A recording studio in a major Canadian city had a problem.
Their condenser microphones started producing a persistent whine. High-end units that had worked flawlessly for years. On the audio spectrum analyzer, it showed up as a clean spike at 4 kHz, with smaller artifacts at regular intervals. When routed to the studio monitors, it sounded like a stuttering tone that would occasionally glitch.
Nothing in the signal chain had changed. Same console, same cabling, same room.
But outside, three cellular carriers had been adding new antenna sectors and frequency bands to rooftop sites less than 100 metres away. The studio suspected the cell towers. They were right about the source — but wrong about the mechanism.
Here’s what the interference actually sounds like. Press play.
That whine you hear is a 4 kHz tone, with harmonics at 2 kHz intervals. The microphones were not picking up phone calls or data. They were detecting something far more subtle — and the explanation starts with how RF energy works.
A quick primer on RF energy
Radio frequency energy is electromagnetic radiation — the same fundamental thing as visible light, just at much lower frequencies. Cell towers transmit RF signals in bands ranging from about 600 MHz to 3.5 GHz. These signals carry voice calls, text messages, and internet data for every phone in the area.
RF energy passes through most building materials. Concrete, wood, glass — it goes through all of them, weakened but not stopped. This is why your phone works indoors. It’s also why sensitive electronic equipment inside buildings can be affected by outdoor transmitters.
The key point: RF energy isn’t sound. You can’t hear it. A microphone isn’t supposed to respond to it. But under certain conditions, some microphones do — and here’s how.
How a condenser microphone works
A condenser microphone converts sound into electrical signal through a clever trick of physics. Inside the mic, there’s a thin conductive diaphragm — essentially a tiny metal membrane — suspended very close to a fixed metal backplate. Together, these two surfaces form a capacitor.
When sound waves hit the diaphragm, it moves. That movement changes the distance between the two plates, which changes the capacitance. To convert those tiny capacitance variations into a usable electrical signal, the mic uses a FET (field-effect transistor) amplifier with extremely high input impedance.
How high? In the classic Baxandall circuit design from 1957, the input resistance is 100 megaohms. Some modern designs push to 1 gigaohm. This is necessary — a lower impedance would drain charge from the capsule too quickly, killing low-frequency response.
Step through the signal path to see how this works:
Notice stage 4. That’s where things go wrong.
The accidental RF detector
At impedances of 100 megaohms to 1 gigaohm, the FET gate and diaphragm assembly form something unintended: a broadband RF voltage probe. Any amplitude-modulated RF energy present at the capsule gets rectified through nonlinear junction behavior at the FET.
Think of it this way. The FET junction acts like a crude AM radio detector — not by design, but by accident. It strips off the carrier wave and passes along whatever amplitude pattern was riding on it. In radio engineering, this is called envelope detection.
The microphone isn’t hearing cell towers. It’s demodulating them.
The mesh screen at the top of the mic is supposed to be an acoustic window, not an RF shield. Woven stainless mesh — the material used in most mic windscreens — has gaps far too large to block cellular frequencies. At 2600 MHz, effective shielding requires apertures below about 1 mm. Most mic screens have openings many times that.
And the problem has a nasty nonlinear characteristic. Try the slider below:
Below a certain RF field strength, nothing happens. The microphone’s internal suppression margin handles it. But once the cumulative RF power from all nearby transmitters crosses that threshold, the artifacts appear suddenly. This isn’t a gradual onset — it’s a cliff. That’s why the studio went from zero issues to a severe problem seemingly overnight. Three carriers adding new bands and sectors pushed the aggregate field strength past the tipping point.
Where the 4 kHz fingerprint comes from
This was the part I couldn’t find documented in publicly available literature at the time of the investigation.
To test the hypothesis that the cellular signals carried an amplitude modulation component at 4 kHz, I used a spectrum analyzer configured as a tunable AM demodulator. By tuning to individual carriers and examining the demodulated audio output, I could see what the microphone was detecting.
