LED Q&A thread

Discussion in 'DIY' started by Gomer, Sep 24, 2009.

  1. Gomer

    Gomer Honorary Member

    I got PM'd a fair bit about LEDs, so I thought I'd make things more efficient, and have info more accessible. I won't have all the answers, but between the many minds on BAR, we get things squared away :)

    If you have any questions, feel free to ask away here!

    I'd like to turn this eventually into some sort of FAQ, so direct questions like "What is the difference between a buckpuck and a boostpuck and why would I use one over the other?" would be appreciated vs questions like "How do I build an LED light for my tank"
  2. sfsuphysics

    sfsuphysics Supporting Member

    Where's the state of the art for household LED lighting at today and how does it compare to CFL bulbs? :)
  3. DirtyDrew

    DirtyDrew Guest

    how do i calculate led lights into my tank? i have a 50 gallon hex that is 22in from the water line to the sand. i want to run sps lps corals and more. any sugestions?
  4. heyitsomid

    heyitsomid Guest

    Why are the XR-E LEDs run at 1amp when cree states they should be run at 700ma, does this do any long term harm (shortening the lifespan) of the led or does adequate cooling asuage this risk? Has anyone run the XP-E on their tanks and if so how much cooling is necessary at 1amp for those LEDs (as I read they are more efficient...more light per watt than the XR-Es). Also never heard of a boostpuck. haha...more questions to come...
  5. sfsuphysics

    sfsuphysics Supporting Member

    Why do you need a current fixed device (buckpuck or similar) with them and not simply a voltage source? Is it a matter of wanting to overdrive them at a fixed voltage (are you even doing that?).

    How close can you pack these buggers?
  6. Gomer

    Gomer Honorary Member

    Keep em coming :) I'll get to these tonight!
  7. Gomer

    Gomer Honorary Member

    If we ignore lab prototypes....
    It's a 2 part game. 1) the lumens/watt efficiency at a given temperature (kelvin rating) and 2) how efficient is the light delivery system.

    Phillip's Energy Advantage t8 lamp at 4100k are about 95 lumens/watt (just looking at bulb power, not including ballasts/drivers etc as that is another issue). Cree's easily availaible warm whites of similar kelvin ratings are probably in the 85 lumens/watt range, and their higher kelving (6-7000k) brand new R4 XP-G's are probably 30+ % more efficient.

    But the other part is lighting "extraction" and getting the light where you want. Fluorescensts are very floody and light everything...ceilings, walls....everywhere. LEDs are by design direction and can be used with optics. For spot lighting, LEDs kill Fluorescents. For broad room lighting, it varyies. If you don't care to light the walls and ceiling, then LEDs should shine.

    Check this out:

    Also worth noting:
    Cree Repeats as Lighting for Tomorrow Grand Prize Winner
    There is no "formula" like watts per gallon. However, A very rough estimate is you need 50% the power equivelent to MH for similar PAR (based on what I have gleened from others) Of course, YYMV with reflectors on MH and optics on LEDs, LED chosen, and ratio of LEDs. Most suggest optics for >18" tanks, especially with SPS. For AIO tanks, you ca generally run optics free. I'd crudely guestimate that you'd want something like 30 or so high flux LEDs with optics at 700-1000mA

    The voltage-current curve isn't linear. A small change in voltage can have a large change on current. Also, the Vf of any given LED varies. One may have a Vf of 3.3V and another 3.6V for a given current. If you drive these two LEDs at 3.6V, you might toast the 3.3 Vf LED by dumping...oh..1.5A (just tossing out numbers) through it. Also, temperature effects current. At a fixed voltage, you can send different amounts of current through an LED at different junction temp. As temp goes up, the LED will draw more current...and heat up more..and pull more current etc. This is known as Thermal Runaway.

    If you KNOW the Vf and want to drive a single LED, you can usually safely underdrive it with a voltage regulated source.

    You have 2 limits. A thermal limit, and a bond wire current limit. I think that you can run upwards of 1.5 amps through the bond wires before they fry. The Junction (the die) is the other thermal limit. You CAN run at 1amp if you can effectively pull the heat away from the die. For all intensive purposes, this requires active cooling.

    There are many many DIY projects using Crees, and many people driving at 1A. You'll have to hunt around :) NanoReef has a large and long standing following on these projects. There are also a few good threads o Reef Central.

