EMI Sources and Fixes
An audio amplifier picks up a whine that changes pitch when a phone charger is plugged in nearby. A microcontroller project resets every time a relay switches a motor. A temperature reading jumps by several degrees whenever the LED strip in the same enclosure turns on. An AM radio hisses everywhere on the dial until the new solar inverter shuts down at sunset. All of these are electromagnetic interference (EMI): unwanted energy from one circuit getting into another, by conduction along wires or by radiation through the air.
The number of noise sources in homes, workshops and vehicles has grown enormously. Switch-mode power supplies are now in virtually every device that plugs into an outlet. Solar inverters, motor drives and battery chargers switch kilowatts at tens of kilohertz. LED lighting has replaced incandescent bulbs. Each of these, if poorly designed or installed, can interfere with radio reception, audio systems, sensors and digital circuits — and the circuit you build can be either a victim or a source.
This lesson teaches you how to recognize, classify, locate and fix the most common EMI sources, and how to make your own circuits less susceptible. Armed with this knowledge, you can systematically identify what is causing a noise problem and apply the right solution.
Categories of Interference
Interference comes in many forms, and classifying it helps you understand both its likely origin and the best approach to fixing it. Three classifications are especially useful:
Conducted vs Radiated Interference
Conducted interference travels along wires. A noisy device drives high-frequency currents back into its power leads, and from there through the building wiring, a shared DC supply or a signal cable into other equipment. A switching supply with poor input filtering spreads its noise to every device on the same circuit and turns the building wiring into a large, inefficient antenna.
Radiated interference travels through the air as electric, magnetic or electromagnetic fields. The noisy device's circuitry, cables and wiring act as unintentional antennas, and nearby circuits, cables and antennas pick the energy up. Close to the source, magnetic or electric near-field coupling dominates and falls off very quickly with distance; farther away, the radiated field falls by 6 dB for each doubling of distance.
Narrowband vs Broadband Interference
Narrowband interference occupies a small portion of the spectrum — a steady tone, a carrier with sidebands, or a set of discrete harmonics. A switching supply running at a fixed frequency produces a fundamental and a comb of harmonics at exact multiples of it; a microcontroller's clock does the same.
Broadband interference spreads across a wide range of frequencies, from a few hundred kilohertz to tens of megahertz or more. Arcing, commutator motors and switching supplies with very fast edges or poor filtering produce it. On a spectrum analyzer it appears as a general rise in the noise floor; in audio it is a hiss, buzz or crackle.
Intentional vs Unintentional Radiators
Regulators classify RF-emitting devices by whether they are designed to radiate. Intentional radiators (Wi-Fi routers, Bluetooth devices, radio transmitters) emit on authorized frequencies. Unintentional radiators (computers, power supplies, appliances, LED lamps) emit RF only as a byproduct of their operation. Almost all the sources in this lesson are unintentional radiators, subject to emission limits such as FCC Part 15 in the United States and the EMC Directive in Europe.
Common interference sources around a typical home. Each has a characteristic signature, and each can reach equipment through the air, the mains wiring or its cables.
View LargerSwitch-Mode Power Supplies
Switch-mode power supplies (SMPS) are now the dominant way to convert mains AC to low-voltage DC, and DC-DC converters do the same job on almost every circuit board. You will find them in phone chargers, laptop power bricks, LED drivers, computers, televisions, routers and battery chargers. They are popular because they are efficient — a good SMPS converts 85–95% of its input power to output power, compared with 50–70% for a linear supply — but the switching that makes them efficient also makes them noise generators.
A switch-mode supply switches a transistor (usually a MOSFET) on and off at a high frequency — typically 50 kHz to 2 MHz. Each transition drives a sharp change of current through the circuit's loops and a sharp change of voltage across its switching node, and those fast edges contain energy far above the switching frequency. The harmonics of a 100 kHz converter extend well past 30 MHz, and the ringing on a fast switching edge can reach hundreds of megahertz.
