Watercooled Smart IC COB LED Build

[MMJ Dreaming 99]

I do not use CoBs. I use strips of red, white, and blue discrete LEDs with dimmers on each string of 16 or 21 LEDs.
LEDs, especially red, are very temperature sensitive. The cooler they run the more photon flux produced per watt of electricity.
The heatsink is also the frame of the fixture. The copper pipe's cost is less than an aluminum heatsink.

May I ask where you get those strips? Do they provide the horsepower to match HPS in flower?

 

[ssj4jonathan]

Linear integrator is not a thing. Regulator is a thing.... Thermoregulating is not a thing... [130C]That is too hot for any chip

I copied the term linear intergrator from the video posted by DiodeGoneWild. To date he is the only expert on these things, so whatever he had to say, I will back up... Watch his video for more information about these linear driver chips, I linked it on page 1.
BP5132H chips are termed "smart" because they are capable of thermal regulating the current when things get too hot. I didn't pull this idea out of a hat, it's coming straight from the horses mouth. Also straight from Brightpower's data sheet, it states thermal regulation test conditions were measured at 130C.

I have a feeling NoFucks2Give is casually skimming through the LED threads, looking for blarring inconsistencies and chances to show off his works. Anyways, are you still willing to test the light output on these things. I'm going to be buying another batch soon. Let me know

 

[CobKits]

A 1° C rise in temperature reduces the life span 5%. So a 10° C rise will reduce the life of a semiconductor 50%.

so what youre saying is that the chips we use that are LM80 tested to 50000 hours at 85C Tc will outlive us when operated at the typical 45-55C Tc?

temp is a factor but lifespan isnt really even on the map of what most people should be thinking about when building a light

 

[ttystikk]

so what youre saying is that the chips we use that are LM80 tested to 50000 hours at 85C Tc will outlive us when operated at the typical 45-55C Tc?

temp is a factor but lifespan isnt really even on the map of what most people should be thinking about when building a light

That's how I read those charts too, that lifetime would be extremely long when run at the temps we usually achieve.

I'm looking for 8-10 years from my chips and drivers, personally. That's 87k hours if I run them 24/7 the entire time.

 

[CobKits]

in reality more efficient chips will be <$1 in that time and you'll still have your waterblocks and drivers

 

[ttystikk]

in reality more efficient chips will be <$1 in that time and you'll still have your waterblocks and drivers

Right, so when it makes sense to upgrade I most certainly will.

 

[NoFucks2Give]

have a feeling NoFucks2Give is casually skimming through the LED threads, looking for blarring inconsistencies and chances to show off his works

What?????????? You think wrong!!! I can stop viewing your thread if you like.

Translated to English: Linear LED Driver with Overtemperature Protection
In the video when he said "regulator" it sounded like he said integrator. Regulator can be used in place of driver. This chip uses a linear regulator configured as a constant current driver.
When the driver gets too hot, it dims the LEDs until the driver's temperature drops back down within spec.
That is not normal operation, it is protecting the chip from burning up. The thermal shut down triggers at 130° C.

In the video he said the LEDs operate at 222V.
In that case you have seventy four 40mA LEDs each with a forward voltage of about 3V in series, giving the ~222V forward voltage of the string of 74 LEDs.
A linear regulator is very efficient when the input voltage is very close to the LED forward voltage. The loss is (Vin - Vf) x current plus the efficacy of the LEDs.

 

[NoFucks2Give]

May I ask where you get those strips? Do they provide the horsepower to match HPS in flower?

I design the PCB and have a PCB house fab them for me. The strips are just strings of LEDs. I connect a separate driver with dimmer to each string.
You can put 64 white (16 x 4) or blue LEDs which can be driven at 1-2 amp.
Or 63 (21 x 3) red, amber, or orange which can be driven at 1 amp.
The LED footprints accommodates OSRAM Olsen SSL and Cree XLamp

It would depend upon how many strips and how many LEDs were put on the boards.

You can put just 16 or 21 LEDs per PCB.

Matching an HPS 1000W can easily be done with 196 LEDs or less. 3 strips min. 12 strips max.

The strips are only 0.35" wide. Just enough to mount the LEDs and screw it to a heatsink.

 

[NoFucks2Give]

so what youre saying is that the chips

That is the rule of thumb for integrated circuits, not LEDs. Remember the days when PCs would last maybe 3 years and now they last forever. That's because they ran much hotter back then.

I mentioned that because he was saying something about the driver chips kicking in and doing their "thermal regulating" the current when it reaches 130°C where it's the over temperature protection circuit kicking in. 130° C is too hot for any silicon integrated circuit.

 

[NoFucks2Give]

I'd very much like to hear your thoughts on the design.

I'm not sure you do. I am very critical.

I think the CoBs should be evenly spaced not doubled up. You would get much better uniformity with even spacing.
I do not like optics. First off they have a max transmittance of 0.9 meaning a 10% loss.
I think you get better uniformity without optics allowing you to get closer to the canopy which more than makes up for the wall loss.

I do not know what your thermal design looks like. What is between the plate and the water?

Have you ever used a small chiller? Like ones that sell for around $300-$400. 5 ton sounds like enough to air condition a large house. Or a hell of a lot of LEDs.

 

[ttystikk]

I'm not sure you do. I am very critical.

