What's new
  • ICMag with help from Phlizon, Landrace Warden and The Vault is running a NEW contest for Christmas! You can check it here. Prizes are: full spectrum led light, seeds & forum premium access. Come join in!

My $30, 20w DIY LED

Phychotron

Member
I've decided to start making LED grow lights for my mother/clone area. I've ordered a few components and was wondering if anyone would like to review my electronics to see if I made any mistakes or could make any obvious improvements.

I wanted to start out cheap and refine my techniques before I attempt a flower light.

Driver: 600ma 18-33V (20w) ebay generic ($6.35)
Diodes: Cree XTE 2900-3200K, 5800-6500K on aluminum pcb (5 each, $11)
Aluminum: 300x25x12mm finned (2, $5.32)

The heat sink is rated for about 12w. I was thinking of putting 4-5 diodes per bar and the one driver for both. I'm going to be driving the diodes at 600mA (3-1500mA) as they recommend which puts the diodes at 2w each (1-5W). I might end up getting a higher current driver later but I wanted to check out how hot it got first.

I'm also going to be doing this passive cooling for now but might add more diodes to each bar and strap a fan on top of them.

For mounting the whole thing I was going to build a wooden rectangle that allows the 30cm aluminum bars to just drop down into a slot. Its only going to be 2 bars right now but in the future if it has 10 bars I'll be able to swap out the individual bars of different color diodes.

Here's the bad part though, I went cheap on mounting the diodes to the aluminum, for now, and will be using double sided thermal tape. It has a few advantages for prototyping compared to epoxy and drilling. I do have access to a drill press and may eventually hard mount them.

Spent a little over $20 on materials and expect to spend a few more dollars getting the rest of the components together.

Right now I'm waiting for the parts but if anyone thinks I might need something else please let me know. Thanks

Heres my cheap stuff:
http://www.ebay.com/itm/13183623107...49&var=431187205669&ssPageName=STRK:MEBIDX:IT
http://www.ebay.com/itm/29188344876...49&var=590885475890&ssPageName=STRK:MEBIDX:IT
http://www.ebay.com/itm/252483697654?_trksid=p2057872.m2749.l2649&ssPageName=STRK:MEBIDX:IT
http://www.ebay.com/itm/221988168755?_trksid=p2057872.m2749.l2649&ssPageName=STRK:MEBIDX:IT
 

Dion

Active member
so u will mount 5 leds with 1 driver per sink right? or have i misunderstood you

the adhesive is fine
the driver will run 10(if specs are correct)
heatsink will handle 10 leds max at 600mA


it will be really hard to solder once u attach those leds(personal experience) so id advise you do ur wiring then stick those leds down

for anything less than 20w those sticky pads are great btw

I've always used them for single diodes, can't be beaten

sweet little set up dude

how many bars will you make? are you doing 10 per sink+driver combo?

EDIT: ok i re-read the thing, yeah 5 per bar is fine. 10 per par is right on the edge and seen as you already have the sinks may as well keep em cool and get more spread

don't forget to post up in the led porn thread once ur done
 

Phychotron

Member
Right now its 10 diodes, 1 driver and 2 aluminum bars.

Ideally it will be as modular as possible. Once I have enough bars and diodes to justify a fan I think It'll end up being about 10 diodes per bar or about 20w so that I can secure the power supply to the bar and be able to just unplug it from a 120v rail when I need to swap it out. I think with a 200mm fan it would cover around 7 bars. I'm not sure how many bars I'll end up with but probably multiples of 7 or 8 to maximize cooling with the larger fans.

In the future since there are no lenses I'll probably figure some sort of mounting system to hold a bunch in place without having to glue them.


It'll be awhile before i get the parts but I'll make sure to post a few photos of the build.
 

