Lime LED's are dubious

[NanoFox]

I've been reading about the recent addition of LIME led's fromplaces like Steve's such as here:

http://www.nano-reef.com/topic/346477-luxeon-rebel-es-lime-pics-and-par/

I'm in no way putting down the work done, but the "aesthetic" considerations without real spectra always leaving me wanting more. I'm a scientist and not an artist, so that might have something to do with it.

It seems that the usual mix is ~3 limes to 2x4000K +1x2700K... the color temp of the whites is a mix of 2 phosphors, so this mix is really just 3x ANSI 3500K. I wanted to see the spectrum of this "composite white".

I took the emission spectra from Luxeon for Royal Blue, Lime, NW (3500K) and CW (5000K) leds, and digitized them, then played with mixing in EXCEL. When I compare the standard 1:1 Lime+3500K versus the bog standard CW I get the following spectra. (I added RB to the composite since people also use more RB with these designs).

Notice that 1xNW + 1xLime+ 2xRB gives you almost exactly the same (normalized) output as a standard 5000K CW. I wont argue that having separate channels adds value - more control is always nice, it's just that all these "aesthetics" arguments seem to me to be mostly navel gazing.

In terms of precision control of CRI, the correct decision is probably to put Lime + 2700K on separate channels as "white", so you can tune the effective color temp of the white channel, then balance against RB/CB for color temp. But this is no real (spectral) improvement over just picking the color temp of the whites to suit your tastes to begin with and keeping it simple. It seems to me that most folks are just defaulting back to ~4700K for the white channel.

Using something like OCW to fill in the color gaps should still be advisable as this actually does add new wavelengths around 480 and 670nm which are mostly missing in these spectra. What we REALLY need as a 480nm LED to fill in the gap, rather than CB+Cyan, of which still doesn't quite get it right.

 

Just my $0.02

[ajmckay]

Is the blue "Mix" line the ~3 limes to 2x4000K +1x2700K? That's really interesting how spectrally it compares to a 5000K. I've been out of the loop for a while so I built my LED array based on the 2:1 RB:NW + OCW and I enjoy the color plenty.

 

I do think that keeping lime on a separate channel would be beneficial if you have enough channels in your controller to handle it.

 

Thanks for your thoughts, it's an interesting first post, so

 

BTW: I LOL'd at your thread title...

[NanoFox]

The basic idea here is really just that different people have different preferences for aesthetics. That's perfectly natural, and something like L/WW (or even CW/WW, which just as good since we add RB anyway) in different ratios let people tune-in their preferences. It's great to have that option, but there's so much room for subjective evaluations (and our eyes are TERRIBLE spectrometers) that you really need to add some data... ideally, an actual spectrum. Even something as silly and simple as this:

 

http://physicsbuzz.physicscentral.com/2013/02/turn-your-phone-into-spectrometer-for.html

 

If we had real spectra for our lights, the we could hone-in on the actual requirements to get good color from our tanks, and then design accordingly. But those ingredients need not be the same as the ones for color temp.

 

BTW, that was the original idea behind OCW. I've been at this for a while, just tend not to make much noise. I was the original guy pushing Milad at LED GroupBuy to try Cyan in the first place... This was me about 5 years ago:

 

http://www.reefcentral.com/forums/showthread.php?t=1888163

[Mr. Microscope]

to Nano-Reef and thank you for making your first post thought provoking.

 

Really looking forward to see what Ben and Evil have to say about this..

[GHill762]

to Nano-Reef and thank you for making your first post thought provoking.

 

Really looking forward to see what Ben and Evil have to say about this..

lol, looks like he's been here since aug '11, but creepin in the shadows, so not exactly "new"..

 

interesting thread, gonna follow along and try to keep up

[evilc66]

Certainly an interesting idea to put to a test. I may have to rig up something along the lines you are questioning and put it over the spectrometer (Ocean Optics USB4000-VIS) and see the real results.

 

And before anyone gets too wound up in this discussion, please note that the spectral plots in the datasheets are guidelines for expected output. Specific CCT and CRI will change that curve based on the phosphor recipe.

 

BUT, I wouldn't call lime's "dubious", as I have seen the real world change in using them. I've tried many different CCTs with different CRIs, and none of them have produced the same results as adding lime to the mix. The light is brighter, and crisper than just using whites alone.