I think the CoBs should be evenly spaced not doubled up. You would get much better uniformity with even spacing.
I do not like optics. First off they have a max transmittance of 0.9 meaning a 10% loss.
I think you get better uniformity without optics allowing you to get closer to the canopy which more than makes up for the wall loss.

I do not know what your thermal design looks like. What is between the plate and the water?

Have you ever used a small chiller? Like ones that sell for around $300-$400. 5 ton sounds like enough to air condition a large house. Or a hell of a lot of LEDs.

I'm always interested in constructive criticism.
1. The paired COB chips allow for placing the driver between them, making for a thinner fixture. Most of my modules were built like that. Also, paired chips allow for the modules to more evenly light the 2' x 3' target beneath, which then allows more mounting options and more even illumination when running many of these modules end to end in a larger array;


2. These are glass optics, so transmittance is closer to 96%. Less light is wasted to the sides and more hits the target, more than making up for losses. Finally, they protect the chip from damage from rubbing, overspray, etc in real world use.

3. The chiller in question also controlled the environment in the growing space, the temperatures of the RDWC system and the rest of the house, so it was well utilized. Larger single chiller units are far more efficient- even running at light loads- than lots of small ones.

Yes, there were a lot of modules lol

 

[GrnMonStr]

@NoFucks2Give: Where do you get the "strings of LEDs" from? I am thinking about doing a LED light build I am only looking for about 250W max and I have low head room, so I really don't want high intensity.

 

[GrnMonStr]

@ttystikk: That is a awesome DIY COB set up!

 

[NoFucks2Give]

2' x 3' target beneath

Why are they vertical?

They give you a serious look. What is the temperature of water and the CoB?

They look like a good balance of cost and effective. Very nice. Any idea on approximate parts cost for just the cooling?

 

[ttystikk]

Why are they vertical?

They give you a serious look. What is the temperature of water and the CoB?

They look like a good balance of cost and effective. Very nice. Any idea on approximate parts cost for just the cooling?

Vertical to maximize yield vs floorspace. Allows for aisles for tending the plants.


Water temperatures between 55-60F, Tj was between 60-65F. 3/16" extruded aluminum between the chip and constantly flowing water, actively scavenging heat.

The chiller then took this heat and rejected it into a hot water circuit that drove heating and dehuey of the growing space plus home heating and domestic hot water.

The modules cost about $400 each altogether, the chiller was in the neighborhood of $10k- a cost offset by the fact that the house needed Central AC in summer and the gas furnace for heating was very inefficient- and broken.

The point of the facility was a small scale model of a commercial facility to test designs under real world conditions using a fully scalable approach.

 

[NoFucks2Give]

Where do you get the "strings of LEDs" from? I

I solder them. I have been doing low volume. I moved out of a house with a electronics manufacturing line. I left it behind figuring I would never do manufacturing again.

I moved over a year ago. I am renting a house and I may have to move again soon. I need to buy another pick and place robot to put the LEDs on the board. But I need some stability.

For now I use a syringe to put the solder on the pads, drop the LEDs on with a tweezers, and put the boards in my kitchen oven to solder.

This is my garage March 4, 2016 at about 2:00am. The moving van was packed. This pic was taken minutes before I left my house for the last time. This is what was left behind plus a bedroom that was converted to a workroom has a lot of stuff in it too. Like a manual pick and place machine.

 

[ttystikk]

I solder them. I have been doing low volume. I moved out of a house with a electronics manufacturing line. I left it behind figuring I would never do manufacturing again.

I moved over a year ago. I am renting a house and I may have to move again soon. I need to buy another pick and place robot to put the LEDs on the board. But I need some stability.

For now I use a syringe to put the solder on the pads, drop the LEDs on with a tweezers, and put the boards in my kitchen oven to solder.

This is my garage March 4, 2016 at about 2:00am. The moving van was packed. This pic was taken minutes before I left my house for the last time. This is what was left behind plus a bedroom that was converted to a workroom has a lot of stuff in it too. Like a manual pick and place machine.

View attachment 3973659

I don't think I could bear to leave equipment like that behind...

 

[NoFucks2Give]

mall scale model of a commercial facility to test designs under real world conditions using a fully scalable approach.

I like that. I like that a lot. I never thought to do it like that. I always thought there was too much wasted cubic space.

The distance between plants and lights is at a premium. How close can you get the lights to the plants?

 

[GrnMonStr]

I like the vertical SCOG! I don't think I have seen anything like that. wow......

 

[NoFucks2Give]

don't think I could bear to leave equipment like that behind...

It's worse than that. It's a $300,000+ house. I had $800,000 in code enforcement fines. I have this issue with authority. I sold it for $15,000 and walked away. And walked away happy. I really did not give a fuck. The fines were from a few minor things, motorcycle trailer, left over roofing tiles stacked next to house, and some over grown trees. Each one of those $150 per day. For years. I owed about $100,000 on the mortgage. My daughter thought I did not leave enough behind. When she moves, she leaves almost everything behind.

. 3/16" extruded aluminum

Too thick. Thinner is better. In the thermal flux formula the thickness in this case is the length. The area is the side of the extrusion.

Also if you put some obstructions inside to give the water flow some turbulence, big plus in convective heat transfer.