I will be writing more about broadband antennas for AM BCB in the future, but one of the issues I have is the dual problem of using an inexpensive low dynamic range RTL-SDR dongle as a receiver and having two local stations right next to each other on the dial. Granted, this is a much easier problem to deal with than in a big city where there might be many strong stations spread across the dial. At least if one’s worst problem stations are at either the low end or high end of the dial, filtering becomes a practical option.
I live within 2 miles of the transmitters for 1 kilowatt stations on 1450 and 1490 kHz. While I’m lucky in that the station on 1490 only broadcasts from 5am to midnight even though it is licensed for 24-hour operation, with the stations so close to each other, the frequencies can interact and create intermodulation products at 40 kHz intervals below 1450 and above 1490. (In fact, 40 kHz is the closest allowed separation for stations serving the same community; why our only two AM stations are this close on the dial is beyond me.)
Anyway, when using an untuned loop antenna for broadband coverage, the signal tends to be quite weak, requiring significant amplification to get decent signals at all broadcast frequencies and to pick up trans-Pacific or trans-Atlantic signals if so desired. But, strong local signals can overload with amplification and cause problems at other frequencies. I find that I can’t otherwise amplify a broadband loop such as a Conti Superloop due to the strong 1450 and 1490 signals. A more expensive SDR with a higher bit rate A/D converter would probably help with the dynamic range, but this seemed like a good opportunity to experiment with filtering.
It’s possible to build one’s own filters out of discrete components, but the frequency precision required for something like this makes the project quite challenging. Thus, I went with off-the-shelf filters from Mini-Circuits. They seem to have the best options at lower frequencies like this. I narrowed it down to two filters, the LPF-B0R7+ and the LPF-B0R8+. These are both surface mount filters as opposed to “plug-and-play” filters. Unfortunately, they have to be purchased in bulk, at least 10 units. Thus, the cost is about the same ($80-90) as a cased filter with connectors. (Note that the plus sign at the end of the part number just means that it satisfies the international RoHS guidelines for hazardous substances.)
The following plot uses the data available on the Mini-Circuits site, showing from top to bottom the response of a single LPF-B0R8 filter, two LPF-B0R8 filters in series, and a single LPF-B0R7:
You can see that while the LPF-B0R7 kills 1450 and 1490 kHz, it also wipes out too many lower frequencies. A single LPF-B0R8 is actually pretty decent, but I wanted to make sure I really diminished 1450/1490 and had a sharper rolloff. Plus, having to buy 10 units, I had plenty to spare, so I build a double LPF-B0R8 filter. When you multiple numbers together, the logarithms add, so the loss depicted on the graph for the double filter is exactly twice the value for a single filter.
The data sheet for the filter indicates which of the solder pads on the bottom of the filter are input, output, and ground, so doubling up is just a matter of making sure that the output for one filter is connected to the input of the second filter. I put the filters on perfboard, soldering from below to get enough solder through the plated hole up to the ground pads to attach the filter to the board. One has to be a little careful not to overheat the component while soldering. I ran 20-gauge connector wire from the “hot” lead of an SMA connector to the input of the first filter, another wire from the output of the first filter to the input of the second filter, and a third wire from the output of the second filter to the “hot” lead of the output SMA connector. I then ran “ground” wires from two of the ground posts of the input SMA connector to the corresponding posts of the output connector, making sure to make a solder connection through the perfboard to the ground pads of each filter along the way. The following pictures show the result from the top and from the bottom:
It turns out that the filter works pretty much exactly as expected. Here is a screenshot from the gqrx SDR software of AM BCB from 540 through 1490 kHz:
You can see the general signal dropoff at high frequencies along with the dropoff in noise level. The stations at 1450 and 1490 kHz are still fairly strong, but without the filter they would by far be the strongest signals in the band and at this gain level would overload the RTL-SDR dongle. The signals beyond that are basically buried in the noise, which is collateral damage in this case. However, that’s a small part of the band and there are very few stations beyond 1600 kHz in any case. I can use a tuned loop in this area to sample individual frequencies.