DIY MakersLED Cree CXA2530 build grow light

[malendro]

Heatsink: MakersLED 24"
Controller: MakersController
Driver boards: Makers 5Up
Drivers: LDD-1000H
Chips: Cree CXA-2530. 2700k 85CRI
Chip Holders: 180720-0002 By Molex
Thermal paste: ArticSilver5
Additional Cooling: 92m Delta ASB0912H 52CFM

Goals:
370W maximum output
Professional Clean install
No drilling on the shiny metal

Lets get things started with the LEDPorn

Space LED holders evenly.
Route all internal wires.


Below is the input/output side:
PC fan Molex style connector and pins were to route power.
Controller Signal was routed using something like the 3.5mm adapter shown.

The circular hole was drilled with lots of care and precision for proper alignment so as to limit stresses on the board's components, and the square hole was carved with an Xacto knife for a tight pressure fit. The power input for all the fans is also visible in this picture, it's the white connector on the lower left corner.






Here is the complete mockup.

 

[malendro]

Part 2

Part 2

And now there's my stick o' chips




The backside where all the cooling takes place.
Power to the fans is routed externally, and are not controlled by the 5up driver boards.




And finally the business end:





Last the power supply:

do yourself a favor and buy a quality power supply.

get something like this.



I made the mistake of buying a china one like shown below first:
the first they sent was DOA and after opening it up the stuff inside just looked extremely shoddy.

 

[malendro]

Results

Results

In action:

 

[malendro]

Update- dead chips

Update- dead chips

A second chip now seems dead. I'm taking steps to safeguard the life on this baby.

I ordered 12g of artic silver 5 , 4 times what I used to originally put the light together. I'm planning of being much more generous with it on every single chip once the 2 replacements are in.

I'm not overdriving this things they are capable of handling 1.6A @37V.

My drivers are able to put out only 1000mA.
I originally had the drivers at 80% by PWM.
Now they are preemptively set to 60% until I apply the extra silver.

I thought I had been thorough enough last time, I applied a couple of drops of the stuff and slathered it everywhere on the contact patch with a little spatula that came with a similar heat transfer kit, but I tried to do so in a very thin layer. Supposedly the less you need the better off you are.

I have not done the math but I don't believe that the possible xtra thickness of the artic silver that you can stick in there will have a significantly detrimental effect. It will probably be an extra 1C difference, compared to those that might already have enough.

If this keeps my chips from dying I will be so happy.

 

[luvdemshrooms]

Can you see where the LED failed? Did it leave a dark spot on the LED? Is the dark spot over the flat contact portion of the heat sink or is it over one of the slots?

Have you measured the temp of your LED at the spot provided by Cree?

While the LED backplate is 23.85mm and the die area is 21mm, the flats between the heat sink slots is 14.35mm. This leaves a portion of the die area above the slots. As there is no contact where the slots are I'm betting your failure point on the LED is above the heat sink slots.

These puppies run hot and need either a full contact at the back or greatly reduced power. When running the CXA's you definitely need to measure the temps at the location Cree provides.

I personally would not use that heat sink. It was designed with smaller LED dies in mind.

 

[positivity]

Sucks the chips are burning out.

The setup is nice but your surely taxing the makers light duty heatsink. I've run a similar amount of power on a 24" heatsink usa t slot. It's quite a bit more aluminum than the makers.

Think you could make it work well though.

Add a sheet of 1/4" aluminum. Use some cheap thermal grease and screw it down in a few places so there's good contact. It should end up giving you a nice flat plate to work with that will work perfectly with 1/4" screws.

I'd space the leds out as much as possible to spread the heat out. Get a bit of metal between the leds so the heat gets a chance to dissipate before running into another led.

That's what I'd do....

Actually planning on doing it with my t slot so I don't get those empty spaces under the led.

Something like this but maybe skinnier to fit between the channels......http://www.ebay.com/itm/1-4-X-6-ALU...627?pt=LH_DefaultDomain_0&hash=item53fb530163

I'd also try prolimatech pk3 thermal paste under the leds. It performs a little better than artic silver 5 and the consistency is easier to work with.

Some ideas...hope ya get it working. Could always just split the design onto two heatsinks..

 

[malendro]

To Luvthemshrooms:

Can't see where the LEDs failed , there are no dark spots, or any signal of discoloration on or around the die.

I had not attempted to measure temperatures yet, but I just purchased an IR thermometer to better assess the situation. A detailed update with temperature profiles will be posted during the weekend.

When you say I have to measure the temp at the location that Cree provides? I guess you are talking about the gold dot near the die. You know if I can use non-contact iR to measure at said location, or do I have to use a thermocouple?