How much noise reaches the outside world depends on the design: input filtering (common-mode chokes and X and Y capacitors), output filtering, snubbers that slow and damp the switching edges, and a layout that keeps high-current switching loops small. Good supplies carry all of these; cheap ones, particularly unbranded units sold without regulatory testing, may have almost none.
The characteristic signature of SMPS noise:
- A comb of harmonics at multiples of the switching frequency, or broadband hash if the frequency wanders with load
- Usually strongest at lower frequencies, below about 10 MHz, and decreasing higher up
- In audio equipment, a high-pitched whine or a buzz that changes with the load on the supply
- On an oscilloscope, sharp spikes on the output or ground at each switching edge
- Constant while the device runs, and gone the instant it is unplugged
The quickest way to identify an SMPS as a noise source is the on/off test: unplug devices one at a time while watching the affected equipment, a spectrum analyzer or a portable AM radio tuned between stations. When the noise drops, you have found the culprit. Remember that the noise often travels through the building wiring, so moving the device farther away may not help on its own.
Remedies for SMPS noise include:
- Replace the offending unit with a better-quality supply carrying genuine regulatory compliance marks
- Add common-mode ferrite chokes on its input and output leads (see the ferrite chokes lesson)
- Install an inline mains EMI filter on its power cord
- In your own designs: keep switching loops small, add an input filter, use a snubber or a gate resistor to slow the edges, and consider a converter with spread-spectrum switching
Solar Inverters and Charge Controllers
Solar photovoltaic systems are now common on homes, workshops and off-grid projects. A system consists of panels, a charge controller or maximum power point tracker (MPPT), often a battery bank, and an inverter that converts DC to AC for the home or the grid.
The inverter is the main noise source. It switches high-power transistors (IGBTs or MOSFETs) at 10–100 kHz while handling 2,000–10,000 W. Even if only a tiny fraction of that power escapes as RF noise, the absolute level is large, and the long DC cables from the panels and the AC wiring make effective antennas.
A poorly filtered inverter can raise the noise floor across the entire range from about 1 MHz to 30 MHz for tens of meters around an installation, ruining AM and shortwave reception and interfering with any sensitive receiving equipment nearby. MPPT charge controllers in off-grid systems use the same switching technology at lower power and can cause similar, smaller problems — including in the battery-powered projects they supply.
Reducing solar inverter and charge controller noise:
- Fit ferrite common-mode chokes on the DC cables between panels and inverter or controller
- Install an inline RF filter on the inverter's AC output before it connects to household wiring
- Bond the inverter chassis to the safety ground if it is not already
- Route DC cable pairs together (positive and negative side by side) to minimize loop area
- Ask the manufacturer — reputable makers publish EMI application notes, and some offer firmware or hardware updates
- When choosing equipment, prefer inverters and controllers with good independent EMC test results
LED and CFL Lighting
The move from incandescent to efficient lighting created a new category of interference. Compact fluorescent lamps (CFLs) and LED lamps both use small switching power supplies, and the same switching noise that makes an SMPS a problem applies here.
Compact fluorescent lamps use a high-frequency electronic ballast, typically 20–60 kHz, to drive the gas discharge tube. Without adequate filtering, the ballast produces harmonics that are strongest below about 10 MHz.
LED lamps vary enormously. Lamps from established manufacturers use well-filtered drivers that meet FCC Part 15 Class B or the equivalent European limits. Budget lamps from unverified sellers often have minimal filtering and can radiate significant noise, as well as producing visible flicker from poorly smoothed drivers.
LED strip lighting with cheap inline drivers and PWM dimmers is a frequent culprit in workshops and projects. The PWM dimming itself switches the full LED current at hundreds of hertz to several kilohertz, and long unshielded strips radiate the fast edges. Outdoor floods and decorative lighting with cheap drivers can also be troublesome.
The most practical approach to LED lighting noise:
- Replace budget lamps with units from established manufacturers that state Class B compliance
- Use the on/off test room by room and lamp by lamp to find the specific units causing problems
- In LED strip projects, keep the PWM wiring short, twist the supply pair, and add a ferrite at the controller
- For hardwired fixtures that cannot be replaced, ferrite chokes on the power leads can sometimes help
Characteristic power line noise: broadband hash across the HF range with a 120 Hz comb structure and bursts synchronized to the mains, caused by arcing at damaged or contaminated insulators or hardware on utility poles.
