Table of Contents
Write a comment
welcome to another edition of Coonrod corner today's topic a basic tutorial of micro PCB based filters here's your host John Keane hello and welcome to [ __ ] rods corner my name is John Coonrod I am a Technical Marketing Manager for Rogers corporation today I'm going to spend a few minutes talking about filters and I'm gonna give you a basic tutorial or overview for RF filters made with high frequency printed circuit board technology to begin with anything that is Wireless that transmitter receives is going to need a filter of some type and there's many different types of filters out there and they all have different functions but basically the filter does as the name implies and it's going to filter out energy from one frequency and it's not gonna allow frequency through at another frequency so pretty much blocks energy at one frequency and allows for energy at another frequency to go through in this drawing I'm showing a low-pass filter response and on the y-axis is loss and on the x-axis is frequency and you can see a loss at zero means obviously no loss so all the energy goes through at that point and a loss of 80 DB although all the energy is blocked pretty much because you have a huge amount of loss so if you look at the x-axis frequency axis what you see is the energy from the frequency 0 to F 1 all the energy is going to be allowed through by the filter and that's the pass band so the term there is pass band and that's passing energy through a band of frequencies with a loss very low and then to the right of that is going to be the stop band and that's where the filter is going to be blocking energy in the frequency range from F 1 to F 2 and you can see the loss there as an example is 80 DB so it's pretty much shutting off all energy in that range of frequencies there are several different RF filter functions three of the most common are a low pass filter high pass filter and bandpass filter these that can be made with several different attributes on a printed circuit board by the design of the the features on the board but to begin with let's take a look at the response of a low-pass filter the first run is a low-pass filter and is similar to what we already discussed but essentially what it's showing you is that low-frequencies that's going to pass the energy through and that's how it gets its name low-pass and the higher frequencies it's going to block the energy now the next drawing is a high-pass filter and it's pretty much the opposite of the low-pass filter so at high frequencies it's going to allow the energy to pass through the name high-pass and at low frequencies that energy is going to be blocked and then finally the bandpass filter is shown and in this particular drawing you can see it's actually got to stop bands there's a stop band at lower frequency stop band at higher frequency and the actual pass band or the region of frequencies that the energy has passed through is centered at f1 now how these filters are implemented in a print circuit board are normally done by resonators so a resonator is a structure on a printed circuit board that will generate a lot of energy at a very specific frequency so at the bottom of this picture you can see a top view drawing of the PC board with their resonator element and the knurled lines are really the feed lines that's what's feeding the energy to the resonator and the resonator itself is the wider feature there and what happens is this wide feature will set up a standing wave at a certain frequency and generate a lot of energy at that frequency and it will essentially resonate at a specific frequency now above this is a screenshot from a network analyzer showing a microstrip resonator and in this case it's resonating about nine point nine gigahertz so resonators are used to put together a filter function and this is done by putting together several resonators very closely a lot of times edge to edge and it called edge coupled resonators or edge coupled filters and a microstrip format that's where the term comes from a microstrip edge coupled bandpass filter is very common and I've shown that on the drawing here so here in the drawing you can see that there are multiple resonators put together side by side and each resonator will resonate at a specific frequency and if you can imagine one resident there could be slightly different for a resonant frequency than another and as you blend these different resonators together they couple energy to cause a resonant peak that's stretched you might say in frequencies so you can get a band of frequencies that will resonate or you can get a pass to end of a filter there's actually a lot more to that story but that's a good way to think about it anyway so now let's look at a actual measurement of a microstrip Ben pass edge coupled filter and really what you're gonna see is it does not have the very distinct sharp drawing as what I've been shown before from the stop band to the pass band there's actually some curvature there and that's normal in the picture you can see a top view of the microstrip bandpass filter and you can see resonators here now they're a little different than what I've drawn before because this is a little bit more detailed but again these are resonator or links of conductor elements that will resonate and join together and coupled together they perform a filter function the top picture is a screenshot of this microstrip bandpass filter being tested from a rage about 1.5 gigahertz to 2.5 gigahertz and again you see it doesn't have that immediate stair-step when going from the stop band of the pass band and that's normal for bandpass filters and actually this is a pretty good bandpass filter when you look at the transition from stop band to pass band this particular filter is centered at about 2 gigahertz and you can see marker for there assess 2.0 5 gigahertz and all things considered with fabrication tolerances that's actually pretty good this concludes this session of Qunu rods corner and thank you for watching for additional information and technical tools if you are not already a member join the Rogers technology support hub and gain access to calculators technical papers and more of Qumran scorner and other informational videos Rodgers technical information is also available at your fingertips with Iran mobile app available for the iPhone iPad and Android devices check it out today
Thanks for your comment Darren Feddersen, have a nice day.
- Kacie Cotti, Staff Member
today we're going to be talking about RF filter design now these days the actual design of a filter is pretty straightforward you've got tons of design tools and simulations that you can run to put together you know pretty much a perfect RF filter the trick is actually realizing that same simulated performance in the real world take this filter for example this was designed using a filter design tool it's set up for 50 ohm input and output impedances and it is a second-order Butterworth bandpass filter and you can so you have to tuned elements here and they are coupled together with a coupling capacitor and the input and output is also coupled with coupling capacitors as well so the actual component selection for this is all done by the design tool obviously this is set up to have a six megahertz 3 DB bandwidth with the center frequency of a hundred and thirty seven megahertz and on paper and in the simulator this works wonderfully but when you go to actually build this in the real world there are a couple things that can trip you up pretty quickly the first thing to consider is the cue of your components because most design tools and simulators assume practically lossless components which you're certainly not going to get in the real world especially when it comes to your inductors the capacitors usually have very high q but the inductors can be lossy so just because your simulation says that you'll have very little or no loss in your pass band don't expect that to happen when you actually build the filter typically you can expect 1 to 3 DB of loss in the pass band depending on primarily the q of your inductors the next thing to consider is stray capacitance component tolerance and tolerance over temperature and time because components do wear out and they do drift over time so if these two capacitors here which are part of these two tuned circuits changed by one Pico farad that can knock this filters passband completely out of the range of frequencies that I actually want to pass now you can easily pick a picofarad of capacitance just from the stray capacitance on the board not to mention these inductors if you want high Q inductors will typically be hand-wound so they're not going to be exactly a hundred nano Henry's or whatever your nominal value is there so these are gonna change the center frequency of your band pass filter and they can change it quite drastically so what I typically do is I make these capacitors variable just in trimmer capacitors so you can really dial in the center frequency of the filter now optionally if you've hand-wound these coils you can kind of pull the coils apart so they're little spread apart a little bit more or push the turns together so they're a little closer and you can vary the resonance of these tuned circuits that way but I actually find that to be very temperamental and really a pain to do so I much prefer to have some good quality trim caps in here to tweak the center of the bandpass filter if necessary now while these two tuned circuits are going to determine the center frequency of your van pass filter this coupling capacitor is actually going to control the shape of your bandpass filter now an ideal bandpass filter is nice and flat through the pass band and then it sharply cuts off in the stop band but that's only gonna happen if this coupling capacitor is just right so it's called critical coupling if your capacitor is too large here you're actually going to be over coupled and what you'll see is instead of a flat pass band you kind of have this double hump in the pass band likewise if this is too small you'll have under coupling and you'll you only have one peak in the pass band but it'll be very narrow and you typically have a lot of loss in your pass band as well it's also very common as you probably noticed for this coupling capacitor to be quite small in value this is only supposed to be 0.4 Pico farad's now you can achieve a small capacitance in a couple of ways obviously you could put a bunch of larger capacitors in series however I prefer to just take two insulated wires and twist them together because really all the capacitor is is too conductive pieces of metal that come close to each other but don't actually touch so the tighter you twist them together the more capacitance you have and if you kind of untwist them a bit you get less capacitance so it's a very cheap and easy way to give yourself a variable very low value capacitor here so you can really dial in that coupling capacitor just right so I have the filter built up on the bench here and you can see I've used a couple of good quality air variable trim caps here those are in parallel with these inductors here forming the two tuned circuits and these inductors are just six turns of eighteen gauge enamel wire on a three sixteenths inch diameter for the input and output coupling capacitors I just have a couple of 3.3 picofarad MPO types and of course we have the coupling capacitor in between the two tuning elements formed by these two wires here so let's hook this up to the spectrum analyzer and see how it performs so I have the filter hooked up to the spectrum analyzer and right now I'm showing a span of 100 to 200 megahertz here and you can see the 3 DB cutoff points are at 134 point 5 megahertz and 141 megahertz and at the center frequency here which is just about 137 megahertz we have just about 1 DB of loss here so that's pretty darn good and this filter matches almost exactly what was predicted in the simulation now of course it did take a bit of tweaking to get it to this point obviously I have three capacitors that I can vary here if I modify the air variable capacitors I can change the center frequency up or down as I need to and if I play around with that coupling capacitor I can kind of change the shape of the band width there in the middle so you can see I'm under coupled here so I'm getting a bit of a narrower band width but I'm also getting a lot more loss so if I kind of twist those wires back together I start getting a little less loss I can kind of play around with that till I get it just right now as you can see there is a sharper cutoff here down towards the lower frequencies then there is towards the higher frequencies that's actually expected when you have a bandpass filter that's coupled with capacitors typically this is the response you'll get you'll get more attenuation on lower frequencies than you do on the higher frequencies and in this particular application that's actually desirable these stronger frequencies that I want to reject are actually down below my centre frequency but if you wanted to have steeper attenuation on the higher frequencies instead you could use inductors to couple the elements together rather than capacitors so this filter seems to be performing pretty well at least in the 100 to 200 megahertz range but if I change my frequency span to the full span of the spectrum analyzer which is up to 1.5 gigahertz you can see that up at higher frequencies we're really not getting a lot of attention at all in fact as we approach 1 gigahertz here we're only getting about maybe 5 or 6
Thanks Glen your participation is very much appreciated
- Kacie Cotti
About the author
I've studied jurisprudence (philosophy of law) at Alverno College in Milwaukee and I am an expert in international organizations. I usually feel alone. My previous job was medical equipment preparer I held this position for 5 years, I love talking about slacklining and knitting. Huge fan of Noomi Rapace I practice curling and collect collectible card games.
Try Not to laugh !
Joke resides here...