Finnputer All articles
Opinion

Your Leftovers Are Jamming the Channel: A Lunchtime Spectrum Audit

Finnputer
Your Leftovers Are Jamming the Channel: A Lunchtime Spectrum Audit

Photo by Photo by Jakub Żerdzicki on Unsplash on Unsplash

The microwave is the most honest appliance in your kitchen. It has one job, one frequency, and one weakness: the door. Your Wi-Fi, by contrast, has eleven overlapping channels, a dozen neighbors with default settings, and a mesh node you bought at 11 p.m. out of spite.

Which makes the microwave a useful suspect, because it is the rare suspect that will confess on schedule. Set three minutes, press start, and watch what happens to your latency.

A true story about radio astronomy and a snack

For years, the Parkes radio telescope in Australia recorded mysterious broadband bursts nicknamed perytons — brief, powerful, and maddeningly irregular. In 2015 the team traced them to staff opening the microwave door before the timer finished, releasing a burst of 2.45 GHz energy into the dish. The universe did not send those signals. Somebody's lunch did.

You are running a smaller version of that experiment every day, except the dish is your living room and the perytons are your video calls.

What's actually leaking

A microwave magnetron runs at roughly 2450 MHz. The cavity is a Faraday cage and the door seal is a choke — a carefully tuned gap that traps the energy. US rules require new ovens to emit no more than 5 milliwatts per square centimeter measured about two inches from the surface. That is small. It is also not zero, and it is broadband, spiky noise sitting about dead center between Wi-Fi channel 6 and channel 11.

Here's the uncomfortable detail: channel 6 occupies roughly 2427–2447 MHz and channel 11 occupies roughly 2452–2472 MHz. The microwave's mess lands in the gap and smears into both. That's why channel 9 was never popular and why the oven is a nuisance even when your router is 'on a clear channel.'

A router radio hears this as a raised noise floor and a stream of corrupted frames. Your Wi-Fi doesn't drop. It retries. Retries look, from the sofa, exactly like buffering.

The tools you already own

You do not need a spectrum analyzer. You need to read what your radio already measures.

Linux: iw dev wlan0 survey dump is the gold standard here. It reports channel active time, busy time, receive time, and transmit time per channel. Busy time divided by active time is your congestion percentage — measured, not guessed. A microwave in the room will move that number in a way no neighbor can imitate.

macOS: Option-click the Wi-Fi menu, open Wireless Diagnostics, and use the Scan window. You get channel, width, RSSI, and noise, plus a crowded-channel chart. No spectrum, but plenty of evidence.

Windows: netsh wlan show networks mode=bssid for a quick inventory, then NetSpot's free tier or Vistumbler for a map you can actually look at.

Android: Wi-Fi Analyzer is fine for seeing who's on channel 6. Treat its noise floor with suspicion.

Optional, if you enjoy this sort of thing: a $30 RTL-SDR and rtl_power will give you a genuine sweep of 2.4 GHz. The microwave shows up as a broad, ugly hump. It is one of the most satisfying graphs in domestic computing.

The lunchtime experiment

This takes twenty minutes and requires a mug of water in the oven — never run one empty.

  1. Park the laptop two meters from the oven, on the counter, with the lid up and nothing metal nearby.
  2. Ping your gateway at a steady rate: ping -i 0.2 -c 300 192.168.1.1. Ten minutes of baseline.
  3. Note your AP's channel and width.
  4. Start the ping log, press start on the oven, wait three minutes.
  5. Repeat at six meters away, because distance is the whole point.
  6. If your AP offers both bands, repeat on 5 GHz and watch the problem evaporate.

Typical results look like this:

Condition p95 latency Packet loss
Baseline, oven off, 2 m 8 ms 0%
Oven on, 2 m away 140 ms 3%
Oven on, 6 m away 22 ms 0.2%

Your numbers will differ. The shape won't: a wall of latency during the timer, and a clean recovery afterward. If your survey tool shows busy time jumping from 25% to 70% in the same window, you have your culprit and you can stop arguing with your internet provider.

Reading the numbers like a Finn

There are three possible verdicts and only one of them is about shopping.

The oven is guilty. Latency spikes only during heating, busy time spikes with it, distance fixes it. Move the access point out of the kitchen — even a wall's worth of separation helps enormously, because leakage falls off fast.

The oven is innocent, the neighborhood is not. Busy time is high all evening and doesn't care about your soup. That's congestion, not radiation. Set the 2.4 GHz radio to 20 MHz width on channel 1 or 11, hard-wire anything that doesn't move, and put the streaming devices on 5 GHz.

Your microwave is fine and your problem is elsewhere. Excellent news, and you found it for free. Now go look at the three access points your neighbors named 'NETGEAR' and 'NETGEAR-2.'

Fixes, cheapest first

Move the access point. Narrow the channel. Use Ethernet for stationary devices. Separate the IoT clutter onto its own SSID or band. Replace a 1998 oven whose door seal no longer seats. Only after all of that does a mesh node enter the conversation — and if it does, place it with a meter in hand rather than by walking around until the app shows a sad face turning happy.

What not to do

Do not run the oven with the door open. Do not defeat the interlocks, no matter how curious you are. Do not download an app claiming to be a radiation detector; your phone's Wi-Fi chip is not a Geiger counter and the privacy policy is worse than the physics.

Measure first, buy second. That's the whole doctrine, and a lunch break is enough time to practice it.

All Articles

Related Articles

One Button Is a Feature: The Finnish Elevator Test for Your Smart Home

One Button Is a Feature: The Finnish Elevator Test for Your Smart Home

No Words, No Excuses: Your README Could Learn Something From a Billy Bookcase

No Words, No Excuses: Your README Could Learn Something From a Billy Bookcase

The Distributor Is Out of Stock. The Fisherman Is Not: Sourcing Components When the Supply Chain Says No

The Distributor Is Out of Stock. The Fisherman Is Not: Sourcing Components When the Supply Chain Says No