sfsuphysics
Active member
Hmm, definitely required, I don't know about the "probably not much" part. I mean depends upon how hot you are willing to let them go. Hopefully your thermal solution has a cut off if temps get too high? (i.e. fan failure)
LEDBrick Project - DIY pendant w/ pucks
- Thread starter theatrus
- Start date
Very cool build theatrus - I'm subscribed.
Have you looked into aluminum core PCB's? That seems to be the way to go for heat dissipation.
Curious how you chose your LED's - specifically the deep blues, hyper reds and ambers. What was preferable about the deep blue and hyper red?
Why ambers over broader warm whites or neutral whites?
Have you looked into aluminum core PCB's? That seems to be the way to go for heat dissipation.
Curious how you chose your LED's - specifically the deep blues, hyper reds and ambers. What was preferable about the deep blue and hyper red?
Why ambers over broader warm whites or neutral whites?
theatrus
100-mile-commuter
I'm figuring the realistic loading is more about 30W. Still required of course.
The board has an on-substrate thermal sensor (assuming it conducts mostly to the ground plane and not to the air, to be determined still).
Last edited:
theatrus
100-mile-commuter
This is a single-layer Al core PCB. Its not any fancy process (the "china special"). If I attempted this on FR4, it would be thin and chock full of vias.
Different spectral points, deep/royal blue (450nm) vs "blue" (~480nm). Ambers so I don't end up with even more blue in the mix. Reds to end up with something at the top of the spectrum which most whites won't hit.
Since I've done goofed up the thermal mounting for the UVs, I'll probably have a handful of boards left over, if you don't mind a non-ideal UV channel.
Its a little off topic, but I'm curious to know how you figured out the efficiency of the emitters.
....its on the data sheets. 30% is typical for latest gen LEDs at rated currents. Drop the current down and 40% plus is attainable form at least Cree and Luxeon......
Either I'm looking at the wrong data sheets or I have no clue how to read them, always a possibility. Royal blue and red show radiant flux which is what I thought you needed to calculate efficiency. Everything else is shown as luminous flux. from which you can't calculate efficiency. What am I missing?
theatrus
100-mile-commuter
White LEDs are also effectively royal blue LEDs, but with the phosphor added. Of course the conversion is not going to be 100% (from what I've been reading, about a 10% Stokes shift loss), so a white is somewhat less efficient in radiated energy vs a similar generation royal blue.
(Note my 30/40% numbers are ballparked, its worth noting that green LEDs and cyan LEDs are less efficient than most others at this point in time)
(Note my 30/40% numbers are ballparked, its worth noting that green LEDs and cyan LEDs are less efficient than most others at this point in time)
Last edited:
That's the number I was looking for. Just figured out what stokes shift was a few days ago, but did not see anything on what sort of energy loss there was.
Do you not also need to figure in the loss for internal reflectance of the phosphor? That's another ~20% if my (failing) memory serves me.
theatrus
100-mile-commuter
You would, I'm not sure of how much is going to bounce back (and how much of that is pure loss). I do wish this was a more direct data sheet spec, instead of lumens/W
I found a study by Phillips the other day comparing remote phosphors to to direct coating phosphors. That's where I got the 20% (I think) number from. Unfortunately I did not save a link and my memory for these things is not very reliable. 
Yes, it would be nice if there were a more direct data sheet, but then I would sleep more and read less and where is the fun in that.
Edit: Sigh. It was a Cree paper, and the 20% was the lumen output gain from the remote phosphor. They mention internal reflection loss, but do not put any numbers to it.
Yes, it would be nice if there were a more direct data sheet, but then I would sleep more and read less and where is the fun in that.
Edit: Sigh. It was a Cree paper, and the 20% was the lumen output gain from the remote phosphor. They mention internal reflection loss, but do not put any numbers to it.
