RTL-SDR bandpass-filter and DIY coax-notch filters

Some days ago I decided to install an APRS RX-only iGate here in my apartment. The iGate simply takes the APRS signals from the air and forward them to the internet. I live almost in the center of the city and I have a free sight over the buildings. Good preconditions for a „fill-in“ iGate I thought. In this project I learned more on designing HF-filter than on Linux and Raspbian, what actually was my aim at the beginning of this project.

On Silvio's website I found a manual (German manual, english manual) on how to use an Raspberry Pi and a cheap RTL-SDR dongle (DVB-T stick) for this purpose. Sounds like an easy job. After I had everything set up (thank you Silvio, DM9KS) I received nothing within the next 12 hours. I knew that something must be wrong. I live in the capital city (Saarbrücken, Germany) of my state with a population of about 200k, so there should be some APRS signals in the air.

I googled a bit and found out, that those 10 bucks RTL-SDR dongles have a wide-open frond end with no filters in it. I had no clue what to filter out, so for me the most logical thing was to look for a 2m bandpass-filter. I wanted the APRS frequency – which is 144.800 Mhz here in Europe – to pass the filter and everything else should be attenuated. So far so good.

I called my friend Peter, DC2ZO who is an telecommunications engineer. I asked him if he can help me to develop a 2m bandpass-filter. He had one already made in his stock and only a bit later I had the filter in my hand. I had to fine-tune the filter for 144.800 Mhz, which I did with my network analyzer, having Peter on the phone answering all my questions. After some time I had the filter set to 144.800 Mhz with 0.8db of insertion loss and 20db attenuation on everything else. I was happy that I tuned my first filter (built by Peter, DC2ZO) and inserted the 20db bandpass-filter between the antenna and the RTL-SDR. After some time, I did not receive any packets.

I knew that I had two very strong 100kw (!) radio transmitter just in a line-of-sight. They transmit 101,7 Mhz (Radio Salü) and 103,7 Mhz (UnserDing) from „Sender Riegelsberg-Schoksberg“, so I knew I have to take them out. But how? I called Peter again and he told me about the ¼ wavelength coax-notch filter (also called „stub-filter“). I googled a bit and found a very nice calculator for the coax-filter and also a short manual on how to solder coax-t-connections if you don't have any T-adaptors on hand.

My iGate was running during the night, and I went to work with my APRS tracker on. I received nothing with my 20db bandpass-filter, so I started just after work to make the notch-filter to take 101,7 Mhz out.

After some tests with the 101,7 Mhz filter I still received nothing, even with a 59db attenuation for this frequency.

Peter was pretty sure that there must be some something else with big power in my area. I set up the spectrum analyzer and found 2 big peaks besides the 101,7 Mhz radio station. It was 178.352 MHz (DAB+ 5C) and 202.928 Mhz (DAB+ 9A), both pushing another 50kw each from the same tower as 101,7 Mhz.

Peter suggested to make 2 additonal coax-filters and solder them to the existing one for 101,7 Mhz. I did the (kind of messy) soldering, remember I had no T-connectors on hand and set up the network analyzer again.

After slowly cutting piece by piece I tuned the second stub for maximum attenuation on 178.352 Mhz and then I tuned the third stub for maximum attenuation on 202.928 Mhz. Surprisingly this worked and the 101,7 Mhz stub was still good. I had 33db-35db attenuation so far whithout the 2m bandpass-filter in line.

After this I had to re-tune the 2m bandpass-filter, which I added in series again. The analyzer showed 12db loss, that was far to high for the APRS signals. I was able to tune the filter for a 4db loss on 144.800 Mhz and a attenuation of 50.16 db on 101,7 Mhz, 58.38 db on 178,35 Mhz and 49.32 db on 202,92 Mhz.

I went to bed and the next day I received my first APRS packets which came from DC0VZ who passed my iGate from about 6km distance.

I later tested this myself and it works pretty good so fa:

Thank you Silvio, DM9KS for the idea and Peter, DC2ZO for answering so many questions and also for the 2m bandpass-filter.

Update 03.11.2016: If you do not have a network analyzer to tune the coax-stubs, use the calculator which I linked above only. If you use the exact length, you will be very close to the frequency you like to attenuate. That should help in most cases.