PDA

View Full Version : Noisy LED bulb



J Tiers
03-16-2011, 11:10 PM
So a friend of mine who lives off-grid brings in an LED MR-16 type bulb.... His issue was that when the LED bulbs are on, they completely wipe out AM, FM and weather radio reception in his place.....

He has one type, a warm white that he likes from Home Despot for $30, and a bluer type that he doesn't like, but that is 8 bucks delivered from a chinese source.

Both are to blame for this noise. Both are CE marked, which may be a simple "self-certified lie", at least as far as the EMI situation.

Whan I checked the 8 buck one out, it produced full quieting of an AM radio several feet away from the bulb and wires, completely wiping out a local AM station that the radio was actually tuned to..... all you heard was either near silence at the 550 khz end, or strong hash noise at the 1600 kHz end of the dial.

Investigation with the equipment here produced the information that it gave a strong wide-band noise output up to at least 5 mHz, which was as far as my easily-brought-in radios would go.

Interference with FM was also noticed, but not as strong. He plugs his FM radio into the 12V DC wiring at his place, which also runs the 12V lights. If he plugs into the inverter output 120v wiring, he has less problem with the noise, presumably due to the isolation between them.

These little bulbs apparently use a totally unfiltered SMPS to reduce 12V input (AC or DC) to 4V or so to run the LEDs.

At 12V input, they are as noisy as sin. With 9V DC input, they are pretty quiet, no doubt due to the variation of SMPS duty cycle with input voltage.

I intend to use the spectrum analyzer at work to check them out in detail, but even without that, I am shocked at the amount of RF noise this little 4W LED "bulb" puts out.

I had figured that OK, a capacitor across the DC input would probably do the job.....

Uh-Uh.... not even close. The direct broadcast noise, presumably from an inductor in the SMPS, evaded any line filters I tried, so that even though the DC lines were filtered and quiet, the radio still picked up strong interference. It took a pretty decent Corcom line filter to clean up just the wires, with no effect on directly emitted RF noise.

I do NOT like this....................

Here is the perpetrator..... known as an "LX051S"

http://img.photobucket.com/albums/0803/jstanley/LEDmr16_2.jpg

http://img.photobucket.com/albums/0803/jstanley/LEDmr16_1.jpg

Paul Alciatore
03-16-2011, 11:38 PM
Report it to the FCC. Seriously. That is a hazard to the entire RF spectrum.

beanbag
03-16-2011, 11:42 PM
My mom had recently sent me a link to a scare documentary that talked about how compact fluorescents throw out horrible radiation as well.

DougC_582
03-17-2011, 08:00 AM
My mom had recently sent me a link to a scare documentary that talked about how compact fluorescents throw out horrible radiation as well.
It's been known for a long time that any kind of fluorescent bulbs generate strong interference to AM / shortwave listening.

I was really hoping that LED bulbs would not, but it looks like the store-bought kind may not be useful there. Have to build yer own with a stepdown transofrmer, a couple rectifying diodes and a resistor I spose....

Evan
03-17-2011, 08:38 AM
I have the same bulb in front of me. The output is about 6 volts since the diodes are wired series-parallel. It is filtered with a small electrolytic which can be verified by waving your fingers in front of the bulb when it is on. The inductor is about .1" in diameter and the lead to the diodes about an inch long. The heat sink shell is electrically isolated from the circuitry so it cannot act as an emitter. I find it hard to believe that such a small inductor with such a short length of wire can transmit much on the AM frequencies.

Using a handheld multiband radio on a local AM station I get mild interference from my track lights within 18 inches or so. The track lights in question run from 120vac so they have indvidual ballasts for each light. The noise is coming from the ballast since the one I took the bulb from is still making noise without the bulb installed.

Correction: I also have a string of three that are running from 12vac. Those bulbs are making some RFI but it only interferes at a distance of less than 18 inches or so. We have a pretty weak AM signal here as we are in a deep valley and the transmitter is on the other side. Incidentally, quieting is produced by a carrier, not by RFI. It isn't a term used to describe AM receivers since a carrier will beat with another AM signal. Quieting applies to FM discriminators since they will lock on the strongest signal and ignore the rest.

