My DIY LED Lighting project
- Thread starter pdelcast
- Start date
yousmellsfishy
New member
Is the brightness of the LEDs prone to fading? I have moonlights on my tanks that are about a year old and are only a mere fraction of the brightness they were when they were new. Here's a pic of what I'm talking about. These are the same LED's - new on the left and old on the right (as if you needed to be told
)
)
pdelcast
New member
Well, for this project it was VERY expensive. The LEDs are about $4-$5 each, the circuit boards for LEDs about $2000.00 total (or about $12.00 each) Parts and circuit boards for the ballasts were in the $1000.00 range. Machine tooling was around $1000.00
As a commercial product, you could expect to pay about $4 a watt. The lifetime of the LEDs is about 50,000 - 75,000 hours, so the cost is reasonable given the lifetime.
pdelcast
New member
<a href=showthread.php?s=&postid=12826653#post12826653 target=_blank>Originally posted</a> by yousmellsfishy
Is the brightness of the LEDs prone to fading? I have moonlights on my tanks that are about a year old and are only a mere fraction of the brightness they were when they were new. Here's a pic of what I'm talking about. These are the same LED's - new on the left and old on the right (as if you needed to be told
Usually LEDs do not dim significantly over time -- your LEDs are either getting too hot or they are being over-driven.
If they get over 120C junction temperature, they dim REALLY fast, and the dimming is permanent. So, to keep LEDs alive for long periods of time, most manufacturers recommend that you keep the LED as cool as possible.
Hi you guys may want to check out nano-reef.com This guy and a few other guys like myself have begun our own projects or research with high power leds at a bit cheaper prices. Especially taking into accountwavelength, lumen production, and construction the circuts and more.
http://www.nano-reef.com/forums/index.php?showtopic=134394&hl=
http://www.nano-reef.com/forums/index.php?showtopic=138608
http://www.nano-reef.com/forums/index.php?showtopic=134394&hl=
http://www.nano-reef.com/forums/index.php?showtopic=138608
pdelcast
New member
Are you getting LEDs at less than $1.00 a watt? That's about what I'm paying right now.
Of course I am using Cree LEDs, which are fairly expensive --- but they are also VERY high in lumen/watt. (what I'm using is about 90 lumen/watt.
You link is still broken -- please try again. I'm interested in seeing what others are doing.
easttn
Premium Member
Patrick,
Your LED bars look very nice and similar to something I built in a past life.
We supplied US-LED with 500K+ (?) of these things for comerical track lighting. Your customer is using the same basic design; same physical size, voltage and current requirements, but a different LED supplier.
We used LumiLed back in 2003 -5 because it was a customer requirement. At that time we were beta testing CREE; glad they made it to production volumes.
Your thread has me asking a couple of questions.
Your bar(s) have 6 "cool white" LEDs, and 6 450nm blue LEDs; therefore you only have 21W/bar, or 84W/array. The royal blues are not additive. The numbers you are giving are also very dependant on Cree's capability to deliver the LED brightness you are paying for.
The MCPBS (metal clad printed circuit board) are sufficient for heat dissipation in our application as we require substantial airflow in the canopy; the heatsink is not necessary for 3.5W LEDs. If your customer is requiring additional thermal management, I'd review the contractual agreements.
These LEDs will not be that bright as they are point source. They will not not kill your tank. If you have SPS, they will not do well. If you have LPS I have no idea as these were not my obsession.
If you are promoting this as a project then I think it is very important to explain the LED binning situation to the folks reading this thread. Cree (and all other manufacter) have an inability to predict the exact chromaticity (color), luminous flux (brightness), and Vf (voltage drop) of each LED prior to the production process. Therefore, they must test and bin every LED they make and provide a matrix of how to purchase their products. Here is Cree's binning. You may not see the difference, but over time the LEDs will self destruct if they do not have the same Vf in parallell paths.
If you are a manufacturer, send a PM. I can give you some "things to avoid" as you engage your contract manufacturing organization.
This is not a slam. We had quite a bit more capacity wrt chip-shooters on 12 SMT lines. I'm really trying to help, just ask the folks I sent LED arrays to in the past.
Your LED bars look very nice and similar to something I built in a past life.
We supplied US-LED with 500K+ (?) of these things for comerical track lighting. Your customer is using the same basic design; same physical size, voltage and current requirements, but a different LED supplier.
We used LumiLed back in 2003 -5 because it was a customer requirement. At that time we were beta testing CREE; glad they made it to production volumes.
