It's a very simple problem to solve.

Buy luxeon, cree, bridgelux or any other reputable brand from a real supplier (ie: you won't get a counterfeit from digikey, steves leds, mouser, etc..) and you will get exactly what the datasheet specifies.

Otherwise, buy a bunch of unknown chinese stuff and test all of it, then select the best parts and plan based on what you end up with. In the end, you might save money but you'll spend much more time. And you'll end up with lots of useless parts.

You obviously don't enjoy finding treasure among junk as much as I do. But, thank you for the list of American LED sellers. I will buy a few 3W LEDs from them just to compare the LEDs with the Chinese counterparts.

It doesn't take much more time to install two drivers in place of the one driver that was originally planned. I just don't want to pay for an extra driver.

I went through the same thing buying Cree XT-E 5w white LED's on ebay premounted on stars. They also seem to be running at a higher voltage than even the lowest BIN real Cree parts so I am suspecting I may have bought fakes. They are working though and growing veg's very well so not a complete loss or anything. For my latest build I bought real Cree COBs from Mouser even though it seemed I was paying 2-3 times more than what I could get on ebay. Well, they are exactly to spec and I am getting about twice the lumen output per watts vs. the XT-E's (due in a large part to learning to run them at lower current and buying the best BINs). Thinking the electricity savings running these things 16 hours a day will quickly offset the additional part costs. I haven't found anybody on Ebay that even lists the BIN/part no's so that tells you something too I guess.

That is the reason I use the dc-dc boost convertors hacked to a current source it will make every voltage between Vin+3V -> 60V by the desired current. The current will be 1.25/R for your 700mA is that R=1.25/0.7=1.8 omh

The resistor dissipate 0.7x1.25~0.88W

Some explanation by the picture, I replaced the output cap 100mF/50 by 100nF/63V the max voltage of the IC is 60V. Than I put a resistor between the - out and the sense point of the IC (right pin of the potmeter) and put a little heatsink on top of the IC. The diode is only used when I'm testing protect the circuit against stupidity (reverse voltage).

I know there are extra losses with using two steps 220 -> low voltage and the low voltage to the Led voltage, but is much safer to work and play with.

Just got my 3 and 10 watt full spectrum leds, I'm curious what they will do probably I will mix those with some cool white, easier on the eyes.

By the way, I tested 3 Watt leds more than a year ago and the spreading was something from 3.1 to 3.5V most where 3.2-3.3V by 500mA.

scorpio_bafh, my objective is to do this as cheaply as possible and tell forum members how. One bargain that I found was this:

http://www.ebay.com/itm/Effective-High-Power-Driver-For-1-3pcs-Leds-1-3x3W-Light-Lamp-12V-500MA-600MA-/371221542634?var=640355428665

This is the $.99 driver that I used to test the 3W LEDs. It looks like crap, but it stays cool to the touch during operation. I previously had success using a similar $1.34 10W 900 ma driver:

http://www.ebay.com/itm/201413623531

And, notice the free shipping. These can be mounted by gluing with silicon caulk. American sellers don't give you free shipping on small orders because they believe that their labor is worth too much for that ... minimum wage ... blah, blah, blah.

rooftopbklyn, I ordered these from Steve's LEDs for comparison with the Chinese counterparts:

http://shop.stevesleds.com/SemiLeds-Deep-Red-3W-Newest-Version-8794102499.htm

http://shop.stevesleds.com/Philips-Luxeon-ES-Deep-Red-3-Watt-LEDs-RARE-Version-8794102435.htm

http://shop.stevesleds.com/Philips-Luxeon-ES-ROYAL-BLUE-3-Watt-LEDs-8794102476.htm

And, I was charged $4.00 for shipping.

Incidentally, the owner of that website must be insane for putting those large, obnoxious images at the top every webpage. I have to scroll past them to view the serious content on every webpage I look at! In a perfect world, karma would give that person carpal tunnel syndrome so that his mouse movements are excruciating.

Hi aruzinsky

I use these convertors:

http://www.ebay.com/itm/DC-DC-Adjustable-Step-up-boost-Power-Converter-Module-XL6009-Replace-LM2577-L5RF-/381480756678?hash=item58d20665c6:g:UXwAAOSw8-tWVxNw

Those are much more flexible than the ones you showed, you can go too about 14pcs of 3.5Volt leds the only hack, I think, really necessary is to add the current control resistor to change from a constant voltage to a constant current source the other hacks I make are more extra's I like to add such as the heatsink for higher currents and the Cap. for higher voltages.