Phychotron

Member
PrivateMessage said:
hey mate just saw the LED bar your building im curious how did you work out the leds to the driver ? i was wondering how you did your math

I'm no expert but its simple: Volts * Amps = Watts

The diodes run at 300-1500mA, but we want to run them at 600mA (.6A, constant current) based on manufacturer recommendation.

Then you look at the voltage on the diodes, my diodes are 2.9-3.6V but 3.3 is the median I'm aiming for.

So we can figure out watts per diode, one diode, at 600mA.

3.3V*.6A = 2 watts (1.7-2.2 watts based on voltage range)

So about 2 Watts each diode is the power we need .

Mind you these are "5watt" diodes only being driven at 2watts. If I wanted to make it brighter I would get a 1500mA constant current driver with the same output voltage but with considerably more heat to manage and much shorter diode lifespan. It would also consume 2.5x the power.

So now you want to run 8-10 diodes in series you need a bigger constant current power supply, still at 600mA but with a larger voltage output.

Figure out your voltage requirements by just adding up all the voltage/watts diodes you want. if 1 diode is ~3.3V, 10 diodes will be about 33V. I found a power supply with a suitable range out output (18-33V) to supply the voltage without supplying too much voltage. It comes out to around 20 watts, which is what the driver is marketed as.

At 1500mA, a 20w driver would only drive around 3 diodes (9v) and would take a 50w driver to light all 10.

Once you know the current your driving at you'll be able to easily figure out the voltage/wattage of the driver you need.

Hope this helps, It was such a long reply I just copied it to the main post. This really is the most confusing part of building an LED. This is for diodes in Series, I'm not sure if Parallel has the same requirements.
 

Phychotron

Member
Got the driver and diodes but still waiting for the heat sinks. Of course I'm not going to wait around to wire them up and test the circuit so here is where I am so far.

It was pretty straightforward wiring everything. Just a matter of determining the diode spacing and placement and cutting the wire to length, soldering and plugging into the power supply.

Supplies so far, the small aluminum heat sinks are for testing until the real aluminum shows up.
picture.php




The components spaced out for a 5x2 configuration, alternating the 3000K and 6000K diodes.
picture.php


I'm using a polarity protected JST connector between the diodes and the power supply. To wire the power supply I'm using some orange 120v connectors I picked up from Home Depot. To test I have another end that plugs right into the wall socket. When I expand the design and more power supplies are added I will make a wiring harness that they can all clip into.

I also made sure to tin all the diodes/wires and let them cool before soldering them together. Although the iron was only on the star pcb for a second or three it still heated them up to the point that they would be almost too hot to handle.

The Moment of disappointment..

picture.php


Got it all soldered together, those tiny aluminum heat sinks barely underneath all them (which is hard cause the wires wont allow them all to sit perfectly flat) and I plugged the power supply into get absolutely nothing...

Turns out the orange plug on the power supply end wasn't put on properly, I didn't strip the wire enough to make contact. After fooling around with that for another minute or two and confirming that I was getting power before plugging the diodes in it was ready.

picture.php


and it turns out she's pretty damn bright. The photo above is interesting because if you look at the cabinets the camera picked up a ghost image of the diodes and you can tell the different spectrum of the diodes by the orange/green dots.

picture.php

Without a way to hold the light steady I just sorta held it above the plant, a lot of the diodes were all twisted and some were still in the cabinet(being held by a bottle of hot sauce).
 

Phychotron

Member
The photos came out upside down for some reason, but here's what it did to my apartment.

picture.php



I can only imagine how its going to look when I go full size with it.
picture.php


I'll post more when I get the heat sink and the mounting pads. I'll have to find a way to secure the two bars together to make a single unit, and I'm already considering getting another 10 diodes and drivers and putting 10 diodes per bar.
 

Phychotron

Member
I know its breaking my heart waiting for the rest of the stuff. I still need to figure out a way to hold/suspend the heat sinks. I might start a much larger frame that will allow me to just drop in the modules as I piece them together. I just got the 'fever.'
 

