Identifying Sources of Radio Frequency Interference Around the Home

Most people associate radio frequency interference (RFI) with electricity that is arcing through the air, as when a power connector is corroded or has become loose. But in recent years, many consumer devices have become widely available that contain circuits that are intrinsically noisy. These devices contain Class D amplifiers and power supplies that use silicon controlled rectifiers (SCRs). Digital circuits in computers and computer networking equipment are also intrinsically noisy, because of the rapid acceleration and deceleration of electrons at high processing speeds.

This page describes the results of some tests with a portable spectrum analyzer in my attempts to identify the sources of RFI around the house. Radio interference was found in a number of unexpected places. Each type of interference has a characteristic signature that can aid in finding the source. In some cases, the RFI can be easily eliminated. In others, the only sure way to eliminate the interference is with the repeated application of small-arms fire.

Computer networks

The biggest source of interference was from the computer network which, in this case, consists of a small number of PCs connected through a network switch with Cat 5e cable. The computer network produced a comb of carriers spaced about 30.5 kHz apart that made reception difficult above 17 MHz (see figure below). This RFI was a problem only when a computer was plugged into a network hub or switch. Clamping large ferrites over the network cables and the switch's power cable had little effect. Changing to a D-Link DGS-1016D network switch, which is in a metal box instead of the plastic case used by the previous hub and switch, reduced the RFI from S7 to S3 (where background noise was S2) on some frequencies. However, other frequencies still had RFI at S6-S7. The RFI was reduced from S7 to nearly undetectable levels when the Cat 5e cable was replaced with shielded Cat 5e cable.

On a radio, this interference just sounds like an unmodulated carrier. There is no buzzing sound, but often a faint whine or hum.

The above figure shows the interference from the network switch in the VHF region. In this example, the antenna was positioned in close proximity to the switch. The amplitude of these spikes is actually fairly small, and a switch or hub by itself produces only a small amount of interference. However, unless shielded patch cables and wiring are used, the network cables act as an antenna and can broadcast this interference over a wide area, easily reaching 25-50 feet from the source.

VCRs and DVD Players

Interference on upper HF was also produced by a Sony SLV-N900 VCR, and was present even when the VCR was turned off. It turned out that this VCR is supplied with no ferrites on the power cord. I cut the power cord (after unplugging it) and wrapped it several times through a large toroidal ferrite. This reduced the interference from the VCR from S8 to undetectable levels.

No spectra are available for the DVD player, as it is in the shop having a number of small holes repaired (and to have some small pieces of lead that somehow got inside it removed).

Uninterruptible Power Supplies

Some UPSs produce a very interesting and distinctive pattern of interference that is time- and frequency-dependent. For example, an APC Back-Ups RS-1500 uninterruptible power supply (the newer model with the LCD display on the front) produced radio interference in longwave and in the upper part of the HF band. When heard on a radio, the RFI consists of periodic bursts of buzzing sounds about once every three seconds. The burst is repeated every 69 kHz; however, in the lower frequencies the bursts remained at a constant frequency, while in the HF region, they swept upwards in frequency. The interference is not conducted through the power cable, but was emitted from the front and sides of the unit. It was strong enough to be detectable 15-20 feet away. The only solution was to get rid of the UPS. An older model Back-UPS XS-1500 did not produce radio frequency interference.

This upward sweeping is somewhat similar to the "mooing" cow-like sounds that are often heard in the longwave region. (These sounds are obviously not real cows; it is well known that cows only use Morse code on this band.)

APC Back-Ups RS-1500 in longwave region (click to enlarge)

APC Back-Ups RS-1500 in HF region (click to enlarge)

I finally got rid of the RS-1500 and replaced it with a Smart-UPS 1000. We have several of these in the lab and they are very reliable and relatively interference-free.

