LED lights interfering with VHF/DSC/AIS

An alert has been raised that use of LED lights on ships can interfere with communication:

This is apparently not an entirely new observation:

It appears that the cheaper lights cause the most problems.

I was responding to an RFQ for a small boat recently and the client expressly forbade the use of LED lights on the vessel at all due to the potential of interference with various electrical devices and signal detectors to be installed. I thought it was fascinating.

The actual LED isn’t noisy. But the electronic drivers used for high efficiency and power supply tolerance can be (likewise the electronic ballasts in compact fluorescent lamps).


LED Drivers for Dummies: https://www.savemoneycutcarbon.com/learn-save/introduction-led-drivers-lighting/

The latest fad is to use DC generators and drive motors on ships with diesel/electric propulsion:

Thus eliminating the need to use Variable Frequency Converters, or convert AC to DC via SCRs for the main motors on board. Combining this with DC/AC converters to supply a 440/230V AC grid for lighting and equipment requiring such power supply and a 24V and/or 12V grid for instruments etc.

I don’t know if LED lights can be driven directly from the 12V grid though??

LEDs are fixed-voltage devices like any diode. Vf runs between 1.8V and about 4.5V for different types and colors. They need some sort of current limiting (a series resistor for small ones). Many power LED drivers also are buck-boost devices that will supply constant output over a considerable range of input voltage.

In my “real” job my company builds optical encoders In which LEDs are an integral part of our optical systems. In some space and military applications the specifications for Electromagnetic Interference (EMI) for our encoders are incredibly stringent. I’m on the mechanical design end rather than the electronic, but I know our guys routinely design circuits including LED drivers in which EMI is virtually undetectable. Since our stuff is normally low volume and not low cost, I don’t know how hard it is or what it costs; I just know it’s doable.

So… would all the LED backlit displays in the wheelhouse next to the VHF’s be a problem then too I suppose?

would all the LED backlit displays in the wheelhouse next to the VHF’s be a problem then too I suppose?

Individual cases. A great deal of electronic equipment including lighting LED assemblies and CFLs employs some form of switching power supply. This is bad news for RF noise, because in order to make the supplies small and cheap (which means making transformers small and cheap) they have to use high switching frequencies, some of them approaching two megahertz. And because what they do is switch stuff on and off abruptly at these frequencies, the result is approximately a square wave with a great deal of harmonic (multiples of the nominal frequency) energy. Ask @Emrobu about Fourier analysis of square waves.

The LED backlighting in equipment undoubtedly uses current-limiting resistors rather than active drivers; so I doubt they’re an issue themselves. But the gear they’re in may well be; and the noise may be radiated, conducted, or both.

Computers generate a great deal of noise aside from the power supply, because again they’re in the business of switching many things on and off at rates up to a few gigahertz.

It’s the job of the mfr to keep any such noise contained within the case of the device (and running a device with the [metal or conductive-coated plastic] case open allows radiated noise out). In the US, electronic gear has to meet one of two “Part 15” standards for emitted noise – one for commercial gear and one for household. The levels allowed for household gear are IIRC about one tenth of those allowed for commercial.

Suppressing noise once it’s been generated is painful and expensive, so mfrs may skimp. Testing is expensive and time consuming as well – it begins with constructing a room entirely lined with copper mesh, with few or no openings larger than the mesh size.

It’s do-able. I suspect the cost curve approaches vertical as you approach zero noise.

1 Like

If its too expensive to make circuits that don’t do this noisy Marconi trick, why not put filters on the radio to attenuate it? The brilliant thing about regular waves is that they can be subtracted. If the frequency is very different from the signal, it can be filtered. If you can model it, you can attenuate it, I believe.

1 Like

You can filter some noise, certainly. For example, fancy HF receivers in the '80s had a woodpecker filter specifically designed to combat the Russian’s over-the-horizon radar installation, which operated on whatever band was best for propagation without regard to who or what else might be using the band. Here it is with WWVB (Hawaiian time station; you can tell it’s not Fort Collins because of the female voice) in the background. This stuff could cover an entire band… https://youtu.be/_Bjcsoqmmcw The filter didn’t remove it entirely, but it made a considerable difference.

