# Calculating specs for power supply of array of high power LEDs

I doing an experiment with plants, for which I want to use commercially available LED lamps. They provide the light I need but the area they cover is too small. So I am trying to replicate a lamp of high power LEDs but I have straggle deciphering the necessary power supply, and I wonder if the specifications can be calculated?

The image below is the circuit I want to replicate. There are four arrays of 30 LEDs connected in series (orange outline in image; each LED is claimed to be 10W) and it includes two power supply (based on the image it appears each power supply is 50W). Each power supply feeds two of the four LED arrays (one circuit is shown in red, in the image below).

Using standard values for 10W LEDs: LED forward voltage=3.5v, LED current = 1.05Amps

For the sake of simplicity, I do calculations for each of the two independent circuits (i.e., two LED arrays and one power supply, red outline in the figure). So these are my calculations so far:

Expected voltage driver=Number LEDs * LED Voltage

Expected voltage driver=30*3.5

Expected voltage driver=105v

Expected power driver=Number LEDs * LED Voltage * LED current * Number of LED arrays

Expected power driver= 30*3.5*1.05*2

Expected power driver=220.5W!

Yet the only information in the power supply it says it is 50W. I wonder where my error is? The circuit works, so I wonder if perhaps the LEDs are lower wattage (e.g. 3W instead of 10W) or if this type of circuit uses a PWM so that only some LED's are ON at a given time?.

• " I wonder where my error is?" "Using standard values for 10W LEDs: LED forward voltage=3.5v, LED current = 1.05Amps" You answered your own question. Dec 4 '16 at 22:09
• 1 A * 3.5V = 3.5W Dec 4 '16 at 22:10
• yes but each array has 30 LEDs, and there are two arrays, powered by this power supply. So the total current demanded from the circuit should be: 30LEds*3.5vol*1.05Amps*2Arrays, no? Dec 5 '16 at 0:04
• The error may in attempting to replicate quite possibly THE worst LED grow light design ever. Feb 28 '17 at 6:56

LEDs work absolutely fine at below their rated power. So if it's a 50W current limited power supply, it will give as much power as it can, and the LEDs will glow at the corresponding brightness.

• That's a poor manufacturing method, and a poor idea to work on for a new device. Dec 4 '16 at 22:14
• hmm, this is a big possibility. These devices have been critiqued extensively for not delivering the light they are suppose too. So this low power supply, may be why.. Dec 5 '16 at 0:07
• It is actually not true. Some LED, especially phosphore based gives sometimes a minimum operating current, I do not know exactly why but I've seen it in datasheets. Perhaps to avoid local phosphore coating degradation. Oct 8 '18 at 12:27

Expected power driver=Number LEDs * LED Voltage * LED current * Number of LED arrays

Expected power driver= 30*3.5*1.05*2

Expected power driver=220.5W!

Miscalculation. 10 W leds are ~3 Amps at 3.3V. Assuming these are relaying driven at 1W, or 3.3V @ 0.3 Amps, then you have:

30 Leds* 3.5 V * 0.3 A * 2 Arrays = 63 Watts Per 2 Arrays.

Otherwise if they are being under driven then both the Forward Voltage and the Forward Current are different, and the total power needed will also be different.

Measure the actual voltage across any diode and current used across any array in this circuit, and then you can work from that.

## Of course, this still isn't the complete picture:

Obviously, its using two different color diodes. White and I guess blue or green. Different VFs. Also, looks like it has a glass zener or schottky diode across each led? This all affects the circuit and its current draw.

• I did not find LED's of 10w that look like the ones in the photos, but others I found suggest that they use ~1000mA. ebay.com/itm/… Dec 5 '16 at 0:11
• @camilo the same package for 0.5W, 1W and 3W leds. Dec 5 '16 at 1:14

They are 2 Watt LEDs.
See Lower Left Corner of this similar board where it says 132 x 2W.

Just because an LED is in a class of 2W does not mean it is rated to be used at 2 Watts. Luxeon Rebel Color and Cree XPE Color LEDs are 3 Watt LEDs, but not all colors can be run at 3 Watts. The 3 Watt usually referrs to the White LED in the series. White, Blue, and Green LEDs are 3V+ and the Red are about 2V and rated 700 mA max, where the White and Blue are 1 Amp. But all are still called 3 Watt LEDs.

Most LEDs are not driven at Max Current.

Label from one of the LED Drivers

The supply labeled LED Driver, is actually powering the Fans.

• hmm no images... Feb 28 '17 at 17:54
• I do not know what you mean by no photos. There are 4 photos in the post. The LED side of the board, the heat sink side of the board, the fans, and a power supply label. Feb 28 '17 at 18:07
• AWS server was down Feb 28 '17 at 20:59
• 4x50W supplies? in 100 sq in? or 2W/sq.in. with fans and massive baseplate sink is about right Feb 28 '17 at 21:04

The 50W power dissipation limiting factor is Tjcn and Rjc+Rca thermal resistance of the MCPCB.

This primitive layout cannot possibly dissipate more than 50 Watts without force air cooling and rear fins.

I would suggest you buy lamps , cheaper and and better designs exist all over the web. or ask me.

This is what a 100W heatsink looks like with forced air cooling. The difference should be obvious to an engineer. Learn Ohm's Law for thermal Resistance of each interface from junction, Tjcn to case ,Tc to sink Ts, to ambient Ta. Each interface has resistance in ['C/watt] [Read]2

This cheap design has inadequate power and heatsink.

# My Analysis

This size chip might be able to handle 3A peak and 1A continuous with adequate heatsink about 5cm square flat alum per LED , but this density of LEDs and no sink perhaps only 300mA and that will be burning hot.

LED's are no long rated by power since many users misunderstand specs that all chips are rated by 25'C using an infinite heatsink then derated for junction rise according to use of heatsink and ambient temp.

• 1Amp Blue & White LED's have ESR ~ 0.3Ω above 2.8V
• 1Amp Red LEDs have ESR ~ 0.6 Ω above 2.0V ( for I>0.3A) and 1.4 Ω above 1.8V (for I<300mA) (similar specs)
• There are 14 White LED's out of 30 ( visible phosphor)
• The rest are a mix of Red and maybe Blue
• There is an SMD R in series with each LED to normalize and raise the ESR of each LED above ( something <=1Ω)
• there are 4x30S modules with unknown wiring.
• if all in series 120S with an average of 2.5V would be 300V which can be easily driven with cheap rectifier and current limiter from 230Vac rectified to 365V pk.

They have done something primitive like this and there is not much that can be improved with the layout or power and certainly not efficient if the resistors are hot. (I would rate this product as 1 star in horticulture lighting)

• If you dont capiche, then ask an intelligent question. Dec 5 '16 at 16:32
• What about Tpcb? Thermal vias? Feb 28 '17 at 7:01
• MCPCB is only way to go here for substrate. After this many options. Rth is the critical spec to design Feb 28 '17 at 15:39
• Posted an answer with the heat sink used for that board. Horribly designed board. But it works, and runs cool. Feb 28 '17 at 16:23
• not sure of your point without photos, define what you mean by cool for Tjcn. ( solder not melting?) Feb 28 '17 at 17:56