This document describes a compact wideband bandpass filter design using stepped-impedance resonators and interdigital coupling structures. It analyzes a 4-pole Chebyshev filter design with four stepped-impedance resonators interdigitally coupled to each other. The design provides strong coupling in a compact structure with a second passband far from the first, addressing needs for next generation wireless communications.
Compact Wideband Bandpass Filter Using Stepped Impedance Resonators and Interdigital Coupling Structures
1. Compact Wideband Bandpass Filter Using
Stepped-Impedance Resonators and Interdigital
Coupling Structures
EE 5601
Akshay Soni
Ashutosh Mehra
2. • All same color holes (not black) are shorted together.
• All black holes are grounded.
Wideband Bandpass Filter
• Next generation wireless communications need.
l = 90
• Strong coupling
High Z TL
• Close spacing
Low Z TL
• Fabrication difficulties
• Interdigital structure
•Strong coupling between stepped impedance
resonators.
• Second passband is far away from the first
passband.
• Provides a very compact structure.
• Folding provides additional size saving.
• We analyze the 4-pole Chebyshev Filter design
• Four stepped impedance resonators
• Each Interdigital coupled with others. Shorted All fingers
shorts together
to in bottom metal
ground layer
• Specifications
• Passband ripple of 0.05 dB
• Center frequency of 1.0102 GHz
• Fractional BW of 48%
3. • All same color holes (not black) are shorted together.
• All black holes are grounded.
Wideband Bandpass Filter
• Next generation wireless communications need.
• Strong coupling 1 4
• Close spacing
• Fabrication difficulties
• Interdigital structure
•Strong coupling between stepped impedance
resonators.
• Second passband is far away from the first
passband.
• Provides a very compact structure.
• Folding provides additional size saving.
• We analyze the 4-pole Chebyshev Filter design 2
• Four stepped impedance resonators 3
• Each Interdigital coupled with others.
• Specifications
• Passband ripple of 0.05 dB
• Center frequency of 1.0102 GHz
• Fractional BW of 48%
4. • All same color holes (not black) are shorted together.
• All black holes are grounded.
Wideband Bandpass Filter
• Next generation wireless communications need.
Top Metal Layer View
• Strong coupling
• Close spacing
• Fabrication difficulties
• Interdigital structure
•Strong coupling between stepped impedance
resonators.
• Second passband is far away from the first
passband.
• Provides a very compact structure.
• Folding provides additional size saving.
• We analyze the 4-pole Chebyshev Filter design
• Four stepped impedance resonators
• Each Interdigital coupled with others.
• Specifications
• Passband ripple of 0.05 dB
• Center frequency of 1.0102 GHz
• Fractional BW of 48%
5. Issue of Multilayer and
Grounding Published Response
• The multi-layer structure is not well explained
• The grounding methodology is also not given
• We did a series of simulation, each going for
over 30 hours.
Simulation 1: Bottom layer forced ground Our IL = -20*log(mag(S21))
Result
Eqn
30
conductor 20
10
dielectric 0
-10
dB(S(2,1))
dB(S(1,1))
-20
-30 75
No
vias -40 50
IL
-50 No 25 spurious
-60 passband 0
response
0.6 1.1 1.4
-70 response
freq, GHz
-80
0 1 2 3 4 5 6
freq, GHz
6. Issue of Multilayer and
Grounding Published Response
• The multi-layer structure is not well explained
• The grounding methodology is also not given
• We did a series of simulation, each going for Eqn IL = -20*log(mag(S21))
over 30 hours.
Simulation 2: With ground and many Our Result
grounding vias
0
-10
-20
-30
dB(S(2,1))
dB(S(1,1))
-40
75
-50 Ripples in 50
IL
-60 passband 25
0
0.6 1.1 1.4
Nice
-70 spurious
freq, GHz
ground -80
0 1 2 3 4
response
5 6
freq, GHz
7. Issue of Multilayer and
Grounding Published Response
• The multi-layer structure is not well explained
• The grounding methodology is also not given
• We did a series of simulation, each going for
over 30 hours.
Simulation 3: With ground and less
grounding vias Our Result
Eqn IL = -20*log(mag(S21))
0
-10
-20
-30
dB(S(2,1))
dB(S(1,1))
-40
75
-50 50
Bad passband
IL
25
-60 0
0.6 1.1 1.4
-70
freq, GHz
-80
0 1 2 3 4 5 6
freq, GHz
Hi Good evening i am ashutoshmehra and I am going to present along with akshaysoni on the paper – topic – The author of this paper is chi yang chang from nctutaiwan and was published in IEEE MICROWAVE AND WIRELESS COMPONENTS LETTERS in september 2009.
The structure implemented in this paper is similar to constructing filters using coupled resonators that we had learnt in chapter 8. There we studied narrowband application of the design. To realize a wideband filter strong coupling is required which is impractical because of fabrication constraints. The author here implements the design in an interdigital fashion which due to crosscoupling between 4 resonators provides wideband response. The folding of the resonators provides additional coupling along with real estate saving. The design allows the spurious second order passband to appear far away form the required central frequency. On the right the figure on the top illustrates a single stepped impedence resonator. The low impedance portion of the resonator is then divided into multiple fingers. The high impedance portion is then folded.
4 of these folded resonators are interleaved together to give the top metal layer structure as shown.
We analyzed this 4 pole chebyshev filter design. The specifications are …
The author does not clearly explain the multi layer structure of this design. The structure in the paper has a break in the ground plane, i.e. the same metal used in the ground plane is used to via down to connect the top level fingers. We tried multiple simulations with different substrate layer settings in ADS Momentum. These incorporated different grouding strategies which we established by many simulations was a very important feature of the design. This was not clearly explained by the author. Simulation 1 below shows using 2 metal layer structure with bottom metal layer forcefully grounded. The results on the right show a non-physical response with the S12 going above 0 dB.
Simulation 2 has a dedicated ground boundary which is inherent to momentum and we use again a 2 metal layer structure where we via down to the middle layer to connent top level fingers as shown. We use multiple vias to connect second metal layer to grounding boundary. The results on the right graph illustrated close similarity with the published result for passbands, but have ripples due to resonances throughout the response. We think this incorrect response is a result of either improper grounding or simulation conditions used for the density of the mesh structure used for evaluation.
Simulation 3 tries to cover the issue of optimum number and density of vias that should be used to connect to ground. Clearly the response in the previous design is better than this one.