Upgrade and replace burnt out LEDs in AI Lights

[jrp1588]

The test unit doesn't have any more bad LEDs, but I'm replacing all of them for funsies.

I think it uses:
XP-E RB
XP-E B
XP-E Green
XT-E Cool White
Osram Deep Blue
Osram Deep Red

I'm planning on replacing the RB and Deep blue with XP-G3, Red with XP-E2 Red, Green with XP-E2 Green, and the whites with some kind of XP-G3 to be determined.

 

[Lingwendil]

For the whites, do you want warmer or closer to stock? whatever you choose go for a 90CRI type, something like either a 4000K, 5000K, 5700K, or 6000K are readily available in good bins, and will have great color rendition compared to the stock whites. I would be inclined to go for a 5000K or 5700K on the whites, leaning towards the 5000K a bit more, particularly these- https://www.mouser.com/ProductDetail/Cree-Inc/XPGDWT-U1-0000-00FE3?qs=e4fOZ%2b9ijA8gHQ4deMVyLA==

The Royal Blues, use the ones I linked earlier.

The Osram deep blue and red might be a different mounting profile, if I'm not mistaken they are smaller? I'll have to look into that in a bit. Otherwise It sounds okay, although Osrams are generally very good. I would be inclined to leave them be if they are in good working order. The Osram deep Blue hits a different peak than the Royals do, so good for variety.

The greens are ~ meh. I don't really like them, but they can be helpful for "tuning" the color a bit. Since they don't really have any meaningful contribution otherwise, I would leave them if they work well as they are.

 

[jrp1588]

Ok, so I'm looking at these data sheets. Flux, if I understand it correctly, is equivalent to lumens, right? Higher lumens would be indicative of a higher bin? CRI determines how well colors are rendered. I'm seeing an inverse relationship between the two. Would I be trading usable light and efficiency for more vibrant colors?

Looking at the numbers, it seems like the 80 CRI diodes would be a good option. They only have a modest drop of around 10 in Flux, but have 10% better CRI. The 90 CRI diodes drop by 40 in comparison. Is it worth it?

 

[jrp1588]

Ok, using RapidLED as a jumping off point for part numbers, and looking at the data sheets, I think I've selected my diodes. Looks like Rapid sells pretty high-binned stuff, but I can definitely save money this way unless shipping is really high. I think I'll leave the Osram Deep Blues alone. Maybe the Reds as well if they use a different mount as you suggest.

XPGDWT-H1-0000-00KE3 Cool White
XPGDRY-L1-0000-00601 Royal BLue
XPEBBL-L1-0000-00301 Blue
XPEBGR-L1-0000-00G03 Green
XPEBRD-L1-0000-00901 Red

 

[jrp1588]

Still torn between 80 and 90 CRI. I'm doing some reading on it (on a pot growing forum funny enough), and it seems the more complete spectrum makes up for the lower light output. Plus it'll look nicer. I'll need to mull that one over a bit.

 

[jrp1588]

I hate to monologue, but I'm coming around to the 90 cri whites. My white channel currently only peaks at 55% anyway, and even the new 90CRI diodes would be more efficient than what's in there currently, and they would look better than the 70CRI average. A few 'Grower' forums I've come across really seem to tout the 90's even with the drop in lumens.

Sigh, now that I've dipped a toe into all this, I think I'll have to custom build my next light. haha

 

[Lingwendil]

Definitely go for the higher CRI whites, it's a pretty big difference in color rendition, and the spectrum is a little more "complete" so to say. The difference in output is a negligible trade-off for the better looks, in my opinion. It's not like we run these at full-blast often anyway

That selection of parts looks good to me. Where are you ordering from? I find Mouser usually ends up a smidge cheaper on shipping, at least for smaller items like this.


Fun stuff huh? It gets to be sort of addictive. Especially if you get a nice slotted heatsink like the MakersLED or RapidLED units, then swapping emitters gets pretty easy when new and interesting ideas come up

I homebrew all my lights these days. I just can't help myself, and it's easy to get exactly what you want, or try new ideas easily.

Before you know it you'll be designing PCBs and having them made!

 

[jrp1588]

It's not like we run these at full-blast often anyway

Yeah, that was my main thought. Alright, you've convinced me. I think the one you suggested was actually considerably cheaper anyway, and I do feel like my tank could use some better color pop on a few corals.