When you use few leds you need a buck convertor to lower the input Voltage.

http://www.ebay.com/itm/LM2596-DC-DC-Adjustable-Buck-Converter-Step-Down-Module-Power-Supply-1-23V-30V-/262104670178?hash=item3d06a7dfe2:g:o70AAOSwv0tVFO8c

I still wonder about your test of those powerleds, it is strange that both are nearly 1 Volt to high at the same current, have you try to test, for example, at 200mA ? it should give nearly the same 1 Volt difference.

The schematic of the hack

Looks like you have the right spectrum covered for sure aruzinsky.

What is your ratio of red/blue going to be? I ended up going 8/2 with mine.

You may want to add some white led's on a separate circuit for viewing- the "pink" light you will get will make it hard to tell how your plants are doing.

I am on my third build/grow test for my DIY LED's now and instead of getting the individual LED's mounted on stars with all that wiring/soldering like my first build, I am using these boards:

http://www.dx.com/p/30w-10-cree-xp-e-led-1000lm-red-blue-plant-grow-light-module-white-22-26v-190272#.VmLoTeJ0ac0

and then driving each board with one of these:

http://www.dx.com/p/water-resistant-20w-led-constant-current-source-power-supply-driver-90-264v-186892#.VmLpPeJ0ac0

Prices come down pretty good when you get 5 or more. I am using 6 and plants are thriving under them. Very intense lights. So far in my tests these seem to be outgrowing my XT-E's and COBs. Also, pretty basic wiring, just solder two connections to power supply. Only have to worry about making sure you have adequate cooling.

Hi Stu,

You lose pretty much power with those parts, some because of false promise from the seller and also from using a to low current driver, 0.65A instead of 1.0A

>> http://www.dx.com/p/30w-10-cree-xp-e-led-1000lm-red-blue-plant-grow-light-module-white-22-26v-190272#.VmLoTeJ0ac0

The board need 24 - 26 Volt and 1000mA result is ~ 25 x 1.0 = 25 Watt not the promised 30 Watt. (a rough calculating of the LED voltage give indeed ~ 25Volt)

>> and then driving each board with one of these:

http://www.dx.com/p/water-resistant-20w-led-constant-current-source-power-supply-driver-90-264v-186892#.VmLpPeJ0ac0 "

With this driver the power will be 25 x 0.65 ~ 16 Watt

>> Only have to worry about making sure you have adequate cooling.

Yes cooling is very important special the red LED's are very sensitive for heat the efficiency will really go down and also the longevity

.

Ditto, stu zone. You would be better off with standard 30W, 900 ma drivers. I use these $8.08 30W drivers to operate three 10W LED chips in series:

http://www.ebay.com/itm/Water-Proof-Power-Supply-LED-Driver-for-3x3w-10w-20w-30w-50w-100w-SMD-LED-Light-/291463138119?var=590550313365

But, I just realized that I can use 650ma 20W drivers to operate my out of spec 3W LEDs, maybe, with the aid of scorpio_bahf's upconverters for 12V operation.

scorpio_bahf, I just ordered 5 of those upconverters. Thank you.

I'm using the 650ma to get better efficiency, passive cooling and longer life. After studying the datasheets I realized I was being "penny wise and pound foolish" running my earlier LED's at max power. Running these things 16 hours a day everyday really adds up and the efficiency really drops as you get to max current. I think they call it a 30w board because it is 10- 3w leds. The same way a 3w red led is not actually 3 watts. Yes, 16w per board was my calculation too, for 96w total in a 2x2' area. They are just mounted to a 1/8" thick sheet of aluminum. I have a fan connected through a thermal switch in case they start to get hot but it never comes on- probably will in the summer though when my basement heats up!

Hi

aruzinsky, Stu

Special the red LED's efficiency drops with temp, see graph. in link

http://www.ledsmagazine.com/content/dam/leds/migrated/objects/features/4/8/1/MarlFig4.jpg

Luckily the blue ones are much better and those are used in the LED's with the fosfor coating to get a wider spectrum. And Stu you are right the efficiency drops also with the current, I use most of my LED's at 60-70% of max current.

aruzinsky, did you check the picture for the output cap. of the convertors ?

