Maker Pro
Maker Pro

dog bone filters

P

Philip Newman

Jan 1, 1970
0
Hi,

For my project, I have to design some dog bone structure filters to
work at 10 GHz and 140 GHz. Using an IEEE transaction by C. Rauscher
as a reference point, I have managed to simulate a 5-channel
multiplxer at 5.2 GHz using these structures. However, I can scale up
for 10 GHz, but I want a different fractional bandwidth, and I am
finiding it difficult to multiplex 3 of these strcutures together.

How can I design dog-bone structures from scratch? Going from centre
frequency and knowledge of substrate to be used for microstrip. Roll
of is important because the filters may interact with each other.

Any ideas as to where I can start? I have looked on the web and
nothing has come up on google.

Cheers

Phil
 
P

Patrick Bolton

Jan 1, 1970
0
Phillip,

I think that the quick and dirty answer is use the optimiser in the
modelling tool you have, ensure that the seed values are not too limiting
and let it run overnight. What modelling package are you using?

Patrick
 
P

Philip Newman

Jan 1, 1970
0
Agilent's Advanced Design System

i have managed to scale up my 10GHz to 140GHz and am now in the process of
changing the substrate and optimising.

cheers

phil
 
N

N. Thornton

Jan 1, 1970
0
Hi. What is a dog bone structure filter? I googled and got not one hit
- for electronics that is :)


Regards, NT
 
P

Philip Newman

Jan 1, 1970
0
its a filter structure that uses microstrip as its implementation. its
called dog bone beacuse it uses hi-lo structure, that is two thick sections
with a thinner section in the middle, so it looks like a dog bone.

the thinner sections represent an inductance, while the thicker represents
capacitance. it is a way of implementing lumped element filters in
microstrip, thus allowing for much more compact strcutures. dog bone also
allows the satelite passband to be 3x the original band, way out of the
frequency band of interest and not interfering with your filter.

i just dont know how to design them :eek:(

phil
 
N

N. Thornton

Jan 1, 1970
0
Philip Newman said:
its a filter structure that uses microstrip as its implementation. its
called dog bone beacuse it uses hi-lo structure, that is two thick sections
with a thinner section in the middle, so it looks like a dog bone.

the thinner sections represent an inductance, while the thicker represents
capacitance. it is a way of implementing lumped element filters in
microstrip, thus allowing for much more compact strcutures. dog bone also
allows the satelite passband to be 3x the original band, way out of the
frequency band of interest and not interfering with your filter.

i just dont know how to design them :eek:(

phil


thanks. I got it all except the last sentence :) That one had me lost.

So there msut be a reason why you cant calculate C and L values as if
they were discrete components, then translate that into your
microstrip dimensions for each section of the CLC filter. You can tell
I've not done microstrip design :)

Regards, NT
 
P

Philip Newman

Jan 1, 1970
0
yeah, but when we did it the filter had a huge passband and was at the wrong
frequency etc

all sorts of problems.

satellite passband is quite difficult to understand, only through
simulations did i understand it myself!

Phil
 
M

maxfoo

Jan 1, 1970
0
its a filter structure that uses microstrip as its implementation. its
called dog bone beacuse it uses hi-lo structure, that is two thick sections
with a thinner section in the middle, so it looks like a dog bone.

Eagleware calls them "Stepped-Z Bandpass"
note: element values are unrealizale below ~20% bandwidth.
 
P

Philip Newman

Jan 1, 1970
0
Eagleware calls them "Stepped-Z Bandpass"
note: element values are unrealizale below ~20% bandwidth.

ah. why!?

phil
 
P

Philip Newman

Jan 1, 1970
0
riiight

can u reccommend a simple software that can design these filters? in the
example i got from eaglewear shows a simple sim package

phil
 
M

maxfoo

Jan 1, 1970
0
riiight

can u reccommend a simple software that can design these filters? in the
example i got from eaglewear shows a simple sim package

Free or $$$?

Caltech's Puff software is pretty good for free. I got my copy from a
CD that came with Cotter W. Sayre's book, "Complete Wireless Design"
its probably available online...do a search for it.
 
P

Philip Newman

Jan 1, 1970
0
i cant find it anywhere for download. is it possible you could ftp it to me
or something?

cheers

phil
 
C

Charles Edmondson

Jan 1, 1970
0
Philip said:
Hi,

For my project, I have to design some dog bone structure filters to
work at 10 GHz and 140 GHz. Using an IEEE transaction by C. Rauscher
as a reference point, I have managed to simulate a 5-channel
multiplxer at 5.2 GHz using these structures. However, I can scale up
for 10 GHz, but I want a different fractional bandwidth, and I am
finiding it difficult to multiplex 3 of these strcutures together.

How can I design dog-bone structures from scratch? Going from centre
frequency and knowledge of substrate to be used for microstrip. Roll
of is important because the filters may interact with each other.

Any ideas as to where I can start? I have looked on the web and
nothing has come up on google.

Cheers

Phil
Ok, so you are describing using stepped impedances as discrete devices,
i.e. wide stips for caps, narrow strips for inductors.

You will find reams of material on this in Matthai, but I don't have a
copy anymore! :-( When I took the class (from Matthai) 10 years ago,
went into a lot of detail. Basically, you read the values from a table
for the filter you needed, and scaled the dimensions to the frequency of
interest!

For modern use, take a look at Microwave Office. You can get a 30 day
demo for free, and should do what you need and a whole lot more!

Charlie
Edmondson Engineering
Unique Solutions to Unusual Problems
 
P

Philip Newman

Jan 1, 1970
0
cheers, my colleague had a copy of that book, but it was far too
complicated! we have a 5-channel contiguous multiplxer design from a paper
writte by Chritan Rauscher which started at 5.2 GHz. we scaled this upto
10GHz, and it works okm but we are unsure as to how to design the matching
impedance sections as shown in the diagrams given on his paper. i cant
submit the diagrams, but i can email them to you if you prefer?

cheers for the advice about microwave office, i am trying to get a copy now!

phil
 
P

Philip Newman

Jan 1, 1970
0
maxfoo said:
Eagleware calls them "Stepped-Z Bandpass"
note: element values are unrealizale below ~20% bandwidth.

what do you mean by unrealizeable? in that you cant implement it in
microstrip? or its just not possible for whatever reason?

phil
 
Top