Homemade *bright* LED light setup

[torrent99]

Originally Posted by Haku

Yeah soak test is a good idea, I'll be running the LEDs from a stable 12v source, 25watt DC-DC converter which is adjustable so if I find the LEDs are burning out I can simply lower the voltage. I stuck to 120ohm resistors which is for 3 3.2v LEDs in a row each drawing 20mA, whereas the LEDs I bought are rated for 3.4-3.6v, undervolting them will extend life & reduce brightness but heck there's going to be 300 of them so it can't be all that dim!

BTW, I just found the ultimate LED bike light: 100 watt LED (but it costs £240+)

100 Watt, that's just silly!

The other idea I had for your circuit was to common each of the LEDS at to it's neighbour. So instead of 20 independent strings of 3 in series you have 3x20 LEDS in parrallel arranged in a series string of 3x 20 parrallel LEDS. (language is failing me here!). That way if a single LED fails it won't take out the other 2 in its serial string as well, the extra current will simply be shared amongst the other 19 in the parallel block. Should be pretty easy to mod your PCB to get this extra resilience. (Having had to retrofit failed LEDS I'm paranoid about resilience! )

[rog_london]

Originally Posted by torrent99

100 Watt, that's just silly!

The other idea I had for your circuit was to common each of the LEDS at to it's neighbour. So instead of 20 independent strings of 3 in series you have 3x20 LEDS in parrallel arranged in a series string of 3x 20 parrallel LEDS. (language is failing me here!). That way if a single LED fails it won't take out the other 2 in its serial string as well, the extra current will simply be shared amongst the other 19 in the parallel block. Should be pretty easy to mod your PCB to get this extra resilience. (Having had to retrofit failed LEDS I'm paranoid about resilience! )

No, you can't do that. LEDs don't behave sensibly in parallel, in that there are minor differences in operating characteristics, which means that some take a LOT more current than others if you connect them in parallel. If you vary the current through a single LED and monitor the voltage across it, you'll notice that the voltage does not vary much with quite a large change in the current through it. In other words, like any other diode, it tends to behave as a constant-voltage device (though not quite). Therefore you need a series circuit to keep the current constant through each LED. Even then you may have to 'tune' the resistor if you mean to drive the LEDs close to their maximum rating, as you will get variations as in Haku's 60-LED circuit.

Of course there are LEDs which are designed to work in parallel (since this can be a useful characteristic) but they are more than just LEDs, incorporating some sort of regulator circuit to make them 'pseudo' constant-current, and you pay more for those.

Rog.

[luxlighting]

Now Led Bulbs such as MR11,MR16 ,GU10 ,E11,E14,E17 ,E27 etc because of its light source is LED .( Light emitting diode ) .with advantage of environment ,long life lighting and no -fading .Have already replace halogen in wide range such as hotel ,resterant ,bar ,project .home .garden etc .

Especially use CREE XR-E LED ,reach 107Lumens/W ./350mA drive .a 3*1W bulb is enough to replace a 60W halogen .


Auto light 12V also in our producing list .( Wedge bulb ,BA9S ,Festoon etc ) .

Want to see more performance of these bulbs please ask us at any time .

Best regards

Danny Duan
Shenzhen Lux lighting tech co.,ltd
e-mail : luxlighting@126.com ,duanjian_2006@yahoo.com.cn
Skype ; sinocolighting
Msn:gtl-guangzhou-danny@hotmail.com
.

[Haku]

The circuit boards arrived last Friday, no solder resist so they're cheaper and quicker to have made:



Last Saturday the resistors arrived and today the 1500 LEDs+push switches arrived:



So I went ahead and soldered up a couple of boards which took 2-3 hours:



When soldering them I was testing each new set of 3 LEDs would work ok with a slightly used 8x alkaline D cell pack (11.5v) and that was bright enough during the daytime, now it's dark I rigged the two LED bars up to a regulated 12v 3A supply - holy [expletive]! you can't look at them direct or you'll get camera-flash blindness.

The most annoying thing I discovered is that after a few minutes they get too hot to touch (alkalines can't put out much amps which is why they didn't get warm on the 8x D pack), so I may have to lower the voltage down below 11v and possibly use lower ohm resistors so they don't dissapate so much heat. Two bars are drawing 10 watts from the 12v PSU, I tried 8x rechargable AAs which were 9.5v and the bars together drew less than 4 watts but weren't as bright.