This one with a 35V output Cap

http://www.ebay.com/itm/DC-DC-Adjustable-Step-up-boost-Power-Converter-Module-XL6009-Replace-LM2577-/310717070508?hash=item48582e3cac:g:IycAAOxy9DRSDNm2

and this one has a 50Volt output Cap (those I bought)

http://www.ebay.com/itm/DC-DC-Adjustable-Step-up-boost-Power-Converter-Module-XL6009-Replace-LM2577-/371054648639?hash=item5664949d3f:g:nx0AAOSwGvhUJ3NQ

The last one give you more headroom and as I wrote before I even replaced the 50V with a 63V,

I glued a heatsink 22x10x10 on the XL6009 which is half of the following 22x22x10

http://www.ebay.com/itm/181846949696?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

I roughly calculate the input current as follow

output power/input voltage X 1.1

Has to be within the limits, 2A without heatsink 3A with. I go for save and not higher than 2A with the heatsink

I use cheap 60mm fans like this one,

http://www.ebay.com/itm/60mm-12V-Internal-Tower-Desktop-Cooling-Cooler-Fan-Computer-PC-CPU-Case-Quiet-y/281712669139?_trksid=p2045573.c100033.m2042&_trkparms=aid%3D111001%26algo%3DREC.SEED%26ao%3D1%26asc%3D20131017132637%26meid%3D4b984f0385894e0ba08d61d8a76eefef%26pid%3D100033%26rk%3D1%26rkt%3D4%26sd%3D281712669139

Since they only use about 2W, I keep them on while the LEDs operate.

My bad. Yesterday, I realized that I made a mistake in the measurement procedure. I connected the LEDs to the driver using tiny jumper cables (the kind with alligator clips at the ends) and measured the voltages ahead of the jumper cables. The voltage drops across the jumper cables added 0.63 volts to my measurements. I didn't suspect that each cable had a 0.5 ohm resistance. Now that I corrected this error, about 90% of the LED forward voltages fall within specification. It's still a good idea to measure the forward voltages and discard the out of spec LEDs.

Hi aruzinsky.

Puzzle solved, must be something it was too far off in my opinion.

I just ordered some little DC-DC convertors for regulating the fan speed, I used LM317's before but when coming from 20-24Volt to 6-12 they became pretty hot now I found these:

http://www.ebay.com/itm/331530620180?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

I can't build as neat as those for that price, maybe I can even use those for driving one or maybe two 10 Watt LED's, a pity the max voltage of the IC is only 26V but the IC is good It is synchronous (MP2307) instead of the asynchronous (XL6009 and LM2596)

scorpio_bafh, why are you using 20-24V instead of 12V from an old computer power supply? By now, everyone should have some old computer power supplies, Heck, I have four or five of them, good for about 15A each. I also have two waterproof 4A, 12V power supplies.

Hi aruzinsky,

That is because of the current limitations of the XL6009 boost convertors I'm using. That is 3Amp max with heatsink, 2Amp without, for example

I have a string of 3Watt leds 50V total and 0.7Amp add up tol 35Watt. Using 12Volt input the current will be 35/12x1.1 =3.2Amp at 24Volt input current will be 1.6Amp

Yes Pc power supplies have enough current and old ones are laying around everywhere.

Maybe I can connect two PC power supplies serial as long the low voltage side is not connected to the mains ground.

scorpio_bafh, have you considered using an adjustable step up constant current supply like this,

http://www.ebay.com/itm/DC-DC-250W-Constant-Current-Boost-Step-up-Module-Mobile-Power-Supply-LED-Driver-/251715894855?hash=item3a9b6fbe47:g:TDQAAOSweW5U9BhR

, together with a 12V computer power supply? Using resistors to limit current is wasteful of power.

Hi aruzinsky

The serial resistor is not a current limiting but a current sensing resistor as you can see in the schematic I posted earlier, it is a pity that the sense voltage is 1.25V for the XL6009 and not for example 0.5V, so I will loose 1.25 Volt x Current in the resistor, at 0.7A is that 0.87W this is independent how high the output voltage is. I found a hack to lower the sense voltage but did not try it out yet, it can also make the current variable (I hope)

I like to have smaller units so the unit you pointed out has too much power/current, The LED light I'm working on at the moment will be 5x 10w full spectrum grow LED's, 6x 660nm red and 4x royal blue. The red and blue are separate adjustable to shift the ratio between the colors

Yes I'm thinking about to use PC power supplies and will try to connect two serial to get 24 Volt, have to check if the mains is free from the low voltage side.

I have just checked a power supply, the low voltage 0 and 12 Volt are free from the mains but not from mains ground and housing so you have to put electrical and mechanical wise the supplies separate and don't use cables with a ground wire, Or better remove the mains ground pin in the power supply connector to avoid mistakes.

