Considering a DIY LED build For flowering only: Design considerations...

Here's what I'm doing:
https://www.rollitup.org/led-other-...custom-1w-panel-observations.html#post9121145

in addition to the four 24w panels, I'm doing two cree xpg WW floods (10X cree xpg WW) and then adding 1X XP-E Red (630nm) and 2x phillips rebel es 660nm FR to fill out the reds for flower.

Thanks, so how did you come up with this graph? Is there a modeling program of some kind to do this?

ledspecs.com

caretak3r said:

ledspecs.com

Click to expand...

Great resource! Thanks

The Ledengin spectrum tool is good too, you have to download it to your computer. You can find it here.

It's easy to tweak and add other LEDs and power levels.

SupraSPL said:

I agree separate modules will be the way to go for the coverage you need. It also allows for airflow around the heatsink and into the canopy.

Check out heatsink USA. You can passively cool the system all you need is the right amount of surface area. For top bins running soft I use 17in2/watt to aim for a Tj of 50c. Passive cooling has the advantage of higher efficiency (fans consume electricity), quieter and simpler design. Also the heatsink will benefit from your circulation fan in the grow room. I would be willing to bet that my passively cooled LEDs run cooler than most actively cooled designs. The downside is the initial cost, $130 for the heatsinks to run 200W of LED and that was a friend price. They are also heavy, 4 modules at 6 pounds each. Those downsides are acceptable to me but may not be for everyone.

If you are planning on running the LEDs at soft levels (700mA) and using top bins, you can replace your 400HPS with 200w of LED, coverage and yield will be no problem. I encourage 700mA as a good compromise of efficiency and upfront cost. 1A is the highest I would run an XML2 in a grow room and since XTE is much cheaper I recommend those over XML2. Also the XTE is available mounted on a 10mm star but the XML2 you would have to buy stars and reflow. Illumination supply sells XML stars for $0.50.

XML2 - 3000K - T3 bin - running at 50c
.35A 42.6%
.7A 38.7%
1A 36.4%
1.5A 32.8%
1.75A 31.6%
2A 30.1%

XTE - 3000K - R3 bin - running at 50c
.35A 41.5%
.5A 38.3%
.6A 36.8%
.7A 34.6%
1.05A 28.6%
1.5A 25.5%

Click to expand...

Read your thread:

SupraSPL said:

If you run the XML at 1500mA it will be 17% less efficient and 20% more expensive than 2 XTE's at 700mA. But I can understand the laziness part to each his own

Click to expand...

Compelling argument for running on less current/better efficiency.

Wondering how lazy I am.

SupraSPL said:

XML2 - 3000K - T3 bin - running at 50c
.35A 42.6%
.7A 38.7%
1A 36.4%
1.5A 32.8%
1.75A 31.6%
2A 30.1%

XTE - 3000K - R3 bin - running at 50c
.35A 41.5%
.5A 38.3%
.6A 36.8%
.7A 34.6%
1.05A 28.6%
1.5A 25.5%

Click to expand...

Are you sure about the R3 bin? According to the XT-E datasheet, there is no R3 bin with 3000K tint.

tebos said:

Are you sure about the R3 bin? According to the XT-E datasheet, there is no R3 bin with 3000K tint.

Click to expand...

I think that was discussed on Supra's thread

OK, so here are conceptual design schemes I am seriously considering at this point. I haven't had time to start plug in all emitter combinations into spectrum modeling graphs yet, but will start playing around with that soon:

Certainly convinced that modules of some sort will be the way to go, probably secured into one effective "panel" with a frame rather than a single massive panel. Not sure yet whether several parallel bars or 4 rectangular chunks. Advantage is better airflow "through" the panel for both heatsink cooling and canopy air circulation

variation 1- xm-l predominately WW with a few CW thrown in for more solid blue range. run around 1750ma