Rosco-P
03-17-2011, 10:54 AM
He has one type, a warm white that he likes from Home Despot for $30, and a bluer type that he doesn't like, but that is 8 bucks delivered from a chinese source.

Both are to blame for this noise. Both are CE marked, which may be a simple "self-certified lie", at least as far as the EMI situation.

What about the findings on the $30 HD bulb? The one I saw seemed to be intended as a direct replacement for PAR bubs in recessed cans. The color temperature was a little on the cool side, but it did have a high CRI.

Evan
03-17-2011, 11:31 AM
As a point of reference I just measured the 20 watt LED bulb I built using a standard Chinese off the shelf emitter and switch mode power supply. It's installed in a can on a track light. It produces detectable interference only directly in front of the emitter at a maximum distance of about 2 inches.

RenoF250
03-17-2011, 02:39 PM
I have the same bulb in front of me. The output is about 6 volts since the diodes are wired series-parallel. It is filtered with a small electrolytic which can be verified by waving your fingers in front of the bulb when it is on. The inductor is about .1" in diameter and the lead to the diodes about an inch long. The heat sink shell is electrically isolated from the circuitry so it cannot act as an emitter. I find it hard to believe that such a small inductor with such a short length of wire can transmit much on the AM frequencies.

Using a handheld multiband radio on a local AM station I get mild interference from my track lights within 18 inches or so. The track lights in question run from 120vac so they have indvidual ballasts for each light. The noise is coming from the ballast since the one I took the bulb from is still making noise without the bulb installed.

Correction: I also have a string of three that are running from 12vac. Those bulbs are making some RFI but it only interferes at a distance of less than 18 inches or so. We have a pretty weak AM signal here as we are in a deep valley and the transmitter is on the other side. Incidentally, quieting is produced by a carrier, not by RFI. It isn't a term used to describe AM receivers since a carrier will beat with another AM signal. Quieting applies to FM discriminators since they will lock on the strongest signal and ignore the rest.

????
How does waving your hand verify it is using an electrolytic? That only shows the bulb is flickering.

The case being unconnected DOES allow it to radiate, it is capacitively coupled to the noise inside. If you don't want it to radiate it needs a low impedance path to ground.

What is your theory as to why your bulbs/setup are producing so much less noise than the OP? What you are seeing is what I would expect of a bulb like this.

Do either of you have pics of the bulbs disassembled?

Evan
03-17-2011, 04:34 PM
Waving your fingers in front of a bulb is very good way to see how well the DC supply is filtered. Wave your hand in front of your TV to see what I mean. Lots of flicker there. These bulbs don't produce any noticable flicker which indicates decent filtering.


The case being unconnected DOES allow it to radiate, it is capacitively coupled to the noise inside. If you don't want it to radiate it needs a low impedance path to ground.


That isn't possible with an MR16 bulb. There is no ground available. Capacitive coupling will be very slight. The wires are twisted together and run through a hole in the center of the heat sink shell to the plate that has the emitters bonded to it. It is very symmetrical which reduces coupling. I will takes some pictures shortly.

Evan
03-17-2011, 07:27 PM
http://ixian.ca/pics8/mr16a.jpg

http://ixian.ca/pics8/mr16b.jpg

J Tiers
03-17-2011, 10:05 PM
I have the same bulb in front of me. The output is about 6 volts since the diodes are wired series-parallel. It is filtered with a small electrolytic which can be verified by waving your fingers in front of the bulb when it is on. The inductor is about .1" in diameter and the lead to the diodes about an inch long. The heat sink shell is electrically isolated from the circuitry so it cannot act as an emitter. I find it hard to believe that such a small inductor with such a short length of wire can transmit much on the AM frequencies.

See picture below, where I have circled the "loopstick antenna".......a term some may remember from the 1960s. Just as ferrite core coils were used to PICK UP AM in "transistor radios", the principle of antenna bidirectionality allows them to be a decent emitter also.

http://img.photobucket.com/albums/0803/jstanley/mr16a.jpg



Incidentally, quieting is produced by a carrier, not by RFI. It isn't a term used to describe AM receivers since a carrier will beat with another AM signal. Quieting applies to FM discriminators since they will lock on the strongest signal and ignore the rest.

