DESIGN OF PARALLEL COUPLED MICROSTRIP BAND-PASS FILTER

Filters occupy important acts in several Radio Frequency microwave applications. Several applications such as wireless communications still challenge RF/microwave filters with strict requirements such as smaller size, higher performance, lighter weight, and lower cost. Microstrip Filters for RF/Microwave Applications offers a unique and comprehensive treatment of RF/microwave filters based on the microstrip structure. One of the most common methods in designing microwave filters is using of parallel-coupled microstrip. In this paper simulate and fabricate by using Ansoft Designer a two resonator microstrip band-pass filter suitable for Wi-Fi applications. The results of simulation were quite good.


INTRODUCTION:
Filters occupy important acts in several Radio Frequency microwave applications. It is used to separate or combine different frequencies. Filters are employed to select or confine the Radio Frequency microwave signals within assigned spectral limits due to the electromagnetic spectrum is limited and has to be shared [1].
The traditional design of parallel-coupled line filters, however, suffers from the spurious response at twice of the design frequency, which not only degrades pass band symmetry but also deteriorates rejection levels in upper stop-band. It is due to that phase constants of the even and odd modes of each coupled stage are not identical [2]. It has been shown that a micro-strip coupler filter may have more bandwidth and better isolation characteristic. The equalization can be achieved by using capacitor compensation, suspended substrates, corrugated coupled lines and an overlay dielectric. Furthermore, both of [3,4] used parallel-coupled microstrip filters and considered one of the best and common approaches for implementing a planar filter.
The band pass filter with suppression of spurious responses has been designed by using a flat dielectric overlay. For a single-stage coupler with an overlay, the thickness of the overlay dielectric can be easily determined by a full-wave transmission line program, e.g., the spectral domain approach. The thickness of the overlay dielectric, however, generally depends on material constants and geometric dimensions of the coupled micro-strips.
A multi-order filter consists of several cascade coupled stages. If the overlay dielectrics for these stages have different dimensions, fabrication of the whole filter will be tedious and difficult. Our objective is thus to design a filter with a uniform overlay dielectric, so that each stage is suppressing the spurious responses.

BAND-PASS FILTER BY PARALLE COUPLED LINES
This filter or as it known as parallel-coupled filter as shown in Figure 1. The strips are arranged parallel beside each other, where they are coupled with certain coupling factors [5]. The normalized admittance inverter is given by [6,7]:

Filter design:
In this part the filter design will be explained, simulate, fabricate and testing a coupled line band-pass filter that be suitable for Wi-Fi applications with 100MHz bandwidth. Our aim is also to design it as small as possible so we are using 2 or 3 resonator by the given "g" values for a Gaussian filter.
In this paper we use the CER substrate with properties εr=10 and 0.64 mm thickness and τgs=0.0035. By having the substrate information and applying previous equations on them we can get the values of even and odd characteristic impedance (Zoe, Zoo). then we use the figure 2 to find S/h and W/h for each stage.

Fig. 2 Even-and odd-mode characteristic impedance design data for coupled Micro-strips with an overlay dielectric [8]
.  These results are verified using any simulator software as in figure 2 to see the expected results before fabrication.

Fig. 3 Micro-strip Band-Pass Filter Design
For designing the circuit, designing tools are used. In this environment we can simply put our resonators, connect them together and also connect the microwave ports into its ends. The schematic would be as bellow: Then  As shown, the simulation results are near to expected results, because its center frequency is very near to 2.4, the S12 peak is -1.52 dB at 2.4GHz and the bandwidth is around 160 MHZ. Although some verification is needed to improve the bandwidth we postponed it now and try to fabricate a prototype to see what will happen in fabrication.
In this part it should make its layout and save it in Gerber format so that be compatible with CNC machine software. The 2D and 3D layout is as bellow: Port1 Port2

Fig. 5, 2D Layout of Ansoft Result
Everything was looking good and fabricated, but after fabrication the experimental result was quite different to simulation result. The problem that doesn"t mentioned in the fabrication is impedance matching between 50 ohm connectors and resonators. This effect would be big in small cases as ours.

Fig.6 Filter Design
As known, the transmission line width is most important and its physical length has less effect, the samples prototype proof this idea, so it is added two 5 mm length transmission line to our design. By this act the simulation result is changed thus we have to do some improvements also including bandwidth improvement that we forgot it last time. The new schematic is as bellow: And its result as following picture: And the result below is satisfied the qualification:

Fig. 9 Ansoft Result
As shown, the center frequency is around 2.4 GHZ, the bandwidth is around 130 MHZ and the S12 max value is -2dB at 2.4GHZ.

CONCLUSION AND FUTURE WORK
In this paper, the two resonator microstrip bandpass filter suitable for WIFI applications has been simulated and fabricated. The simulation results are quite good with some of difficulties occurred in fabrication. One of them is impedance matching and soldering part. Others maybe the limitation of CNC machine in fabricating narrow lines so that although of respecting its limitations some prototypes were corrupted. Ansoft Gerber file output so that if the differences between two consequent resonators are high as figure bellow, the distance between corners become very small and although it has no effect on simulation result it will corrupt the real result.  Due to this limitation, the design changing so that having less effect of these limitations, for instance different shapes needing narrower lines neglected and proceed with simple linear shape with big enough lines.
Another problem that faced after fabrication was changing the center frequency and bandwidth of the filter. After investigation, this effect would be due to difference between substrate characteristics in datasheet and real that is quite common in sample substrates normally used in used student projects. For instance although its copper thickness in datasheet is considered around 18 microns in some parts it was more than 70microns so that it reached CNC machine surface cleaning limitation and can not be removed by machine. If considered its copper thickness equal to 18micron -as in its datasheet-the physical length of resonators should be 12.1mm that mentioned it in simulation and fabrication, but when increasing the thickness to 70micron -as in real-the resonators length should be increasing to 12.41mm.
Another effect that didn"t considered in our design was this fact that the substrate dissipation factor is measured in 10GHZ while we was using it in 2.4GHZ, so further compensation should be done to improve the result.
To overcoming this problem we have to use standard substrate instead of sample substrate, and making several prototypes to achieve the exactly desired results.