Advantages: KISS. simplicity in LED/driver selection. less to mount solder

Possible disadvantages: Less efficient than xt-e run at 700ma, could be a little lacking in deep/far red, not as good even spread mixing these higher output emitters. Regarding penetration, My grow space dictates a controlled level canopy, Scrog/mainline etc, so theoretically, even spread is more important than deep penetration.


variation 2-as above but change one string/driver to 700ma and add a string of deep red/ far red and possibly change some WW to NW or even CW for blue and added overall output since they are more efficient than WW. This next bit is not KISS, but this also allows the possibility to switch or dim the extra red in order to start flowering with proportionally more blues to keep internodes shorter, and crank up reds into flowering cycle.

Possible advantage: more controlled red/blue balance amidst the whites, and adjusting for plant need transitioning into flower and through flower cycle

Disadvantage: those specific red/blue proportions for FLOWERING still remain elusive to me! More complexity, but one switched or dimmable driver circuit is not a huge amount of extra work/complexity


variation 3: xt-e rather than xm-l driven at 700ma (as spec'ed by SupraSpl) white spectrum as version 1 above

Same advantages/disadvantages as version 1 + better spread and blending, probably can also keep closer to canopy that higher power xm-ls

Extra disadvantage of double the components to mount/solder,


variation 4: Xt-e and possibly other brand/model roughly same variation considerations with spectrum as #2 above also possibly switched or dimmable deep red/far red driver circuit

advantages: better efficiency, easier to mess with different proportions of discrete emitters, better spread and blending, probably can also keep closer to canopy that higher power xm-ls

Disadvantages: twice the work mounting/soldering,



SO, now I'll start playing with spectrum graph modelling and look at relative build and ongoing energy costs.

Life happens, I have been distracted from this for a while, but back at it.

Running some different combos in the graphs I still remain confused how much blue is too little or too much in the context of CW, NW. WW.

I've been looking at some of the grow threads and seeing good results with Area 51 which with all strings on is 2:1 White:red and tags recommendations for 3:2 with the apache. Area 51 uses 4000K for whites, and unfortunately Apache is mostly smoke and mirrors, but one pic/graph seems to show their 4:1 panel as 6100K whites and 630nm red.


Area51 panel:







My Combo #1:

This graphing web page, unfortunately does not have cree XT-E or XPE-2 in the higher bins I'll use, but does have XP-E. All the LEDS I'd order have higher output so I think the relative underlying white/full spectrum and blue bump would look just about the same the same as this graph. This combo has higher red spike and tapers off more down into lower green range.

96 total LEDS, All cree to keep ordering simple and they have really good efficiencies:

2:1 overal ratio White:red,

3:1 ratio of WW to CW in the white mix, CW added for bumping blue to keep stretch down

Actual parts:

48 XTE WW R3 BIN 3000K
16 XTE CW R5 BIN 6500K For blue kick and a bit of added full spectrum efficiency
32 XPE-2 RED P3 BIN right about 630nm

Huh, bizzarre, couldn' t upload the third graph, try again continuing in this post

(Please read previous post, this is continuation of same)

View attachment 2705435


Hmm, this is graphing with XPE's and actually turns out identical to the Area51 graph, I guess with XTE whites and XPE-2 reds it would still be almost the same anyway just more output

In either case, I will put one or two of the red strings on a switch to switch off and lower red for first week or so of flower to help limit stretch.

My question remains whether additional flux/overal spectrum and blue will help or hinder FLOWER production

In any case, LEDS driven by 8 Mean Well lpc-35-700, 12 LED per string

5 or 6 water cooled 1.25" aluminum rails as heat sinks each 36" long spaced for total fixture size 36"x18"

With 5 rails this would put the rows nominally 4.5" apart, and with 6 rails 3.6" apart.

Whaddya all think regarding the spectrum concerns, and the rail spacing 5 vs 6 rails?

Aw jeez, guess I wore y'all out.

In the meantime, today under HPS. P1 about ready to chop, P2 a little longer: