What's new
  • ICMag with help from Landrace Warden and The Vault is running a NEW contest in November! You can check it here. Prizes are seeds & forum premium access. Come join in!

LED FAQ) Building and DIY

Arf

Member
80x COBs over 4'x5'? That's gonna be mighty bright.

Not when you are running them at 10Watts each. That's 40W per sq ft which is less than many people run HPS at. My rule of thumb is to run HID at 40-50W/sq ft. and LED at 30-40W/sq ft. Now that's not based on super cropping which can give you an even canopy and get efficient LEDs like CXB3590 down to just over 20W/sq ft.


Wiring 80 COBs would be a pain, even at 10 per 100W driver. I wouldn't do it, I would use 20 x CXB3590 running at 40W each. But then again, I don't do commercial sized grows like 4' x 5'.

A single 2' x 2' grow is heaps for personal use, that would be 16 x Chinese 100W COBS running at 12.5W each which works out to 200W if you turn it up full, normally run at about 150W. In the past I would have used a 250W HID for that space and had a lot more shadow. That's the big thing about LEDs, you get rid of the shadow a single HID casts.


Some actual measurements.
https://orchidborealis.blogspot.com.cy/2015/11/diy-chip-on-board-cob-leds.html

Vero 29 4000K running at 42W has a PPFD of 280 at 12"
CXA3070 3000K Z2 bin running at 41W has a PPFD of 250 at 12"
Chinese 100W 3000K running at 41W has a PPFD of 175 at 12"

From another source I have th following measurement:

CXB3590 3500K CD bin running at 48.9W has a PPFD of 365 at 12"

Because the CXB3590 is running at 20% more watts, it's not a great comparison.

I certainly wouldn't run the Chinese COB as high as 41W like the tester did.
So you can see the Chinese COBs are far from useless, it's just that if you have more money you can get better COBs if you want to, but expect to pay 10x the price.
 
Last edited:

CannaBruh

Member
No experience with anything other than the 3590s, I think 1 of those over 1sqft is fair at 50W, what would you say the minimum be 20W?
 

Arf

Member
No experience with anything other than the 3590s, I think 1 of those over 1sqft is fair at 50W, what would you say the minimum be 20W?

Depends how even your canopy is, if it's a nice flat SCROG 20W is fine, but if it's an arrangement of different plants at different heights, you will need more power or the lower parts wont be well lit. When everything is even, the lower you can hang your lights, so the lower the power you need to run. It's a matter of trial and error, a normal lux meter helps, it wont tell you PPFD, but it will tell you if the light is bright enough. About 40,000-50,000 lux at the grow tips with a CXB3590 seems to be good, the lower buds need to get at least 20,000 lux or it gets a bit popcorn.

Some people have made LUX -> PPFD conversion tables for popular Cree LEDS at different color temps. I don't have any links, but I remember seeing them around.
 

jikko77

Active member
something like this:

Here are the multipliers to convert illuminance in Lux to PPFD for some common light sources:
Code:

Source Multiplier
Sunlight 0.0185
Fluorescent (Grolux) 0.029
Metal Halide 0.0141
Fluorescent (865/840) 0.0135
High Pressure Sodium 0.0122

Figures provided by Apogee Instruments.

So, we can multiply the Lux value by the multiplier in the above table to find the PPFD in µmol m-2 s-1.


Worked example:

Code:

60 cm x 60 cm space (0.36 m²)
250 Watt HPS lamp (33,200 lumens)

Lux = (total lumens) / (total area in square metres)
= 33,200 / 0.36
= 92,222

PPFD = Lux x factor
= 92,222 x 0.0122
= 1,125

but is innacurate and inprecise...
by the way, not all par meter are accurate either.
 

CannaBruh

Member
50W over 1sqft with a standard scrog where the tops aren't allowed to deviate from the screen other than to cone up and stack above the screen. I've pulled this off with 600W HPS with the glass of a coolhood right ontop of the canopy. Tradeoff with the LEDs COBs now are cost when considering how hard to run them and the number to spread over an array above a canopy.
 

Arf

Member
50W over 1sqft with a standard scrog where the tops aren't allowed to deviate from the screen other than to cone up and stack above the screen. I've pulled this off with 600W HPS with the glass of a coolhood right ontop of the canopy. Tradeoff with the LEDs COBs now are cost when considering how hard to run them and the number to spread over an array above a canopy.

About 350W of CXB3590 will match a typical 600W HPS.

Here is a shoot out yield video using 1200W HPS vs 1200W LED

https://www.youtube.com/watch?v=IKXEFAhSQoM
 

Dawn Patrol

Well this is some bullshit right here.....
Veteran
A 4' x 5' space is massive, as I mentioned before in post #1028, the Chinese COBS are essentially the next step up from using CFL.

Would you try lighting a 4' x 5' with CFL? You could certainly do it, but you would need a lot of them, and you would be dealing with a lot of heat. Same goes for the Chinese COBs, in that space I would expect to use about 80 of them. They work best around one per a 6" x 6" square.

Dude, you originally responded to a post by DTOG. While I don't personally know the man, I've seen the pics of his work and he probably clones in a 4' x 5' space! He's not somebody that's growing 12 1' plants in solo cups in his bedroom closet or that grows with CFLS.

You seem to be real good at telling everybody else what they should do, how about posting some pics of your setup and what you've got going on?