Regardless of that generally correct statement.... "quieting" is exactly what was happening, I used the term for that reason..... The local station was replaced by silence.

The little buck regulator presumably is operating somewhere in the low AM, near 550 or a submultiple of it.... There was a carrier from it in that neighborhood which was so strong that it literally "quieted out" the local AM station, leaving "dead air". The signal was strong enough that it simply overpowered the other one, and I assume the radio has some form of AGC which dumped out the weaker signal.

At higher frequencies, the signal was such as to produce a noisy "hash", which was very evident right up through 5 mHz. Confirmed by interrupting power periodically to check if the noise was affected. It was........... it went away with no power going to light. Most stations were wiped out, the local 50,000 watt clear channel AM was just distinguishable through the strong noise.

At some distance , as in several feet, the effect was less, but still very strong.

My friend gets the effect with any radio operating from the same 12V DC power line.

He does not use AC power for the lamps. I understand that in general, and that may or may not include Evan's setup, a power-pack is used with them which probably includes some form of filter.

The unit is voltage-sensitive.... which means that the duty cycle of the regulator affects the harmonic structure. The thing must be operating with a long "on" pulse at 9V, but at 12V it may be a much shorter "on" pulse with more harmonics.

I don't have any idea what happens with AC yet.... That would be 12V RMS, but would generally be either highr or lower than an actual 12V.

Evan
03-17-2011, 10:12 PM
That isn't a decent loopstick for AM frequencies. It's just too small. Maximum range with the bulb shown is about two feet off channel. I don't run mine from DC so I don't know about that but if you wrap a few turns of the supply wire through a toroid it will take care of any RFI on the supply. If you want to get fancy you could put a ferrite bead on each supply wire inside the bulb. The bulbs are really easy to take apart. The bezel snaps off with a finger and 4 tiny screws hold on the base.

J Tiers
03-17-2011, 11:06 PM
That isn't a decent loopstick for AM frequencies. It's just too small. Maximum range with the bulb shown is about two feet off channel. I don't run mine from DC so I don't know about that but if you wrap a few turns of the supply wire through a toroid it will take care of any RFI on the supply. If you want to get fancy you could put a ferrite bead on each supply wire inside the bulb. The bulbs are really easy to take apart. The bezel snaps off with a finger and 4 tiny screws hold on the base.

The antenna works well enough to be picked up by another similar (but larger) ferrite core antenna at several feet distance.... I haven't calibrated it in detail.

Since neither of the cores is anywhere near AM wavelength dimensions, they are relatively similar in efficiency of reception/broadcast, roughly in proportion to overall dimensions.

In any case, theory aside, it is for sure noisy.

As it happens, I have a relatively complete collection of various cores, toroids, lossy ferrite clamp-on cores, cores on wires, beads, common mode chokes, line filters, and etc..... you get the point. I have a whole boxload, leftovers from doing lots if EMI work at a prior employer.

While you might think that a core such as you mention would do a decent job, in fact, they did not. I tried quite a collection of them, with little result.

Even the simpler types of line filter failed. At first I tried just an inductor and capacitir, or a lossy core and a capacitor, then a "legitimate" simple line filter. No cigar, it changed the spectrum a bit, but still the power-on vs power-off test showed plenty getting through at the low end

I had to go to a two-stage line filter to get the wires on the "far side" of the filter to be quiet when tested with a radio. The noise appears to be really wide-band, and while the simple cores/beads did OK at the HF end, they failed (as you might expect) to do much at the low end.

I'll know more when and if I get time to check it with the lower frequency spectrum analyzer at work. At the old place they had a 1.8 gHz EMI type S.A. including EMI software to run it and the whole quasi-peak averaging "official" response filters, etc , which might tell an even more interesting story, but they got rid of it. We actually have their old screen room, but someone else nabbed the analyzer and antennas when it was sold off.

He'll likely be OK with a good line filter on the wires to his array of lamps. He has 5 or 6 of these lamps on one DC line, and the RF noise from the lot of them is really very impressive.

Evan is correct that at some distance, the emission from the little core is not significant. It is the shorter range interference that is a hassle.

That type of "dog-bone" core is THE most noisy type.... known to everyone who does EMI work as a notorious public enemy. There are shielded ones, but they are a little more expensive.