Last edited:
The simpler way to look at it is how many watts of heat is being generated vs how many watts are being fed to the LED (still not all of the equation though but between stokes shift and that is enough to make my head hurt when trying to account for the rest)........most of what isn't lost to heat is emitted as the useful light we are after, not all, but most of what's not accounted for as heat.....
I found a study by Phillips the other day comparing remote phosphors to to direct coating phosphors. That's where I got the 20% (I think) number from. Unfortunately I did not save a link and my memory for these things is not very reliable.
Yes, it would be nice if there were a more direct data sheet, but then I would sleep more and read less and where is the fun in that.
Edit: Sigh. It was a Cree paper, and the 20% was the lumen output gain from the remote phosphor. They mention internal reflection loss, but do not put any numbers to it.
Remote phosphor is pretty cool and gains in efficiency over comparable white LEDs with phosphor on board if the design of the reflector array allows the blue light to be reflected back to the phosphor over and over again instead of being absorbed and turned directly into heat........
Very cool technology that I hope will soon hit the reef market since it's far cheaper to make a remote phosphor coated lens tuned to emit reef specific light than it would be to make an LED with the phosphor on board which requires much more specialized manufacturing......
oh, well, enough side tracking.
:beer:
theatrus
100-mile-commuter
I have a list of various PCB outfits which I've used for various reasons (with varying levels of English skills), but for these I ended up being lazy and simply using http://www.goldphoenixpcb.com
Agreed. Thanks for the answers folks.
theatrus. I'm looking forward to the finished product. Its fun when someone does something different.
theatrus
100-mile-commuter
I'm relatively terrible with precision in the machine tool arts, but I did find a few spare cycles to drill and tap the heatsink:
The four corners will be replaced with nuts and threaded rods, but for testing the LED board energized screws will do nicely. All screws are M4x0.7 (yes, I went metric).
Now it has the "brick" form factor :celeb3:
The four corners will be replaced with nuts and threaded rods, but for testing the LED board energized screws will do nicely. All screws are M4x0.7 (yes, I went metric).
Now it has the "brick" form factor :celeb3:
theatrus
100-mile-commuter
Alright, found some time this evening and did the first emitter brick build! :bounce2:
Step 1: Stencil the PCB:
Step 2: Place parts:
Step 3: Since this board is a great heat conductor, I opted for just tossing it in my Reflow Skilletâ„¢ (Hamilton beach large skillet, c/o Amazon. Flat bottom!). The astute reader will notice my cyan channel is outright missing - which is correct, I forgot to order that. Welp. Maybe I'll carefully tack in something else on the Rebel footprint later (Warm whites?)
Success!
(not pictured, a bunch of probing, and realizing the top-right UV LED (of the very spendy 405-410nm bin, is shorted at the pad. I futzed with it some, but decided to press on anyway - its likely a PCB short I didn't see and cut away)
I started building the cable harness (got better as time went on) and tested each channel at 10mA from the bench supply:
Left to right, my wiring was improving. This is a nice silicone jacket 24AWG stranded.
Cool White
Step 1: Stencil the PCB:
Step 2: Place parts:
Step 3: Since this board is a great heat conductor, I opted for just tossing it in my Reflow Skilletâ„¢ (Hamilton beach large skillet, c/o Amazon. Flat bottom!). The astute reader will notice my cyan channel is outright missing - which is correct, I forgot to order that. Welp. Maybe I'll carefully tack in something else on the Rebel footprint later (Warm whites?)
Success!
(not pictured, a bunch of probing, and realizing the top-right UV LED (of the very spendy 405-410nm bin, is shorted at the pad. I futzed with it some, but decided to press on anyway - its likely a PCB short I didn't see and cut away)
I started building the cable harness (got better as time went on) and tested each channel at 10mA from the bench supply:
Left to right, my wiring was improving. This is a nice silicone jacket 24AWG stranded.
Cool White
ZoaAddicted
New member
That's a lot of green. Are you going to put optics?