    A boosts puck raises the voltage so that Vout > Vin. (ie, you can drive 3 LEDs in series (>10V req) with 3V power supply. A buck puck does the opposite. Boost pucks are often more efficient but I usually see 350mA current limit ones on turn-key pucks.
  8. sfsuphysics

    sfsuphysics Supporting Member

    Groovy Tony, Yeah I've seen that PDF before with the lighting comparison, don't know about pricing but I've seen some of those bulbs going off at upwards of $80 for a fricking bulb... well screw that! Not like I'm lighting coral yanno! :D

    I ask because I saw a couple of LED flood lights at costco for around $10 or so, although they had what looked like the 5mm leds inside of them (maybe 50 or so?), but it was lacking on any information of output. They had a 3.5 and a 5 watt version of each. Thinking of doing some sort of recessed lighting in the living room, so thought maybe those would be useful, but who knows.. its costco I can get one then return it if I don't like it :D
  9. Gomer

    Gomer Honorary Member

    All I would suggest is:

    If you want to compare an $80 Cree light to a $10 LED light

    1) consider output
    2) consider color rendition
    3) consider lifetime
    4) consider efficiency
    5) consider the $/lumen over the life of the bulb.

    What you will likely find out is:
    The output is way less
    The color is way blue with a poor CRI (would you light your house with an 8K lamp with poor CRI?
    Although the 3mm/5mm LED has a long theoretical lifetime, it isn't engineered for such and sucks in the lifespan department.
    Efficiency is probably 50% of the cree
  10. Gomer

    Gomer Honorary Member

    I just looked up a possible comperable lamp

    Warm White: 3000K , 150 Lumens (compare to 15W)
    120 Volts - 4.5 Watt
    Best case 150lumen/4.5 watts = 33 lumens/watt.

    Cree LR6 3500K lamp with a CRI of 92 is
    650lumens/12watts = 54 lumens.watt.

    For equal light, you need 4.3x of the 5mm light. 4.3x$24 = $103
    For equal light, you need 64% more power.

    I don't have lifetime info on the 5mm light.
  11. sfsuphysics

    sfsuphysics Supporting Member

    Well its not about "equal light" it's about trying to justify a product to pay itself off, or at least come close, in a reasonable amount of time vs current CFL technology NOT incandescent technology (which I see a lot of the "Eco Bulbs" doing). Who knows maybe one of them would be a good fuge light :D
  12. Gomer

    Gomer Honorary Member

    It is about equal light :) I wouldn't recommend an LED night light to replace a 10x T5HO setup. I'd recommend something which puts out comperable light. If you don't equalize things, then what is the point? You can't compare 1 unknown variable absolutely with 3 independent unknowns in the system ;)

    If you want to light a room, you want to be X bright. You can do it with 4 crees or 18 of those other bulbs. If you had 4 of those other bulbs, the room would be too dim. You also need 4.3 x the sockets or ceiling holes to accomidate it too ;)

    Have fun playing with this: http://www.creeledlighting.com/calculator.aspx
    You will want to adjust for current prices and you can swap incan numbers to reflect fluor.
  13. sfsuphysics

    sfsuphysics Supporting Member

    Well you have to understand it isn't about Cree vs Equivalent amount of other LED bulbs it's Cree vs CFL bulbs. And honestly at around 900 Lumens at less than a buck a bulb, it would be very difficult to justify the cost of the Cree.

    Playing around with that thing, making the fixture cost go to zero.. I mean really now? Fixture cost? Do these Crees come as a fixture rather than bulb? With a $1 CFL, labor costs go to zero because my time isn't that precious :D Changing the average daily use from 18 hours to 5. i.e. make it more realistic. Then the payback time is 371 years 5 months and 4 days

    Which really tells me what I believed all along, those LED guys like to make up outrageous claims on numbers to justify the buy back period of their bulbs. I mean really $5.40 for an incandescent bulb? 18 hours a day usage on 10 fixtures? What fricking museum in the arctic during winter do those guys live in!?
  14. Gomer

    Gomer Honorary Member

    You have to understand that I DO understand ;) YOU were the one who tried debunking a Cree fixture with another LED option hence the prior post comparing LED to LED.

    Case 1: The business. Replacing a bulb for any sizable business turns into a facilities cost. How many mid-large size businesses have the normal workforce (vs facilities) change the bulbs? You don't want to even know what the facilities charge is over where I work.
    18hrs is realistic for many places. There are tons of businesses that run their lights at every hour there isn't sunlight plus a few overlap.

    Case2: The home. Replacing a bulb..no problem. Free. 18hrs..no way...BUT, you incur another expense. Lamp life. Typical lamps in a house are only left on for a short period of time compared to a business. I made a little plot for you using data from "Economics of Switching Fluorescent Lamps" IEEE Transactions on Industry Applications Vol 24, No 3, May/June 1988

    Attached files /attachments/sites/default/files/Graph2_0.jpg
  15. Gomer

    Gomer Honorary Member

    More numbers for you mike

    Suppose you, a homeowner, run your lights 6 hrs a day and are at the PG&E averate tiered rate of 0.17643/ kwh

    It The cree booth demo room uses 9 fixtures at 180watts PC/108 watts LED
    I just looked up a typical spiral compact bulb from Lowes.com ( part# 252475). It is 23watts, but It is probably safe to say that the life is similar. Posted life is 8000hrs. Given the results from the publication above, the 6hr burn lifespan is 4400hrs.

