Ceiling / LED Awesomeness
From Interlock Rochester Wiki
(Difference between revisions)
Dhraakellian (Talk | contribs) m (→LEDs: picking at the nit a little more) |
(→Status) |
||
| (6 intermediate revisions not shown) | |||
| Line 2: | Line 2: | ||
* Fluorescent lights are not l33t enough | * Fluorescent lights are not l33t enough | ||
* Mood / situation lighting is important, especially when you're here late at night | * Mood / situation lighting is important, especially when you're here late at night | ||
| - | |||
| - | |||
== Proposed Solution == | == Proposed Solution == | ||
| Line 10: | Line 8: | ||
=== Diffusers === | === Diffusers === | ||
| - | * | + | * We are going to staple white fabric between the beams. Once its up, we're not going to take it down. |
* Diffusers would be mounted between the beams. The bottom face of the beams would be exposed. | * Diffusers would be mounted between the beams. The bottom face of the beams would be exposed. | ||
* Diffusers would be mounted above the piping system, so as to preserve the paint job, and allow access to the space by the water from the sprinkler heads, should the need arise. | * Diffusers would be mounted above the piping system, so as to preserve the paint job, and allow access to the space by the water from the sprinkler heads, should the need arise. | ||
| Line 16: | Line 14: | ||
=== LEDs === | === LEDs === | ||
| - | * 6 | + | Each LED will ultimately be *individually addressable* |
| + | * 6 of 4 LEDs each, per section of ceiling (as defined by the intersection of beams) | ||
* Each cluster would consist of 4 LEDs, one each of red, green, blue, and white | * Each cluster would consist of 4 LEDs, one each of red, green, blue, and white | ||
| - | * Each cluster would be fed by a Category 5 network cable ( | + | * Each cluster would consist of 4 LEDs plugged into a 2x4 IDC (no soldering!) |
| + | * Each cluster would be fed by a Category 5 network cable (cheaper) or ribbon cable (nicer) | ||
| + | * Each section of ceiling would terminate to a 25-pin IDC connector. Two of these would plug into a 50-conductor ribbon cable, bundles of which would run down the North/South center beam in the ceiling. 50-conductor cables would plug into surface-mount or edge connectors on the control system | ||
=== Control System === | === Control System === | ||
| - | + | We plan to create a control circuit, and fab out the manufacture of a PCB. PCB would need to have 10 50-conductor IDC pin blocks on it. | |
=== Costs === | === Costs === | ||
| - | * | + | * Found a source of 10mm bright LEDs. bestshop2008hk on eBay would provide 800 LEDs for $120. |
| - | + | * Ribbon cables and IDC connectors would run about $200 | |
| - | * | + | * PCB could run $50 or so. |
| - | + | ||
| - | * | + | |
| - | + | ||
| - | + | ||
=== Q & A === | === Q & A === | ||
* How bright is this thing? | * How bright is this thing? | ||
| - | ** | + | ** Each LED puts out 20-40K MCD. Based on some really fuzzy math to convert that to lumens, we'd roughly have the equivalent light of a 40 watt lightbulb, per color, per ceiling grid (each grid is about 40 sq feet). TL;DR... pretty bright. We'd probably want to supplement with additional incandescent lamps or directed workspace lights; this project is intended to be bright, but not bright enough to do fine work. |
| - | + | ** A (rough) detail of the light output, not modeling reflectance/absorption issues | |
| + | *** An incadescent bulb makes roughly 12 lm/W, so a 100 W lightbulb is about 1200 lm. A T12 34-watt fluorescent tube will do ~2500 lm. But these outputs are basically cast in all directions (viewing angle 2π). A lot of this light will be wasted and unusable for task lighting, but a good share will be reflected/diffused and be useful. | ||
| + | *** The bright whites are rated at a bottom of 30 cd, and a viewing angle lower bound of 25°. Converting the sphere cap of a cone of a vertex angle to solid angle (sr) is: 2π(1-cos(ϑ/2)). So the output per white LED is about 4.46lm. A red, green, and blue triad could put out about 60 cd, or about 9 lm. So a 4 LED cluster could put out about 9 lm. | ||