The new tryout (work in progress) 65 Watt maximum, 350x40mm, 1x XL6009 for the 5x 10W LED's, 2x small down convertors for the 3W star LED's, heat sinks 22x22x11mm for the 10W Led screwed under the LED, a blower and a plastic duct. Not on the picture switches for the red and blue string and a convertor to control the fan speed

Nice. I don't see screws mounting the LED chips. Are you using thermal adhesive, and, if so, what kind? If you use thermal adhesive, are the PCBs on the 3W beads necessary? Why not just glue 3W beads without PCBs to the heat sink?

I discovered that it is much easier to solder wires to the 3W bead tabs than to connectors on the PCB. In case you ever make a spotlight, solder beads on the PCB interfere with the proper mounting of lens holders.

Hi aruzinsky ,

I think the thermal paste is a better thermal conductor than the clue so the led is put on the star with paste and the star is glued to the Al. strip. It is also easier to change a LED you don't have to mess with the wires. You right you cannot see the screws because there not there yet, they came Thursday now I have to find a 1.8mm drill I known I have one but where !

The thermal paste silicone based, I usedit also to glue the heatsinks to the dc-dc convertors

http://www.ebay.com/itm/390875533588?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

I have been experimenting with "thermal silicone plaster."

http://www.ebay.com/itm/10g-HC-910-Thermal-Adhesive-Glue-Gel-Tube-Heatsink-Plaster-CPU-GPU-Mainboard-/311414023150?

After it hardens, it is still soft enough to pry off the star, if needed. I can't use screws for my spotlights because the screw heads prevent proper mounting of lens holders.

scorpio_bafh , I tried your XL6009 hack to power just one 3W LED and the LED immediately burned out. Since the minimum output voltage is 5V, is that supposed to happen? Also, since the burned out LED is an open circuit, won't that cause the output voltage to go high enough to damage the module?

I ordered some 10 ohm 10 watt resistors to use as test loads so I don't burn out any more LEDs.

Incidentally, I used silicone heatsink plaster to mount 5W 660nm LED beads, without PCB, directly to the heatsink and they work fine. I checked one with a spectrometer and it has an unusually broad band. You may be interested in trying some:

http://www.ebay.com/itm/1W-3W-5W-White-Green-Blue-Royal-Blue-660nm-Red-High-Power-LED-Bulb-Chip-Bead-/111647091896?var=410658989908

Hi aruzinsky

Sorry I was maybe not clear about the convertors but a boost converter can not go below his input voltage for that you need a buck convertor. When you see the schematic there is a only coil and a diode serial to the output it can not regulate the output below the input, it must be even something like two or three volt above the input. There is an over voltage protection in the IC so I think it is still OK.

the buck or down convertor

http://www.ebay.com/sch/i.html?LH_BIN=1&LH_FS=1&_from=R40&_sacat=0&_nkw=buck+convertor+lm2596&_sop=15

An absence of open circuit protection is a problem. I think that I will stick to prebuilt constant current modules and I think that you should do the same.

Hi aruzinsky

I can see prebuild is easier and not error prone but I like the flexibility of my hack and now I have figured it out how to do it I will stick to it and even try to improve it, I'm working on a hack to make the current variable from 10 -> 100%

Using for example 19V note book adapter I can connect a string of 10 -> 17 pcs of 3 watt LED's 700nA or a string of 3 -> 5pcs of 10W LED's 900mA and easy change the current by using another current sensing resistor..

You are wrong about the absence of an over voltage protection, there is an in build over voltage (60V), over current (4A) and over heat protection in the IC see data sheet

http://lib.chipdip.ru/164/DOC001164799.pdf

scorpio_bafh, I ordered these constant current buck converters,

http://www.ebay.com/itm/Constant-Current-LED-Driver-Adjustable-Lithiumion-Batteries-Charging-Module-2Y/281807048030?_trksid=p2047675.c100005.m1851&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D20131003132420%26meid%3Dc86bd25d3d3e48a181b3997a209ef769%26pid%3D100005%26rk%3D2%26rkt%3D6%26sd%3D171991622283

http://www.ebay.com/itm/NEW-Buck-Converter-Constant-Current-4V-35V-to-1-23V-30VDown-Module-LM2596-01WF-/121763677855?hash=item1c59ae469f:g:UGUAAOxy79VReMTy

but, I will also try this circuit,

http://www.pcbheaven.com/userpages/images/LED_driving_and_controlling_methods_14.png

after the step-up power module. The source-drain saturation resistance of power mosfets is low, maybe, 0.2 ohms, so, at 700 ma, the input voltage can be set 0.6 + 0.7*0.2 = 0.74 V above the output voltage. And, open and short circuits are not a problem. Incidentally, these days, power mosfets are cheaper than regular power transistors. I ordered these:

http://www.ebay.com/itm/10Pcs-IRF520N-IRF520-Power-MOSFET-N-Channel-TO-220-GM-/262136735618?hash=item3d08912782:g:sC8AAOSw5IJWe4ao

