DISCLAIMER im not an electrical engineer. Im not giving practical advise for you to follow here. It’s purly theoretical. If you burn down your house with a flawed deisgn LED light, or a badly exectued good one or any other way, thats on YOU. Im merly pointing out shortcomings of designs imo. And pointing out posible theoretical ways you could further research to fix those shortcomings.
@RichardRoundTree quite a few of the points you make seem reasonable. And i don’t disagree with.
The following statement though seems way off.
Granted good Led die’s and the chips are expensive. And saving money there will cost you efficiency.
But no-one builds there lighting systems at 4% of typical capacity of the led die. For comercial well tested and designed products most are run fine at typical recomended currents.
The 0.4v per led (die) in the series string is also BS. 0.4v is not enough potential to overcome the band gap for visible light. Let alone for blue light to further convert into other wavelengths by the use of a phosphor.
The beta max in a VHS also does not make much sense. You can calculate back from lumens and the spectral graph to µmol using the eye’s sensitivity curve. It’s some complex maths but it’s accurate and based on sound physics.
The physics and electronics of good led lighting design are complicated and require maths. Most people don’t have this knowledge (of physics/maths) and thus default to #of led chips, “wattage per chips” number leds per color or other metrics that can not be used as “blanket statement metrics” and thus the statements made are at best only applicable to highly specific situations but more usualy just plain BS.
i hate blanket statements, but gonna make one thats usualy true.
“Advise about leds given by anyone without them linking and referencing to a data sheet should not be trusted, same goes for a lack of maths”
Even when a person links to a data sheet they should be able to explain anything on that datasheet before you could consider them even vaguely knowladgable. This usualy holds true for any electronic part. (from resistor to fets to leds)
@dimitri_7a
most of the things you say seem reasonable. I do not know how you did the conversion from lumen to µmol, without spectral data (could not find it in the data sheet, the data sheet is quite limited, especialy compared to for example one from cree)
I am worried about that thermal design.
The datasheet uses 65c as recomended temperature on the measuring point on the led itself. This is quite low. assuming at least 25c ambient. This leaves 40 C temperature differential. And since the PCB and also the dubble sided tape have quite a high termal resistance you will loose a large part of your termal differential to those alone. And then you also use a quite primitive passive heatsink.
Im not saying it does not work. I am verry curious though what you would measure if you measured temperature on the termal measuring point on the led pcb.
The data sheet also states a max temperature of 95 c which again is quite low. Especialy comperaed to for example cree’s 125c max temperature.
This all leads me to assume in your use case your overloading the leds. they probably run at a higher temperature than rated. This will cost you some efficiency and quite a lot of expected lifetime. To make it more complex, you should not measure LED lifetime as in till it brakes. But from start till when it only emits X% of origional light (x is usualy 80%)
If you keep that design. you could improve it with better heatsinks. Or a thing layer of TIM (cpu paste for example) and the appropriate pressure on the PCB to ensure good TIM contact. Another easier fix would be to run the leds under rated power. Below rated power they run a little more efficient. But can also handle higher temperatures for longer. And they produce less heat so the temperature differential devided by the thermal resistance can end on a lower number and still be fine.
Another thing that worries me in that design is the load balancing. I see no way in which they ensure the paralel strings are evenly load balanced.
LEDs can be roughly modeld as a constant voltage drop with a small resistor in series. This means if there is even a small diference in the voltage drop between paralel LEDs, the only “buffer” is that tiny resistor in series. And thus the LED string with the lower forward voltage drops can take significant higher share of the current. And thus overheat and burn out. At that point the remaining LED strings each get more power, and likely 1 of them burns out, and thus cascading into failure.
Putting them in series balances load. But can increase the required voltage to dangeours levels. So as a DIY you should not do this.
Putting them paralel with a seperate CC driver like for example https://www2.mouser.com/datasheet/2/260/LDD_L_SPEC-1291586.pdf
can work. Although these CC drivers are usualy Buck design and have a minimum voltage drop ov 3v or somthing like that. So your AC/DC power supply would have to be 36v version.
You can also roughly load balance them by increasing the series resistance. In this case running the leds at 600mA (1/3 of max rated current for DIY seems wise) seems reasonable to me. Id aim for 2-5% of the forward voltage drop (go 5% if you measure them all to be really close go 2%) to drop over your resistor. This would give you a 1.667 ohm resistor. You can “make” this resistor from 3 paralel 5ohm resistors or 6 paralel 10 ohm resistors. use 0.5+watt rated resistors for safety, it’s overkill but for a diy you don’t need to save the 0.10 dollarcent.
you would thus place this extra resistor in series with each string. And than put the strings (each having it’s own extra series resistor) paralel.
An more efficient and elegant option would be to use a few opamps and a fet in it’s lineair region as a CC source (or as a variable resistor adjusting itself to increase/decrease to cause a constant current flow)
you would have to oversize your fet and cool it appropriatly.
There is also an IC that does this. but i can’t find the part number atm.
i quite like your choice of power supply brand. Going for a good power supply brand and product line is a good safety choice. It does however NOT compensate for other bad safety choices, so keep that in mind.
Alternatives might also be from meanwell “ELG line” usualy a bit cheaper and comparable in most regards.
Also going for a AB version can be nice for flexibility. You can use the B version for dimming if your going passive resistance load balancing. And the A or blank version for CC drivers (either fet based or switching based). The B line is also nice for switching the light on and off by dimming to 0%. This avoids inrush problems that i discus next.
Keep in mind the inrush current on your power supply.
Most power supplys have incrush currents above 60 amps (100s often). This can easily start a fire when switched by a cheap inapropriatly rated for max 16 amp relay in time switches.
Using the B power supply and diming from 0 to 100% with somthing like an arduino, seems a lot safer to me.
Meanwell also has an inrush current limiter that might be intresting.
@nil some things your saying do not make scientific sense to me. I see no reason why a LED light could not outpreform another lighting technologie. That it has not in your experiance is not “proof” that it’s inposible.
All the electrical energy has to go somewhere. in these cases it goes either into light, or into heat. LED’s radiate/convect that heat away via a heatsink. other technologies usualy radiate it on the plants. That is a significant difference. if it is positive or negative depends on the enviroment.
compare it to me planting a seedling apple tree. pruning it really small each year. And than claiming all apple tree’s never give fruit. This might have been my experiance but is definitly not a true statement.
ps, data sheet link here
https://cree-led.com/media/documents/CMU2239.pdf
it is quite extensive. and cree also has other good design documents. If you can read such a document, and not have any questions and recognise all the maths and termenologie you’d build a much better and safer light.
. This means the heatsink gets to no more than 140°F in stale air, when I add a small circulating fan it probably drops to ~110°F. I don’t know if IR thermometers work well on aluminum.