The heat problem also means I can't use the intended polymorph plastic as the shell/backing for the light because above 60 degrees C it becomes soft & pliable (it's how you can re-form its shape again and again). Going to need a toohpaste tube size of thermal paste & piece of aluminium as the backing. I wonder how good Milliput is as conducting heat...



The only thing that worries me about building such a monster LED array is that a year down the line I'm going to be thinking "I could do with a bit more light..."

[Haku]

Soldered up a 3rd 'bar' today and have been doing some testing with a bench power supply and my multimeter that can measure volts & amps at the same time to also show watts.

After having to swap out 4 LEDs because they burnt out in a bit of soak testing at 12v, I discovered that at 12v the 3 'bars' were using 15.8 watts and were getting burning hot, but dropping the voltage down to 10v didn't significantly reduce the brightness and the 180 LEDs were then using only 5.6 watts.

So the full 300 LED array shouldn't draw more than 10 watts at 10v, instead of the anticipated 24 watts at 12v, and furthermore I can probably get away with using a 10 watt DC-DC converter I already have instead having to order a 25 watt DC-DC converter from DimensionEngineering. Which is nice

[rog_london]

Originally Posted by Haku

So the full 300 LED array shouldn't draw more than 10 watts at 10v, instead of the anticipated 24 watts at 12v, and furthermore I can probably get away with using a 10 watt DC-DC converter I already have instead having to order a 25 watt DC-DC converter from DimensionEngineering. Which is nice

Most LEDs give almost their full output with well under the maximum rated load, so I'm not surprised your array behaves as it does. Also, the large change in power consumption for the fairly small reduction in applied voltage makes me think that probably your resistors are not dividing the load very evenly and you may be over-running some groups of LEDs while under-running others. This will be due to the manufacturing variances among the LEDs rather than discrepancies in the resistors. The problem can be at its worst near to maximum rating, so your decision to reduce the applied voltage may be a wise one for several reasons!

Rog.

[Haku]

I was worried that running the array at 10v would adversely affect the brightness as I've finished soldering all 300 LEDs for my light today and couldn't properly test it until a couple of hours ago when it got properly dark.

After charging up a couple of RC car battery packs & putting them in series to power the 10 watt DC-DC switching regulator then taking it out for a test, I can honestly say I can't bloody wait to get it permanently fixed to my bike:



It's like a car headlight! Drivers are certainly gonna know when I highbeam them


edit: WOOOOW!!!
I temporarily stuck it over my existing light and went for a spin along some dark roads & cycle tracks with all 300 LEDs powered by the two RC car battery packs, it's fricken unreal, it's like you've got your own personal piece of daylight in front of you! even saw a deer at the end of the journey

[Haku]

Slow progress the past few weeks, a combination of no spare cash for parts & ill for a week.

Anyway, I've now managed to order a bunch of PICAXE chips + parts needed to make them run so I'll be able to start soldering the control circuitboard for the bike light, with a pile of spare PICAXEs for other project ideas.


I went out a few more nights with the 300 LED array blutacked & taped to my bike, it's too much fun! I discovered that one set of traffic lights that never sees me unless there's a car stopped behind me can be triggered by high-beaming the LEDs and you can get cars to easily dip their highbeam by putting all 300 LEDs on when they're still some distance away. But the 'Blue Peter' fixing method got too much hassle to maintain so I took it off and won't put it back on until it's waterproof & finished. I plan to put some 2mm thick acrylic over the front of the LEDs to stop muck sprayed from the front wheel from getting inbetween them, also making it very easy to clean the muck from the light.

[z0mb13e]

Hey Haku,

Interesting project, how is it coming along?

I figured you might be interested in something like this given you are familiar with microcontrollers...

SpokePOV: LED Bike Wheel Images

Have you looked at pulsing your LED array rather than running it at a constant current? LEDs can produce more light and last longer if pulsed. Something your PICAXE can do easily. Not sure what frequency would work best but might be something to explore.

[monster]

why don't you guys just use MR16's or MR11's they are way better than any amount of led's stuck to a reflector. the amount of time and effort you have to put into this, i would be gutted if one broke. with MR16 you can just replace it for 50p. the efficiency is not all that different and we have big batteries so why not used them?

my bike puts out more light than most motorcycles.