110v LED project

[BlakVBR]

I have been lurking, reading and studying up on the wealth of information here. Now I need to ask for some direction. My goal is to make a DIY highly efficient 110v based LED lamp that includes all the following wavelengths:

425 nm deep blue
460 nm blue
480 nm blue
630 nm red
660 nm deep red

The application (which is probably obvious from the spectra) is intended to be used for indoor growing lamps. Despite this, the lamps will not be near water so if I have to run 110v lines around I have no problem with that. The plants will be 100% organic so they won't be exposed to rainwater and must survive the long dim winters.

I realize that in order to get all those wavelengths on one lamp I will probably have to use 5mm leds and put the lamps directly over the plants. As long as the electrical efficiency can still be good, I am OK with that style. If not, maybe I can use 5mm for the odd wavelengths and high power LEDs for the more commonly available wavelengths? Or I could scrap the odd wavelengths and try to cover them by using high power white LEDS?

Can anyone point me in the right direction, especially as far as finding a value point for good efficency? TIA!

 

[evilc66]

At an absolute minimum, I would use 1/2W narrow angle superflux LEDs, and not 5mm. You would have to run thousands of them to make it worthwhile, and even then it's questionable. At least the superflux LEDs stand a fighting chance to work. Ideally you need to be looking at high power LEDs. Finding all of those wavelengths is going to be very difficult. Edison has a number of "odd" wavelength LEDs that will hit some of your choices, but they are difficult to find in the US.

Using a neutral and/or warm white LED might be a better solution just from the fact that it has a broader output, and far greater luminous output than other LEDs.

 

[BlakVBR]

Keeping all that in mind, Im starting to think that it might be best to try and use LED to replace the blue fluoros that are typically used to augment HPS lamps and maybe also add 660nm. That certainly would be a more realistic project to start with.

 

[BlakVBR]

Is there any reasonable way to connect them directly to AC power line? Most of the AC-DC adapters I see are 25-30% efficient. Are there any better options?

If not, I have an assortment of AC-DC adapters I have been collecting and a few old laptop power supplies. Is there any way I can determine whether they are constant voltage or constant current? Can I still use them somehow if they are constant Voltage?

TIA I can't seem to find the answers to these types of questions anywhere.

 

[bhvm]

Rather than spending extra cash on Odd wavelenth LEDs, Buy the ones that are easily available- and add in a bunch of whites to compensate for more!

Also, High power LEDs will last longer and more efficient than 5mm or superFlux.

Always use high power LED.

 

[Firecop]

I'm just getting started in aeroponics, myself (the legal, tomato-y kind)

Check this out, I've learned tons:

http://www.greenpinelane.com/

Having said that, my research into the topic (Crees, mainly) show that "warm" Cree XR-Es have substantial ~600 nm emission, with 60% ~450 nm production. I just ordered 24 of the older, cheaper 7A chromacity.

I'm looking at using an old PC power supply's 3.3v rail to run a long string of these. As far as I know (I'm not an EE), laptop PS are constant voltage - the ratings are listed in watts, indicating a variable current. (I'm pretty sure I'm right on this, but I've been wrong before).

I hope that helps.

 

[lolzertank]

You can get Luxeon Rebels (or the upcoming color Cree XP-E) for the 460nm, 480nm or 630nm wavelengths. 425nm and 660nm are harder to find since they're nearly IR and UV.

 

[BlakVBR]

Thanks for all the input!

I found a cool white Cree XR-E Q5 on DX that says it can do approx 175 lumens at 700mA with forward voltage approx 3.7. So power consumed by the LED is 2.6 watts. If I was lucky enough to find a CC power supply that is 33% efficient, that means 7.8 watts total would be required. Therefore 22 lumens per watt overall. Was that math correct?

Cheap compact 6500k CFLs typically get 50-60 lumens per watt (if we believe the advertised 26 watts of consumption and the advertised output of 1660 lumens).

So if all that math was correct, it looks like I am better served to augment the HPS with 6500K fluoros, unless there was a heat issue that requires the LEDs.

 

[BlakVBR]

This power supply claims up to 88% efficiency:

http://www.computronics.com.au/meanwell/eln-60d/

Is that a different type of technology like PWM or something along those lines? 88% I can live with

 

[Firecop]

Most newer computer PS that I've seen are ~80% efficient at their optimal loads. When you're calculating efficiencies (for economic reasons, I'm guessing), don't forget to factor in the initial costs.

I'm lucky enough to have a 6.5kW solar system, so I don't feel bad about a less-than-perfectly-efficient lighting system.

If cost is not much of an issue, the newer T5 fluorescent light bulbs are approaching 100 lm/watt: 56 watts used/5000 lumen output.

The stuff I've been researching shows that not much is truly known whether high lumens outside the most efficient wavelength is better than lower total output in the "sweet spot". Experimenting is half the fun!