[NanoFox]

I have to agree with you Evil... nothing better than real life data. We always have beware of biases like the "placebo effect" and such, and data always wins over impressions (That's nice spectrometer BTW). I'm certainly not trying to get anyone "wound up", just trying to open a question. In fact, what I mentioned above is what I'm planning on using on my own next experiments. I think the idea of of being able to tune the color temp of the whites precisely is exactly what is needed to improve CRI independently of color temp.

 

In the attachment is the CIE color chart...

 

What you can see is that by mixing a pure monochromatic 450nm (RB) with ~550nm (Lime) can get you very close indeed to the 10-20k color temp that most people are looking for. But real sunlight is very broad band indeed and contains wavelngths that are completely missing from the L+RB spectrum (mostly wavelngths between 460 and 500nm and again after ~650nm. High color temps are not "missing" these wavelengths but they are just overhwhelmed by the shorter ones. So when this goes missing, you notice it as poor CRI.

 

What's great about the Lime led is that it's very broad so gives much better CRI that most colored LEDs, but it only lets you move along the line bewteen the two. The addition of the WW which peaks at around 610nm helps you move off the 450-550nm line, but neither one adds any wavelengths that are missing from the other (The WW has quite a bit of green or it wouldn't come close to the right color and is actually "centered" around 580nm). Since they are both broadband and very near in color, they "mix" very well and avoid giving you annoying polychrome shadows. The 550nm range iis right near peak sensitivity for the eye, so will certainly look "bright" and they are very efficient!

 

The real missing pieces are bandwidth red and in the blue We make up for this with violet and CB/Cyan emitters in the blue. The dip from LEDs at aroun 480nm is something that's missing from sunlight, so nice good to have, but your eye is at its weekest here so not a show-stopper (no good emitters here anyway). But we don't have a good solution for the red. What I really want is an emitter at around 620nm- with a nice long tail like the lime has. (PC Amber doesn't quite cut it). Then I would do away with whites entirely. The deep-red are narrow band and a bit obnoxious but do help IMHO and are good for growing algae (and other things).

 

I think what all this really points to is that the modern "white" led is really a very impressive bit of engineering in a tiny and relatively inexpensive package and make it super-easy for DIYers to get nice results. We shouldn't shun them out of hand.

[jedimasterben]

The takeaway with lime is that it's not to help with CRI for our applications (especially as blue-heavy as they are), it's to help brighten up an array to the eye, there is no LED out there right now that can do it the same way. Eight lime can nearly overpower eight royal blue Luxeon M at the same current.

 

I'm in no way putting down the work done, but the "aesthetic" considerations without real spectra always leaving me wanting more. I'm a scientist and not an artist, so that might have something to do with it.

Not everyone has the money for or access to a spectrometer, but if you're offering I'd love one

[NanoFox]

Well, I've got very nice spectrometers at work, but they aren't exactl "portable". The one Evil mentioned though is extremely portable and just about perfect for this, but costs about 2k. Not the kind of thing I could afford but certainly a custom LED fixture maker should be able to afford one and write it off as a business expense.

[GHill762]

Benny you should get one and chalk it up as a business expense then charge for your consulting services. We all know you've had a hand in steering several companies in the development of their fixtures.

[NanoFox]

Actually... I just spotted these guys. Holy cow, Batman... that's a broadband LED

 

http://www.electrospell.com/GatewayPages/WhiteLEDs.html

[jedimasterben]

The problem is that I don't make money when I do that. Plus, I don't have a business, so I can't have business expenses, just an angry wife LOL

[GHill762]

[NanoFox]

I also think that Ben is quite right in a way. The Lime offers exceptionally high brightness in this spectral range for the power input. A CW led still puts out half of its power in the blue range... However, if you're doubling the number of RB to compensate, then it largely negates that... meh. In the end, of course it's always up to the owner of the tank to decide what looks best to their particular eyes with thier particular animals. Evil's point about binning and manufacturer variations is also very important here, and that's a big benefit to having many truly different channels to tune as so many manufacturers are now doing.

[NanoFox]

OK, on last post on this... Evil's comment binning got me to thinking, so I dug through the datasheets. The choice of the white LED is somewhat critical.

 

The Luxeon ANSI whites @ 3000K and below use a different phosphor (650nm) than the higher-K whites (610nm), and this gives them a significantly better CRI. The LXW9-PW30 is pretty much the king of these and has a CRI of 95! By comparison, LXW8-PW35 (3500K) offers 40 more lumens (up to 185 from 145 @ 700mA) but the CRI drops to only 85... that's a VERY noticeable difference.