Phychotron

Member
Still waiting for the heat sinks to come in the mail but I decided to upgrade the design a little.

Pretty much all lights from china that I've seen have some sort of aluminum sheet with separate finned heat(s) glued on.

I purchased a 1" x 36" x 1/8" aluminum bar to mount the diodes on. I'm then going to add the finned heat sink to the bars using some sort of thermal glue. It gives me something to do now, but for mounting it seems easier to later drill holes into the flat bar instead of the finned piece. It also gives me a larger heat sink so that I can add more diodes to each bar more safely.

I spaced them out for 10 diodes per bar, but only putting on 5 each for now since I have the materials. in the future design I'm going to have slightly longer bars and 12 diodes per bar. Having the bars longer than the fins will allow for better mounting options, as it takes an extra half inch per end to mount.

I got a yard-sale sign post to make the frame out of for $1.50. Aluminum bar was $7
picture.php


I used the sharp end of the stake to make a diode press. I drilled a small hole in it for the diode and pressed all my weight down.
picture.php

picture.php


The frame with salvaged wire hangers. They're a bit too short so I might replace them later.
picture.php
 

Phychotron

Member
And here it is nearly complete.
picture.php


I decided to use all three bars since It's easy to just tack some diodes and drivers to it and upgrade it in the future without much extra effort.

picture.php


As soon as I turned it on it seemed MUCH less bright, and I think the frame is having some effect, since its absorbing the lights around the sides somewhat.

I ran it for awhile but decided its not worth cranking it until I get the finned aluminum.

Here it is over a plant as an example of its color temp. I'll try and keep everyone posted on what this thing can actually do with a clone or two in veg.
picture.php


I might just order a few more diodes and drivers and try and pump this thing up another 40-50 watts. I already have some spare fans/drivers from salvaged lights that I can easily slap on.
 

Phychotron

Member
I accidentally left the light plugged in and realized it's been running for the past day with my other light.

Since none of the diodes were burned out I decided to use the cheap $10 IR thermometer to check the diodes.

It was a little difficult to get an accurate reading at first but I realized it was just me not doing it consistently. Its so bright that I just scanned the whole area around the diode and then checked the max temperature after I pulled it away. Even with sunglasses I'm still seeing spots. You have to look a little and these things are bright.

The temperature was around 140-150F, or roughly 65C. Its not as bad as I thought it would be with just that but not what I wanted. According to one site, here, they're running at around 80% efficiency.

I have it unplugged for now until I get the actual finned aluminum and run some more tests. The back of the aluminum was 75F, roughly room temperature. However I need to read to see if the aluminum needs to be calibrated for the thermometer. The star PCB's have a coating that should read accurate.
 
Last edited:

Dion

Active member
thats good news

I've used simple aluminium u chanel/ c chanel things to sink monos pretty well too btw
 

Phychotron

Member
The plot thickens...

After thinking about the lower temperature on the back of the aluminum all day I decided to add a piece of tape to change the reflectiveness and see if I got a different read from the meter. I used a very thin piece of first-aid tape and compared the two temperatures. I've actually never seen a cloth tape so thin but the thinner the better in this case.

Before it was 71ºF (22ºC) but just by adding the tape the meter was reading 120ºF (49ºC). I then got to thinking about it while doing other things and realized that the PCB might have the same issue. I tested the center and it went from 162ºF (72ºC) to 175º (80ºC). Not nearly as bad but still some discrepancy that needs to be addressed.

According to THIS site the temperatures are comparable to commercially available residential bulbs. That doesn't mean I want to stop there. I want to get them as cool as possible and for that I will be testing some fans that I'm going to rip out of a broken LED fixture.

I'll try and get that worked on tomorrow. The finned aluminum has yet to come in the mail. I just ordered some cheap thermal paste to adhere the fins to the bar so I'll even be able to test the difference between just setting the fins on top compared to paste. Who knows how long that will take from China.