View LargerPower Line Noise
Power line noise is one of the oldest sources of interference, and it remains common. Unlike the device-based sources above, it comes from the utility distribution system itself — the poles, wires, transformers and hardware that bring electricity to homes and businesses.
Causes of Power Line Noise
Power line noise is generated by arcing — small electrical discharges where a connection is imperfect or where insulation has degraded. Common causes include:
- Damaged or contaminated insulators: Insulators can crack or collect a conductive layer of salt, industrial dust or other deposits. When wet, the contamination creates a leakage path that arcs at line voltage.
- Loose hardware: Bolts, clamps and pole-top hardware loosened by years of vibration and thermal cycling make intermittent contact that arcs.
- Corroded ground connections: Corroded ground clamps on poles and transformer pedestals arc at the current they carry.
- Failing transformers: Internal arcing in a distribution transformer that is beginning to fail can produce extremely strong noise.
Characteristic Signature of Power Line Noise
- Harmonic structure: Because the arcing is synchronized with the AC cycle, the noise repeats at 120 Hz (100 Hz on 50 Hz systems). On a spectrum analyzer in zero-span mode it shows regular bursts; on an oscilloscope triggered from the mains, bursts appear at the same points on each half-cycle.
- Sound: On an AM radio it is a harsh, rasping buzz at the power-line rate.
- Weather dependence: Arcing at contaminated insulators is much worse in fog, rain or heavy dew. Noise that rises dramatically in wet weather points strongly to the power lines.
- Seasonal variation: Arcing at loose hardware can be worse in cold weather, as metal contracts, or after storms.
Locating and Reporting Power Line Noise
Power line noise can travel a long way along the lines before radiating, so the place where it is loudest is not always obvious. A portable AM radio tuned between stations makes a good locator: its internal ferrite antenna is directional, with a sharp null along the rod's axis. Walk along the line, watching for the noise to peak as you approach the source and fall as you pass it, and note the pole number. Utility companies are generally obliged to correct noise from their equipment; report it with the pole number, the frequencies affected, whether it is worse in wet weather, and recordings or analyzer screenshots. Follow up if the first contact does not resolve it — larger utilities have power-quality departments that understand the problem.
Electric Fence Controllers
Electric fence controllers apply short high-voltage pulses to a fence wire — typically about 0.3 milliseconds long, once or twice per second. The sharp edge of each pulse is a broadband RF impulse, and the long fence wire radiates it efficiently.
- Very regular clicks at 1–2 per second, on every frequency at once
- On a spectrum display, a broadband rise of 10–20 dB at each pulse
- Steady regardless of weather and time of day
- Much worse when vegetation touches the wire or an insulator is cracked, because these cause arcing at each pulse
Remedies include a filter or RF choke at the controller output, repairing arcing points along the fence, and replacing old controllers with modern EMC-compliant designs whose pulse shaping reduces high-frequency content.
Motors, Welders, Dimmers and Computers
Brushed Motors
Brushed DC and universal motors — in power tools, vacuum cleaners, blenders and many hobby projects — arc at the commutator as the brushes switch the current between windings. The result is broadband noise and sharp spikes on the supply. In projects, a 100 nF ceramic capacitor directly across the motor terminals, plus capacitors from each terminal to the motor case, removes most of it; long motor leads should be twisted together.
Arc Welders
Arc welders generate intense, broadband, impulsive noise during operation, and TIG welders that use a high-frequency arc start are especially troublesome. The interference stops when welding stops; if it affects nearby equipment, the practical cures are distance, separate mains circuits and filtering on the victim equipment.
TRIAC Light Dimmers
Phase-cutting dimmers chop the AC waveform with a TRIAC, and the sharp turn-on edge in each half-cycle produces conducted and radiated noise, mostly below a few megahertz. It is often heard as a buzz in audio equipment on the same circuit. Dimmers with built-in filtering, or dimmers designed for LED loads, usually solve the problem.
