Noise From Mains LED Spotlights Swamping AM Reception

[DigitalNoMore]

Hi All,

Just wondering if anyone has had similar problems?

My house is just a couple of years old and fitted with mains LED spotlights throughout. They are fine until you try listening to any AM radio wavebands. The noise from them completely swamps reception on all the AM wavebands. I've tried 4 different radios with the same result. It's the same on both battery and mains so the noise is clearly radiated rather than mains conducted.
Presumably the noise is emitted by the switched-mode PSUs in the spotlights.

It's hard to see how these spotlights could possibly meet the relevant EMC regulations!!

[Radio Wrangler]

A lot of stuff doesn't really meet the EMC regs, though plenty does.

The levels set in the regs made the antiquated assumption that there would probably only be one interfering device within range, not the dozens we now have, so even compliant devices are problems.

The levels set in the regs were designed to be just tight enough to protect your neighbours, not you. They were assumed to be 10m away (30m from commercial premises) You have stuff much closer.

The noise is conducted from the lights, then the mains wiring radiates what it's had conducted into it. Your set gets the radiated noise via its antenna. The set needs its antenna to get the wanted signal, so there's nothing much can be done at the radio end. The place for filtering is between the switch-mode power converters in the lamps and the wiring.

If you fix your own, you'll hit a noise floor from your neighbours...

David

That's why my aerial is 100 foot down the garden. I have some good quality LED lights, I can use a MW radio in the room they are in until next door turn on theirs!

[DigitalNoMore]

I worked on the design of state of the art telecomms equipment for years and was heavily involved in getting equipment through worldwide EMC regulatory testing. It was difficult enough getting equipment to meet regulations even when money wasn't much of an object. Obviously, margins are very low when it comes to the design of LED bulbs and so corners will inevitably be cut and barely adequate noise filtering will be designed in. Not an excuse, of course, but the root of the problem I suspect.

If indeed the bulbs are EMC tested individually then clearly this is inadequate and a more realistic testing scheme is needed. Good luck to anyone trying to devise one though!

The regs. say things (domestic) only have to meet the 'essential requirements' (whatever that means) not a specific level. Great get out clause.

[Herald1360]

The essential requirements are that it doesn't interfere with other equipment and isn't itself upset by normal of RF to be found in its expected environment. One way to demonstrate this is to test against recognised national standards for EMC, but even a pass there does not guarantee that the essential requirements are met in all circumstances.

Getting anybody to actually enforce the legislation is the big problem.

[G6Tanuki]

Originally Posted by Herald1360

The essential requirements are that it doesn't interfere with other equipment and isn't itself upset by normal of RF to be found in its expected environment. One way to demonstrate this is to test against recognised national standards for EMC, but even a pass there does not guarantee that the essential requirements are met in all circumstances.

And what may be put forward by a designer as a "test sample" often ends up somewhat different once it's been tested/approved then gets sent to production with an instruction "make a million of these, by next week, for no more than $0.57125 apiece".

It's not uncommon to find cheap stuff like LED lightbulbs, phone-chargers, laptop PSUs with the spaces marked-out on the PCB, and the holes drilled for the leads of the EMI-filtering components which may have been fitted for the test-sample (which may actually have complied with the regs when built that way) but these filter-components are absent and replaced by simple wire links.

[emeritus]

There are certainly LED bulbs available that produce almost negligible amounts of hash. The bedside table lamps at our last visit to Warners were excellent in this respect, with no discernible interference on AM until the radio was about a foot away. There was no interference detectable by my Sony transistor on any waveband with the radio positioned as in the photo. That was unfortunately the only good thing about the lamp. Although marked GLS, and of traditional GLS shape, the light output was so directional that most of it (over 80% as measured by my camera's exposure meter) was directed up to the ceiling. It would have been fine in a pendant fitting, but was pretty useless as a bedside reading light. Made by BELL I think.

[Nymrod121]

Not wishing to go O.T. but there seems to be a much wider interference problem that extends beyond the A.M. bands ...

We have a D.A.B. radio in our kitchen, the RF input being from a simple half-wave vertical dipole formed from the 'separated' inner and outer of a short length of 75ohm coaxial cable. This is suspended behind a wall-mounted cupboard with the coaxial inner portion being topmost.

It works better than the original 'F' connector-equipped telescopic. Even so, some stations appear to be 'nearer the digital cliff' than others (SmoothUK vs the BBC National Services for example).

Recently I replaced the 6 off 'cool white' LED GU10 ceiling downlighters (4 watt) with 'Diall' branded 'warm white' types of the same notional power consumption. The technology has obviously improved in the last seven years because the Diall lamps give a considerably brighter light - however, all D.A.B.services are obliterated as soon as these new ceiling lights are switched on.

The pictures herewith show the outcome of a very crude test using a Band III magmount antenna (mounted on an oven shelf rack for both mechanical stability and to act as a sort of ground plane!), the output being fed to my Tektronix spectrum analyser (7L12 in 7603 storage mainframe). Input / reference level, span, resolution & video bandwidth controls being adjusted to obtain the best possible visual display of the D.A.B. broadcast band.

Two traces were obtained using single shot sweeps, one with ceiling lights off, then with them on. The noise floor appears to rise by approximately 1/2 division i.e. ~5dB. Clearly this is more than enough to swamp the marginal signals being received by the crude dipole located at the back of the cupboard.

The point of interest to me is the amount of wideband hash being generated at Band III ...

Just for interest, I'll go back and do the same thing on Band II and will post the results later.

Best wishes
Guy

[Nymrod121]

OK ... I changed the magmount antenna vertical to a notional Band II dimension.
The three blips are BBC Radio 2, 3 & 4 from Sutton Coldfield.
There is an increase in noise floor when the Diall LED units are switched on, however the traces can be seen to diverge (increasing level of 'hash'?) as the sweep goes further up the band.

Not very scientific tests, I agree - but it's prompting me to think about going back to GU10 halogen bulbs ...

Sorry about the rotated photos (doh!!)

Best wishes
Guy

[Scimitar]

It is worth looking at the power units. I replaced standard 12V lamps with LED and noted a flicker, also the RFI became bad enough to affect 70cm FM. I think it may be the power supplies that are now too lightly loaded with the LEDs and to prove it, I have bought a specific 9W power supply, designed for LEDs. Just got to get around to fitting it now.

I have other RFI from the house that as yet I haven't traced. Turning off all the power drops the 40m hash from 20dB over 9 to around S8. Individually turning lamps and other equipment off makes no difference. A little more head scratching to do!

[G8HQP Dave]

Diall is the B&Q brand for lighting. Like most retailers at the budget end of a market, they now seem reluctant to use their own name on their own products.

You may find better results from a proper branded product, although these days even that cannot be guaranteed.