"Home made" LED grow lights...

MajorCoco said:

Hey Rasser. That looks like some serious power. I did see that PWM comment on your thread, but it looked a bit advanced for me...I'm not to worried about dimming my LEDs at the moment. I always need more watts, not less. So I didn't look into it too closely..

How do PC power supplies limit the power they put out? I've been told that LEDs need to have current controlled drivers, and I'm not very good at electronics...the reasons for avoiding voltage controlled power seem pretty solid though. Do you know about that and whether all PC power supplies would work? You sound like someone who knows about electronics!!

Click to expand...

Lets say you would use 3 pc psu's for your 30V light's witch has a max draw of 3.7 amps,
if you connect them directly to 36V then the LED array will draw much more current than
the max 3.7A and properly burn off, so regulation is needed.

Traditionally you did that in big resistors or transistors(a electronic resistor) limiting the current
but waisting a lot of energy in heat. With the fast switching transistors and mosfets using PWM the loss in small.

PC power supply's don't limit current, but voltage and thats not good for LED's -a little fluctuations in voltage can results in the LED draws to much current,
and if one led in a string short circuit the rest will consume the power and maybe burn off.

So some current limitations is needed. and that is the function of the PWM switching mosfet
with some feedback circuit measuring the draw current.

In reality one could just fit a amp meter and adjust while looking at that, to stay within
the operational range between, no light in one end, and burned off in the other.

It would be nice the get so close as possible to the actual voltage of the leds
and that is the case with the modules in my Apollo 6 light, running with the build in inverters
I measured 35,7Volts if I remember correct.

In you case with 30.4 volt needs
I would say 4 PC PSU 12V+12V+3.3V+3.3V = 30.6 - it could properly run directly without anything else that way.
Or 3 PSU's running 12V+12V+5V = 29V -a bit dimmer than normal but would not need any imitating current.

bassclef said:

I would not recommend driving LEDs with PC power supplies. I looked into it before building mine and it was just too much of a hassle. You only have a few rails of 12V which would only drive 3-4 lights each. Also you have to install a voltage regulator as LEDs are current controlled devices. Too much current will fry them real fast, generally a PC motherboard does all that work. I think. Anyway it's not a good idea unless you really know your stuff.

If you're building LEDs I recommend Meanwell drivers. They have plenty (dimmable & non dimmable) all the way up to 48V at a variety of currents. You can also run strings in parallel to half the current. They are rock solid with overvoltage protection using zener diodes. And they're cheap.

Click to expand...

Cheap?!! My drivers are meanwell, and they cost more than the basically cutting-edge LEDs they're powering. It's crazy.

I did find some circuit diagrams for power LEDs using two high power transistors and a dozen resistors and capacitors as the main current controls (total cost was a few bucks), but they were only designed for 3W LEDs, and don't think transistor capable of controlling 85W exist!

MajorCoco said:

Cheap?!! My drivers are meanwell, and they cost more than the basically cutting-edge LEDs they're powering. It's crazy.

I did find some circuit diagrams for power LEDs using two high power transistors and a dozen resistors and capacitors as the main current controls (total cost was a few bucks), but they were only designed for 3W LEDs, and don't think transistor capable of controlling 85W exist!

Click to expand...

Have a link for that diagram ?

bassclef said:

I would not recommend driving LEDs with PC power supplies. I looked into it before building mine and it was just too much of a hassle. You only have a few rails of 12V which would only drive 3-4 lights each.

Click to expand...

When you say 3-4 lights you do mean 3-4 ps. of 85 watts light arrays i hope.

I was just looking at what the graphics card in my gaming PC is consuming, It's a GF470GTX and according to this:
http://www.behardware.com/articles/817-3/the-energy-consumption-of-93-graphics-cards.html
it's drawing 220 Watt while running futuremark3D, and it is suppled by +12V so I know a 500W PC psu can deliver some current
and if nothing is used on the +5v(28A) and +3.3(30A) drawing full current from all of the 12 volt rails should not be a problem.

Looking at the specification of a 25£ 650W PC PSU

Output Current
+3.3V - 28 A
+5V - 28 A
+12V1 - 18 A
+12V2 - 18 A
+12V3 - 18 A
+12V4 - 18 A
-12V - 0.8 A
+5VSB - 2.5 A

I'm reading it like this
The power supply has 4 12v rails and you can draw a maximum of 18 amps from one of the rails
but you can properly not draw 4*18A(72A*12V=864W) from the 12V system of a 650 watts PSU
I think the main wattage label is to be respected and properly 40% under that.