Computers and Peripherals
- USB 3.x: USB 3.0 signalling produces broadband noise around 2.4–2.5 GHz, which is well known for degrading nearby 2.4 GHz wireless mice, keyboards and Wi-Fi, and GPS reception. Shielded cables, short connections and keeping wireless receivers away from USB 3 ports help.
- HDMI and display cables: High-speed video signals radiate from poorly shielded cables and connectors.
- Ethernet: Unshielded network cables radiate harmonics of their signalling; shielded cable or ferrite chokes near sensitive equipment reduce this.
When Your Circuit Is the Victim
Interference only matters when something is susceptible to it. Knowing how EMI gets into circuits lets you design and build equipment that ignores it.
- RF rectification in audio circuits: The input transistors of op-amps and audio amplifiers act as detectors for strong RF signals, turning them into audible noise — the familiar “dit-dit-dit” buzz from a mobile phone near a speaker. A small RC low-pass filter at each input (for example 1 kΩ and 100 pF), plus a ferrite bead on the input lead, stops the RF before it reaches the semiconductor junction.
- Hum and ground loops: Mains-frequency hum usually enters through ground loops between equipment, not through radiation. The shielding and grounding lessons later in this module cover the cures.
- Microcontroller resets and glitches: Spikes from switching relays, motors and solenoids couple onto supply and reset lines. Flyback diodes on inductive loads, local decoupling, a filtered reset pin, and keeping high-current wiring away from logic prevent most of them.
- Noisy sensor and ADC readings: Long sensor leads pick up interference. Twisted pairs or shielded cable, differential inputs, an RC filter before the ADC, and averaging all help.
- Touch screens and capacitive sensors: Cheap chargers inject common-mode noise that causes false touches and erratic readings, which disappear on a good charger or battery power.
Locating an EMI Source
A systematic approach finds sources far faster than random searching.
Step 1: Characterize the Noise
- Which frequencies, or which equipment, are affected?
- Is the noise constant, periodic (regular clicks) or variable?
- Does it change with time of day, weather, or what else is switched on?
- When did it start, and did anything change around that time?
Step 2: The On/Off Test
For noise that might come from within the building, switch off circuit breakers one at a time while watching the affected equipment, a spectrum analyzer or a battery-powered AM radio. When the noise stops, restore that circuit and unplug its devices one by one to find the source. For noise inside a piece of equipment or a project, power its sections or loads one at a time in the same way.
Step 3: Direction Finding and Near-Field Probing
If the noise remains with everything in the building switched off, the source is outside. A portable AM radio or a small loop antenna has a figure-8 pattern with sharp nulls: take bearings from two or more positions and walk toward where they cross. Inside equipment, a near-field probe connected to a spectrum analyzer pinpoints the exact cable, board or component that is radiating, as described in the EMC pre-compliance lesson.
Step 4: Frequency Analysis
The spectral signature often identifies the source type:
- Regular impulses at 1–2 per second = electric fence
- Bursts synchronized to the mains, worse in wet weather = power line arcing
- Comb of harmonics at a fixed spacing = switching supply or digital clock at that frequency
- Broadband hash that appeared with a new installation = inverter or charger
- Noise only when certain lights are on = lamp driver or dimmer
⚖ Experiment: EMI Hunting with a Spectrum Analyzer
This experiment teaches systematic EMI location using a low-cost spectrum analyzer as the main diagnostic tool. You will identify noise sources by their spectral signatures and use a portable radio to locate them.
- A low-cost spectrum analyzer such as a TinySA, running from its internal battery
- A short wire or telescoping whip antenna
- A small portable AM radio
- A few snap-on ferrite cores
- Connect the wire to the analyzer input and set a span of about 1–30 MHz with a resolution bandwidth of 10–30 kHz. Note the noise floor with as many devices switched off as possible.
- Switch devices on one at a time — phone charger, laptop charger, LED desk lamp, television, LED lighting. After each, watch the display for 30 seconds and note any change in the noise floor or new peaks. Take a screenshot when you find a source.