Telephones

You might expect a cell phone or portable wireless phone to be a potential source of interference. But many non-wireless telephones also contain oscillators that emit an unstable sine wave around 25.6 MHz. These telephones aren't a major source of interference, since this signal is only on a single frequency, but they are easily detectable up to 50 feet away. Somewhat disturbingly, on one phone, a Nortel 9516CW, I found that dialing a number produced a series of frequency deflections. The size of the deflection was different for each digit, allowing the number you are calling to be easily identified from a distance. If other phones exhibit this behavior, it could pose a widespread privacy risk.

The image at left shows the deflections when three digits were dialed, and the image at right shows the phone signal measured from an outside antenna about fifty feet away. Note that the phone's frequency continues to drift during the measurements. On a radio, it sounds like an unmodulated carrier that drifts by a couple hundred Hz the first twenty minutes after the phone is plugged in.

Fluorescent Lamps

Fluorescent lamps produce a characteristic series of spikes in the longwave and lower MW bands. This noise has a pronounced 60 Hz hum or buzz.

Different types of electronic ballasts have markedly different RF spectra. In this test, I measured the noise with a low-gain antenna one foot away from the ballast. Because my neighbors have lamps that eradicate all reception below 600 kHz between the hours of 5 PM and 1AM, all the measurements shown here were taken before 4 PM local time.

Lutron dimming fluorescent ballasts emit their energy in a sharp band with harmonics extending above 500 kHz into the AM radio band. The Tu-Wire was the worst, with peaks up to -70 dB in my setup. When the lamp is dimmed, a small amount of broadband noise is also evident between the major peaks. Non-dimming ballasts, such as the ballast in the GE F21T5 fixture shown here, tend to have broader bands, with markedly reduced intensity above the third harmonic. The sixth harmonic (around 320 kHz) is already down to -120 decibels. This indicates that the ballast in this lamp has a relatively low THD.

Fluorescents with magnetic ballasts also produce small amounts of radio frequency interference over a very broad spectrum, but the levels are very low compared to electronic ballasts (the y axis in the figure is in decibels, which is a logarithmic scale). Light dimmers also produce broadband interference. The light dimmer in this case was farther away, about ten feet from the antenna.

Remediation: Some lamps can be rewired to reduce RFI (See here. )

Treadmills

Just recently, I found a new source of radio frequency interference: a Precor exercise treadmill. The motor generates very strong interference that blankets the radio spectrum from 0 to 30 MHz (see figure below). The antenna in the first two graphs below is a monopole about 30 feet from the house.

Even though the closest point of the antenna in these measurements was outdoors and over 30 feet from the treadmill, noise was increased by up to 20 dB above background across a broad region from the AM band through the HF. It's impossible to pick up any AM broadcast stations within 15 feet of this treadmill when it is running. It's possible that the treadmill would interfere with DSL traffic as well.

The treadmill even radiates in the VHF region. This interference is not coming from the control panel or the heart monitor, but from the drive motor. This is apparent when listening to the noise with a radio: a low-pitched hum can be heard that changes when the treadmill is stepped on. The figure above shows the spectrum in the VHF just above the airband used by commercial aviation. The narrow spikes in this figure are from the local computer networking equipment, and are about the same size as a typical aircraft signal. I don't want to single out Precor; most likely, other treadmills have the same problem. But one can't help asking, where was the FCC when these machines were being tested?

Measuring closer to the source (4 feet away), it's evident that most reception up to 30 MHz is blocked. Because the antenna in this case is indoors, numerous spikes from the computer network are also visible between 10 and 23 MHz.

A higher resolution FFT shows that the interference consists of a large number of evenly-spaced, high amplitude spikes, indicating that it is the power supply, not the motor itself, that is producing the interference. It has always been a difficult challenge for engineers to design an efficient and lightweight variable-speed power supply. Unfortunately, manufacturers have chosen to use pulse-width modulation (or "chopped DC power"), similar to what is done with AC in cheap light dimmers. However, because of the higher power, the interference produced by treadmills is much worse than even the worst light dimmer.

---

Back