I might remember to do a side by side of the two bars between a sanded and factory-direct joins. I took 150 and 300 grit sandpaper to the bars before I mounted everything so that I could get the best possible contact (but not the star pcb's as I already had them soldered). It was pretty easy to tell when it was flat by getting it wet and watching the surface tension break patterns. The 300 grit sandpaper was much needed and I made sure to go back and forth, then side to side at half pressure, then back and forth with lighter pressure until the stroke marks from the previous direction disappeared.

I have access to a drill press and tap set but might just get some self tapping metal screws and secure these proper with some thermal paste under it. When I mounted the bars to the wood frame the holes I drilled were too small and I used the last few of those scews I had from another project. Not sure if it'll secure the diodes tight enough but I'll be able to cut any excess off with a hacksaw and file/sand the ends flat.
 

Phychotron

Member
My finned aluminum came in the mail today... but they only sent 1 instead of 2. Looks like it'll be another 4 weeks before I can finish it.

I should just order some more parts and expand it out.
 

Phychotron

Member
Since I have the one I've got it on a side by side to see how much its able to do on its own just by taping it to the aluminum bar. This is just a test until I get some thermal paste. I want to demonstrate the difference between the two. I'm also going to repeat this test to see if the top end changes once I sand the finned aluminum down a bit more flat.

Here you can see a considerable sized gap between the flat bar and the finned aluminum.
picture.php


I tested the two sides by scanning the temperature around the center diode. After a few minutes I'd check again. For the most part the finned aluminum keeps it about 10ºF cooler with a slightly slower ramp. I didn't record the time between scans but it was about 2-4 minutes.

Finned, Plain
101, 132
126, 127
135, 145
163, 171
173, 180
172, 185
173, 188
165, 182ºF

Not sure why it would drop at the end, but it would do that sometimes and I'd rescan it and it'd be back up, so its most likely my error or the cheap meter.

When taking the screws out of the bars to fit the fins on I realized that the screws were tapped pretty well. I did use metal-self tapping screws so its no surprise. I might end up taking off the double sided thermal tape and use the screws to hold both the diodes to the bar and the bar to the fins. I just need to drill my holes in the right spots to get the fins to work with the screws and keep it from wanting to offset.
 

Phychotron

Member
I cant find my Kill-A-Watt meter so I used a multimeter to try and figure out the power consumption. The voltage of the circuit of lights was 30V just by sticking it in parallel to the diodes. I found the hard way that when I tried to test the current by doing that the lights all turned off, then back on when I removed the meter. A quick search online revealed that the meter needs to be in SERIES in order to measure the current. So I fashioned up a double ended JST connector with the black wire soldered together and red open, to be connected to the two ends of the probe. Much to my amazement the light turned right on and was reading .62A

.62*30=18.6watts for the diodes. I'd like to see how that compares to the Kill-A-Watt which will also factor in the driver as well.

I ordered more finned aluminum, another driver and 10 cree royal blue diodes. I'll do some more work on it as the parts start to come in. I even ordered the finned aluminum from another vendor who estimated the shipping a month faster than the slow-poke I first ordered from. It should be all done by the end of the month.
 

Phychotron

Member
thats good news

I've used simple aluminium u chanel/ c chanel things to sink monos pretty well too btw

I was looking at that and wondered if the shape would help. I see there are a few different options when it comes to aluminum from the hardware store. The only problem i see with the C bars is they make them thinner than the flat one's. I'll have to pick them both up and compare weights.

I ordered some more diodes and drivers once again. This light is so handy around the house that I need to replace it when I finally put it in the veg room.

The new parts are just some crappy 1w diodes (50 for $5) and three dimmable 21w 350ma drivers ($12). I'm going super cheap on it and will probably test out the C channel aluminum. I'm going to just use a little thermal paste and crazy glue to secure them. When one burns out I'm just gonna chisel it off and replace it. So ~60 watts for around $25. I'll keep that one posted as well.
 
Top