- For each noisy device, narrow the span around the frequencies it affects most to see the shape of its noise.
- Clip a ferrite core onto the noisy device's power cord, close to the device, and see whether the noise drops. This demonstrates suppression of conducted, common-mode noise.
- Tune the portable AM radio between stations near the top of the band (around 1600 kHz). Walk around the building and outside, rotating the radio to find the direction of maximum noise, and follow it toward any source.
Different devices produce distinctly different signatures. Chargers and power bricks raise the noise floor or produce combs of lines across several megahertz; LED lamps with poor drivers produce similar but often narrower interference. Ferrites usually reduce, but do not always eliminate, conducted noise. Most people discover at least one noisy device they were not aware of.
Regulations and Standards
Emission limits exist so that equipment can coexist. Knowing them helps when you choose components and equipment, design a product, or need to resolve an interference problem.
FCC Part 15 — Unintentional Radiators
In the United States, FCC Part 15 limits emissions from devices that are not designed to transmit. Class B limits apply to products used in homes; Class A limits, about 10 dB less strict, apply to commercial and industrial equipment. Equipment that exceeds the limits may not be marketed, and the FCC can require an operator to stop using a device that causes harmful interference to radio services.
FCC Part 18 — Industrial, Scientific and Medical Equipment
ISM equipment such as induction heaters, microwave ovens and some medical equipment is covered by Part 18, which allows higher emissions in designated bands but still requires that harmful interference to radio services be corrected.
European and International Standards
In Europe, the EMC Directive requires all electrical equipment to meet both emission and immunity requirements before it can carry the CE mark. The harmonized standards include EN 55032 (CISPR 32) for multimedia and IT equipment, EN 55014 for household appliances and tools, EN 55015 for lighting, and the EN 61000-4 series for immunity to static discharge, radiated fields, fast transients and surges. Many other countries adopt the same CISPR standards.
Resolving Interference in Practice
Most people generating interference have no idea they are doing it. A polite explanation — with a demonstration of the problem and a suggested fix such as a better-filtered supply or a few ferrites — resolves most neighbor-to-neighbor cases. Power line problems go to the utility. The regulatory route is for cases where diplomacy fails or a commercial operator will not cooperate. For your own designs, testing early with the pre-compliance techniques in the test equipment module is far cheaper than fixing a product after it fails formal testing.
Frequently Asked Questions
Is it legal for a device to cause interference?
Devices sold in most countries must meet emission limits, such as FCC Part 15 in the United States or the EMC Directive in Europe, and a device that causes harmful interference to radio services can be required to stop. In practice, enforcement against individual products is slow, and many cheap imported devices sold online do not meet the limits. The most effective approach is usually practical: identify the device, then replace it with a compliant one or add filtering and ferrites. Report power line noise to the utility, and reserve formal complaints for serious cases where diplomacy fails.
How do I tell power line noise from switching supply noise?
Power line noise is synchronized to the mains: a spectrum analyzer in zero-span mode shows bursts at 120 Hz intervals (100 Hz on 50 Hz systems), an AM radio gives a harsh buzz, and the noise is usually much worse in wet weather. Switching supply noise is not tied to the mains frequency; it appears as a comb of lines at multiples of the switching frequency, or as a uniform broadband rise, and it stays the same regardless of the weather. The on/off test is decisive: if switching off the main breaker removes the noise, the source is inside the building.
Will a ferrite on the power cord cure the interference?
Often partially, sometimes completely. A ferrite on a cable suppresses common-mode current — noise flowing in the same direction on all conductors and using the cable as an antenna. If the noise also leaves on other cables (USB, HDMI, Ethernet, speaker or sensor leads), or radiates directly from the enclosure, a ferrite on one cable is not enough: treat every cable attached to the device. If a device still causes problems after that, it probably has a failing or poorly designed power supply. See the ferrite chokes and cores lesson for how to choose the right material and number of turns.
Test Your Knowledge
Answer the questions below to check your understanding of this lesson. Every answer can be found in the lesson above.