Your Software Defined Radio around the home

Foundations of Amateur Radio

What would you do if you found that at random times your garage door opener didn't work, or the Wi-Fi network dropped out, or you couldn't switch off a light with an RF controller?

That's the position I found myself in and the times at which this was happening were madly unpredictable. One moment everything would work fine and the next all things radio would just stop.

As a radio amateur you're likely nodding your head and thinking, radio interference, there's some direction finding in your future. Sure enough, that's the case, but before that, I needed to know if the interference was random, if it had a particular pattern and how widespread it was, since it seemed to impact multiple different devices using different parts of the radio spectrum.

Initially I focussed on getting a recording of it. I turned on my radio, tuned it to a 2m frequency and recorded the noise. Only one problem. There was no noise. All I could see was an extreme signal strength, but it wasn't showing up as noise.

I enrolled the help of my RTL dongle and recorded some raw data, essentially capturing a 3 MHz slice of noise centred around 147 MHz. All that revealed was that there was noise. I already knew that.

At that point I decided that a bigger hammer was needed. Something you can do if you have a $5 RTL-SDR dongle and some free software, in my case I used a tool called rtl_power and a visualisation tool called gnuplot.

rtl_power is a nifty piece of software. It takes measurements and averages out the power level across the measurement range. To make it work, you specify a starting frequency, a stopping frequency, how big a step to use to average, how often you want to measure and for how long.

For my little investigation I started with measuring between 0 and 1.7 GHz, at 1 MHz intervals, every 2 minutes for 10 days. That creates a big CSV file that you can process with gnuplot into a picture that tells a thousand lies.

Seriously, it showed me that the interference was very wide, 0 to 300 MHz, it occurred every 20 or so hours, lasted up to six hour at a time. There were other things happening as well, similar patterns, but across an even larger frequency range, from 0 to 600 MHz, but in shorter duration and of lesser strength.

Based on the times alone, I can immediately, almost certainly, eliminate any source under my control.

Based on the timings I can also determine that the noise is likely not created by an automatic process, given that they vary in duration and the way they're clustered around specific times.

The variation of the interference allows me to determine that there are at least three separate types of noise, each with specific characteristics and times, sometimes overlapping.

It's too early to tell if this pattern will continue. One possible next step is to set up the same measurement tool and powering it from a battery. Once I've got that working, I expect to turn off the house power during an interference session and determine if the noise is coming from my house, or if it's an external source, which seems likely.

Once I've determined if it's in house or not, I can start either eliminating gadgets by switching off specific power circuits, or I can start direction finding and locating a nearby source of pain.

At that point I can decide what to do next. That said, at the moment it looks like several televisions around me are creating an RF noise storm of epic proportions.

I've documented all of how I did this and you can find it and the scripts I created on the web at vk6flab.com.

One thing that has happened since I started documenting my efforts is the idea that we could collectively as a community make measurements like this and document the state of our RF space and how it changes over time. I plan to update my code to incorporate this idea, perhaps log in 24 hour blocks and generate a chart over that time, perhaps make it into a video.

One challenge ahead of us would be to come up with a universal way to calibrate our various dongles, so we all report the same signal level in the same way. One thought is to use the sun as a global calibration, but I'm not yet sure how that might be implemented.

One thing's for sure. If you've ever wondered what use can a $5 RTL dongle possibly be, this is one thing that you just cannot do with a traditional radio. That's not to say there's a place for both in the world, just different tools for different problems.

I'm Onno VK6FLAB