| + | *** Another issue is LED output degradation. Red LEDs last thousands of hours with under 10% decay, but blue LEDs decay faster, and white LEDs using a phosphor decay much more rapidly. For example, I have a datasheet of a cheaper orient part (China Young Sun YSL-R547, their website that used to be at http://www.100led.com/ doesn't seem to be responsive), and the white LEDs are rated to lose 25% (L75) at 1000 hrs, the decay becomes more gradual, but the 50% is around 3000hrs. This is pretty lower-end, but most T style LEDs (5mm/10mm round) are not designed for long term high output/high power operation. According to Narendran, heat damage to the phosphor and epoxy breakdown are the primary causes, so the decay can be widely variable with the part. Lowering average operating current can extend longevity. High power general lighting LEDs use a package with a heat-sinkable slug. SMD packages can do better, and high power LEDs in appropriate packaging have better thermal properties and larger junctions. | ||
| + | * Is the emphasis on providing dark room effects, ambient light toning, or general ambient lighting? | ||
| + | If it is the latter, it would be difficult and possibly impractical to use general purpose high-brightness LEDs. It takes a lot of 25° 30cd LEDs to produce light comparable to office lighting. For example, let's estimate decent general area lighting of 250lx with efficient recessed fixtures might take roughly about 350lm/m^2 (lm of actual light output). Standard office lighting is anywhere from 300-450lx, so 350lx (ideal) is certainly not extremely bright. This would be roughly 35-40 RGB triads for every square meter. The Illuminating Engineering Society of North America (IESNA) Lighting Handbook, provides a systematic approach to estimate light output requirements much more accurately. Low-cost surface-mount 1W LED would produce about 60 lumens and cost about $0.80 a piece. | ||
=== Purchasing === | === Purchasing === | ||
| - | If you'd like to contribute, please | + | If you'd like to contribute, please give money to Steve Smith. He'll be coordinating an eBay order or two. |
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
| - | + | ||
=== Assembly === | === Assembly === | ||
| - | Once the stuff has all arrived, we're gonna need to | + | Once the stuff has all arrived, we're gonna need to hook up a bunch of LEDs and figure out a control mechanism. I'll need help with the circuit design. |
| + | |||
| + | |||
| + | === Status === | ||
| + | |||
| + | This project has similar aspects to the use of [http://www.adafruit.com/products/322 these addressable RGB LED strands]. Several folks bought 50-count strands of these in late 2011 and have [http://interlockroc.org/2012/11/26/blinky-lights-test/ goofed around with them a little bit], though no mood lighting just yet. | ||
Latest revision as of 22:41, 9 December 2012
Contents |
Problem(s)
- Fluorescent lights are not l33t enough
- Mood / situation lighting is important, especially when you're here late at night
Proposed Solution
Based on some discussions with other members, and random strangers on the internet, I'd like to set up some kind of diffuse LED lighting.
Diffusers
- We are going to staple white fabric between the beams. Once its up, we're not going to take it down.
- Diffusers would be mounted between the beams. The bottom face of the beams would be exposed.
- Diffusers would be mounted above the piping system, so as to preserve the paint job, and allow access to the space by the water from the sprinkler heads, should the need arise.
LEDs
Each LED will ultimately be *individually addressable*
- 6 of 4 LEDs each, per section of ceiling (as defined by the intersection of beams)
- Each cluster would consist of 4 LEDs, one each of red, green, blue, and white
- Each cluster would consist of 4 LEDs plugged into a 2x4 IDC (no soldering!)