The 4 kHz component was present on carriers across multiple bands — 700 MHz, 1.9 GHz, 2.6 GHz — from multiple operators using different equipment vendors. The demodulated audio had the same spectral fingerprint everywhere: a dominant 4 kHz tone with components at 2 kHz intervals. It stuttered on and off, matching exactly what was heard in the studio.
Here’s the actual measured spectrum from the interference recording:
The even-harmonic pattern — strong at 2, 4, 6, 8, and 10 kHz, weak at odd multiples — is consistent with full-wave rectification at the FET junction. The 4 kHz component at 35.9 dB dominates, standing roughly 20 dB above the noise floor.
I confirmed this signature at multiple cell sites across a wider geographic area. The pattern appears to be an inherent characteristic of the cellular signal structure in the bands and technologies deployed during this period. I want to be precise: this observation was empirical and should not be generalized to all cellular deployments everywhere.
The investigation
The first thing I did was disconnect the microphones from their cables, leaving the cables plugged in but unterminated. The artifacts vanished. The console was silent and clean. The interference lived inside the microphones.
Reconnecting brought the artifacts back. We started muting individual frequency bands on the nearest cell site. 700 MHz off — no change. 850 MHz off — no change. AWS band off — still there. Even with every radio on the nearest site shut down, the whine persisted. Other operators on adjacent rooftops were still transmitting.
Only when all carriers from all operators were simultaneously muted did the artifacts drop below the noise floor.
Testing different microphones told the rest of the story. Certain pencil-type condenser models were highly susceptible. Others from the same studio inventory showed nothing at all. Moving a susceptible mic even a few inches changed the interference level dramatically. A walk-test with a battery-powered preamp showed interference dropping steadily with distance from the cell sites.
Ferrite chokes clamped onto the balanced mic cables had zero effect. In a properly balanced and shielded professional audio system, common-mode rejection on the output cables is already excellent. The cables weren’t the ingress path. The RF was entering at the capsule and front-end amplifier, before the balanced output stage.
Here’s what proper shielding looks like at the microphone level:
Recognizing this in the field
If you suspect RF interference in your audio setup, work through this diagnostic:
The key diagnostic fingerprint: condenser-only susceptibility, artifacts at regular kHz intervals (especially a dominant 4 kHz spike), ferrites on output cables having no effect, and a threshold-like disappearance only when aggregate RF power is removed. If you see that pattern, the problem is front-end RF detection, not cable pickup.
What actually fixes it
The regulatory framework is clear. Under Canadian spectrum policy and FCC Part 15 in the US, radio-sensitive equipment must accept interference from licensed transmitters operating within their authorized parameters. The carriers were compliant. The fix had to come from the equipment side.
In order of effectiveness:
Choose microphones with better RF immunity. Some models in the studio’s own inventory showed zero susceptibility. Designs that use lower-impedance sensing circuits — bridge or discriminator topologies instead of raw high-Z FET inputs — inherently reject RF. For most studios, mic selection alone solves the problem.
Improve shielding on susceptible microphones. Work with the manufacturer on RF shielding around the high-impedance circuitry. All components above about 1 kilohm should sit inside an effective Faraday screen with proper ground bonding. The sub-1 mm aperture guideline at 2.6 GHz is a starting point, but real shielding effectiveness depends on seam continuity, connector bonding, and the acoustic window material.
Control placement and exposure. Even moving a microphone to a less exposed position in the room can help. RF field strength varies spatially due to multipath and reflections.
Accept the trend. As carriers deploy more bands and densify urban networks, cumulative RF field strength will keep rising. Equipment designed before this era may need manufacturer updates.
The broader lesson
High-impedance analog front ends have always been susceptible to RF. Guitar amplifiers pick up AM radio stations. Hearing aids buzz near cell phones. The classic GSM buzz in speakers was the same phenomenon at a different frequency.
What changed is the RF environment. Urban areas now have dramatically higher aggregate field strength across a wider range of frequencies than even five years ago. Devices that had adequate suppression margins before may not today. The onset is sudden because the detection mechanism is nonlinear — everything is fine until it isn’t.
For anyone designing equipment with high-impedance analog inputs intended for use in urban environments: intentional EMC design is not optional, even when the product is “audio only.” The RF is there whether you designed for it or not.