DiY LEDs - How to Power Them

[T-Time]

I'll come back to Your data in the morning, when I'll have a fresh mind

I'm about 2 weeks from ordering the parts so its still planning phase for me. I have a bit of time then to wait for Your calculations

For heatsinks I was planning for either 120mm or 133mm passive heatsinks from jerry at kingbrite.
Water cooling sounds nice but it adds unnecessary complication to the whole system. I even have a pump (different project, dwc down the road) and wouldnt be afraid to do that but I rather add a bigger heatsink.

If You're willing to show some photos of Your watercooled system, that would be cool

 

[NoFucks2Give]

what drivers do i need for 12 vero29 gen 7 36V (class d)?. looking for around 700-900Watts

Two HLG-480H-C2100
will push over 2.1 Amp up to 280V
Current adjustable from 1 -2.1 Amp for Type A, or Type B Dimmable from 2.1 Amp down to near zero
280V / 36 = 7 CoB @ 40V.
Two strings of six Vero 29 @ 36V.

 

[NoFucks2Give]

a light with LDD drivers and board (3 or 4 channels)

What do you need help with? Connecting the LDD to the µController? Programming sunrise to sunset?

 

[NoFucks2Give]

Your watercooled system, that would be cool

In my line of work I use PC boards with mono LEDs. The liquid cooling I developed was for PCBs.

Then I was running thermal experiments and used CoBs to generate heat. That gave me an idea for cooling CoBs.

I like to put ideas on paper first. By paper in this case, I mean SVG graphics.

Basically the CoB's are mounted on a 0.062" thick bar of copper. The copper bar is in contact with a 1/2" water pipe for its entire length.

Water pipe costs about $0.80 per foot.
The cooper bar is under $2.00 per foot when purchased in single quantities.
I expect if I buy by the Ton, it will cost a lot less. Like one tenth, where it would be $0.20 per foot.

I originally was soldering the bar to the pipe. That was an ordeal. Closing the gap between the pipe and the bar was a problem.
I was going to make a die to hold the pieces in place and squeeze them in a vise.
Then I used some C-Clamps in a thermal experiment and this idea was born. The C-Clamps worked better than solder.
The screws and bracket replaced the C-Clamps.




Looking at the formula for heat transfer



q is the the thermal flux
k is the conductivity of copper (400)
A is the copper bar's cross sectional area .

In this case the heat is flowing through the copper bar so the Length is 0.062"
The Area is 1" x 15" (the 15" between CoBs)
Essentially, zero thermal resistance.
When I would touch the bar exactly opposite where the CoB is mounted I could not feel any heat what so ever with a water temp of 25°C with a 60 watt CoB.

This was how I clamped the pipe and bar together. I had a hard time getting the clamps to no fall off. That's why I had the extra copper between the clamp and bar.
I have since improved the clamping .

This was setup in my garage because I ran out of room in my lab.




The above pic was cropped from the below.
In this photo below, the case temperature of the CoB was 45 as seen on the meter. Now it's the same temp as the water.





I've tried all kinds of water reservoir setups but found this simple dual pump (in case one fails) works fine and is very inexpensive.
My first on is on the floor using 4" PVC. It was a monstrosity. I may still use the 4" PVC as the reservoir because I want a sealed system.
In this experiment I was using a 5 gallon bucket.
I may add a UV light to kill any algae or bio-organisms. Now I am using chlorine.







I may add something like this to cool the water.
Firsts I want to try an ice maker.
Ice water takes the case temperature below 25°C.
With a passive heatsink 100°C would be doing good.



I ordered this milling machine Friday to make the brackets and drill and tap the holes:

 

[T-Time]

As much as I like Your idea theres a one problem with it. Heat from the lights is never lost even it that kind of system. Its just shifted somewhere else. So running it inside my small tent wouldn make any sense. I would have to take the heat of out of the room and I cant do that. There's a noise restrictions as well unfortunately.

 

[mahiluana]

Water cooling sounds nice but it adds unnecessary complication to the whole system.

Not at all - with my watercooled * coolmac - system * i`m able to rewin 60-65% of the inverted electr. power in form of hot water with an 80L heatexchanger (EUR 160,-). I use it in the bathroom and kitchen.

With ~ 330W lamp / ~15h a day - I was able to stop the use of oil or gas waterheaters and isave ~ 500 EUROs / year in my heating bill. ( which is more than the electr. bill of the lamp)

The system is of course watertight, no noise and can take out up to 80% of the lamps heat from your grow room.

So this is not only world record in energy-efficiency(~85%)

The second record is - the lowest Tj of all comercial led lights. (20°-40°C)

The system can cool up to 2000W lamppower with only 3W waterpump.

LIGHT - HEAT COUPLING is three times more efficient than any other conventional, aircooled led light.

So what problems do you see ???

for more info have a lock to my content or google: lumen-laden

 

[NoFucks2Give]

Heat from the lights is never lost

Using any heatsink has the same issue. Except with the liquid cooling you have the option to remove the heat from the tent and use it or lose it where ever. With a passive heatsink you do not have that option.