 

So, if I re-do the previous graph with this new LED, I get a much better result. Now, the red extends well out past 650 and will give much better color rendition, but the addition of lime shifts the color balance back toward yellow rather than orange, making it centered like a "neutral" white while still achieving good red output - nothing like the old 1:1 RB-CW mix we started with.

 

This is not possible using only High-K whites and makes for a big advantage for using lime (but only when used in combination with the CORRECT low-K white). The second curve also adds Cyan (Cy) and cool-blue (CB). This looks remarkably like a high-temp blackbody spectrum and is laudable.

 

I would therefore revise by initial assessment from from dubious to something more like very promising, but only if designed correctly.

 

 

[evilc66]

That looks a lot more like what I would expect from a good combo with lime. Now, if only we could fix that 480nm hole

 

Actually... I just spotted these guys. Holy cow, Batman... that's a broadband LED

 

http://www.electrospell.com/GatewayPages/WhiteLEDs.html

I have some samples of those LEDs somewhere. Low efficiency though.

 

And I wasn't trying to imply that you were the one getting wound up over this topic. We have some very passionate members on this forum and things can escalate quickly when the wrong emphasis is placed on a topic. Just wanted everyone to come into this discussion with a level head

[DaveFason]

To add, I have been using limes for some time. I prefer adding a NW in there as well. However the NW is on the warmer side.

 

One other factor for your average DIY guy is getting the bin you want. We can pick apart datasheets for the "best" however getting the actual bin hat you want can be very hard. I have to buy reels in order to get the correct binning and still have troubles getting some.

 

Goof info!

 

-Dave

[markalot]

Goof info.

 

My thinking is Lime is to brighten up the display to our eyes in order to avoid the common issue with LED fixtures, frying everything because it looked dim to our eyes but was actually putting down 500 PAR.

[fredfish01]

The thread title needs to be updated to dubiously promising.

 

Thank you for doing this NanoFox, particularly the second set of plots. Its funny you should post that one as this is a combination I have been considering (but using a 27K warm white,LXW9-PW27 , from Steve's). The reasoning being pretty much the same as yours.

 

It looks like adding in the Cyan pulls the green peak back to 500 nm. That has me wondering whether that colour should really be there.

 

So, Evil. When can we expect to see some real plots? :D

[jedimasterben]

It looks like adding in the Cyan pulls the green peak back to 500 nm. That has me wondering whether that colour should really be there.

Lime doesn't replace cyan, as it does not cover that spectra well. There are pigments that still need that spectra to express.

[DaveFason]

The funny thing is I have been using...

• 2700-3400K WW and NW
• Lime
• Cyan
• Blue
• Royal
• Violets

For some time now. I have always thought that cyan and blue working very well in our arrays. While some people put FAR to many. Just adds a little "POP". The lime was the icing on the cake.

-D

 

-Dave

[fredfish01]

Dave. Now that you've added separate channels for control, your chips are Da Bomb!

 

I want to try something with more smaller clusters to get even better light distribution. I'm aiming for 7 LEDs per cluster.

 

It seems like 2 royal blue, 1 blue, 1 cyan, 1 lime, 1 warm white and 1 violet would be about as good as I can do.

[jedimasterben]

Dave. Now that you've added separate channels for control, your chips are Da Bomb!

 

I want to try something with more smaller clusters to get even better light distribution. I'm aiming for 7 LEDs per cluster.

 

It seems like 2 royal blue, 1 blue, 1 cyan, 1 lime, 1 warm white and 1 violet would be about as good as I can do.

I agree, that is realistically the smallest it can be broken down to and still pack in as much spectral coverage.

[evilc66]

So, Evil. When can we expect to see some real plots? :D

Soon-ish. I have a pair of Dave's new boards that I'm going to be setting up here soon, but I still need to get the driver/control board soldered up. It won't offer the most granular control over the light output (some colors are ganged together on a channel), but it at least will put most, if not all of the required LEDs that NanoFox was using in his example over the spectrometer.

[fredfish01]

Soon-ish. I have a pair of Dave's new boards that I'm going to be setting up here soon, but I still need to get the driver/control board soldered up. It won't offer the most granular control over the light output (some colors are ganged together on a channel), but it at least will put most, if not all of the required LEDs that NanoFox was using in his example over the spectrometer.

Very cool! I love stuff like this and you don't often see measurements that are not part of some company's marketing litterature.

 

Would it help any to have LEDs that are all on separate channels? I was planning to get four sets of 3 ups from Steve's soon. I could have them shipped to you to measure first if that is of any value.