The basic problem though is that your radio is receiving a signal, or trying to, that may be down around a microvolt or less. It’s very easy for nearby noise sources to be many times that.

I should have said that FM deals with interference differently than AM, and I don’t understand it nearly as well. But a basic characteristic of FM receivers is that the strongest signal around “captures” the receiver, and even slightly weaker (less than 2 dB difference for a good FM stereo) signals will be rejected. I do not understand at all how this happens, but it does (which is why aeronautical radio uses AM; so that when one signal steps on another it’s obvious that it happened).

In seismic you have the advantage of knowing basically what the signal will look like, so noise rejection is perhaps easier. Also we can run our guesses on noise rejection, look to see if we are harming the signal, undo it, redo it, make it very pretty, add colours, add repetition of nearly the same signal with different noise…It takes days. I suppose a radio is expected to just produce the signal right now, no waiting on the math or the human fiddling.

Maybe just a lead plate around your noisy circuits, then. Which makes it hard to see the LEDs. Yes. I begin to appreciate the problem. So we need the real data to look at. What does this LED noise look like?


Here’s what [some random] power supply sounds like on HF AM…run this into your scope and it may give some idea.

Incidentally, for a cheap thrill** hold an electronic flash close to an AM (BCB) radio and turn it on so it starts charging up.

**for some value of “thrill”

my favourite modern experimentalists illustrate the point, best part is after about 1:15
1 Like

unfortunately I don’t have an ocilloscope. ooo, but there’s one at school. I can play with it in January. The tool I know how to use to model the problem and the solution is promax/seispace which is owned by Halliburton. You can get the full spectral analysis and design a filter just for your specific problem. I don’t have a promax license. someone would have to make it into an SEG file, which I don’t know how to do.

A university with a geophysics program, or any seismic data processing centre can help you. Next time you are in Houston, go hang out at Sherlocks on a Wednesday night. That used to be were all the shorebased geos would hang out. Except for the ones in Sugarland, they have a different haunt. There’s also some in Denver and Calgary, but I don’t know where they drink.

1 Like

This is in LSB mode. More narrow band but the noise is pronounced. AM is more broad banded and so is the noise. 90% of the noise is from switching power supplies. Another 10% is from VFDs.

Most of the noise comes from switching supplies from laptops, wall warts, ect. In 2003 I was in a new-build program and battled many noise sources. The biggest source then were DIN rail mounted power supplies in the various PLCs. It took over a year to rectify most of the noise.


So how is it done?

Good catch, thanks.

For non-radio types, an AM (amplitude modulation) transmitter sends a “carrier wave” at the desired frequency, then uses the audio signal to control the strength of the carrier. Because physics, when you electrically combine two signals you end up with the originals, plus one that’s the sum of the two frequencies and another that’s the difference of the two.

So suppose you were transmitting on ten megahertz (WWV wouldn’t be happy…). Looked at one way (with an oscilloscope set to display audio frequencies) you’d see the carrier as an apparently solid band across the display; but with both positive and negative edges of it reproducing the shape of the audio signal. Looked at a different way, with a spectrum analyzer, you’d see a big spike at the carrier frequency, and two smaller spikes offset above and below the carrier by the instantaneous audio frequency. Those upper and lower spikes are called sidebands.

An AM receiver takes in all this stuff, removes either the upper or lower half of it, throws away the carrier, and passes the remaining audio signal to the audio amplifier and speaker. This is easy but inefficient, because you’ve thrown away more of the signal power than you’ve kept. Also the signal takes up more space in the frequency band than necessary.

This inefficiency led to the development of SSB (or strictly speaking Single Sideband Suppressed Carrier). The carrier and one sideband are suppressed in the transmitter, so all the power can go into the remaining sideband. The missing carrier frequency is generated inside the receiver and Bob’s your uncle.

The virtue is increased range for given power and smaller bandwidth required. The snag is that tuning the receiver is more difficult because even the slightest mis-tuning will change the pitch of the received audio.


Some of the noise sources were eliminated with a simple toroid choke on the output. A few of the noisy switchers were replaced with linear supplies. The DIN rail supplies were upgraded to more efficient modules. A faint, but irritating intermittent noise that took me over a year to find turned out to be the charge regulator on the running light solar cells.