I'm glad I found this forum, some very helpful stuff here. As for your comment on shipping aruzinsky, looked into it and I guess the Chinese government subsidizes shipping to help boost their economy. I was curious how so many of my $1-$10 ebay purchases could include free shipping.

"I just ordered some little DC-DC convertors for regulating the fan speed, I used LM317's before but when coming from 20-24Volt to 6-12 they became pretty hot now I found these:"

scorpio_bafh, Have you considered running two 12V fans connected in series directly off 20-24V?

" I guess the Chinese government subsidizes shipping to help boost their economy."

Chase Hart, see

http://forums.gardenweb.com/discussions/3596871/my-petty-revenge-on-liberals?n=36

Hello scorpio_bafh,

regarding your hack of the boost Circuit. Can you help me understand how it works? It looks to me as if the LEDs are being operated at the difference between the out+ and the reference voltage of 1.25 V. If i am right then you are driving the LEDs still using a regulated voltage set by the potentiomenter.

It looks to me as if that is the reason why aruzinsky burnt his LED. He drove it with a high voltage and the current went to infinity ;).

i am still interested to understand how to convert such a boost voltage regulator to a current regulator though.

Thanks for the info.

Hi ogaetej

Sorry I have missed your post, in short what happen to aruzinsky is that he used a boost convertor while needed a voltage lower than the input that is impossible. Look at the circuit, when you do that there is only a coil and a diode between the source and the load. the idea of the constant current is that when the current through the extra resistor (connected between ground and the sense pin) result in 1.25 V the IC will control the current. when the needed output voltage is lower than the input you have to use a buck convertor. Don't forget to adjust the potentiometer at something near the needed voltage otherwise the "hack" could fail. ( when the potentiometer is nearly zero it will control the IC and not the current sense resistor)

Hi scorpio, my point is that in your hack the LEDs do not get a regulated current. Observe that you are just connecting R1 and the 1k resistor in parallel. So in a sense , your hack and the original circuit are equivalent. Both are voltage regulators. The difference is that in the original, the leds are driven by the regulated voltage set by the potentiometer and the 1k resistor. Whereas in your hack the LEDs are driven by the regulated voltage (now set by the potentiometer and the equivalent resistor set by 1k and R1 resistors) minus the 1.25 V reference.

For a current regulator the correct way is suggested in this post:

https://sites.google.com/site/lcd4hobby/7-backlight/b-step-up-converter-as-led-driver

regards.

Hi ogaetej,

I disagree, the hack makes the circuit really a CC source, the value of the 10k pot and 1k resistor don't count anymore when there is a string of leds and a 2.5 Ohm resistor parallel. The negative side of the leds are connected to the FB pin and the current sense resistor between the FB pin and ground. When the current times 2.5 ohm reached 1.25V the XL6009 will regulate the current. I know it works because the pot don't work anymore, sometimes I even remove it so I can't by accident turn the pot to zero (it happens once and I blow the pot, it didn't like high amps)

On the Picture a 5A CC/CV buck, the second one a modified buck using the LM2596 and the bottom one a modified XL6009 boost, all with added heat sink.

Hi scorpio,

thanks for your fast reply. I still think your are not really regulating current through the LEDs. Look at the poti, it's is getting the same voltage as the LEDs, and there will be current flowing through it and it will dissipate power unnecessarily. But most importantely, the current flowing through R1 is the sum between a the LED current and the poti current (assuming infinite input impedance of the feedback pin of the xl6009). The sum of both currents would be set by the parallel connection of the 2.5ohm and the 1k resistor (which is pretty much the same as having the 2.5ohm resistor only). So you are indeed setting the current through the R1 but you have no control over the LED current.

It would be interesting to measure the current through the LEDs. I bet you it is not what you intended to set. But hey, it is working for you. Thanks for the chat, it has really helped me to understand how this little circuits work.

cheers

Hi ogaetej

It is a CC but not adjustable with the pot, you set the current with the value of R1, for example 1.25 Ohm resistor gives 1A it is best to set the 50K pot at max or remove it.

Regards