The major design criteria is the 100% copper, very low thermal resistance path from the LED's thermal pad to the liquid. This makes the LED run cooler, which gives a boost in photon flux.

 

[NoFucks2Give]

So what problems do you see ???

I see no problems, very nice. I like your stuff.

What I like best is how close the fixture is to the canopy. I don't think enough understand how important that photon travel distance is to PPFD.

I like the eye bolts and paint brushes too.

I use two 3W pumps in case one fails. I could have the second kick in when the first fails.

You are using aluminum? Copper has twice the conductivity. I like copper. Except it's more difficult to machine, drill, and tap.
Then there is the issue with CoBs and their aluminum thermal pads.

NEVER MIND, I just saw another post of yours where the thermal pad is in contact directly with the liquid. That's ingenious. I had to think about the thermal dynamics for a second or two. I think it beats going through a heat spreader. Although the thermal resistance of a TIM if applied properly, is nominal due to the "Length" (thickness) of the material. Where Length refers to the L in the denominator of the heat transfer equation.

But you do lose some cross sectional area of the thermal pad where you mount and seal. When that gets factored in it may get interesting. It may depend on the thermal conductivity of the sealant. Still an ingenious design.

I work with mono LEDs where the thermal pad is soldered to the PCB copper. I need to adjust the individual wavelengths for horticulture research applications.

 

[mahiluana]

You are using aluminum?

Yes - copper is also much more expensive and heavy - as you can see and imagine in the pic
I could seal the optimized chip platines even on a plastic tube - no need of any conductiv materials - this patented heattransfer happens as close as possible to the place where the heat is produced.


In winter i like to use the lamp as an radiator during the night - that`s why i still use alu-tubes

 

[NoFucks2Give]

much more expensive and heavy

Copper is about twice as expensive aluminum for the same size rectangular bar.

no need of any conductive materials

I'm still struggling with that. I do not know the answer. Got to brush up on my thermal dynamics.

Where I am stuck is regarding the low conductivity of water (0.6) vs. copper 397, and aluminum 238 W/m ∙°C

I can find very little regarding the forced convection of water over a flat surface. The surface area in contact with the water is a major consideration for forced convection heat transfer. The emissivity is a major convective factor. Polished aluminum has low emissivity vs anodized aluminum.

The question is if you were to attach a thin plate with much more surface area to the CoB thermal pad you would be increasing the convective surface. Does the thermal resistance of the added plate get canceled out by then increased surface area?

The transfer of heat between aluminum and water is relatively slow compared to aluminum and copper or aluminum and aluminum.

So I was thinking that possibly conduction from the thermal pad by aluminum or copper would create greater thermal flux than aluminum and water. Although liquid forced convection is a very efficient method of heat transfer, at least when the liquid is flowing in a pipe.

Because the flat surface is offset from the inner surface of the rectangular tube, there is less turbulent flow. Possibly you could put a baffle to direct the flow directly on the LED surface creating lost of turbulence. For example if you made the cutout by cutting 3 of the edges of the square cutout then bent the slug I think it may greatly improve the thermal transfer. More so than the roughing the thermal pad surface.

I don't know. I understand it's small stuff as it works well now. Over the years one of the most helpful things I learned is the small details add up in the end and combined can make a significant contribution. Kinda like a restaurant. What is the difference between a shit hole and a very successful store? The small details, but lots of them combined.



Regarding your rough surface I found this link https://people.csail.mit.edu/jaffer/SimRoof/Convection/#Abstract

It about roofing, but its gets into convection very deeply. I just skipped over the math and read a few things. Like: "the Reynolds numbers for this size smooth plate are low enough that the boundary layer would not progress to turbulence."

Turbulence is good for convection. Now I'm thinking about roughing up the surface of my water pipe. I had seen something a long time ago where it is a big help.

It's not that I think your design is deficient, I like it a lot, I just wonder about stuff and try to help if I can.

We should run some experiments comparing mine and yours.

 

[mahiluana]

run some experiments

what do you want to compare ?

 

[GBAUTO]

Automotive radiators use what they call 'turbulators' on the inner surface of the coolant tubes to maximize heat transfer-wonder if anything like that is available?

 

[NoFucks2Give]

what do you want to compare ?

We could start with whether the surface being in direct contact with the water is better than expanding the surface area with a thin plate.
Or we could just compete on the most efficient and cost effective thermal management design.
I am also interested in the heat exchanger to recover the heat from the water for reuse and cool the water at the same time. Or vice versa. I also use PV panels to power some LEDs. The heat reuse and the excess solar power have some common. The current common method of storing excess solar power is hot water. I'm thinking greenhouses in places like Canada and the Netherlands that have energy problems associated with with growing in the winter where a heat recovery would be commercially viable. I live in Florida and the water comes out of the tap warm so hot water would do me little good.

Thermal experiments have become almost a hobby for me. I am running thermal experiments all the time. I like to come up with a hypothesis and see if it pans out. My hypotheses usually have to do with cost reduction or improving efficiency.

The way I look at things is the impossible is only impossible until someone does the impossible. Which I have done only 1.5 times.