Before I decided to use this heatsink, when questioned, the heatsink manufacturer did say that their heatsink should be able to handle the power, since their customers use this heatsink with the Lumia chips that can dissipate 70W each, even if they have a large footprint that leaves non contact areas thanks to the T-slot construction.

 

[luvdemshrooms]

Yes, the gold dot. I have no experience with non-contact measurements of LED's.

I suspect you'll find the LED needs to have full contact (at least for the die area). I missed a small spot when applying thermal compound to my CXA3590 and that spot is exactly where it burned out, except mine was visible. Probably as I run 113 watts through them.

Another possibility is to stagger the LED's so that all the heat from the 10 LED's isn't concentrated onto the one flat.

Your current configuration:
0 0 0 0 0 0 0 0 0 0

My suggested configuration:
...0......0......0
0.......0......0......0
......0......0......0

Edit: Argh, the site kept reformatting my layout to remove the spaces it didn't like. Kindly disregard the periods.

Additional edit: Which LED failed in the string?

 

[cocogrower]

Very neat and nice construction! I'm really sorry that you're running into problems..

I think the advise you've got so far is great! But i think you maybe should consider buying a new heatsink to be on the safe side or driving them very low! 350ish - at least if the heatsink feels hot to the touch. Not saying that lovethemshrooms solution won't work, i'm sure it will, especially if your sink has some weight?

I Just hooked up a few 2530's, but each one on one smaller (270 gr) heat sink, together with 4 3-watters. If i run them on 350 mA, the temp on the measuring spot gets to 50 degrees C. Driven on 700 mA the heatsinks feels pretty warm, and the measuring spot goes up to 80 C. My measurements are so far based on passive cooling. But my point is that you can expect your cob to be at least 20 C. hotter than the sink, probably significantly more if you have a poor connection between heatsink and cob..

For some reason i cant enlarge some of your photos.. Is your soldering tidy? If strand or solder is touching the sink it's not good... You can use a multimeter to check this if you are in doubt..

 

[malendro]

What do you use to measure temperature at the dot?

 

[cocogrower]

I have a multimeter: Multimeter UNI-T UT132E. It measures temps with a tiny little metal ball in the end of a wire. Does it feel uncomfortable to keep your fingers on heatsink for a few secs? it did for me when my temps were hitting 80 on the chip.

 

[luvdemshrooms]

What do you use to measure temperature at the dot?

I use thermocouples. Arctic (of Arctic Silver fame) makes Arctic Alumina Thermal Adhesive. It's a two part compound that will permanently attach the thermal couple to the test point on your LED. I left a thermal couple attached to he LED in each of the lights I made. This way I can easily monitor the difference over time and after any changes to my cooling system.

I bought a digital thermometer from Amazon. It came with two thermocouples. More thermocouples are there available there for a reasonable price.

Edit: Cree uses a temp temp of 85c for their published binning data on their CXA line. While you can run them hotter, lifespan will be reduced. With the thermalcouple attached start your testing at a lower current. Mount your fixture in the same way it will be used (room size, any fans that run, door closed? Your situation will vary). Allow the temperatures to stabilize and when they have slowly dial up your current until you reach the desired temps. To be safe I added a termostatic switch to my heat sink which shuts off the power to the driver if the temperature rises too high. I used a 75c model. You can find them at Digi-Key. Enter THERMOSTAT 75 DEG C NC FASTON into the search field. Mount the thermostatic switch as close as possible to the LED/heat sink mating point. It's cheap insurance. Use a thermostatic switch rated for 10c higher than your desired max LED temp, so setup your driver current first.

 

[cocogrower]

I use thermocouples. Arctic (of Arctic Silver fame) makes Arctic Alumina Thermal Adhesive. It's a two part compound that will permanently attach the thermal couple to the test point on your LED. I left a thermal couple attached to he LED in each of the lights I made. This way I can easily monitor the difference over time and after any changes to my cooling system.

I bought a digital thermometer from Amazon. It came with two thermocouples. More thermocouples are there available there for a reasonable price.

Edit: Cree uses a temp temp of 85c for their published binning data on their CXA line. While you can run them hotter, lifespan will be reduced. With the thermalcouple attached start your testing at a lower current. Mount your fixture in the same way it will be used (room size, any fans that run, door closed? Your situation will vary). Allow the temperatures to stabilize and when they have slowly dial up your current until you reach the desired temps. To be safe I added a termostatic switch to my heat sink which shuts off the power to the driver if the temperature rises too high. I used a 75c model. You can find them at Digi-Key. Enter THERMOSTAT 75 DEG C NC FASTON into the search field. Mount the thermostatic switch as close as possible to the LED/heat sink mating point. It's cheap insurance. Use a thermostatic switch rated for 10c higher than your desired max LED temp, so setup your driver current first.