 

[jrp1588]

Doing some more digging. 2 of my fixtures definitely use XM-L white chips, so they'll have to be upgraded to XM-L2. It looks like they only sell the 90CRI XM-L2 cool whites in packs of 1000, so I think my hand may be forced into going warm white. Highest K value I could find was 3000K, and they're still pretty pricey.

XMLBWT-00-0000-000UT20E7

 

[Lingwendil]

The XM-L2 don't seem as easy to find in small numbers unless you go really low K, or low CRI... I'm not super familiar with the XM-L lines enough to really point you to all that much else for them On those I would just do something in the XM-L line that has a good K value and leave it, as 3000K will be pretty hard to keep from getting purple from the large amount of red unless you have lots of green to offset it, which can have other wonky side effects when trying to get a good look.

The 75CRI 4500K is about ~3.16 each-

https://www.mouser.com/ProductDetail/Cree-Inc/XMLBWT-00-0000-000LT50E4?qs=5YqktSXERipTJV/afROeGg==


I can't seem to find any of the 85CRI stuff in singles, they want to sell them in full reels only it seems. Looks like warmer stuff or 75CRI is the best you can really find on those really. The warm stuff might work well if you offset it enough with green, or like a whiter tank (like some of the SPS guys do, maybe?) but caution, there be dragons

 

[jrp1588]

Yeah, pretty much lines up with what I was finding. The ones I bought at RapidLED are 65CRI. Old as these Vegas are, I imagine the whites are similarly low. I do have 7 good pucks that can use the XP-G3 whites. Maybe I could do some puck swapping and run half warm white on the XM-L2 and half cool white with the XP-G3?

 

[Lingwendil]

Maybe. The color mixing may not be all that great without a diffuser of some sort, but that would cut down on the intensity and shimmer a great deal, so not all that fun of an option. it's worth a shot, but I wouldn't be able to tell you how well it would mix without trying it otherwise. Cheap enough to try, I suppose.

 

[Gungo]

If my old AI's burn out leds I'll just send them to you guys so you can replace them for me!!! Thank you very much for sharing this!

 

[Lingwendil]

Heh. I'm not set up to reflow PCB stuff anymore, but I used to do it a bunch. I like through-hole parts on PCBs nowadays, and get my LEDs mounted on stars for easy swapping around as I see fit or get ideas to try out. I've got a weird knack for finding parts and deciphering datasheets so I try to help as I can

 

[jrp1588]

If my old AI's burn out leds I'll just send them to you guys so you can replace them for me!!! Thank you very much for sharing this!

Let me get a fully successful resoldered light under my belt first. Haha. I'm trying again this weekend.

 

[jrp1588]

After a lot of data sheet fun, I was able to find ONE 80CRI LED in a 4000K neutral white that I think may work that they'll sell individually. Better than the 65 or 70CRI they seem to default to anyway.


XMLBWT-02-0000-000HT40E5

 

[jrp1588]

Alright, new results. I reflowed all the remaining LEDs on my test unit. I didn't damage anything this time, and all LEDs light.

The AI Director software is very angry though. When I first connected the modified light to my ramp schedule, it thought the light had all these channels that it never had before, which caused it to only light up a couple LEDs. After a handful of resets, power cycles, and connecting and disconnecting it to Director, it finally read the channels correctly...but I can't say I trust it to come back on correctly if it loses power again.

I will say, the new light is noticeably brighter. So much so that I'm a little concerned about the corals that are under it. I might try to put it on a separate ramp schedule that's toned down a bit, but I'm kind of afraid to touch it.

I'm going to run it over the tank for a while to see how it acts before I consider doing this to the rest. Does anyone know how the software figures out what channel is what? Maybe the XP-G3 and the XP-E2 Red have a significantly different voltage or something than the xt-e white and royal blue and the Osram Red? Would I have better results swapping in more direct replacements?

 

[oreo54]

Remeber this "problem" before but no real answer.. Could very well be the V(f) differences in the strings...

 

[jrp1588]

Well that's unfortunate. I assumed something on the pucks themselves or their individual driver boards were what told the brains of the light what LEDs were on it. As long as the light holds its programming now that it's reading the channels right, I'll probably go ahead and do the rest of them eventually. It's not like my tank loses power very often. Occasionally having to fiddle with the light a few times after a power cycle wouldn't be the end of the world in exchange for increased output and efficiency.

 

[jrp1588]

Here's some shots of yesterday's project. Heat on my stove was turned to about the halfway point, but again, I recommend starting on low and working up gradually.