FM Bandstop Filter

[alterbaron]

A fun little project inspired by playing around with an rtl-sdr dongle.

An FM bandstop filter or "trap" is a circuit that severely attenuates frequencies in the FM broadcast band (88-108MHz) while ideally allowing other frequencies to pass through unabated. This is my first attempt to design and build one.

I started off by playing around with a few circuit ideas in LTSpice, eventually settling on the design below:



The parallel LC circuit on the left has a resonant frequency of about 108MHz, i.e. the end of the FM band.
Similarly, the parallel LC circuit on the right has a resonant frequency of about 88MHz, or the beginning of the FM band.
Each of these parallel LC circuits presents an (ideally) infinite impedance at resonance, strongly attenuating signals in the vicinity of 108MHz and 88MHz, respectively. These circuits should have relatively high Q in order to give a nice, sharp roll-off.

The series LC circuit in the centre has a resonant frequency of about 98MHz, which is right in the middle of the FM band.
This circuit has a low impedance to most signals in the FM band, and shunts them to ground. The Q for this circuit should be a little lower to make sure it can present a low impedance to the majority of the FM band.

Here's the prototype:


The board is single-sided copper clad FR4 (or at least something similar to FR4 that I got for cheap on eBay). I used a hobby knife to scratch away the copper to form islands, which I could then solder to.

The yellow trimmer capacitors can be adjusted between 10 and 40pF, which is just right for this circuit. Coils are 5 to 6 turns of 28AWG magnet wire on a 1/8" drill bit. The coils are easily adjusted by stretching out or compressing the turns.

Here's a sweep of the thing from 50MHz to 150MHz after a quick adjustment (source power 0dBm):


Attenuation is between 35dB and 50dB, which I think is pretty alright! 

Here's a sweep from 10 to 990MHz:



Loss at the higher frequencies is about 3dB, which isn't great, but not as bad as it could be.

I'll eventually put this thing into a nice little aluminum box for shielding purposes.

Any tips / suggestions on how to improve the performance of this circuit? I'm still quite new to RF stuff, so any and all advice is greatly appreciated! 

[Howardlong]

I'd say that's really pretty good, especially for an RF newbie.

I just picked up one of mine that I use with a $1000 receiver, only to discover that the reason it hasn't been working so well recently is that and L and C have mysteriously broken off.

When I make these I try to get away with fixed lumped parts. In this instance you're rather at the mercy of the tolerances and construction techniques.

To minimise layout parasitics and losses at VHF and above I tend to only use surface mount, typically 0603, fixed lumped parts, and when in doubt I tend to use wirewound inductors for the higher Q. Typical applicable values for 0402 inductors tend to have lower Q than 0603 at VHF which is why I use 0603. Sometimes I need to tweak the values, but for this application it's good enough.

Worrying about 3dB losses at UHF may or may not be a problem, it depends what you're using it for. I hazard a guess you won't notice it unless you're doing EME or other weak signal work, but I doubt you'd be using an rtl dongle if that was your poison!

[German_EE]

If you look at the sweep of your filter you can see three troughs at the bottom, one for each tuned circuit. Unless you really want a wide bandstop filter you should adjust each tuned circuit until all three are at the same frequency, attenuation will then be better.

Response outside of the FM Broadcast band will partially be due to coupling between the various elements. Screens between each tuned circuit will help.

[KJDS]

If you're looking to use this for 144MHz and above, then a five section lowpass quasi-elliptic will probably produce lower loss. Three shunt LC sections with two caps on the top line. You still use three inductors as per your current design, but two more caps.

I'll try and run a design later if I find some time.

[alterbaron]

I played around with the tuning this evening, but couldn't manage to squeeze significantly better performance out of this circuit.

Decided to put it in a nice little aluminum enclosure for shielding purposes.
The enclosure is an Eddystone (Hammond) 11451P http://www.hammondmfg.com/pdf/11451PSLA.pdf

Drilled a couple of holes for the BNC connectors, mounted them in the case, then dropped the board in. The board was attached to the connectors with some solid copper wire.





For a quick and dirty test, I connected a length of wire "antenna" to the input of the spectrum analyzer and recorded the received spectrum from 50MHz to 1GHz. I then connected the FM bandstop box between the antenna and the analyzer and recorded the spectrum again.

Before:



After:



The FM broadcast band (the really strong band at the far left of the "before" image) has been attenuated by somewhere around 35dB. Signals from other bands >150MHz were attenuated by a few dB, as expected.