Last night wandering thru Lowes I spied a LED array lamp sold as a replacement for a 12V 10W halogen. Now since I have several lights of this type in my kitchen that never work as they burn out too frequently to keep up with (and they ain't that cheap either even from IKEA) I bought one of these as a replacement:

Out the door for $8.98. That's cheap if I never have to replace these. I still have to test the importaint thing: how good is the light? It's a nice warm color but with a 67 lumen rating will they be bright enough compared to a 10W halogen which has a rating of 160 lumens. I'll test that tonight in my kitchen at night.

But first let's tear this sucker down! There's no case per say to open, the device is a white PCB with parts surface mount attached. The wires sticking out the back appear to be heat sinks, the pins are all common but unconnected to the rest of the circuit. Each pin has a tab going under a LED to form a thermal path.

The board itself was very dirty! I could see lots of obvious flux residue, and stuck in this were some black fibers. I couldn't read any part numbers on the ICs till I cleaned it.

Just from the outside measurements it's a nifty circuit. It runs bidirectionally so either polarity DC or an AC are good. Current draw at 12VDC is about 130mA, and there is current limiting so even up to 20V it still draws 130mA. As the part heats up this current does drop some, so there is good thermal regulation too.

Here's what I pulled out as a schematic:

No suprises here. U3 is a bridge rectfier for AC use. U1 and U2, whild unidentified parts have to be current source regulators; their output tops out at 9.6 VDC. All LEDS are in 4 parallel chains of 3 LEDs each for a total of 12 LEDs.

If anyone is good decoding top marks the regulator has an L2, small underlined S, and the number 11 on it's side like so:

Links:
LED light at Lowes
LED at Lexxus (manu)

 

Great description, if those were available over here I might buy some just for the parts.
I had a look at the LM317 datasheet and in current regulator mode Iout = Vref (1.25V)/R which would give 83mA per regulator, so it might be something similar to that.

 

It's a linear triple output regulator somewhat similar to the Supertex CL320.

 

Tried the light last night in my kitchen. While the light it gives is not as bright as the halogen bulb it replaced it is a pleasing shade of white that is warm, and does give good illumination without being overly harsh. I might prefer it to be slightly brighter but I imagine I'll forget about that in a few days of getting used to it.

When just coming into regulation (only point I took data) there is only .3 volts across U1 or U2 pins 6-5, and .23 across R1 or R2. So U1 & U2 are some sort of low dropout (LDO) regulator with I suppose an current set by the resistor.

 

So what do you suspect each LED is drawing for current after subtracting power loss in the rectifier, resistor, and regulator, maybe 120mA/4 or 30mA and 3.2V Doesn't seem to warrant any heat sinking.

 

Interesting to read your teardown.

For a minute, I thought this was a 10 watt LED light for under $10, but I was disappointed to see that it's not a 10 W LED light but a “10 watt replacement” whatever that means ... actual power only 1.5 W or so max. It seems like the cheapest you can get for LED lighting is on the order of $1.00 USD per watt, so I thought that 10 watts of LEDs with driver circuitry for less than $10 was pretty great. (Of course lumens are a more useful measure, but I will generally assume that lumens/watt is in the same ballpark for decent power LEDs...)

Your the energy savings will be significant over time even if you were to install twice the number of LED lights compared to halogens to get about the same amount of light output.

 

I did not attempt to open a path to measure actual current to the LEDs. Obviously some current may bypass them out pin 1 of the regulators but I would assume this would be a small fraction of the output current.

With 12V and 127mA in (hard data I took) these things are seeing an actual input power of 1.5W, of course a fraction of this does get emitted as light energy but the majority stays as heat. Keep in mind these things may be sealed inside a small enclosure made of plastic and glass, and neither is a good heat conductor. So the temperature can soar inside.

Where the heat off the pins is going I don't know, I do know that with a competitive price driven commodity item no one is going to spend the money for 12 pins and then solder them onto a board unless they are absolutely necessary.

I like it. I'm going to buy some more. And since they draw so little current, I may do something creative to switch them on and off, like a capacitive touch panel dimmer switch or such.

 

It's interesting that the promotional image you posted in #1 says “Available in AC and DC voltages”. Do you suppose the DC version has no rectifier? But then the socket must be polarized to prevent badness. Do you suppose they means 12 V DC versus 24 V AC, just referring to the “usual” AC or DC voltage in these systems instead of to the fact that it is actually AC or DC current?

What would they change in the design to suppose different voltages? You discovered that the device does linear regulation of current, so in theory you could connect any higher input voltage and it would be fine except it would become an increasingly effective heater and eventually a firestarter.

To answer my own question, I guess if you want to run off 24 V power with reasonable efficiency, then you need to put more LEDs in series so that you have to drop less voltage, right?

I imagine we won't see a product at this price/brightness point that uses a switching regulator for higher efficiency and wider input voltage range, at least not at this time.

 

With 12V and 127mA in (hard data I took) these things are seeing an actual input power of 1.5W, of course a fraction of this does get emitted as light energy but the majority stays as heat. Keep in mind these things may be sealed inside a small enclosure made of plastic and glass, and neither is a good heat conductor. So the temperature can soar inside.

I didn't consider that it might be in an enclosure. From your post it sounded like a bare board, but it makes sense that the product might be sold also as part of an enclosed bulb unit. Is you unit mounted vertically? It looks like light output would be mostly to the sides, which is usually not the direction you want it. A diffuser bulb around it would help but of course would be horrible for thermal issues.

 

Do you suppose this "Pins"/heat sinks just dissipate heat from the PCB board.

 

Do you suppose this "Pins"/heat sinks just dissipate heat from the PCB board.

Wouldn't you say that is the very definition of a heat sink?

 

Can anyone read Nexxus Lighting's U.S. Patent Number 7,911,797 without thinking: what is novel about this?! I don't understand how their claims are uniquely different and non-obvious improvements to the techniques used in virtually every PCB design that deals with power devices.

Nexxus Lighting, Inc. (NASDAQ: NEXS) today announced that it has been awarded a U.S. utility patent (U.S. Patent Number 7,911,797) - APPARATUS AND METHODS FOR THERMAL MANAGEMENT OF ELECTRICAL DEVICES, patenting the use of thermal pins (Selective Heat Sinks™) to dissipate heat away from components mounted on printed circuit boards (PCBs).

PRNewswire link

The patent system is beyond broken.

 

I thought I'd add some contrast to the LED array discussion. ErnieM, I hope you don't mind the addition to your thread....if so I can delete the post.

I recently bought a pack of 5 LED arrays from the take your chances" store ebay.
They were about $11.50 with free shipping. (about $2.25/array)

They are also AC or DC capable but do not have any voltage regulation and the basic resistive current regulation.

My belief is that they are merely high brightness SMT white LEDs and pass a current of 20mA( did not test). If this is so, then I lose 2V on the resistor and and have 10V/3 or 3.33V on each LED. This is reasonable only if you have a regulated voltage source of 12V. But the add says... use it for your car or RV...etc. which could lead to well above 12V, not to mention an unregulated wall-wart. This could push near 4V across each LED and lead to premature burnout. They are also packed closer together than your array and draws 2W without any heat sinking. I turned one on for 20 minutes or so in open air and really felt no heat at all, but as you say, if they are enclosed in a housing the heat factor could change.

 

iONic, have you tried your LED arrays? Are they as bright as you might expect for a 2 watt LED light? Five of these for $12.00 is a good deal if your application doesn't require precision (lack of regulation means applications are more limited).

 

iONic, have you tried your LED arrays? Are they as bright as you might expect for a 2 watt LED light? Five of these for $12.00 is a good deal if your application doesn't require precision (lack of regulation means applications are more limited).

I did light one up with a small 12V LA Batt and they were pretty bright @6500K.
Without a 2W LED I couldn't really say how it compares. I could easily read a book with one at 4 - 5ft. They would light up the underside of a kitchen cabinet just fine... all I'd need would be a regulated 12V source at 1A to drive them all.

Maybe I can drill some tine holes in the PCB and stick some sowing needles in for heatsinks. NST (Needle Sink Technology).

 

colinb: On Nexxus website they picture both the AC and DC models the same. That would kinda have to be, as there is no polarity on the base so both need the rectifier and limiters. Looks like they change the LED current somewhat between Ac and DC models. DC is always on, AC will wave brief blink outs every cycle, so they tweak the AC current higher it seems.

Also, I use these for the enclosures in my kitchen. It's not cool with the wife to have bare bulbs just hanging out <smile>

IONic: Nice find! What seller did you snatch these from? Just from a count of LEDs one would think 24 is brighter then 12.

 

colinb: On Nexxus website they picture both the AC and DC models the same. That would kinda have to be, as there is no polarity on the base so both need the rectifier and limiters. Looks like they change the LED current somewhat between Ac and DC models. DC is always on, AC will wave brief blink outs every cycle, so they tweak the AC current higher it seems.

Another nice engineering feature of your array, reflected in the cost. For all I know the LED's I have in my board were rejects from a manufacturer that did not meet specifications and sold to whomever built my arrays.

Also, I use these for the enclosures in my kitchen. It's not cool with the wife to have bare bulbs just hanging out <smile>

IONic: Nice find! What seller did you snatch these from? Just from a count of LEDs one would think 24 is brighter then 12.

Don't really want to give out free advertisement so just search for the following: (5pcs G4 24-3528 SMD LED Pure white Light Bulb Lamp New) These LED's need to be under-driven in my opinion as there is no datasheet.

I like those enclosures but they'd be a bit expensive for my taste and would add too much cost to the overall project. I have a couple of 5 1W LED units enclosed in a 1ft enclosure complete with 120V power cords. They were about $50, but are more than enough light for a section of counter. Bought them at Lowes.

Bottom Line, if your good with SMT parts, you can remove the 24 LEDs for use in other projects. At less than .10/LED it's not bad.

 

Some measurements:

With Regulated Source of 12.00V

Voltage across the "100 ohm" resistors varied from 1.20V - 1.42V, although resistances varied as well from about 97 - 102 ohms. Voltages across the LEDs varied between 3.02V - 3.16V. But given the current in the resistors and hence the LED's (aprox. 13.6mA) it seems even a tad low, unless it is used with unregulated sources.

Since this does not jive with my previous assumptions I am guessing that I am dropping some voltage across the bridge rectifier (.75V). Increasing the supply .75V to 12.75 brings the calculations back into perspective. There is also noticeable heat (wouldn't want to place it on your cheek unless you like checker-board tattoo's) that might be of concern if contained in a tight enclosure. Given the heat production I would not drive them with more than 11.5V, rectifier included or 10.7V with rectifier bypassed, without adding some sort of heat sinking. The only way I see adding heat sinking is with thermal glue and a heat sink to the back of the PCB. Not sure how well heat dissipates thru PCB.

ErnieM,
Did you feel the heat on your array, if so how would you say it rated? I'm beginning to think I hijacked your thread!!!

 

Do you suppose this "Pins"/heat sinks just dissipate heat from the PCB board.

Wouldn't you say that is the very definition of a heat sink?

Actually, no. Dissipating heat directly from the heat source would be more appropriate I'd think...and more efficient at doing so as well. Cooling the PCB to essentially cool the heat source is an indirect approach as the dissipation characteristics of the PCB are far from equal to the direct contact to the heat source itself.

 

iONic: Did you just quote your own post and answer it too?

Not to worry, I have a special affinity to simple LED lighting and appreciate all comments on this and similar devices. No hijacking seen here, post away!

While I do feel some heat it doesn't feel "excessive," meaning I'm not worried about a thermal tatoo. I just put the device on my bench, LED side down (so I can still see) and stuck a thermocouple onto a "heatsink" pin. In still air it's getting up to a 60°C rise. It might be interesting to see how hot it gets inside my NAN lamp, which should be a sealed unit but I popped the glass so I might get a tiny bit more light from it.

It will be interesting to compare this with the ones you have, I have a set coming to me on that proverbial on a slow boat from China.

The pins on this unit would not be entirely useless, as they are making a thermal path from directly under each LED to the other side. Then a thermal convection air path could pull some heat off them. It has to be better then trapping all the heat against the PCB itself.

However... my next post may also be interesting...