Now that I’m working in the kitchen, I’ve noticed a couple of times my internet practically cuts out whenever Matthieu comes in to use the microwave. I wondered why and… it turns out the microwave shares a 2.4 GHz radio frequency with our WiFi connection. This is apparently such a common problem that Ofcom felt it necessary to tell people working from home.

It turns out your home WiFi router is like a very low-powered microwave, which is why you should always turn off your WiFi when you’re not using it, and hold your phone at least 10cm away from your head during calls.
This is why tin foil hats are back in fashion now ;)

Here’s what the floor plan of our house looks like.

floor plan

The microwave sits between my computer and the router.

I decided to test it properly.

wget -O /dev/null http://speedtest.wdc01.softlayer.com/downloads/test100.zip

This just downloads a 100MB zip file straight to nowhere.

We can see that the microwave obviously slows down the download rate. Does microwave strength make a difference?

I put another mug of milk in the microwave, turned the power level down to its lowest setting, and ran the speed test again.

low power microwave

This time the download completed in 3m13s at 583KB/s. So, it is noticeably slower, but the effect is not devastating.

There are other ways of measuring the WiFi connection in GNU/Linux, you can see some more numbers through this:

cat /proc/net/wireless

This gives values for noise and connection quality. I think those values determine the connection strength shown in the network manager panel icon.

network indicator

So, I wrote a script to output the connection quality to a file every 0.1 seconds, and I went to turn on the microwave again.

network quality graph

The average network strength over that period was 52% and it doesn’t show a sustained drop in connection quality even while the microwave is running. There are occasional drops and rogue data points, but these have been smoothed out by the line of best fit.

The network slows down, but the actual connection between the router and the computer isn’t too badly damaged. This kind of makes sense when you understand that 2.4GHz WiFi is… fairly persistent. It might not be as fast as 5GHz, but it is able to maintain a connection behind multiple walls, over long distances, and through significant interference.

Up until now I’ve been using an official Rasperry Pi WiFi dongle plugged into my computer.

WiFi dongle

It uses the Broadcom BCM43143 chipset, there are free drivers for it available with Debian, which makes it very handy as a sort of all-purpose sonic screwdriver. It just works for whenever you need to connect something to WiFi and don’t want to deal with driver issues.

Meanwhile the network controller on my motherboard is an Intel Wireless 7260, which helpfully also has mainline driver support. Unfortunately, I left the wireless antennas for it in Leicester, so this weekend I ordered some new antennas.

WiFi antennas

I got some really long ones, they’re like drumsticks… the longer the antenna the lower your fertility better the signal quality? More importantly, my internal network card can use 5GHz WiFi, which means I’ll avoid radio interference from the microwave.

With the antennas connected, I ran the speed test again.

Now we see clearly the microwave has (almost) no effect. Yay!

What about the connection quality? I ran my test again over the course of a few hours (while writing this), recording one measurement every second.

graph with antenna

Unfortunately these numbers don’t tell us very much. The average link level was 44%, a bit lower than the 2.4GHz connection.
I can feel the speed boost from the fancy new antennas, and glad to get fast wireless internet even in the kitchen.