DIY LEDs - The write-up

[wfournier]

Take a look at the curve! It is almost a straight line. Their efficiency is considerably higher than previous classes and from what I can read, its efficiency really starts at 750mA. Under this they appear to be less efficient. Check it out, maybe I'm misreading something.

I'm seeing about 140 lumens per watt at 350 ma and about 110 lumens per watt at 1.8A. Now lumens per $ it is certainly going to be a winner. Not sure if it'll beat Bridgelux on a simple lumens per $ comparison but it is certainly a little more efficient than the Bridgelux chips.

 

[Kolognekoral]

Optimal range is really what your corals like. Which brings up a question that will probably spark lots of discussion. I have read of numerous people that had to turn the LEDs down because it was too much light for the coral. Then they slowly increase the level over the next few months. So is this a good idea?

If the corals don't need the light (they were happy before) why waste the electricity? do they grow faster?
Will the make propagating to others harder? They may not have enough light for what your corals have become accustomed to.
Does it make acclimating new corals harder since they are used to less light?

Is this the right place for this discussion? I will start a new thread if deemed appropriate.

Thanks for your thoughts.

Interesting topic! May be best on another thread, but it raises many questions. With LEDs we are clearly raising the bar. but what is happening and are we making things more difficult in the future. I have one section in my tank where the light is too intense for most corals (right under the surface close to a 250W HQI). makes one review the situation.

 

[TheFishMan65]

I would have said the XM-L efficiency starts to drop off around 1 to 1.4 amps. I expect no part of the plot is linear so the best efficiency is a low current. Perhaps the best location might be where the next lower grade LED (XP-G) lumen/watt hits, but then at what current? Or does it ever?

 

[der_wille_zur_macht]

Optimal range is really what your corals like. Which brings up a question that will probably spark lots of discussion. I have read of numerous people that had to turn the LEDs down because it was too much light for the coral. Then they slowly increase the level over the next few months. So is this a good idea?

If the corals don't need the light (they were happy before) why waste the electricity? do they grow faster?
Will the make propagating to others harder? They may not have enough light for what your corals have become accustomed to.
Does it make acclimating new corals harder since they are used to less light?

Is this the right place for this discussion? I will start a new thread if deemed appropriate.

Thanks for your thoughts.

It's a good point but quickly becomes moot. We can "create" the same situation with any other form of lighting - put a 400w MH on a nano, for instance. The key is proper application - having LEDs with insanely high efficiency doesn't mean we can't easily use them, it just means we need to adjust our design processes to use them correctly.

I tried to copy the chart form the Cree XM-L pdf, but can't. The chart is on page 5, to which I reffer. It looks like the efficiency sharts to really drop at 2000mA. Otherwise pretty amazing. I think this brings the efficiency right up, as we must also consider all costs, not just power consumption.

http://www.cree.com/products/pdf/XLampXM-L.pdf

Under this they appear to be less efficient.

It's a simple concave function. It gets worse as you go higher. Yes, it's not AS convex as other LEDs, but there's still a consistent curvature. Again though this is somewhat of a moot point because as stated the curve is very flat.

Not sure if it'll beat Bridgelux on a simple lumens per $ comparison but it is certainly a little more efficient than the Bridgelux chips.

Clearly it depends on the length of time you own the LED. It's a pretty simple relationship - the longer you own it, the more efficiency matters and the less upfront cost matters. At some point, you've paid enough in electrical costs that a more-efficient LED becomes cheaper to own even though it cost more upfront. In the typical generation-to-generation jumps we've seen, where efficiency and upfront purchase cost both increase maybe 20 - 40%, by my math, it often takes less than a year for the "better" LED to pay for itself. Considering that Bridgelux chips are essentially one or two generations behind Cree's best right now, I think it's safe to say that for most people (i.e. people keeping their LED fixture for more than a few months) the Cree products are going to "win" the comparison.

 

[der_wille_zur_macht]

I would have said the XM-L efficiency starts to drop off around 1 to 1.4 amps. I expect no part of the plot is linear so the best efficiency is a low current.

You are correct, no part is linear, so purely in terms of efficiency (lumens/watt) a lower current is "better."

Perhaps the best location might be where the next lower grade LED (XP-G) lumen/watt hits, but then at what current? Or does it ever?

