Author Topic: Microstrip filters  (Read 14276 times)

0 Members and 1 Guest are viewing this topic.

Offline spasmodic1Topic starter

  • Newbie
  • Posts: 6
Microstrip filters
« on: December 23, 2013, 01:44:40 pm »
Hi,

Im looking into designing a bandpass microstrip filter for 2ghz.
Does anyone know of a good starting point to to make these filters ie, a design calculator software (thats free).
I basicly want to make one without going to university and getting a degree in EEing  :-//
 

Offline KJDS

  • Super Contributor
  • ***
  • Posts: 2442
  • Country: gb
    • my website holding page
Re: Microstrip filters
« Reply #1 on: December 23, 2013, 02:20:54 pm »
Look up Denig's paper on interdigital filters and use that with RFSim99

*C. Denig, “Using microwave CAD programs to analyze microstrip interdigital filters,”
Microwave Journal,
March 1989, pp. 147-152


I'm assuming that you're capable of designing a top C coupled bandpass filter, if not then you'll have to work out how to do that first.

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3239
  • Country: gb
Re: Microstrip filters
« Reply #2 on: December 24, 2013, 01:12:54 am »
If you actually want to design and build a real filter PCB and get good results first time then I would strongly recommend you use Sonnet Lite to simulate your finished PCB layout. This is the best EM simulator on the market and you can download a free version that allows 32Mb of RAM space for the simulation.
I use the full version of Sonnet at work and it is the best simulation tool on the planet for this type of task. It will simulate your real PCB layout and the box you put it in.

http://www.sonnetsoftware.com/

I must admit I'm slightly curious how RFSIM99 could be used for this task. I've used this linear simulator a few times over the years and it only allows a limited library of models/components. I didn't think it had models for coupled lines etc.

Am I missing something here? Note. I could find links that referenced the Denig paper but I couldn't find the paper online.

« Last Edit: December 24, 2013, 01:31:19 am by G0HZU »
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: Microstrip filters
« Reply #3 on: December 24, 2013, 01:50:56 am »
I'm reading "Microstrip Filters for RF/Microwave Applications" by Jia-Sheng Hong, M. J. Lancaster
While it is pretty theory-heavy, all calculations are explained and they show a worked example for each type.  Simulate with sonnet like said above

A few of the bandpass designs





Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline KJDS

  • Super Contributor
  • ***
  • Posts: 2442
  • Country: gb
    • my website holding page
Re: Microstrip filters
« Reply #4 on: December 24, 2013, 08:15:05 am »
If you actually want to design and build a real filter PCB and get good results first time then I would strongly recommend you use Sonnet Lite to simulate your finished PCB layout. This is the best EM simulator on the market and you can download a free version that allows 32Mb of RAM space for the simulation.
I use the full version of Sonnet at work and it is the best simulation tool on the planet for this type of task. It will simulate your real PCB layout and the box you put it in.

http://www.sonnetsoftware.com/

I must admit I'm slightly curious how RFSIM99 could be used for this task. I've used this linear simulator a few times over the years and it only allows a limited library of models/components. I didn't think it had models for coupled lines etc.

Am I missing something here? Note. I could find links that referenced the Denig paper but I couldn't find the paper online.

I hadn't checked to see if RFsim99 had coupled microstrip, that was an assumption on my part. I use Hammerstad and Jensen's equations for coupled lines and the whole interdigital design drops out of a spreadsheet without too much effort.

I too couldn't find the Denig paper online, however it must be about somewhere. I'll plough through my papers and see if I can find a paper copy.

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20509
  • Country: gb
  • 0999
Re: Microstrip filters
« Reply #5 on: December 24, 2013, 06:03:34 pm »
What sort of board material are you planning to use? I've heard mixed opinions about how good standard FR-4 is at high frequencies. Personally, I've never designed a circuit running at more than 1Ghz so I've not run into any problems before.

http://croatia-microwave.blogspot.co.uk/2011/02/w1ghz-1296-mhz-hairpin-filter.html
http://share.pdfonline.com/4fd9808779df4c1ea55e5fb69018418a/hairpinfilter.htm
http://www.w1ghz.org/filter/Recipes_for_Printed_Hairpin_Filters.pdf
 

Offline Dreamer1

  • Contributor
  • Posts: 14
Re: Microstrip filters
« Reply #6 on: December 25, 2013, 12:12:20 am »
Hi!

Have a look at this, http://haxr.org/omar/papers/PDF/Omar%20X.%20Avelar,%20Omar%20de%20la%20Mora%20&%20Diego%20I.%20Romero%20-%20High-Frequency%20Filter%20Design.pdf

You design your filter with discrete components then transform them into microstrip using some transforms, it's a pretty easy idea to grasp. Just be careful that you match the input and the output of your line filter properly
 

Online nctnico

  • Super Contributor
  • ***
  • Posts: 28502
  • Country: nl
    • NCT Developments
Re: Microstrip filters
« Reply #7 on: December 25, 2013, 12:57:54 am »
That is a bad appnote. Kuroda identities and Richard's transform often leads to extremely thin traces (in the appnote they get a trace width of 0.031mm).
I recommend building a hairpin filter. The formulas are in the book Marshallh mentioned. Another option is using C shaped structures to make a filter (can't recall the name right now).
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 

Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3239
  • Country: gb
Re: Microstrip filters
« Reply #8 on: December 25, 2013, 01:38:59 am »
By far the easiest and quickest way to make a basic BPF at 2GHz (apart from using a £££ CAD filter synthesis tool) is to simply layout down a basic low order interdigital filter in Sonnet.