 

[NoFucks2Give]

wonder if anything like that is available?

Yes in auto parts stores and junk yards. I had read an article about a guy that went to auto salvage yards to get parts to be used in CPU water cooled heatsinks. But rather than the engine radiator he got transmission coolers and AC evaporators. I think transmission coolers use oil and that would be messy. But the AC evaporators sounded interesting. Small, efficient, and clean.

I use water pipe. I'm thinking some sort of honer to rough up the inner surface. Maybe something like a copper tubing brush that is used to clean the inside of a fitting before soldering on a long attachment in a drill. Like a rifle barrel cleaner on steroids. Or the tool the puts the rifling in the barrel. Or merge the rifling machine with the copper fitting brush.

 

[ANC]

Old Camry heater core, is bout 12cm and has a fan in front.

 

[mahiluana]

whether the surface being in direct contact with the water is better than expanding the surface area with a thin plate.

Imagine the way of the heat - outside of your chip.
Your heat meats your thermal grease - and even if it is a very good one - the heat conductivity compared to eg. copper is very low.

OK. - you learned to keep it very thin and you put it only on very flat and plane surfaces.
But all your heat meet on materials with heat resistance.
Thickness of these materials: glue, grease, aluminium- or copper-heatsinks influence
this heat resistance in a linear function -
until heattransfer happens ( aluminium, copper --> to the air - or water)

I only break down this wall with a drilled hole and - started to dig also on the other side.
The chip-platine itself is a thermal resistance and the area directly under the light emitting surface is the center of heat production.
Scatching the surface over there can double or trippple this area and bring the water closer to the fire

half a year ago I meassured a 50W cob @1500mA mounted on my coolmac without hole.
At 25°C watertemp. my powermeter showed 50W.

Same chip(scratched) and driver mounted on a hole with direct contact to water of 25°C
pushed my powermeter close to 55W.

As you know low Tj has great influence on light efficacy and longlife - but the greatest
benefit is to use the heat in the shower or elsewhere.
***********************************************************************************************

I live in Florida and the water comes out of the tap warm so hot water would do me little good.

I doubt a little bit, that there are no oil or gas waterheaters in whole Florida.
But US-citizens are known to consume 4 times more energy than eg. europeans.
That means in summer you switch on your aircondition and some take out the electric-heater because they catch cold feet.
Together with your orange-blond, stupid superleader you are in a turbo tripple lose-lose-lose situation. America - first in hell...

If you need a watercooler working without any additional energy -
have an eye on my cooling-tower. You need to vaporize ~1L / 600W of heat

here eg. you can find a heatexchanger:

http://www.ebay.de/itm/80-100-120-150-Kombi-Elektro-Boiler-Warmwasserspeicher-mit-Warmetauscher-Solar-/381284978091?var=&hash=item0

 

[wietefras]

Well, it does not work that way. That conversion is wrong.
I am going to guess your 4.64 is 1 Watt @ 555nm.

That conversion is not wrong, but you guessed wrong. That's the QER to go from PAR W to PPF.

 

[T-Time]

You are right , that was QER.
I've used downloaded COB LED calculator from one of the threads here and got compleatly different numbers.
PPF 1832
PPFD 788
so something was wrong.

 

[NoFucks2Give]

That's the QER to go from PAR W to PPF.

No it is NOT! Could you not follow the math? Again?
First off there is no CoB with 61% radiometric efficacy.
Additionally PAR is Quantum and counts photon particles. No such thing as PAR watts. Watts is radiometric and is a measurement
of Radiant Flux. PPFD, Photon Flux, is quantum measured in moles. Watts is neither PAR or a measurement of intensity, irradiance or radiance. Flux is just energy, not moving. Watts/steradian is radiometric intensity. Watts/meter² is radiometric irradiance, and Watts/meter²/steradian is radiometric radiance.

There is no conversion from radiance to irradiance. PPF is irradiance and PAR (PPFD) is radiance. They measure two different things. One is a spherical measurement of an isotropic light source. The other is a directional measurement of a flat area.

Commonly in horticulture Watts refers to radiometric radiance the same as µMole refers to PPFD photon radiance.
You can convert from Radiance to PPFD and you can convert radiance to luminance for a specific wavelength.

What @T-Time was attempting to do was converting from wall watts to PPFD. That can be done if you have the conversion factor for that specific CoB which takes in to consideration the radiance at each wavelength.

20 coba * 35.54W = 710.8 dissipation W
710.8 * 0.6122 efficiency = 435 PAR
435 * 0.9 wall losses = 392 PAR W
392 ÷ 1.5m2 = 261 PAR W / m2
261 * 4.64 = PPFD of 1214umol averaged

0.6122 efficiency is WRONG. A very efficient mono deep blue LED may have an efficacy of 60%.
4.64 is WRONG. 4.64 is the conversion factor for radiometric radiance to PAR at a wavelength of 555nm.

 

[PilouPilou]

That can be done if you have the conversion factor for that specific CoB which takes in to consideration the radiance at each wavelength.

It would be great if you or someone else can tell us the conversion factor for each kind of COB that grower use here... if it's possible of course!