Now you bring up several very real questions, which are more complicated than simply lumens/watt:

1) Is there overlap between XP-G and XM-L in terms of lumens/watt? No. So, in terms of that metric, regardless of current, the XM-L is better.

More importantly,

2) Once you have selected a particular model of LED, how do you look at this graph and choose an operating point, i.e. the current you will run the LEDs at? Unfortunately, this is a very complicated question. Fortunately, even if you get it wrong, as long as you stay within reasonable bounds, you won't be that far off the mark (because the curve is fairly flat). More bad news though, there are many many more factors than meets the eye. You can't make the decision simply on the relative flux/current graph, because you're not operating the LED in an otherwise-consistent environment. As you increase current, you are almost assuredly increasing die temperature, so you need to factor in the relative flux/junction temperature graph as well (page before in the datasheet). Unfortunately we don't really have a way to accurately measure junction temperature, but the implication is very clear - efficiency drops off fast as junction temperature increases - in our case probably more so than the drop from increasing current. So, running at a high current is WORSE than you'd guess simply by looking at the flux/current graph.

Add to this the fact that lumens/watt isn't the only factor in the decision - you must consider upfront cost, unless you're keeping the fixture for a very, very long time. Also, you need to consider design factors - is it "better" for your design to have lots of light from a single point, or less light from more points? I don't think there's a single answer to that question for everyone. Plus, are there drivers that can run the LED at the current you want, in a string length that's convenient? And so on.

To get back to the "good news" I mentioned in the first paragraph, if your head spins, the simple message is, in terms of operating efficiency, i.e. money you spend to run the LED vs light you get from it, lower currents are always better than higher currents - regardless of which LED you're talking about, or what power rating the LED has. Most of us have to factor upfront cost into the decision, since we can't simply buy a ton of LEDs and run them at 100mA. Hence people typically reach a compromise point, where you're running the LEDs at some current lower than their max (to get better efficiency) but still high enough that you're getting good output per emitter. For older LEDs, i.e. XR-E, this meant somewhere around 70% of max most of the time, which worked out to the traditional 700mA target current.

As we get LEDs with higher efficiencies and higher max currents, we have to decide: Do we keep running them at 700mA, which nets much higher efficiency and slightly lower upfront costs, or do we continue to run at around 70% of max, which nets slightly higher efficiency but much lower upfront costs? People are going to have to decide this on their own, based on how long they intend to keep the fixture. Again, if you're going to keep it for longer than a year or two, it makes the most sense to weigh efficiency more heavily than upfront cost.

 

[TheFishMan65]

Thanks for checking on the overlap. After I typed it I started to wonder hence the extra sentence. I just didn't have time to check right now. IMHO more LEDs are better since there will less chance of spot lighting.

 

[der_wille_zur_macht]

I agree, for the average user. It's probably best to reduce counts slightly and continue running near the same current range we've always used.

The upfront cost increases with each new generation of LED have been small compared to the significant increase in performance. I don't think it's really sunk in with most people yet that in typical reef tank operating conditions, the XM-L can essentially produce twice as much light per watt as the XR-Es or Rebels we were all using a year or two ago.

 

[chris023]

Looks pretty good to me.

You might consider doubling the number of U-channels and running twice the LED to LED spacing. Otherwise pay attention to forced air cooling aspects. You might not need it but don't paint yourself into a corner where it would be hard to add.

Sammy if you visualize kcress' suggestion, the LEDs are effectively the same distance from each other, there's just more heatsinking surface area.

I hear ya on more surface area for better heat dissipation, thanks...

Any thoughts on height and optics? I was thinking something like 12" above water and 60 degree optics (would require new canopy for me). Or does 6" above water and 80 degree optics seem like it would work (within current canopy)? 24" deep tank with SPS going from just below the surface to ~2/3 of the way down.

 

[der_wille_zur_macht]

Depending on the exact optics you chose, those two options would probably work out to being about the same, so pick the one you like best (i.e. do you WANT a new canopy?)

Personally I'd lean a tiny bit towards the higher mounting height and slightly tighter optics, because a higher mounting height makes it easier to get in the tank and means less cleaning work on your splash shield (IMHO you'll need one at both heights you posted).

