EMI from LED light fixture often exceeds regulatory limits

Conducted emissions from LED light fixtures at frequencies from 1 MHz to 30 MHz can interfere with amateur radio and several broadcast bands. Radiated emissions reach into the lower cellular bands.

After recently moving to northern Colorado and setting up my EMC lab and amateur radio station in the basement, I wanted to update the room’s lighting. To this end, I ordered two large LED light fixtures to replace the existing ones. Knowing that light fixtures from Asia were particularly noisy, I decided to measure the conducted emissions as an EMC engineer and consultant. Because these fixtures connect to the AC mains, I was also concerned with the coupled radiated emissions from the house wiring.

LEDs, first developed for commercial use around 2009, have become the dominant illumination technology. LEDs are generally driven by switch-mode power supplies sourced from 12 VDC or off-line 110 VAC to 240 VAC. These supplies generate broadband EMI as high as 1.5 GHz, enough to affect LTE cellular and GPS reception. That EMI can cause problems in the 1- to 30 MHz AM broadcast and shortwave bands, impacting amateur radio, marine, aeronautical, and military channels.

This is particularly disturbing for radio amateurs and others using the HF spectrum because the EMI generated can create a high noise level that makes communications difficult to impossible. Compare Figure 1 to Figure 2 to see the difference in noise floor with and without special common-mode noise filtering at the antenna port. The waterfall displays show that the noise floor (circled in red) is lower by 9 dB in Figure 2.

Conducted emissions measurement

The measurement setup shown in Figure 3 consists of a Tekbox TBMR-110M EMI receiver controlled by a PC and a Tekbox TBLC08 line impedance stabilization network (LISN). The whole system is placed upon a Tekbox “roll up” ground plane, model TBGP-250/140, partially unrolled on the tabletop. The LISN is bonded to the plane according to CISPR 15 (Limits and measurement methods of radio frequency disturbances from lighting equipment).

The test results were as expected, failing the CISPR 15 limits by as much as 30 dB or more in the HF communications band (1 to 30 MHz) in Figure 4. Installing the LED light fixture in proximity to sensitive receiving equipment would likely result in high levels of interference.

Removing the plastic cover from the fixture allows access to the LEDs and associated wiring. One experiment involved installing four clamp-on ferrite chokes (made of Fair-Rite 41 material) on the LEDs’ interconnect wiring, as shown in Figure 5.

The addition of clamp-on ferrite chokes within the LED lighting fixture and on the mains line wiring, as well as a Würth Elektronik mains filter board, only succeeded in lowering the emissions by 20 dB to 30 dB, still exceeding the CISPR 15 limits (Figure 6).

Radiated emissions measurement

Radiated emissions can be estimated from power wiring using an RF current probe placed around the wiring. Using a Fischer Custom Communications F-33-1 current probe and looking at the potential radiated emission frequencies between 1 MHz and 1000 MHz using a Siglent SSA3032X spectrum analyzer, we can see a very large broadband noise level, decreasing gradually. This would still create the potential for interference within the major bands indicated (Figure 7). Major services likely impacted include amateur radio, marine, aeronautical, and military HF channels between 1 and 30 MHz.

The yellow trace shows the ambient noise floor of the measurement system. The other three traces show various experiments with filtering. Violet shows the measurement without filtering, blue represents the addition of ferrite chokes, and green represents the addition of a Würth line filter board. There was hardly any difference because so much of the internal wiring radiated.

The average EMI measurement in the HF region of Figure 7 is 45 dBµV. If this were coupled to a 1-m long wire at a nominal frequency of 15 MHz, it would result in a radiated E-field of 34.5 dBµV/m, a moderately high emission. However, if the lighting fixture were connected to 10 m of wiring as in a typical house, the emission level would increase to 54.5 dBµV/m, a very high level of interference.

Summary

This LED lighting fixture would not have been compatible near an area where I would be making sensitive EMC measurements or using my HF amateur radio. Lamp fixtures are necessarily unshielded, so the LED driver module (a switch mode power converter) would need some extreme EMI filtering and shielding. This driver had no line filtering. The lack of mains line filtering is typical for these low-cost light fixtures. Thus, it failed to meet the CISPR 15 emission limits for luminaries and was returned to the vendor.

This appears to be a common problem for similar LED lighting fixtures, including those mounted to the bottom of kitchen cabinets, garage lighting, and other uses. While the fixtures tested included CE marking for the European market (no FCC, though), I can only conclude they had not been compliance tested and were illegally imported into the U.S.

EEWorld Online related content

Assemble your EMC troubleshooting kit and fix your problems
How to use RF current probes for EMI: Part 1
How to use RF current probes for EMI: Part 2
TinySA Ultra: A low-cost spectrum analyzer that fits in your pocket