So from a PSU with the these data I would expect to draw 8-10 amps from each +12V rail givin it a total power of 12V*35-40A= ~450 watt
and by connecting 3 PSU in series and getting 36 volt and 35A 36*35 = 1260 Watt / 85 Watt = 14 LED arrays
( I'm sleepy and on my way to bed, so I might miss something but right now it makes sense)

The data on the mosfet from my power drill says
http://www.alldatasheet.com/datasheet-pdf/pdf/67829/INTERSIL/RFP50N05.html

"50A, 50V, 0.022 Ohm, N-Channel Power MOSFETs"

So in theory that alone should be sufficient to switch the PWM signal,
and I can imagine that a hand power drill is drawing a shit load of amps when running slow under heavy load.

Be careful with cheap PSUs, I've read an article quite some time ago that a chinese-made PSU advertised for 480w wasn't even able to properly output 200w.

You can find the article in French on that page http://www.x86-secret.com/cpchw.php, the category is "ATX 2009" and the pages I am referring to are those starting with "noname". If you're using Google Chrome you can translate the pages by right clicking anywhere and click on translate.

Anyway, computer PSU seem decent DC sources but I did not understand how you planned to regulate the current. Moreover, I've read that high power (1w+) LEDs don't "like" to be run in parallel because there are significant differences in driving voltage from one to another, especially if they are not exactly the same bin. That's why I think the 12v that PSUs can output seem a bit low for that use.

patrikantonius said:

Be careful with cheap PSUs, I've read an article quite some time ago that a chinese-made PSU advertised for 480w wasn't even able to properly output 200w.

You can find the article in French on that page http://www.x86-secret.com/cpchw.php, the category is "ATX 2009" and the pages I am referring to are those starting with "noname". If you're using Google Chrome you can translate the pages by right clicking anywhere and click on translate.[0]

Anyway, computer PSU seem decent DC sources but I did not understand how you planned to regulate the current[1]. Moreover, I've read that high power (1w+) LEDs don't "like" to be run in parallel because there are significant differences in driving voltage from one to another, especially if they are not exactly the same bin.[2] That's why I think the 12v that PSUs can output seem a bit low for that use.[3]

Click to expand...

[0] http://www.corsair.com/power-supply-units/cx-series-psu/builder-series-cx600-v2-80plus-certified-power-supply.html

Here are the numbers for a quality PSU locally priced at 50£ - It properly has more than one +12V rail but the spec is only concerned about the total draw on the 12 volt system


[1] My_LED_PSU_CONTROL.pdf
I've just tested my switching mosfet with a 60W 12V car water heater to sim a load
when running from 0-20% there is no heat on the heatsink, from 25%-75% there is a lot of heat generated(6W maybe), and again from 75% to 100% there is no heat generated
I imagine a bell curve of heat generation.

[2] I think thats true when we talk single LED in parallel my modules have 15 led in series and the inverter thats comes with unit runs two of those in parallel.

[3] Yes when using a single PSU, but connect 2-3 in series and you get 36 volts to play with

Edit: And you can select +12V +5V +3.3V to combine the psu's to the closest voltage of the led's strings/modules.
Just remember the weakest link in the chain is the +3.3v and +5v rails.

MajorCoco said:

Cheap?!! My drivers are meanwell, and they cost more than the basically cutting-edge LEDs they're powering. It's crazy.

Click to expand...

Haha, well relatively cheap. I use the LPC-35-700 and it's $16. The dimmables are more. Obviously you'd need a bunch of them for a big light. But I think it's worth it for the stability and protection. This isn't a very cheap hobby unfortunately.

Rasser said:

When you say 3-4 lights you do mean 3-4 ps. of 85 watts light arrays i hope.

Click to expand...

Maybe I'm missing something. So you have your 12V rail from the PSU and want to hook up a few LEDs at 500mA. At that current each LED draws about 2.2V for red Osrams (according to the datasheet). 12/2.2=5.45 so you'd get about 5 lights on each rail. Using blues/whites you'd get less, like 3 because they run at a higher voltage. Just seems like a big hassle, plus computer PSUs are large and not all of them (especially older models) have 4 12V rails. Plus, in series if one PSU fails, your whole light goes kaput. But it looks like you know more about this kind of stuff than me; I was under the impression that the voltage drop of the light string could not go over the max voltage of the driver, in this case 12V.

bassclef said:

Maybe I'm missing something.