Your thread has me asking a couple of questions.
Your bar(s) have 6 "cool white" LEDs, and 6 450nm blue LEDs; therefore you only have 21W/bar, or 84W/array. The royal blues are not additive. The numbers you are giving are also very dependant on Cree's capability to deliver the LED brightness you are paying for.
The MCPBS (metal clad printed circuit board) are sufficient for heat dissipation in our application as we require substantial airflow in the canopy; the heatsink is not necessary for 3.5W LEDs. If your customer is requiring additional thermal management, I'd review the contractual agreements.
These LEDs will not be that bright as they are point source. They will not not kill your tank. If you have SPS, they will not do well. If you have LPS I have no idea as these were not my obsession.
If you are promoting this as a project then I think it is very important to explain the LED binning situation to the folks reading this thread. Cree (and all other manufacter) have an inability to predict the exact chromaticity (color), luminous flux (brightness), and Vf (voltage drop) of each LED prior to the production process. Therefore, they must test and bin every LED they make and provide a matrix of how to purchase their products. Here is Cree's binning. You may not see the difference, but over time the LEDs will self destruct if they do not have the same Vf in parallell paths.
If you are a manufacturer, send a PM. I can give you some "things to avoid" as you engage your contract manufacturing organization.
This is not a slam. We had quite a bit more capacity wrt chip-shooters on 12 SMT lines. I'm really trying to help, just ask the folks I sent LED arrays to in the past.
BeanAnimal
Premium Member
You mean like the the one that lights up my soldering vice on my electronics bench or the handful of discrete packages that I started (but never finished) to put into my flashlight project?
pdelcast
New member
The brightness is a guaranteed minimum from Cree -- they are binned out (sorted) to meet criteria.
I guess I don't really understand why you think blue LEDs don't use any electricity? I'm not talking watts of brightness, I'm talking ELECTRICAL watts -- each LED uses 1 AMP forward current, and 3.65 volts of forward voltage. That's 3.65 watts per LED, and it doesn't matter what color they are...
Also, there are bars with 12 Whites, bars with 12 Blues, bars with 8 blues and 4 whites, bars with 8 whites and 4 blues, etc. They all draw about 42 watts of power.
Watts are not a measure of brightness, it is a measure of electrical power usage.
Lumens and PAR are brightness measurements. Blue LEDs don't add many lumens, but who cares about lumens? PAR is what matters (photosynthetic available radiation) and blue LEDs add more PAR than whites...
If you were doing this 5 years ago, it seems correct that your ideas of brightness are way off -- LED brightness has increased almost 20 fold in the last 5 years. If you tried using something like this for track lighting, you would blind anyone who looked at the lights.
(These LEDs are not considered eye safe -- you cannot stare directly into the beam from the LED -- it can cause eye damage.)Maybe you were thinking 3.5 Lumen output LEDs, and not 3.5 Watt LEDs? Then your post makes more sense. BTW, these LEDs output 336 Lumens each. (BUT they have a tight beam angle, so the brightness is VERY VERY high.)
Again, you have me thinking you were using 3.5 lumen, and not 3.5 watt LEDs. You can't dissipate that much waste heat on the aluminum PCB alone, and keep junction temps reasonable (at least not in the density we are running.) With the heat sink pictured (THE BIG ONE!) and no additional air flow, the LED junction temps stabilize at about 65C. We wanted to be sure that the LEDs would get sufficient cooling to keep the junctions below 70C if all fans and forced air cooling failed. Anything above 70C, and you compromise the lifetime of the LED.
(edit - I just did the math - - with convection cooling only, and no additional heat sink, the MC circuit board would temperature stabilize at 205C -- WAY too hot for the LEDs.)
Our main designs are for high power motor control -- and we are very good at thermal management. I've done both the simulations and the lab work -- that is really the minimum heatsink for LEDs at that density with no airflow.
Under a PAR meter, at 18"depth, four of these light bars are brighter in PAR than a 400W Metal Halide. This from an independent testing lab, got the PAR results yesterday afternoon.
This explains why we were bleaching SPS corals that were previously under T5 lighting. It wasn't a little too much light, it was WAY TOO MUCH LIGHT for an SPS that was acclimated to T5 lighting.
We are achieving greater than sunlight intensity on some of our setups. Did you look at the pictures of the lighted 28 gallon tank?
That's why we aren't paralleling any LEDs. Bad idea. Each LED string has it's own current regulated power supply. As far as binning, we buy LEDs binned for color and output. I'm not concerned about binning for Vf, because we aren't relying on Vf to control current -- we don't use any resistors for current control. Current control is fully closed-loop regulated for each string individually.
