Project: multiple rebels, huge output

Hey CPFers. I've got a bunch of Luxeon Rebels on order and I have plans for them. Not a flashlight, but a fixed light run on DC from the wall.

What I want to do, essentially, is to create a bloody lot of light. And I want it to be dimmable (continuously not discretely).

I plan on driving about 10 rebels (for starters) up to 1500mA each. I don't think any off the shelf drivers are going to cut it because of the current draw. I'm certainly no electrical engineer, but i've drawn up a circuit that i think *might* work. My power supply is variable voltage and regulated, so I can set it to whatever I want. Voltage spikes shouldn't be too significant, it's a good power supply.

Here's what i have (I based my numbers on evan9162's rebel performance data):



I'm not sure if this is going to work - especially the pot as a dimmer. Ideally I want to have a fair bit of control over brightness at the high end, so I don't want it to be too touchy for making adjustments.

I'd be really appreciative if any of you genius modders can give me a hand!

personally i would stick to a max 1A or 800mA current and just add more rebels for more light as you will loose a lot of efficiency beyond that at which point it would be easier, cheaper & more efficient to use a flourescent tube.

For a 1A max drive current there are many drivers you could use too.

1.5A will surely make the wires heat up good.
the pot will need to be able to handle at least 5A...

going at 800mA or 1A should make it good enough.
like what the guy above me said.

Leds have very sharp, non-linear v-i curves that vary slightly from one led to another. If you have leds in series they all get the same amount of current, but in parallel, even if "identical", they don't necessarily divide up the current evenly. So you should use a separate limiting resistor for each parallel string.

I think I'd go for PWM dimming rather than using a pot, but that's just me.

Hello spaech

welcome to CPF

I agree with the others, don't drive it beyond the 1000mA maximum spec since efficiency and lifetime will be reduced. Also, make sure the LEDs are well heatsinked at this current or they will burn out pretty quickly. 15Watt pot is insufficient for this setup since LED's will be using almost 4W each. Like paulr suggests PWM would be much more efficient for dimming than a pot.

What is your intended application and power source? This will determine how much trouble you should take regarding efficiency: efficiency is far less of a concern for for AC line-powered lighting than for something like a battery powered emergency light.

My general inclination would be to use a microcontroller programmed to operate as a multi-channel PWM controller driving independent MOSFETs for each series string. Using a microcontroller means you can adjust the ballistics of brightness control to your liking (log, linear, log-lin, or anything else) and independent PWM channels allow for more sophisticated dimming such as alternating turn-off periods to reduce flicker or shutting off half the LEDs when running at under 50% power.

Thanks for the comments.

paulr: Point taken about using a limiting resistor for each parallel string.

I'd like to use a more efficient dimming method, but that's probably beyond the scope of the initial experimental phase (and probably beyond my electronics knowledge just now). PWM sounds good, but it'll be more expensive than a pot I'd imagine, and i'd have to switch to AC.

monkeyboy: Thanks for the welcome. I'm not sure a 15W pot would be insufficient - yes, the LEDs are drawing about 4W each at max current, but that's the LEDs sinking that power, not the pot. At max current the pot will be set to its min resistance position, so it should only give a fraction of a volt drop, meaning it won't be drawing much power. As resistance is increased the voltage drop over the pot will increase, but the current will quickly decrease because of the short range of operating voltage of these LEDs. So i don't think the pot should ever be drawing near 15W. I could be wrong with this reasoning though, so let me know if i have any misconceptions here!

enLIGHTenment said:

What is your intended application and power source? This will determine how much trouble you should take regarding efficiency: efficiency is far less of a concern for for AC line-powered lighting than for something like a battery powered emergency light.

My general inclination would be to use a microcontroller programmed to operate as a multi-channel PWM controller driving independent MOSFETs for each series string. Using a microcontroller means you can adjust the ballistics of brightness control to your liking (log, linear, log-lin, or anything else) and independent PWM channels allow for more sophisticated dimming such as alternating turn-off periods to reduce flicker or shutting off half the LEDs when running at under 50% power.

Click to expand...

The application is a plant grow light. I'm going with LEDs over flourescents or HID or whatever because I can control the spectra being emitted to increase efficiency (chlorophyll only really absorbs in red & blue, reflecting the rest).

The power source is a variable voltage DC power supply, powered by AC from the wall. So efficiency isn't much of a concern. However, using microcontrollers to manage PWM dimming sounds great, with the amount of control it offers. It would just be a pretty big learning curve for me to be able to implement a solution like that.

spaech said:

I'm going with LEDs over flourescents or HID or whatever because I can control the spectra being emitted to increase efficiency (chlorophyll only really absorbs in red & blue, reflecting the rest).

Click to expand...

This implies you're using red and royal blue colored Rebels. If so, then you'll have additional control complexities because red Rebels have higher Vf requirements than do their blue counterparts.

If you're not using red and royal blue, then you're barking up the wrong tree. White LEDs are pretty much a waste of effort for grow lights as the phosphor robs 450nm (chlorophyll A blue absorption peak) that plants could use to create a broad swath from 550nm to 775nm that isn't especially useful for photosynthesis. Compare plant absorption spectra (http://www.biologie.uni-hamburg.de/b-online/e24/24.htm, http://www.botany.uwc.ac.za/ecotree/photosynthesis/spectrum.htm) with the Rebel emission spectra (http://www.philipslumileds.com/pdfs/DS56.pdf) to see what I mean.

enLIGHTenment said:

This implies you're using red and royal blue colored Rebels. If so, then you'll have additional control complexities because red Rebels have higher Vf requirements than do their blue counterparts.

If you're not using red and royal blue, then you're barking up the wrong tree. White LEDs are pretty much a waste of effort for grow lights as the phosphor robs 450nm (chlorophyll A blue absorption peak) that plants could use to create a broad swath from 550nm to 775nm that isn't especially useful for photosynthesis. Compare plant absorption spectra (http://www.biologie.uni-hamburg.de/b-online/e24/24.htm, http://www.botany.uwc.ac.za/ecotree/photosynthesis/spectrum.htm) with the Rebel emission spectra (http://www.philipslumileds.com/pdfs/DS56.pdf) to see what I mean.

Click to expand...

Yep exactly, I'm using red and blue at the moment for testing, but plan to add royal blue to the mix later on.


There aren't really any additional control complexities with this setup (in terms of Vf reqs) because I intend to put the different coloured LEDs on separate circuits so I can control the output of each colour independently.

OK, I guess you're right about the pot. Here's a good PWM driver though:

http://www.taskled.com/d2dim.html