini, G. Addamo, R. Tascone, G. Virone, P. Cecchini, R. Mizzoni, F. Calignano, E. P. Ambrosio, D. Manfredi, P. Fino, “Enhanced topology of E-plane resonators for high power satellite applications”, IEEE Trans. MTT, v. 63, No. 10, pp. 3361-3373, Oct. 2015.

[5]O. Monerris, P. Soto, S. Cogollos, V. E. Boria, 1. Gil, C.Vicente, and B. Gimeno, "Accurate circuit synthesis of low-pass corrugated waveguide filters," 40th Eur. Microw. Conj., Paris, 2010, pp. 12371240.

[6]Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron spectroscopy studies on magneto-optical media and plastic substrate interface,” IEEE Transl. J. Magn. Japan, vol. 2, pp. 740–741, August 1987 [Digests 9th Annual Conf. Magnetics Japan, p. 301, 1982].

[7]N.Yildirim,M.Karaaslan,“FilproManual: Available in WEB page: http://www.eee.metu.edu.tr/~nyil/filpro.html.

Fig. 1. Design stages of the LFP using circuit models.

223

Fig. 2. Scattering parameters obtained from the optimization.

224