Thats the route i should take as well, seems easy and reliable! Thought about installing a pmw regulator with thermostat function to dim down or shut down leds in case of fan failure etc. But this seems way easier! thanks for the tip!

 

[luvdemshrooms]

No problem. As LED's for my light cost $72 each, preserving them was towards the top of my priority list. A simple substitution of whatever temp you desire (in 5° increments) during the search will allow for fine tuning of the shut-off point.

 

[luvdemshrooms]

It occurred to me that I left out an important bit. In order to ensure longevity of your LED's, try not to run them above 70% of their max current rating.

 

[malendro]

Good News Everyone!!

Good News Everyone!!





TL;DR: My LEDs are back in the photon making business! It was the drivers.
I noticed after brand new chip would not work when pressed against the terminals of the solderless chip holder.


Keep reading if you want to know about temperature profiles:

Today I received all the supplies needed to diagnose or fix my problem. 2 new cxa2530 came in, along with 1 new LDD-1000H driver, and the best IR thermometer 30$ can buy ( it's actually very nice).

I began with playing with the thermometer and it's setting for Emissivity. The manual recommended 0.3 for oxidized aluminum (the annealing on the sink), gold 0.1 (the dot on the chip), and 0.95 to almost everything else.

After trying all those Emissivity settings: there was too much inconsistency using low ε numbers. I decided to leave it at 0.95, and made sure the measurements where accurate on all materials by turning the light off and allowing everything to equalize. Everything read about the same 22C, both without, or through the plexiglass that comes with the fixture (and I'm pretty sure it's not just reading the surface of the plexiglass).

I also decided that I prefer to measure at the emitting surface of the LED. It is the most conservative measure by far, and measuring the heatsink is not telling me much because it doesn't even feel hot to the touch.

This is the temperature profile running at 0.6A

Ambient temp: 22C
Air trapped by plexiglass: 30C (I put a small thermometer in)
Temperature of the 10 LEDsin C)
62-57-29-56-58-60-59-58-60-59
Highest temperature of drivers: 56C
This measurements where taken by waiting only 3 seconds and recording the maximum observed.
I also touched the aluminum between two active LEDs (about 1cm gap) and it was pretty cold to the touch, didn't even feel like it was over 40C

Note that position 3 is currently dead, since I only got one replacement driver. And position 2 is the one I brought back to life today.

For comparison, and while we wait for my light to equalize at 0.8A, I have these figures for all the other lights I have:

1/2 kessil h150: 75C
Appollo 384w: 87C ( by the way I just noticed 1/4 of this light is dead)
Haight solid state PPF800/400: 40C max on LEDs, 50C on drivers
Kessil h350: 105C (Damn those DiCon Engines are HOT)


So While pushing 0.8A I'm reading

Air trapped by plexiglass: 35C
Temperature of the 10 LEDsin C)
80-93-43-103-93-91-93-94-100-92

Unlike the previous measurements, these were taken by hovering the meter around the diode and recording the maximum temp observed after over 30s, if I had done it as before the readings would have been 10-20C lower.

I'm going to try applying some more of the Artic Silver I'm supposed to receive soon and dialing the power to the point where I get 85C max at any spot.

I want to thank everybody for their kind support and ideas.

 

[cocogrower]

Good that every is working again! I don't have any idea about how those IR thermometers should be used, but hope that you found a way that works with your conditions! Without having to much theoretical knowledge or experience i have to say that your readings sounds a bit high considering the low temps of your heat sink. Maybe someone else can comment on this? Adding a sheet of copper or aluminium udner your leds as suggested would probably help a lot! Do you have any theory about why one of the cobs are so much cooler than the others? Very cool and clean design! with those temps a bit lowered it's a killer lamp!

 

[malendro]

Temperature test with thermocouple

Temperature test with thermocouple

So I bought myself a multimeter with thermocouple today, and proceeded to test the temp profiles. Here are the results.

Measurements were taken with thermocouple by placing it on the gold dot, and waiting for the temperature to stabilize for 1 minute, and recording the highest value observed, even if it was not a stable value, after which the value might have dropped.

Running 0.6A
Temperatures for the 10 LEDs:
47-47-31-51-47-47-45-47-47-49
The third position (31) is the LED that is still off.

Running 0.8A
Drivers: 66C (measured with non contact iR, owed this figure from yesteday)
Temperatures for the 10 LEDs:
70-64-35-72-65-60-63-64-61-64

My hypothesis regarding the LEDs with lower temps: they probably have a better layer of thermal paste than the others. I'm inclined to say they probably have a bit more.

 

[cocogrower]

Great to hear that now everything is under control, congrats! Happy growing!