I went ahead and designed my own light board. May be fun to play with, I’ll get the SolStix but nothing stopping me from a little DIY. I’m already into the rabbit hole far enough I may as well wind up with exactly what I want. I threw together a layout for a 9.5" x 9.5" aluminum panel with 192 LEDs. It’s flexible so I can use most LEDs, including the 301B and 301H, as well as the Bridgelux BXEN family. That footprint is popular so it also opens the doors to “generic” options.
Costing (with HONGLITRONIC HL-AM-2835H421W-S1-08-HR6):
That’s right, 5 complete boards (aluminum substrate 1-layer with white soldermask), custom-fabricated and assembled to my specifications, with cheapo 6000K LEDs are $90.
Now, I understand COMPLETELY that these are chinsy LEDs. A few things lead to wanting to try this, not necessarily price, as I’ve already invested a few hours even finding these, making them cost more to me than 301Hs would. The efficiency is lower than the 301s, probably ~20-30% lower. I think more LEDs, with reasonable thermal design gets around that problem. The next fundamental issue cited by others is reliability. I think the root cause of inexpensive boards failing frequently is the inability for (direct-parallel layout configurations I see everywhere) current sharing. I’m using load balancing resistors. The footprint and the part cost nothing. Curious how it’ll wind up. Each series string uses a 10R balancing resistor, which will cost about 3% efficiency but should allow for each LED Vf to vary by as much as ~100mV and keep my currents reasonably identical. I think that will dramatically improve reliability.
I’m going to try two panels (9.5" x 19" of panel) - about perfect for my space and a standard propagation tray. They’re configured for daisy-chaining so they’ll plug right together with a little jumper.
If they suck, I’ll just build a new version with 301Hs 4000Ks and be done with it.
Here’s the selected cheapo: 2304140030_HONGLITRONIC-Hongli-Zhihui–HONGLITRONIC-HL-AM-2835H421W-S1-08-HR6-SDCM-6-5700K-6500K60mA_C516134.pdf (1.7 MB)
Cheers boys, time for an experiment!