DIY LED 5spectre

[Mastak]

Hi,
I'm workin on this DIY of my own LED grow light. Equipment which I prepared is:
DC 12V 20A 240W Switch Switching Power Supply Driver For LED Strip 110V/220V AC
20pcs of: 3W 620-630nm(VF: 2,2-2,8V, IF: 1100mA)
20pcs of: 3W 650-660nm(VF: 2,2, IF: 750mA)(reverse voltage 5,0V)
6 pcs of: 3W 6500-7000K white (VF: 3,4-3,6V, IF: 700mA)
15pcs of: 3W 470nm(VF: 3,2-3,6V, IF: 650mA)
5pcs of: 1W 440nm(VF:3,2-3,8V, IF:350mA)
2 pcs of Al heatsinks (237*325*30mm, 3,2kg)

Air cooling system:
Basically I'm gonna use my Vents TT 125 Fan (280m3/h), with reduced voltage. Each heatsink gonna be conected through pipeline, to fan. I think you can see it clearly on pictures. I realy hope it's gonna have enought power to cool all diodes. On pictures you can see only one heatsink, in box its gonna be second one next to him.

I've got problem with electrotechnical stuff, with LED array.
On 1. heatsing it'll be:
10+10 red one, 3xWhite, 8x470nm blue + 2x 440nm blue.
2. heatsing:
10+10 red one, 3xWhite, 7x470nm blue + 3x 440nm blue.

I'm using this calc: http://led.linear1.org/led.wiz
and it told me:
the wizard thinks the power dissipated in your resistors is a concern
And also it says that 10pcs of 630nm LED's total power dissipated by the array is 41052 mW. and thats weird because 10*3W led's = 41W ?

Can any one help me with this? I've need to design these LED arrays... And I'm totaly lost. I mean TOTALY


Thx
M.

 

[rives]

41052mw = 41.052w. Milliwatts are .001 watts, so if you multiply .001 x 41052 = 41.052.

The resistors are dissipating 2.662 watts and 8.228 watts, both of which are very high wattage for resistors. That circuit would worry me because of the almost inevitable imbalance between the parallel branches. If you are going to run parallel feeds off of one source, they need to be as closely matched as possible, and individually fused with fast-blow fuses.

 

[trem0lo]

I like your heatsink design, looks good. I think Knna's rule about heatsink size was 9 square inches for every watt, so be careful about overloading them. What type of LEDs are you using?

I think maybe higher voltage drivers that output a constant current (like 24V@500mA) would be easier. That way you won't need resistors in your setup, as the LEDs draw what they need and the power supply handles the rest.

 

[Mastak]

Rives: Thx for response, we'll I'm not happy about that either, but I've already got power supply and LED's as well. What can happen if I don't come up with a scheme which will be stable ? I could use more powerful resistors (10W,20W resistors), but I'm not much happy with the fact, that those resistors will consume that much.

trem0lo: well both heatsinks together have 170 square inches... and I'm planning to use aprox 160-180W but it will be cooled by fan, it's impossible to cool that with passive cooling system... and I'm using LED's from china, ebay...

 

[rives]

Rives: Thx for response, we'll I'm not happy about that either, but I've already got power supply and LED's as well. What can happen if I don't come up with a scheme which will be stable ? I could use more powerful resistors (10W,20W resistors), but I'm not much happy with the fact, that those resistors will consume that much.

trem0lo: well both heatsinks together have 170 square inches... and I'm planning to use aprox 160-180W but it will be cooled by fan, it's impossible to cool that with passive cooling system... and I'm using LED's from china, ebay...

How about using 9 LEDs in a 3x3 configuration, or ordering some additional LEDs? Most resistors are going to change resistance as they heat, and pretty (seemingly) insignificant changes can cause substantial changes in the power being drawn through each leg. This could result in amplifying the problem, until you get a runaway and let the smoke out of the components

 

[Mastak]

So basicaly I have to lower as much as possible power dissipating of resistors ? I'm thinking I'm gonna use only 8 of them (2*4led leg)


Edit:
So I've made some calc and this is best I can get... http://imageshack.us/f/19/slepenice.jpg/

 

[rives]

So basicaly I have to lower as much as possible power dissipating of resistors ? I'm thinking I'm gonna use only 8 of them (2*4led leg)

Getting the power down on the resistors would be good, but the critical thing is to get each parallel leg of the circuit as well balanced as possible so that there aren't discrepancies in the voltage drop, and thus the power draw.

 

[stratmandu]

Fast blow fuses aren't fast enough to protect a diode, no.

 

[rives]

Fast blow fuses aren't fast enough to protect a diode, no.

In general, it is a race that the LED would frequently lose. However, if the power supply and the fusing are carefully selected, they can definitely be made to work, and the successful protection is also impacted by the timing of the failure. Losing a string in a parallel circuit after the fuse is up to operating temperature would have a better chance of a fuse protecting the LEDs than if the failure happened on power up. Additionally, there are specialty fuses that are fast enough to protect electronic components that are restricted to working on 1/2 of a 60-cycle sine wave (thyristors). They are pretty exotic and very expensive, though.

 

[stratmandu]

Sigh. When you say "use a fast blow fuse", these guys are just buying whatever is at Radio Scrap or Frys, which are WORTHLESS for protecting LEDs from overcurrent.

You guys build what you want. If you don't want or need my advice, fine, I'll just shut up, sit back and watch. Good luck. No more comments from me.

What do I know? I have over 20 years experience in electrical engineering and R&D. I make a 6 figure salary with a fortune 500 electronics manufacturer whom I've been with over 15 years. My stuff is on-the-air in over 30 countries. I have a technical Emmy. I don't know shit, so carry on. LOL

 

[rives]

Feeling a little sensitive, Strat? I wasn't going to try and go into the subtleties of circuit design with someone who was struggling with converting between amps and milliamps. Any appropriately-sized fuse in the circuit is better than none, and your blanket statement that "Fast blow fuses aren't fast enough to protect a diode, no." is bullshit.

I can guarantee you that I can design a circuit with parallel strings of leds that will be protected by even a Radio Shack fast blow fuse as long as only one string at a time fails open. Balancing the current limit of the supply, the operating current of the LEDs, and the fuse size against the peak current capabilities of the LED and it's surge current are required. Personally, I don't like parallel strings - that's why I avoided them in my design. The fuse holder is a handy spot to check the current levels, however.

 

[alkalien]

The cheapest fuses will work for this scenario. It is not about a LED getting 10times the current they are rated for but more like 1,5 to 2 times (runnin them at 500mA normally, they are rated for 700mA, fuses 800mA. If one leg collapses 1/R=> 0, the LEDs get 1A for a few seconds and then the fuse blows.). Those are currents they can take for quite some time. Enough time for the cheapest fast blow fuses you can get.

Tried it a few times, no problems. After all, those LEDs aren't that expensive after all, plus, in my experience only one or two blow, the rest will be fine even if you don't use fuses. (Sadly, it's allmost everytime the blue ones that die)

 

[Mastak]

stratmandu: totaly agree with you, and I'm aware the fact, that when I don't use fuse fast enought I lost everything behind it... I try to get something as fast as possible, but as rives said, something is still better than nothing...

Rives: Ok, I think I finally 've got picture of that problem. So what do you think about having second power supply which will supply all blues and white, and they all will run at 3,5V and on second supply will be both red ones, and they run at same voltage(2,2V). Also in strings, I would like to avoid strings as you say but I dont think it's possible in my case...

Really appreciate your responses thumbs up

 

[rives]

I think that two supplies would make your task considerably easier, Mastak. If you haven't purchased the second driver yet, try finding one that has an adjustable current limit so that you can avoid the current limiting resistor in the circuit.