Click to expand...

This one: LED_PSU_CONTROL.pdf my modules use ~36V and I want to be able to dim them,
and since this is MajorCoco home made light thread the talk was about driving his two 85W white LED's or something like it, at least in my mind
but I'm on drugs so I might have run with the assumption that this was the subject.
________


In this post I wrote: "And you can select +12V +5V +3.3V to combine the psu's to the closest voltage of the led's strings/modules"

3 PSU with 3 outputs gives something like 27 combinations so i did a search for a combination generator and made the following table over voltage possibilities:
Lol zero now ok that's a useful combination. I've forgot to remove it just as I did with 12 5 5 and 5 5 12. One 5 12 5 is sufficient.


A PWM regulation is properly still needed but it's best to start with a voltage that's just a bit higher or a match, than way over and waste energy as heat.
The 50V 50A mosfet can handle 140W in heat dispensation so one is not enough alone, but they cost 2-4£ for a 50V 150A so some of those in parallel
will do for my needs.

I did a search for Velleman 8055 and mosfet and I found the simple diagram over how to connect the mosfet correctly without modding the board.


Or what about using some of the boards out/inputs on a light mover
speed control with direction reversing.

bassclef said:

So you have your 12V rail from the PSU and want to hook up a few LEDs at 500mA. At that current each LED draws about 2.2V for red Osrams (according to the datasheet).

Click to expand...

It's another LED but the principle is the same.

Actually is the other way around, if you start from zero and increase the voltage then at 2.2V it should draw 500mA,
and if it starts to draw more then the LED driver turns down the voltage. That's why it's called a constant current driver
vs. a constant voltage supply like the PC PSU, they are just trying to deliver +12V no matter how much current you draw, until it saturate or blow a fuse.

It's observable and measurable when you use the current limiting function on a laboratory power supply.
If you set the voltage to 12V and is supplying a 12V fan or a 12V light, then the amps drawn is displayed
on the amp meter next to the voltage, when you start to limit the current there is a indicator that switch
on when the current limiting function is starting to kick in, and when that happens, the voltage starts to drop on the output terminals to the loads,
adjusting the voltage is the only option, you can't have a fixed 12V and then limiting the current to a load with a fixed resistance.

But don't feel silly I'm educated in electronics although in the late eighties, when I got my lap psu I was scratching my head over it
until I got the dohh and that's properly why I'm writing all this.

Plus, in series if one PSU fails, your whole light goes kaput.

Click to expand...

17 years running PC's and never have I got a busted MB, harddisk, graphics card as a result from a defective power supply
so I would not hesitate running LED directly on the rails from th PC PSU if the voltage suited within a safety margin.
But I'm still planning to use a current sensing feedback loop if I put a serious amount of led on the line.

Edit: And sorry MajorCoco for walking all over you thread.

The current limiting function has allays been in use when you mounted a 5mm LED to 12 volt via a resistor.

On this page there is a calculator for finding the ohm value to limit the voltage to operational levels -Great if you want to use a small led in a car.
http://led.linear1.org/1led.wiz

Using 500mA and 2.2V(1.1W) connecting that to 12V requirer a resistor in series with a value of 22 ohm and capable of absorbing 5.5W of heat.
Not very piratical firing 5.5W of in a resistor to drive a 1.1W LED - It works better with micro amps and 2.5mm indicator led's


The resistor is a passive current limiting device - if the LED suddenly starts to draw more amps then the voltage over the resistor increase and decrease over the LED and the current goes down.

Thank you !

looks awesome, wicked huge heatsinks lol

Rasser said:

But don't feel silly I'm educated in electronics although in the late eighties, when I got my lap psu I was scratching my head over it
until I got the dohh and that's properly why I'm writing all this.

17 years running PC's and never have I got a busted MB, harddisk, graphics card as a result from a defective power supply
so I would not hesitate running LED directly on the rails from th PC PSU if the voltage suited within a safety margin.
But I'm still planning to use a current sensing feedback loop if I put a serious amount of led on the line.

Edit: And sorry MajorCoco for walking all over you thread.

Click to expand...

I'll default to your knowledge then. My formal education is in music so you'll be wasting time arguing with me unless you want to talk about Wagner I've built a couple grow lights with excellent results but they're nothing fancy--just LEDs wired in series. I'm more concerned with the thermal design and spectra than anything else.

I did have a PSU fail once, by plugging in a bad floppy drive. So it can happen.

This is a DIY thread so I think any design discussion is pertinent!

Rasser said:

Have a link for that diagram ?

Click to expand...

No problem. Here is what I found...

http://www.instructables.com/id/Circuits-for-using-High-Power-LED-s/step6/The-new-stuff-Constant-Current-Source-1/

By the way folks..I have no problem at all with DIY chat on this thread, so don't worry about that!

One day we may have to work on a White LED/DIY sticky of some sort to gather all this info together into one place, but for now I think we need more chat and more people to experiment with this sort of thing before we can do that, so all these threads are helping get us there....

Just a brief update on the grow itself....
Even though I was vaguely thinking about harvesting a while ago now, development of the buds has been very slow recently. It is now day 66 of 12/12 (day 55 of actual flowering), well past what the seed company estimated (45-55 days), and I'm still not sure I want to harvest yet based on trichs and the overall fullness of the buds.

That being said, I did find one person on the net who had grown this strain before who stated clearly that he had found that 55-65 days was required for flowering. As usual, he doesn't say if this is 60 days of 12/12, or 60 days of the plant showing flowers, but whatever happens I'm going to cut down the smallest, most done plant either today or tomorrow and start it drying. The other 2 will stay in flower at least another week, maybe even 2 if I think it's needed...just to be sure.

^^^^Good to hear about your actual grow in here major ha.............yeah they need their own thread, but it is still good info for diy'ers

Edit: hey rasser, bass, etc........ask the mods to sticky a diy led thread.........Id be the first to vote for that(sure many others would too/ its great info)..........

I've just looked at that chart at thought, that's a dramatic rise in current on such a small voltage deviation, not good.
But how is that visible when connecting LED in series.

Parallel: using 3V DC supply
2.9V = 450mA
3.0V = 700mA
3.1V = 1000mA

Serial: using 12V DC supply powering a 4 LED's
11.9 = 2.975V = 650mA
12.0 = 3.000V = 700mA
12.1 = 3.025V = 750mA

Serial: using 24V DC supply powering a 8 LED's
23.9 = 2.9875 = 690mA
24.0 = 3.000V = 700mA
24.1 = 3,0125 = 710mA

So by the time we get to 15-30 pcs in serial the voltage variation that is acceptable is very large.

That seems pretty convincing, thank you very much for the explanations
Can you wire up multiple rails from one single power supply in series ?

patrikantonius said:

That seems pretty convincing, thank you very much for the explanations
Can you wire up multiple rails from one single power supply in series ?

Click to expand...

Your welcome.
No unfortunately not, since they have common ground.

And even thou there is 7V between +12v and +5v with witch you can run a 12V fan slowed down on,
you can't draw any serious current that way, the power supply protect itself and shuts down.
I tested it 30 seconds ago on a 12V-0.25W fan and a 12V-60W water heater.
You have to have true "dual channels" typical two windings on a transformer.
__________________________________________________________________________

While we are at it. In case you wonder how to run a PC psu on the table with no mb
You short circuit the green PS_ON wire with any black wire with a paper clip or wire
and keep it during running operation.


___________________________________________________________________________________



Thanks for the diagram MajorCoco, I did a fast calc on 3x3W LED's to see the waste in the resistor and it is:


There is some loss in the regulator all depending how close or far the voltage is to the needed by the led's.
I love the simplicity and it should be relative easy to scale up or mod.
_________________________________________________________________________________

I have found a LED series/parallel array wizard witch uses a resistor as current limiter: http://led.linear1.org/led.wiz
It's good when you have a voltage match, but terrible with large differences, you quickly end up with spending half the power in the resistor.

Ok...I got a little impatient this evening. These plants are definitely finishing slower than I'm used to seeing under an HPS (although I'm used to using a 400W HPS, which is much more than I've got here, so that is probably part of it) so I decided to take the main cola from the small one. It doesn't really look too bad, but the trichomes really aren't looking like she's ready. I'll have another close look at the cola tomorrow, and decide then whether to chop the rest of that plant or not. It's a tiny plant, and the cola only weighed in at 14g fresh....so probably only 3 or 4g dry....so it may be better to chop it now and give the other two plants the light.
I wasn't expecting a good yield off this plant anyway, but I hope it's good smoke!!

The other 2 plants will be going for at least 7 days more yet...I took a tiny bit of popcorn and will test this tomorrow after a 24hr dry..