    The Cree LR6 has a lifespan (70% intensity) of 50,000hrs. At 6hrs a day, that is 23years....so lets just say 5 years of use for a comparison since you want "practical"...I'd go higher, but I'll make this easy on the debate.

    5yrs * 365days/yr*6hrs=10,950 hrs.
    Given the electricity rate above

    cree costs you $208 in electricity
    spiral compact costs you $348 in electricity

    Over 5 years, you will need
    1 cree fixture (and it will last you 5x longer than this number game)
    2.5 spiral compact fixtures

    So there you go. No attic/museum/etc math.
  16. tuberider

    tuberider Guest

    Here's a simple question. The people I've talked to in the industry still roll their eyes at LED technology and claim that it will be another decade before LEDs are the standard due to reliability (mostly) and lack of a track record, do you agree with the timeline?

    Remember I'm talking about the standard, not a novel idea that is trendy.
  17. Gomer

    Gomer Honorary Member

    Reliability is a big issue imo, but not with the "new" high power LEDs (osram, soul,phillips,cree etc) but is with the knockoffs and even more so, 5mm LEDs which are driven hard. The problem most of the time isn't the LED but rather the lack of engineering considerations. Voltage and heat kill LEDs. Light manufacturers that don't truely understand this are to blame. You can take a theoretically perfect LED and kill it by poor implementation. Street lights with dead LED strings are likely dead to voltage or heat reasons and unlikely due to QC of the chip itself.

    A prime example of this is PFO. There are TONS of threads about dead lights. I'm 90% confident it is because of poor engineering on heat extraction.

    But to answer your question (and not comment on the comment lol), I'd say maybe. If it takes a decade, it won't be because of the reliability and it won't be because of net cost. It will be because current technology is expensive up front, while old technology (fluour/incan) is expensive in the long term. In a few years, the LEDs will go down even more, but people still ignore long term costs in favor of short term pricing. As for technology: When I started dabblign in high power LEDs, they were 30lumens/watt. Inside of about 5 years(?) they are up to 130lumens/watt* at about 40% cheaper then the old inefficient ones (*available product, not lab chips).

    just my 2cents :)
  18. sfsuphysics

    sfsuphysics Supporting Member

    Well my issue with long term costs is there's also another flavor that's often ignored by pro-LED people, and that's long term adaptability. Sure I have to keep buying bulbs, pay a bit more for electricity etc, but if my lighting choice decides to change (color temp, CRI, whatever), then I can freely do that within a bulb change time frame and I'm not locked into one particular technology. Much like you said you see LEDs have advanced by leaps and bounds in 5 years, 300% increase in output and cheaper? Yikes! Now imagine if you were an early adopter (lets say PFO fixture) you really are stuck with what that came with if you want to argue the "long term savings" side of things.

    Plus I'm not quite sold on the lifespan of LEDs as great as they sound.
  19. Gomer

    Gomer Honorary Member

    Depending on your electricity rate, it could just be a few years to "break" even. Although wasteful, you could "upgrade" every few years and still be ahead. I know tons of people that use incandecents for canned lighting. My parents for example run a pair of 100watt cans in the hallway.

    300%..scarry huh! (in a good way!) I think we are about 1/3 the way to the theoretical 100% RGB efficiency (and probably something like 1/2 way to the 50% blue/phosophor).

    As for CRI and color etc, you're in for a treat :) Cree has a 4 color dye (white + blue + gree+red chips on a single emitter) each wired indepenently. You could technically dial in your CRI and color :p
    The LR6 cans from Cree actually mix LEDs to achieve their high CRI-Color. You may see current adjusters in the future to dial in user preference. Philips (lumileds) are bound to have some good stuff out in the future as well. Just a waiting game. I'd say that household/business LED users are about at the same point Prius was at year 2-3. It'll take a bit to move mainstream, but we are actually technologically in a good position.

    *I keep going back to Cree just because I am most familiar with them. There are possibly other options out there!
  20. sfsuphysics

    sfsuphysics Supporting Member

    No I dig it, when referencing the state of the art, or where the market is you talk about the ones that move the market not the ones who copy it. If you want skimmer talk you don't discuss what an Odyssea can do :)

    Problem with your Prius argument is that the Prius was actually in the same price range as other automobiles, if it cost ... I dunno $100k+ like the Tesla roadster, then you'd see a slower roll out, and in fact I think that might be a better comparison with LED lighting for the home. It's a hard sell to get someone to pay 90 bucks for a bulb, vs 1 for a bulb, only real selling points are longevity (which I doubt the company guarantees at all) and the mercury issue (which the average person doesn't care/is ignorant over and will regularly throw old CFL bulbs in the garbage). I think these may get out there faster if they had a better market penetration into stores like Home Depot, its a big leap of faith to buy something so much more expensive than a regular light bulb on faith alone with no demo that you can physically see in person.

    BTW, sorry if this is derailing the intent of this topic. Others who have questions feel free to chime in :D

Share This Page