- Each cluster would be fed by a Category 5 network cable (cheaper) or ribbon cable (nicer)
- Each section of ceiling would terminate to a 25-pin IDC connector. Two of these would plug into a 50-conductor ribbon cable, bundles of which would run down the North/South center beam in the ceiling. 50-conductor cables would plug into surface-mount or edge connectors on the control system
Control System
We plan to create a control circuit, and fab out the manufacture of a PCB. PCB would need to have 10 50-conductor IDC pin blocks on it.
Costs
- Found a source of 10mm bright LEDs. bestshop2008hk on eBay would provide 800 LEDs for $120.
- Ribbon cables and IDC connectors would run about $200
- PCB could run $50 or so.
Q & A
- How bright is this thing?
- Each LED puts out 20-40K MCD. Based on some really fuzzy math to convert that to lumens, we'd roughly have the equivalent light of a 40 watt lightbulb, per color, per ceiling grid (each grid is about 40 sq feet). TL;DR... pretty bright. We'd probably want to supplement with additional incandescent lamps or directed workspace lights; this project is intended to be bright, but not bright enough to do fine work.
- A (rough) detail of the light output, not modeling reflectance/absorption issues
- An incadescent bulb makes roughly 12 lm/W, so a 100 W lightbulb is about 1200 lm. A T12 34-watt fluorescent tube will do ~2500 lm. But these outputs are basically cast in all directions (viewing angle 2π). A lot of this light will be wasted and unusable for task lighting, but a good share will be reflected/diffused and be useful.
- The bright whites are rated at a bottom of 30 cd, and a viewing angle lower bound of 25°. Converting the sphere cap of a cone of a vertex angle to solid angle (sr) is: 2π(1-cos(ϑ/2)). So the output per white LED is about 4.46lm. A red, green, and blue triad could put out about 60 cd, or about 9 lm. So a 4 LED cluster could put out about 9 lm.
- Another issue is LED output degradation. Red LEDs last thousands of hours with under 10% decay, but blue LEDs decay faster, and white LEDs using a phosphor decay much more rapidly. For example, I have a datasheet of a cheaper orient part (China Young Sun YSL-R547, their website that used to be at http://www.100led.com/ doesn't seem to be responsive), and the white LEDs are rated to lose 25% (L75) at 1000 hrs, the decay becomes more gradual, but the 50% is around 3000hrs. This is pretty lower-end, but most T style LEDs (5mm/10mm round) are not designed for long term high output/high power operation. According to Narendran, heat damage to the phosphor and epoxy breakdown are the primary causes, so the decay can be widely variable with the part. Lowering average operating current can extend longevity. High power general lighting LEDs use a package with a heat-sinkable slug. SMD packages can do better, and high power LEDs in appropriate packaging have better thermal properties and larger junctions.
- Is the emphasis on providing dark room effects, ambient light toning, or general ambient lighting?
If it is the latter, it would be difficult and possibly impractical to use general purpose high-brightness LEDs. It takes a lot of 25° 30cd LEDs to produce light comparable to office lighting. For example, let's estimate decent general area lighting of 250lx with efficient recessed fixtures might take roughly about 350lm/m^2 (lm of actual light output). Standard office lighting is anywhere from 300-450lx, so 350lx (ideal) is certainly not extremely bright. This would be roughly 35-40 RGB triads for every square meter. The Illuminating Engineering Society of North America (IESNA) Lighting Handbook, provides a systematic approach to estimate light output requirements much more accurately. Low-cost surface-mount 1W LED would produce about 60 lumens and cost about $0.80 a piece.
Purchasing
If you'd like to contribute, please give money to Steve Smith. He'll be coordinating an eBay order or two.
Assembly
Once the stuff has all arrived, we're gonna need to hook up a bunch of LEDs and figure out a control mechanism. I'll need help with the circuit design.
Status
This project has similar aspects to the use of these addressable RGB LED strands. Several folks bought 50-count strands of these in late 2011 and have goofed around with them a little bit, though no mood lighting just yet.