 

[MosMike]

Looking for some feedback on a design

LEDs will be mounted on u-channel.
Planned on 60 degree optics.

I would place the first U-channel closer to the front glass and tilt it towards the rear to light the corals from the front side. This way there will be less light loss on the front of the tank.

 

[chris023]

Depending on the exact optics you chose, those two options would probably work out to being about the same, so pick the one you like best (i.e. do you WANT a new canopy?)

Personally I'd lean a tiny bit towards the higher mounting height and slightly tighter optics, because a higher mounting height makes it easier to get in the tank and means less cleaning work on your splash shield (IMHO you'll need one at both heights you posted).

Thanks, my main concern is frying the corals at a closer range. My current canopy is accessible by lifting the top open via hinges on the back (e.g. like the hood of a car). So I figured I could just replace my T5s with a similar LED rig, if the close range seems doable. Definitely going with a shield in either case.

 

[chris023]

I would place the first U-channel closer to the front glass and tilt it towards the rear to light the corals from the front side. This way there will be less light loss on the front of the tank.

Yeah, that is one thing I would like to implement when mounting. The ability to raise/lower/tilt individual sections. I think I need to figure out how to make an easy, inexpensive, semi light weight, canopy...got any links to DIY canopy directions??? Shouldn't be that difficult to make one...

 

[The Grim Reefer]

Thanks, my main concern is frying the corals at a closer range. My current canopy is accessible by lifting the top open via hinges on the back (e.g. like the hood of a car). So I figured I could just replace my T5s with a similar LED rig, if the close range seems doable. Definitely going with a shield in either case.

I'd be worried about using optics on LED's that close to the corals.

 

[chris023]

I'd be worried about using optics on LED's that close to the corals.

You think no optics would penetrate deep enough (splash guard or course)?

BTW, those T5s in my pic are a combo you recommended to me a couple years ago from ReefGeek (4 ATI Blue Plus, 1 GE 6500K Daylight 1 UVL 75/25 and 1 UVL Actinic White... Changing out every 10 months)... Nothing but BAM! POW! color and growth since.... Almost a shame to risk it all on the LEDs... Thanks :beer:

 

[The Grim Reefer]

You think no optics would penetrate deep enough (splash guard or course)?

BTW, those T5s in my pic are a combo you recommended to me a couple years ago from ReefGeek (4 ATI Blue Plus, 1 GE 6500K Daylight 1 UVL 75/25 and 1 UVL Actinic White... Changing out every 10 months)... Nothing but BAM! POW! color and growth since.... Almost a shame to risk it all on the LEDs... Thanks :beer:

30 inches with no optics aint the answer either. I am opticless at 24 inches tall and get PAR in the 140's at the bottom with the LED's a few inches above the tank. My LED's are spaced 2 3/4 inches apart and the 6 rows of LED's are about 2 1/4 apart over a standard 120.

 

[der_wille_zur_macht]

I agree - no optics that close. You'll get good penetration if you keep your counts and current up.

 

[The Grim Reefer]

I agree - no optics that close. You'll get good penetration if you keep your counts and current up.

You really think he can get enough light to the bottom with no optics?

 

[TheFishMan65]

chris023, I posted this a few pages back that allow tilting. Maybe you can do something similiar.
Aiming LEDs

 

[chris023]

30 inches with no optics aint the answer either. I am opticless at 24 inches tall and get PAR in the 140's at the bottom with the LED's a few inches above the tank. My LED's are spaced 2 3/4 inches apart and the 6 rows of LED's are about 2 1/4 apart over a standard 120.

I agree - no optics that close. You'll get good penetration if you keep your counts and current up.

You really think he can get enough light to the bottom with no optics?

I admit that my knowledge of what constitutes good par numbers is lacking, so I'm a bit confused, is 140 on the bottom not good enough? Because Grim's first post sounds like what he is doing is good for him (24 inch deep, mounted a couple inches above water, no optics). Then Grim, your last post seems to be questioning if it would penetrate enough?

One thing I just read back a few pages from Tagpol when checking out Fishman's adjustable rig:
"I really think that my 60 degree optics, 14" height, and high intensity caused a lot of coral bleaching."

 

[chris023]

chris023, I posted this a few pages back that allow tilting. Maybe you can do something similiar.
Aiming LEDs

Dude... very very sweet...