You can guess the line length or calculate it. Just set the line spacings closer to get more bandwidth and further apart to get narrower bandwidth. Then just adjust the tap point (in Sonnet) to get the best match. Then add more line sections to up the order to get the required shape factor etc.

With £££ CAD SW it can be all done in a few minutes. With the above iterative approach it would take maybe 15-25 minutes to get a decent (finished) design with little or no maths required. Obviously this all assumes the target filter spec is realistic in terms of % bandwidth and shape factor etc at 2GHz.

You 'can' use FR4 for this but I would recommend measuring the dielectric constant of your FR4 PCB by measuring the capacitance of a known area of the PCB material and then work out the Er. Then use this in the Sonnet simulation.

Otherwise use a decent RF substrate but this will cost a lot more to prototype etc.


 

Offline Zero999

  • Super Contributor
  • ***
  • Posts: 20509
  • Country: gb
  • 0999
Re: Microstrip filters
« Reply #9 on: December 25, 2013, 11:59:49 am »
You 'can' use FR4 for this but I would recommend measuring the dielectric constant of your FR4 PCB by measuring the capacitance of a known area of the PCB material and then work out the Er.
Don't forget, you need to measure the capacitance at the frequency of interest as the dielectric constant is frequency dependant. Another concern is FR-4 might be lossy at those frequencies so you might not achieve a very high Q.
 

Offline spasmodic1Topic starter

  • Newbie
  • Posts: 6
Re: Microstrip filters
« Reply #10 on: December 25, 2013, 01:16:40 pm »
Hi

Marshallh and everyone else thanks for the very informative information there i used it to make this dodgy filter today i was aiming for 1ghz so i guess its not too bad for a first effort.
I only have access to FR-4 without ordering some rf stuff. As you can see anything above the 1 ghz mark doesn't look that great, not sure if thats a design thing or the board.

Ill have a crack with Sonnet see how that goes. I made this one with eagle and wasn't by any means convenient to draw up.

thanks again everyone
ho ho ho


 

Offline KJDS

  • Super Contributor
  • ***
  • Posts: 2442
  • Country: gb
    • my website holding page
Re: Microstrip filters
« Reply #11 on: December 25, 2013, 05:45:23 pm »
You'll need to space that out a bit to stop interaction between each half of the hairpins to have any chance of it working well.

FR4 is generally very lossy with microstrip filters.

Dielectric constant cannot usefully be measured just by measuring capacitance as most dielectrics differ between the Z axis and the X and Y axis. What is even worse is that the Z axis effect if different to the X and Y effect as Ze and Zo vary due to the field variation between odd and even mode and I don't think that sonnet can handle non-homogenious dielectrics. one way round that is to build a very simple filter with the required line widths and spacings at about the right frequency and then vary the dielectric constant to match the measured frequency and vary the dielectric thickness in the model to match the bandwidth. These parameters can then be used to analyze a more complex structure.


Online G0HZU

  • Super Contributor
  • ***
  • Posts: 3239
  • Country: gb
Re: Microstrip filters
« Reply #12 on: December 25, 2013, 10:43:21 pm »
Depending on resin content etc FR4 can have a wide range of Er. It is a broad brush descriptor for various laminates.

Assuming this filter is being made by a hobbyist with access to unknown types of FR4 then measuring the Capacitance >> Er from a square will give a good starting figure.

If the Er turns out to be much lower at 2GHz then this will cause a shift in centre frequency on the finished PCB so a better estimate of Er can be quickly reverse engineered in Sonnet and the simulation rerun.

You can use FR4 up at 2GHz for BPF design but it will degrade the insertion loss and temperature stability and a few other things. I usually use Rogers 4003C or 4350 for this type of filter design.

Are you trying for a very narrow % bandwidth?

Your test plot of the 1GHz filter looks a bit dire and it looks to me like you have an issue with the test setup as well as the filter design.  I wouldn't choose to use a Rigol DS-815TG for filter design at 1GHz as I doubt the RF isolation or the port impedance of this instrument will be very good at >1GHz.
« Last Edit: December 26, 2013, 01:59:19 am by G0HZU »
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: Microstrip filters
« Reply #13 on: December 26, 2013, 01:11:08 am »
oshpark 4layer pcbs are done on FR408 which is somewhat less lossy than vanilla FR4. Maybe try that
Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline LukeW

  • Frequent Contributor
  • **
  • Posts: 686
Re: Microstrip filters
« Reply #14 on: December 27, 2013, 05:45:04 am »
It's not really an answer to your question, but I like this guy's remarkably simple, cheap demonstration of microstrip filters and similar devices:

http://scummos.blogspot.com.au/2013/01/rf-microstrip-breadboard.html
 

Offline hagster

  • Frequent Contributor
  • **
  • Posts: 394
Re: Microstrip filters
« Reply #15 on: December 27, 2013, 08:12:36 am »
Eurocircuits offer a RF pool service. They use an Isola substrate which seems to be a cheaper version of Rogers.

http://www.eurocircuits.com/index.php/component/content/article/4-services/184-rf-pool-service-definition

I haven't used this service yet.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf