An introduction to practical RF filter design by building, testing, and tweaking a 137MHz bandpass filter suitable for NOAA APT satellite reception. Additional references and reading materials at: www.analogzoo.com/2017/03/prac...
@@MuhammadAwais-kv9bm it can be both. capacitor and inductor, like Nikola Tesla's bifilar coil. Which has increased voltage difference between its windings.
Yes, such gimmick capacitors would be found in short wave tuning sections of transistor radios ; you would usually see a small stiff metal rod sticking out from the undersurface (soldered to copper lamination, hence grounded), which forms one plate of the capacitor and a part of thin enameled copper wire twisted on to it (like a creeper) that forms the other plate of the capacitor. The drawback would be collection of dirt, if not sealed properly, that would increase the capacitance of the system in due course. And a good demonstration of the gradual cut off on the higher side of the filter because of the coupling is a capacitive type. Thanks 👍
you all prolly dont care at all but does anybody know of a trick to log back into an Instagram account..? I was stupid forgot the login password. I love any help you can offer me!
I watched this video on a whim, I thought filter design was too complicated for me. But this was well presented, and has gotten me excited to play with my own designs. Thanks!
I guess I came late to the party... I just recently discovered your channel by accident and I found the instructional content to be fantastic. So now I'm starting at video #1 and working my way through. Hopefully I'll be caught up by next year :) Keep up the great instructional focus!! rgds,
I like that these constructs show the basic physics and not just engineering. Most super engineering devices kinda hide the physics pretty well in the packaging. I can almost see the field lines here.
I never thought of shielding as being so, so important! I liked your first RF filter with the coils and air capacitors--it looks like old-school wiring. Now, when I was fixing old CRT TVs, I am sorry that I did not open up the terret tuners and collect the small coils and caps. Thanks! Frank
Old CRT tuners won't have variable capacitors. They would have varicap diodes and inductors; you could have some feed-through capacitors but would be very difficult to remove without breaking some.
@@subramanianr7206 real old tuners had cavities with capacitors and were mechanically tuned. I modified several to work as band pass filters with the push buttons set up to tune the center frequency to what I wanted. it was easy to have 10 different center frequencies with a 10 mechanical push button array, I know I am dating my self here lol. but we are talking about the 60s here when most people on here probably weren't even born.
Very, very intresting and detailed class on BPF work. Now, I understand the if cans a bit more. So.... so maybe i'll try to build an old fashion am radio from scratch. thanks to share!
Can you tell more about this filter topology? I mean, with coupling capacitors. What I usually see is one pole composed of series components, another - with shunt elements. What is the difference and how do you calculate coupling capacitors value? And, thank you for getting back to youtube, your videos are really great.
Really informative and insightfull !! A topic completely unrelated,is a suggestion for future content on rf oscillator design,with calculations and a working circuit to prove,the circuit actually oscillates at the frequency required.I have had varying degrees of success,with a lot of different types of oscillator but have never had a circuit reliably oscillate above 30 MHz. This may have a lot to do,with the q of the tank and choice of active device ,but is frustrating when simulated works but never in an actual circuit.
Wondering if there is an actual calculator "On Line" that will give the diagram with the proper values to achieve this in different frequencies??? By the way really nice video...
You're very good at presenting this. It's the cond time I watch :) I would love to see how would you make bandpass filters with SMD components on a manufactured PCB.
Nice video!! I have built many filters like this. the highest frequency I was able to achieve was around 1.2 gHz with lumped elements above that you start needing strip-line components. Have you tried the Dishal's method of tuning. basically you short to ground the section just past the end section you are tuning and use S11 return loss connected to the end of the filter and tune for max --dB dip at center frequency. Then do the same to other end of filter. if it has larger number of poles than 3 then short the third section in from end and tune the second section for double dip with lowest response at the center frequency. do the other end and repeat till all sections are tuned. ;>) Elements of filters around 1gHz are .
I grab all of mine from old junked commercial electronics gear. To an RF hobbyist there are often hundreds of dollars worth of useful parts in a single old chassis. It can be slow and tedious desoldering parts, but I figure it's like getting paid $50 to $100an hour to have fun. :)
RF design is sort of obscure due to details like these. Using a PCB with traces sure would impact a design like this (as PCB traces may present a capacitance of up to 4pF to ground). I see point-to-point construction being used instead. Now I know why. Nice design!
Thank you so much for sharing this practical knowledge. It always helps to design circuits when you have good test equipment too lol, where you can see what is actually happening
The information is very under standing, i want to design a filter using Ka-band frequency (18,32 GHz- 20.87 GHz) using microstrip technology, to have a good return loss and insertion loss. how do i go about it. please in your contribution. from Victor
what a coincidence. I build a few of these diy style doubly tuned filters for an IF Stage. I was wondering where these type of filters fit in with the usual Butterworth and Chebechev bandpass filters? I beleive all of these filters have similar equations albeit with different coupling coefficient arrangements. This explains the asymmetrical response. Butterworth (and other usual ones) for example use both inductors and capacitors as the coupling element, so the zero introduced my the coupling capacitor is cancelled by the inductor leading to a symmetrical response. would love to have your insight on this!
I've seen two instances of filters almost identical to ttys0's demo where there was no coupling capacitor at all :-) One example was an old-school tuned TV preamp, and the other example I noticed today, in a RU-vid vid presented by Mile Kokotov (who has also chimed in to congratulate here, without self-promotion). In those cases, the coupling is apparently provided by inductive transfer between the independent inductors, or whatever messy kind of field gets excited in the filter's enclosure = cavity :-) That's right, the filters along those lines probably work best if sealed in a metal compartment (and would probably otherwise leak energy to the outside).
This was really useful thank-you. I have inherited some unmarked filters and don't have much test gear. What do I need to get so I can work out which bands the filters are for?
Love all your videos. I was wondering what type of variable capacitors you used in this video? Air variable and piston is what I heard but who makes them?
They're not available now except for one possibility of salvaging from vintage VHF circuitry; maybe you can rarely find on the eBay. Some types of piston air capacitors were found in the Broadcast band tube and transistor radio receivers.
Have a question. Why lower frequency rejection is better than higher frequency? Read several papers that indicated when inductive coupling is dominant, TZ will appear at higher frequency; conversely, if electric coupling is dominant then TZ will appear at lower frequency which is the case here. But why is that? What's the mechanism that makes TZ reveals itself in such way? Can someone explain, I appreciate you in advance.
Is there any way to make these filters have a sharper curve? There are some strong VDL2 blips on 136.7 MHz that overload the SDR, Meteor LRPT is much more affected as each time these blips appear the meteor demodulation loses sync. Or is there a way to make a notch filter with a similar design?
I watched the video w/o reading the notes you made first...as soon as I saw the center frequency I knew exactly what your filter was for as I monitor the WX sats as well; I use a very old Hamtronics R139 and Quorum crossed dipole array that has a built in bandpass filter and amplifier right in the mast's tube. Feed it a little 12VDC over the coax and the birds normally give me a clear view from around Cuba to well north of the Great Lakes (the antenna is actually in the upper attic of our house). Maybe you can do a video on your system and show folks the impact of interference from the services on either side of 139MHz and how your filter works in practice? 73 - Dino KL0S
Definitely! I'm in the process of putting together a preamp and a reasonable antenna, which will hopefully work better than just a little rubber ducky. :)
The "WXtoImg" software suite works well for decoding...never know you may want to move up to the GOES birds as well (I haven't gotten to that project yet!). Good luck! 73 - Dino KL0S
Very nice video ! Any ideas on how to plot the frequency response of such a filter on an oscilloscope ? Unfortunately I don't have a spectrum analyzer :(
Dave Jones did a video on how to do exactly that, in conjunction with a function generator: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-uMH2hGvqhlE.html. Your function generator and oscilloscope have to have sufficient bandwidth for the frequencies your testing of course.
Dave suggested a good way to do it; however I really liked how in your video you could read the center frequency of your filter; the corner frequencies etc. I don't think Dave could do that with his solution; thanks for your response though! And again thanks for the video it must have taken a TON of time to make
Use a design tool, or calculate it yourself, to do predistortion when you design to take into account the L/R ratio of the inductors. Results may be better.
For the simulation you can use: www.changpuak.ch/electronics/Direct-Coupled-Resonator-Bandpass.php Center Frequency [MHz]: 137 Bandwidth [MHz]: 6 Passband Ripple [dB]: 0 Number of Resonators: 2 Impedance [Ω]: 50 Inductance: 100 nH use my value
Actually the piece of twisted pair might behave a little like a transmission line, in that the capacitance and inductance would be spread out in space and consequently time... and there's an open ending = resonant stub :-) but realistically, these effects depend on wavelength, and if I recall 13 mm is a 1/4 wave (in the air) at 5-6 GHz, so this is likely out of band for the relevant receiver (and also for the author's spectrum analyzer in the vid). As for inductace brought about by the twisting... try some inductance calculator, some are mentioned in this debate :-) Mind the tiny diameter and long twists... (turn spacing)... 20-30 nH is my gut feeling.
the twisted wires technique is so impressive...but how much is it reliable when using it as really small caps in RF transmitters like 433Mhz DIY tranceviers ? and thnk U so much for this wonderful video :)
Hi! Just one question. Considering that the magnetic field is more concentrated around the center line of the coils which will create coupling at higher frequencies when the coils are aligned like in your first prototype, would it be ok to just make the coils vertical and parallel and skip the shielding?
Not in my experience, no. You really need shielding to provide adequate attenuation at UHF frequencies. OTOH, if there are no interfering signals at UHF that you need to filter out, you may not care so much about the filter performance at those frequencies, and just forego the shielding.
Ideally you would wind the coils perpendicular to each other or on toroids (may not be possible at such high frequency) - that will reduce the unwanted cross-coupling. Shielding may still be required but will be less critical.
Could you please talk about the antenna? How did you get your filter to receive that signal in the first place? How did you connect the antenna? Or you didn't use one?
Same question other people asked - could you _please_ explain how exactly (software, books, etc) this particular schematic was chosen? The reason why I'm asking is that it doesn't look like Butterworth filter, more like two coupled "basic" LC bandpass filters with f0 = 1/ (2 * PI * sqrt(L*C)) = 1/(2*math.pi*math.sqrt((100/1000/1000/1000)*(10/1000/1000/1000/1000))) = 159 MHz that were tweaked (manually or in SPICE simulator) for 137 MHz afterwards considering effects of coupled capacitors. The book "Practical Electronics for Inventors, 4th Edition" by Paul Scherz, Simon Monk describes two types of passive band-pass Butterworth filters (wide-band and narrow-band), none of which looks like your filter. There is an open software Qucs and apparently it uses the same algorithm as described in the book: qucs.sourceforge.net/screenshots.html
You're absolutely right, this filter is two LC tuned filters with capacitive coupling between them. It's often referred to as a double-tuned, capacitor-coupled, or top-c coupled, bandpass filter. Chapter 2 of Chris Bowick's book, RF Circuit Design, covers this topology in a very understandable and practical way if you're more interested in the effects of the coupling capacitors. Many LC filter design tools support this topology; I used Keysight's Genesys software, which is a professional (and expensive!) tool, but Elsie by Tonne Software is an excellent and free filter design tool that supports this topology as well. As for why this particular topology was chosen: top-c coupled bandpass filters tend to have more attenuation at frequencies below the passband, and since 137MHz is just above the FM broadcast band, attenuating those signals (which can be quite strong!) more heavily was an advantage.
@@Analogzoo and TWIMC - turned out Elsie doesn't support this particular topology. It supports others but if you want a narrowband filter in practice you will get high insertion loss with these topologies, 10+ dB. There are several online-calculators based on the paper "Direct-Coupled Resonator Filters" by Seymour Cohn that do the job, e.g. www.changpuak.ch/electronics/Direct-Coupled-Resonator-Bandpass.php# The book "Solid State Design for the Radio Amateur" describes these filters for HF bands well and contains pre-calculated values. The book is available for free on archive.org
Hi, can you give me a reference book on passive filters ? I mean, how did you know how to arrange the lumped components in such way to get a BPF. Thanks !
You can use a noise source, or a tracking generator (a function generator typically integrated into the spectrum analyzer, which is what I used in the video) if your spectrum analyzer is a swept-frequency type. Check out these videos for some more detailed info: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-bbdTRX4_2DE.html ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-Wg3PNgGW_M4.html
What do you use to cut your PCBs btw. ? I am still searching for something that is quick, not messy and doesn't dull sawblades down really fast... For your small value coupling caps, have you ever tried to modify old carbon trimmer pots by gluing a layer of mylar (or similar, maybe with some metal foil on the carbon side) in there and alter the shape of the wiper to make it be a tiny variable cap?
Dennis Lubert The fastest and easiest way I have found to cut any fr4 pcb is simply scratching a line on the top and the button of the pcb using a sharp carbide tip and straight edge and simply snap it with your hands, but the key to do it right is to totally scratch the copper layer until your reach the fr4 material . that wouldn't require any special tools just an old drill bit and straight edge.
Yeah, for small pieces this sometimes works, but the FR4 needs to be sufficiently small and its not easy to get the scratches perfectly lined up. My stock of double sided copper clad is 500mmx400mm and especially the 2.2mm thick ones love to break everywhere but the scored line. My current way is to v score them in the mill, but the glass fibres seem to be really bad for the drill bits
I use tin snips for small pieces. For larger pieces, I have a sheet metal shear that I picked up from Harbor Freight for $50 IIRC. I've also heard people using paper guillotines (the old-school heavy duty metal kind) for cutting PCBs, but the shear is probably going to be easier to get a hold of.
My English is not al the best but a try to say that it was a interesting video because a have pager problem on receiving 137.6210 ( N.O.A.A_15 ) .....but it is strange ..because it is only on saterday evening between 9:48 pm and 10:05 pm ??????
If you're talking about molded inductors, no; they tend to have very low Q values (at least I'm not aware of any high Q molded inductors that would work at RF frequencies like this). Coilcraft specifically sells pre-made "high-Q" inductors (www.coilcraft.com/1008hq.cfm) whose Q values tend to be in the 50-75 range; I doubt many other pre-made inductors would top that. I haven't measured the Q of the air wound coils that I used, but the low insertion loss of the filter suggests that they have Q values in the hundreds (you can get a rough estimate by adjusting the inductor Q value in a simulator until the simulated results have roughly the same insertion loss as the real filter).
I second that - the molded inductors, especially those on feromagnetic cores, tend to have a mediocre Q. And then you can discern between chokes intended for SMPS use (the Q is still relatively decent) vs. for EMI suppression (the Q is piss poor, on purpose). Hardly anything can beat the Q of an air-core inductor - except you can use wire with silver- or gold-plated finish (rather than enamel coating) to cater for skin effect a bit. Also, bare wire with no insulation resonates better than anything insulated - probably due to different and inferior dielectric properties of the insulation. You can possibly get an even better Q from resonant cavities (with polished bare gold-plated finish) but that's a whole different ball game I guess: coupling to a cable takes some know-how, there are harmonic resonant modes etc.