BOM
- 3x
SI-B8V261560WW €26.42
- 1x
ELG-75-C1050A €32.52
- Free 3 day world wide express shipping (how?)
- 1x 30x2mm x 250cm anodized alu strip, ~€6
- Pack of M3 screws and nuts ~€3
- Pack of 2.5mm hooks with universal / wood thread ~€3
- Cardboard, tin foil and scotch tape
Initial build
The parts came in and it's time to test it out. First test: PSU / power level. Turning the setpoint screw lets me pick anything between 25 to 90W power drawn at the wall - huh okay, the PSU clearly can supply more than 1050 mA. The bare PCBs are too wobbly and probably will run far too hot without head spreader, so I got a strip of 30x2mm anodized aluminum and cut them into 59cm strips, drilled 3mm holes and mounted the PCB with a bunch of M3 screws and nuts.
The loops on the back are a bad idea - 2mm pilot hole and 2.5mm wood screw thread eye don't mix well but the shop didn't have M3 hooks. I need to redo them but it will do for now.
Measurements
I made a grid with 10x10cm squares in a 60x60 area ~4 sq ft, a common small grow tent size. This is more HD than Coco's 15x15cm, I wanted higher resolution data. But as I wasn't doing any tent stuff, I'm measuring in a wide open area which leaves out gains from reflection. First test, at 37cm height, shows a big even lit area (but obviously far too big).
Measurements are in lux using a Voltcraft meter. On Android / Pixel 3a, the ambient light sensor on the front also does a good job estimating it within +- 20%, good enough for a quick check.
Going from square to square writing down the numbers results in:
Yeah, not too great as expected, total lumens captured is 1532 vs 4300 rated, 35%. It will definitely do okay as supplementary light at 10DLI in the center but it is such a waste of energy. It is still frikken bright though.
PAR is estimated using a 1:69 ratio from
Par Multiplier Thread #18
Second test is at the same height, with a reflector, a cardboard and foil contraption for prototyping until I send the design off to a metal shop. Much better - 2015 lm / 47% captured.
Third run - lower it so the beam hits just the edge of the sheets on the ground. There is still a lot of light lost on the other direction but you can't fix that without going COB + optics.
Getting there, 60% light captured on the 60x60 area.
Last one. Lowering it again, to 14cm height, to concentrate on the middle 40 cm wide strip.
That's pretty decent... 65% light in capture area, 35% lost to the top/bottom part and reflector loss.
Modularity
You can place multiple strips in a single area, just add more parallel to extend the hot zone. If you scale this up to a 80x60 area with 3 overlapping strips (placed 20cm apart), still no grow tent / reflections gain, you get:
Total score: 12165/48 squares = 253 umol/m2 x 0.48m2 = 121 umol/s total for 75W, 1.61 umol/watt,
outside of a tent. This is so close to a Mars Hydro TS 1000's
1.65 umol/watt measured in a tent.
If I ever need to grow something that requires intense light, 3 strips in a 60x60 tent with reflecting foil would be perfect, I bet it hits 600+ peak in the middle then with barely any cold spots. Another easy way to boost score is to just add 2 reflectors on the top/bottom edge, the sides barely leak any light anyway.
You can modularly extend this basically indefinitely to the sides, just add 1 module per 20cm, but also to the top/bottom - that makes the middle hot zone taller so you get less top/bottom drop in light intensity.
(In hindsight I should have measured a bigger area in the vertical direction so I can simulate extending it to top/bottom too...)
Heat management
At 25W per strip, you don't need much more than just a backing aluminium plate (60x3=180cm2). Use anodized as it has far better emission coefficient. My infrared thermometer says the back is at 45°C and the peak number on the LED side is ~55°C.
Machining the new reflector from aluminium including the sides will increase the surface area to 840cm2 - overkill. It's basically a plate with 2 bends, cost is ~13 euros each which makes them more expensive than the LED module it is mounted on.
Reflector improvements
What you really want is a perfect beam source sending light straight down, like this
A parallel source like that means there is zero impact of distance from source / height on light output and zero losses to the sides. Too bad it is impossible with point sources like LEDs. Narrow lenses get close but it is way too much work to glue one lens on per LED chip.
The prototype reflector #1 has a +/- 45 degrees angle to the sides, which leads to a pretty wide beam and light loss if the plants are not all at the same height. My basil is super tall, followed by rosemary, but the tomatoes and chilies are barely 10cm high, this leads to epic light leaks out of the sides and lighting up the streets, my shelves are next to a window to get 10-100k lux for 10 hours a day of natural light. I went back to the drawing board and came up with a new and improved version that looks more like this - going wide and tall, ~4.5cm down from the LED height and not much wider.
It is almost straight down but it really still needs an angle, a plain 45x50 aluminium U profile leads to internal reflection craziness like image #2, crazy how going from 90 degrees (straight down) to 102.5° changes everything.
Todo
- Prototype reflector design v2.
- If light leak issue is fixed, send final design to metal shop to get rid of that super wobbly / wonky cardboard reflector. My IRL cut and paste skills are so bad...
- Rerun the light measurements with final 100% aluminium reflectors.
- Also measure exactly on axis in the middle and to the top/bottom, do a 80x40 or 100x40 area instead of 60x60.
-> 
. All values were written down/pic taken w/ phone using lux app and using a tape measure.
