A Different Method of Driving Hi Powered LEDs, Using the linear approach

I recently completed a lighting project using a series string of four 30W LED light engines being driven directly from rectified & filtered mains voltage using a single LM317 in current regulator mode. Total of 120W.
I am using an oversize heatsink, so no cooling fan is needed. The unit is fully convection cooled. Chip temp is running around 70°C, with 22° ambient.

http://i202.photobucket.com/albums/aa106/t...525/leds004.jpg

http://i202.photobucket.com/albums/aa106/t...525/leds003.jpg

I have more than 150 operating hours on the LED unit now with no sign of trouble.
The interesting thing is that these LED light engines consist of 30 individual 1W chips connected in series/parallel on one substrate.
This means I am controlling 120 separate LEDs in massively parallel hookup, with one LM317.
The forward voltage drop of the LEDs is such that the 317 sees less than 25VDC, even when line voltage is at it's upper limit. If I did the math right, the system should operate on any voltage between 105VAC & 120VAC. Adding a small fixed resistor would shift the range to 110-125VAC if necessary.
The current in this setup is under 880ma. That same 317 could be used for total powers of up to 200W. Those with 240VAC mains could go up to 400W.

After using this, I wonder why commercial units don't use a similar setup? It's much cheaper than even a simple switchmode driver. About the only disadvantage I can see is that this setup is fairly sensitive to voltage fluctuations...flickers & starting surges seem a bit more noticeable than with other light sources.
The circuit is the standard current reg circuit, with a 10Ω trimpot to set current, & the addition of two .01µf bypass caps to ground, one at each end of the regulator. These are essential...the system tends to oscillate without them. I'm using 800µf of filtering after a full wave bridge of 1N4007s.


BTW: These LED light engines cost $8 each, including shipping, & will last for more than 50,000 hours.

Ed: big pic link

This post has been edited by Sch3mat1c on November 15, 2012 03:53 am

They do! Just aren't available to the general public.

I can't recall the name, but it's a 100% USA company that makes them for RF sensitive environments.... communications stations for commercial/govt./military, research and medical laboratories, etc.


Awesome project, BTW

Man am I out of the loop. Last I checked 1W LEDs were still at least a buck apiece.

Was just going to string together a bunch of normal 5mm white LEDs into an old flashlight reflector as a headlight for my e-bike. If 1W are that cheap though, might go that way instead.

Is heatsinking a particular challenge because of the way you're doing this, or is that just because of how much light there is at the back of the things? I don't recall commercial bulbs being particularly hot or having especially prominent heatsinking.

Not gonna do so great when a heavy inductive load turns off and bumps the line a few volts over the 317's limit, but works great otherwise.

What are the bright glowing things in the first picture??

Tim

^^ They kind of look like reflections from overhead CFL's.. They look sorta 'twisty'.

My guesses would be first because it won't take over voltages of even brief duration very well, second, the UL make take issue with several factors of such designs.

That being said.... that is an awesome way to get a lot of light.

Exactly right. They are reflecting off of the plexiglas splash shield mounted over the electronics.


@Tim:
The system has already absorbed several inductive spikes. The result is a brief flash of extra brightness & nothing else. Apparently the filters keep the voltage from becoming excessive. I also have a suitable fuse on the input.
Note also that the line voltage here is ~113VAC, which means the regulator normally has only ~15 volts across it. This seems to leave plenty of headroom.
If I ever have a problem, the addition of a suitable MOV will cure it.


@Telomere: See above. Also, why would the UL have any issues...it's a simple matter to double insulate the entire assembly, with mica under the heat spreader block. I also have mica under the 317 & the LEDs themselves have isolated bases. To make it up to UL standards would require only the addition of the mica under the spreader block & the addition of side skirts to the plexi shield.
The unit is already wired with a polarized plug.


Now that I see how well this method works, I will be using a lot more LED lighting in the future...I plan to convert much of my shop lighting eventually.
They now produce LED light engines up to 300W, which means setups of up to 1200W are possible.


BTW: My 200W LED lighting setup, with the unregulated auto transformer supply, now has more than 16,000 continuous hours on it without a single failure. The only problems I've had are 2 blown fuses. Perhaps due to surges. The reworked MOT I'm using has been running cool & quiet the entire time.
This has eliminated any doubts I had about the reliability of such linear LED drive approaches.

My advice for bike headlights is to use an MR16 track lighting bulb in a suitable enclosure.
These all run from 12V & are available in a 15° pin spot pattern, in either 20 or 50 watts.
These bulbs make for headlights that actually let you see where you're going.
Nothing else I've seen comes close.
They cost from $3 to $5 apeice & are available everywhere, but the pin spot pattern may have to be ordered from the electrical wholesaler or purchased online.
LEDs aren't really up to headlight use yet, because you would need too many of them to get enough light, plus focus optics to get any kind of decent beam pattern.
Worse, the high power LED light engines I use are tough to focus properly because the emitting area is so big. They are almost impossible to focus into a tight beam.

Heatsinking can be a problem. My designs tend to overcompensate on heatsinks, because I'm not entirely sure just how much heat they will stand & also exactly what thermal resistance I end up with. So, I tend to use larger than necessary heatsinking, because it would be major work to have to upgrade after completion.
In spite of this, all the commercial LED lamps I've seen over 1 watt are mostly heatsink, once you take away the diffuser globes.