How about investigating broadband RF noise from LED driver circuits

[michaelc]

Hi Everybody,

Long post... sorry!

I work in the marine electronics field, mainly in the fishing industry for over 23 years.  I also work with commercial two-way radio systems.

I'm into electronics/computers/IT/amateur radio/home theatre for fun as well!

In recent years I have been finding an increase of broadband RF noise sources finding their way on fishing boats.  The most insidious of these cause a lot of issues for on-board radio equipment.  The symptom generally shows as a decrease in VHF radio receive performance although lower frequency HF equipment is also affected.  HF radio is used less and less these days.   

Various RF noise sources have been identified... some quite surprising!   A well known name-brand marine radar... but only while it's transmitting.   Various cheap LCD TV's. Switch mode power supplies/battery chargers, switch mode phone chargers, microwave oven processors, etc.

More recently a very bad source of interference has been identified... LED globe replacements for navigation lights.  Many of the expensive "vertical filament" incandescent globes have been swapped out to more efficient, longer lasting LED replacements.  But there seems to have been another potentially dangerous cost that has to be paid for this efficiency.

The interference these LED lights generate aren't heard on the VHF radio... the effect is that the extra RF noise affects the radios AGC circuit and impact on the sensitivity of the radio... severely affecting the ability of the radio to receive weak/distant transmissions.

I have had one fishing boat skipper drive himself crazy over a very long time trying to get his radios to work properly... replacing radios, cables, antennas... all to no avail.

I presume that it is the LED driver circuits that are making this interference, square wave PWM circuits can create a lot of harmonics and general noise in the RF spectrum... it is very hard to control and I can imagine that price over RF performance would be a factor in the design of the driver circuits. 

Ferrite cores on the DC supply cable can reduce the interference a little.. but it does not remove it completely.   On many boats... the navigation lights and radio antennas share a very close physical relationship!

I'm wondering if Dave can get hold of a few of these LED globe replacements... use his RF spectrum analyser and do some tests on what effect these globes have on the RF noise floor while in operation.  I would be very interested to see how "RF clean" his LED ceiling light panels are. 

Being an amateur radio operator I am very aware of the increase in radio interference caused by modern electronics... even equipment with a C-Tick certification.

I am also concerned about what this increase of the background RF noise floor will have on other radio equipment in common use... if you think you don't use radio, think again.  You mobile phone uses radio, TV gets to your house (mainly) via radio transmissions,  even ADSL is a form of RF sent down the phone lnes.

Michael.

PORT LINCOLN
SA

[SeanB]

I prefer Led lighting to use simple resistive or constant current DC droppers, as they have a lot less radiated noise. In addition to the RF chokes have you tried adding a few ferrite beads on the leads to the socket, along with a 100n 100V ceramic capacitor across the socket base. This should reduce the noise, and if possible open the light and add the beads and caps inside as well.

I have noted the noise, as at my house i cannot get radio reception reliable inside unless I use an external antenna.

[AndyC_772]

The circuits that are used to supply power for LED lighting are very well known in the EMC industry for being extremely noisy. They're also known for being built down to a price, being generally poorly designed, and (once EMC testing and approval have been done) manufactured without necessary filter components being included.

The problem stems from the fact that an LED doesn't have a nice, roughly straight V-I characteristic like a filament bulb. You can connect a filament bulb straight to a battery and will have a pretty good idea how much power it will draw.

The V-I characteristic of an LED is exponential and varies from one LED to another, so the current though the LED depends critically on the voltage across it. In other words, connect a string of LEDs straight to a battery and they might not light at all, or they might be so bright they'll burn out, and the brightness will vary dramatically as the supply voltage varies.

This means that active control over the LED current is required, so LED lighting uses a switched-mode power supply with closed-loop feedback to regulate the current. This supply has to switch quite large amounts of current on and off fast enough that the eye doesn't perceive any flicker, and this is inherently a noisy operation. However, it isn't perceived as adding any value to the product - it's just a necessary evil - so it ends up having to be very small and low cost. Both these factors work against a designer trying to make a product which is electrically quiet.

You'll struggle to retro-fit filtering to an inherently noisy power supply - it's technically very difficult to achieve effectively. Noise is usually best filtered at source, and that means designing the PSU properly in the first place.

[michaelc]

Right now we are telling boat owners/skippers to remove the retrofit LED lamps and use incandescent globes.  Yachts are very badly affected... by their very nature they are attracted to the power savings that LED's provide... but the mast head LED is quite often only 10cm away from the VHF antenna at the top of the mast.  This KILLS the VHF receiver.

One of the fishing industries that is greatly affected is prawn (shrimp) fishing.  They work at night when navigations lights are on... most of their communications are via VHF radio (plus some have private VHF radio systems P25 & DMR with encryption). 

My main test is to drag my old but reliable radio test set down to the fishing boat... put a fixed antenna outside and generate a progressively weaker signal until I get a noisy but still readable signal on the VHF radio.  Then it's a matter of turning equipment on and off to see what affect it has.

I also use my little Icom IC-91AD radio on WFM mode... the scope function gives a good indication of received noise.

I need to get me a Funcube Dongle or MiniVNA!

[xDR1TeK]

So the LEDs are acting as antennas (transmitters) for the SMPS signal? The only thing I can imagine that a semiconductor can cause is crossover distortion and that is only when in switch mode (not DC bias).
Can you determine the SMPS switching frequency?
Have you added an array of Caps to eliminate the noise? (they have to be large and small together in parallel)
Can you try to swap one SMPS supply with another and see a difference?

If you have solved this, do let us know, it is interesting.

[AndyC_772]

It's not the LEDs themselves that are noisy as. The problem is the long, unshielded wire or PCB trace carrying a rapidly switching current from the supply to the LEDs. That acts as an antenna, and radiates harmonics of the supply's switching frequency in all directions.

Any fix is unlikely to be as straightforward as connecting a few capacitors to an existing design, though they might provide a little useful attenuation. Although the principle is sound, a great deal of expertise and time spent in an EMC lab is likely to be required.