Hanging 80 COB's in that space and the associated cluster fuck of wiring and infrastructure is not practical, or reliable for any normal purpose.

Maybe you caught me on a bad night, but anybody that does anything more than dabble in growing will tell you that good equipment that is reliable and doesn't require constant attention is worth every penny. Certainly initial acquisition cost and the ongoing cost of operation is a factor, but I believe your advice is applicable to very few people in this thread.

Dion has created this thread to advance the use of DIY LEDS and from the start he has recommended Cree COBs for a reason
Chips- I recommend cree Cobs, they are the best and although the price reflects that you get what you pay for, also cree COBs use a ceramic board making thermal management child's play you can buy them from digikey or mouser. I also recomend surexi chips by illumitex they are great and the company has invested enough in R&D to have a nice range of horticultural chips( http://www.illumitex.com/illumitex-leds/surexi-horticulture-leds/ ), these however are no longer in production as they now are just selling prefab lights but you can still buy them on amazon or mouser while stocks last. Others use vero( bridgelux) and you are welcome to try them I never have and never will. Cheap chines cobs off ebay or aliexpress? nah we are chasing efficiency here leave the junk in china.

I'm not completely discounting your ideas but what you're proposing has been done many times before - it just may be much better suited in the microgrow forums.

Many of us have been down the cheap EBay road and found it a great way to learn to build lights, but severely lacking for most practical purposes. My initial chinese LED builds were such shit that I never took pics or posted about them. Doesn't take much searching to find all the folks that tried them and then moved on to Cree or Philips COBs.
 

Dion

Active member
something like this:



but is innacurate and inprecise...
by the way, not all par meter are accurate either.


the best par meter is a plant, you put it under the light and observe

you will find 3 possibilities
1) too much light
2)too little light
3) just right-note that just right will be the widest range
 

jikko77

Active member
t-slot framing eu

t-slot framing eu

no idea where to post, maybe can be usefull to other EU diy'er looking for a modular frame, similar to the 80/20.
it's sold in italy and looks like is ship world wide. for sure they ship in eu.

http://www.openbuildsitalia.com/

something else to add in whishlist sigh ...
 

ballsack

New member
I have a bench variable power supply 0-20 volts. 10 amp max.
Could this be used to with those cheap ebay step up drivers. I could save a a lot on my led power by doing this. I want to use 6 100 watt cheap ebay cobs.
 

Arf

Member
I was wondering if someone could tell me what current a CXB3590 should be pulling at 33volts?

The heatsink is warm to touch, but by no means hot, I don't have a way to measure Tj.

The Cree data sheet for the CXB3590 series, only charts down to 33.5v

I have one in parallel with some other Cree LEDs and it's reading 880mA at 33v on my multimeter, it's on a dimmer driver so I have no absolute reference point. I am a bit worried that my multimeter doesn't like the waveform from the dimmer, and I don't have an oscilloscope.
 

CannaBruh

Member
@32.5V seen at the pads of the 3590 I'm seeing ~470mA both on Fluke8050A and the digital display of my bench supply. @33V it's nearly 700mA

*this was cold, with rising temps that might climb.
 

Arf

Member
@32.5V seen at the pads of the 3590 I'm seeing ~470mA both on Fluke8050A and the digital display of my bench supply. @33V it's nearly 700mA

*this was cold, with rising temps that might climb.

If I look at the CXB3590 data sheet and extrapolate the lines, it should be 650mA at 33v on the Tc = 55C curve, so 700mA might be right depending on the cooling.

I think the drivers might be using a PWM signal instead of steady DC and making my multimeter vomit the wrong data.

I guess the easiest thing for me to do is plug a bench supply into the COB, set the volts and current to say 30W, measure the light output with a Lux meter, then stick it back into the driver and set the dimmer knob to the same Lux reading.
 

Dion

Active member
I was wondering if someone could tell me what current a CXB3590 should be pulling at 33volts?

The heatsink is warm to touch, but by no means hot, I don't have a way to measure Tj.

The Cree data sheet for the CXB3590 series, only charts down to 33.5v

I have one in parallel with some other Cree LEDs and it's reading 880mA at 33v on my multimeter, it's on a dimmer driver so I have no absolute reference point. I am a bit worried that my multimeter doesn't like the waveform from the dimmer, and I don't have an oscilloscope.

http://pct.cree.com/dt/index.html

unfortunately not much info for volts in the 30 range, however because they way less work with voltage id assume that the current would be almost 0 before it got that low, they r strings in parallel within the cob so for a total voltage has to be divied up between them all,sorry i can't remember now how the dies r set up in a cxb3590, u will have to google it(i think about 40-60 leds in there)

@32.5V seen at the pads of the 3590 I'm seeing ~470mA both on Fluke8050A and the digital display of my bench supply. @33V it's nearly 700mA

*this was cold, with rising temps that might climb.


nice work, so they do drop to 30v at 470mA?


If I look at the CXB3590 data sheet and extrapolate the lines, it should be 650mA at 33v on the Tc = 55C curve, so 700mA might be right depending on the cooling.

I think the drivers might be using a PWM signal instead of steady DC and making my multimeter vomit the wrong data.

I guess the easiest thing for me to do is plug a bench supply into the COB, set the volts and current to say 30W, measure the light output with a Lux meter, then stick it back into the driver and set the dimmer knob to the same Lux reading.


why does it matter about the volts? what r u trying to achieve?
 
Top