Oh, and don't let the binning confuse you. I've been trying to get Cree to get me LEDs from the outside bins, and they just don't have any -- The manufacturing process today is so well regulated that 95% of LEDs fall into two color bins, and three output power bins. There just isn't much fallout outside those few bins.
<a href=showthread.php?s=&postid=12837212#post12837212 target=_blank>Originally posted</a> by easttn
If you are a manufacturer, send a PM. I can give you some "things to avoid" as you engage your contract manufacturing organization.
This is not a slam. We had quite a bit more capacity wrt chip-shooters on 12 SMT lines. I'm really trying to help, just ask the folks I sent LED arrays to in the past.
I appreciate your concern --
But I've been doing this for twenty years, eight years with my own company. This project was my own pet project (I have a 300 gallon tank coming, and didn't want MH on it.)I have 1 SMT line with a quad chip shooters, with real-world 50,000 cph rate, and am quite proud that I've achieved that myself without any external financing. 50 full time employees, and about half are family.
Last edited:
pdelcast
New member
Spectrum of the LED bars from lab (All white LED bar -- using a spectroradiometer) The peak in yellow is typical for white LEDs..
Any experts want to give an opinion on the spectrum? (I'm interested in how the spectrum fits to use in a reef tank)
<a href="http://s37.photobucket.com/albums/e92/pdelcast/?action=view¤t=led_08.jpg" target="_blank"><img src="http://i37.photobucket.com/albums/e92/pdelcast/led_08.jpg" border="0" alt="Photobucket"></a>
Any experts want to give an opinion on the spectrum? (I'm interested in how the spectrum fits to use in a reef tank)
<a href="http://s37.photobucket.com/albums/e92/pdelcast/?action=view¤t=led_08.jpg" target="_blank"><img src="http://i37.photobucket.com/albums/e92/pdelcast/led_08.jpg" border="0" alt="Photobucket"></a>
Last edited:
flyingclay
New member
cool ideas. Im interested in leds when the price is more reasonable.I have made leds for my tank to get more viewing time but it is not bright enough to grow corals or light the whole tank. looking forward the seeing more worrk done on LEDS, thanks
Rob_Zamites
New member
Another DIYer chiming in....I'm getting ready to do a DIY retrofit on my NanoWave 9 (actually, it's so butchered up now, I call it the 'Frankenwave') - 5 x Cree Q5's and 8 x Cree XL Blues, gonna drive it with 3 Buck Pucks. I'm very excited by the DIYer's taking on this new forefront in reef lighting!
Great work on the LED bars, those are kickass!
Great work on the LED bars, those are kickass!
pdelcast
New member
OK: Here is the 300 gallon (new tank) lighting:
1150 watts -- 384 LEDs at about 3 watts each LED.
PAR measurements were done with an Apogee Instruments SQ-110 Quantum sensor connected to a Fluke 83 III digital multimeter.
Measurements were:
Above water surface (2.5" from LEDs): 3250 PAR
Just below water surface(4" from LEDs): 2200 PAR
5" below water surface (6.5" from LEDs): 1700 PAR
At top of rock shelf (about 17" deep) : 875 PAR
At surface of sand (about 28" deep) : 450 PAR
All PAR measurements are in umol*m-2*s-1
Edge on (beam pattern visible - 110 degree beam)
Full tank view: (You can't see in this picture how bright this is)
Room lighting totally washed out:
Shot of the Apogee quantum meter and my Fluke meter:
PAR reading at 5" below surface: (343mV - 1715 PAR)
1150 watts -- 384 LEDs at about 3 watts each LED.
PAR measurements were done with an Apogee Instruments SQ-110 Quantum sensor connected to a Fluke 83 III digital multimeter.
Measurements were:
Above water surface (2.5" from LEDs): 3250 PAR
Just below water surface(4" from LEDs): 2200 PAR
5" below water surface (6.5" from LEDs): 1700 PAR
At top of rock shelf (about 17" deep) : 875 PAR
At surface of sand (about 28" deep) : 450 PAR
All PAR measurements are in umol*m-2*s-1
Edge on (beam pattern visible - 110 degree beam)
Full tank view: (You can't see in this picture how bright this is)
Room lighting totally washed out:
Shot of the Apogee quantum meter and my Fluke meter:
PAR reading at 5" below surface: (343mV - 1715 PAR)
Similar threads
