DIY Bridgelux Vero 29 + 3.95" profile heatsink usa

[churchhaze]

Ingredients:
4x (
vero 29 3000k
3.950" profile cut to 4" (almost square)
)

Not everything is thought out yet, but without having physical items in front of me, design was taking too long.

My plan was to run 1 series string at 1.4A, but the 143V meanwell driver may not have the voltage to actually power all 4 cobs at 1.4A unless VF is sufficiently low..

Anyway, here's pictures.

 

[churchhaze]

Here's how tall these massive heatsinks are.



As per bridgelux's recommendations, I tapped M3 holes, and used crush washers on 0.5 pitch machine screws, just long enough to almost make it through the thick heatsink base.

 

[churchhaze]

And here's a test run powered by 4x9V batteries.

 

[churchhaze]

As you can see, as designed, it's a semi-passive sink. Thick, widely spaced fins. If in the future i decide to really crank these things up (2.1A or higher), I can put some 90mm fans on. I may eventually decide to use them that way if I decide brute force is more important than efficiency at some point.

TIM at the moment is shitty zinc oxide based zalman gpu cooler stuff. When I run in production, I will likely replace with something better.

 

[churchhaze]

I don't know about the photon flux, radiant flux, or luminous flux, but the 3000K 80CRI vero look really nice.

Blue renders just fine to me. I usually get the 2700k cree bulbs for home use, if that means anything.

It's hard for me to tell whether the higher luminous efficacy of the 4000k cobs are from efficiency or simply being more tuned to human vision. I doubt the actual efficiency (radiant power/electric power) of the 4000k is much higher than 3000k.

 

[Tazbud]

Good stuff, please keep it going, can you measure the heat somehow? So what will you drive the LED at?
That's one hefty heatsink!

 

[purplegrower02]

He already said it will be driven at 1.4a.

church, did you use the high voltage driver for ease or another reason? I was going to go this route with my cree's but decided on LV with more drivers mounted external.

Also how big is your driver when I looked them up they seemed to be the size of a hid ballast, is this true?

 

[SupraSPL]

Hey CH, very nice install work! I had that profile in the spreadsheet each 4" piece has 1513 in² of surface area. So for passive cooling they are good for about 14Watts each. So if you run the Vero29 at 1.4A that is about 53W and the heatsink be very hot. If you actively cool it you could fit a 92mm fan that really cranks some airflow. That should put you in a decent temp range, each heatsink would be good for about 40-50W.

You are correct about the Meanwell driver, it would be good for 3 Vero29s at 1.4A. If you dimmed it down to just under 700mA, it could run all 4.

PG, the meanwell driver is 228*68*38.8mm (L*W*H), sort of like the size of a brick but smaller.

 

[churchhaze]

Yes, I ordered a high voltage power supply for driving 4 in series for design simplicity, so that 4 current regulators weren't required.

The power supply ordered was the Mean Well HLG-185H-C1400A , but it will not arrive until Tuesday..

The major flaw in my design is that 143V/4 is only 35.75V. I don't think that will be enough forward voltage to run all 4 in series at 1.4A. I might need to use 3 vero 29s in series instead, then find something else at 1.4A to use the remaining voltage. I wish the voltage source was a few volts higher. For some reason, finding high voltage DC supplies is hard/expensive. I thought everyone wanted a high voltage, high current, zapping machine..

The reason my current drivers look so small is because it's 4x 9V batteries in series just to test the setup.

Really though, we sound like pussies when we call 150VDC "high voltage". I think of high voltage as 10-100kV.

He already said it will be driven at 1.4a.

church, did you use the high voltage driver for ease or another reason? I was going to go this route with my cree's but decided on LV with more drivers mounted external.

Also how big is your driver when I looked them up they seemed to be the size of a hid ballast, is this true?

 

[Tazbud]

I'll just say thanks Supra while your here, among a few notable pioneers iv'e just read and re read many of your posts. It's all in the re-reading ; )
I'd gathered that even these size heatsinks may be too small alone, given the size you used. Been thinking of putting smaller sinks in a boxed/ channel (y'know like one of those sinks either end and one in the middle with a fan/vent above) and drawing air out with one larger fan (ie inline fan, hood style)... Draw air in ends and out the middle sorta thing...
I have been thinking too many things though.. sativa madness

Cheers CH, go well!

 

[Positivity]

Yes, I ordered a high voltage power supply for driving 4 in series for design simplicity, so that 4 current regulators weren't required.

The power supply ordered was the Mean Well HLG-185H-C1400A , but it will not arrive until Tuesday..

The major flaw in my design is that 143V/4 is only 35.75V. I don't think that will be enough forward voltage to run all 4 in series at 1.4A. I might need to use 3 vero 29s in series instead, then find something else at 1.4A to use the remaining voltage. I wish the voltage source was a few volts higher.

The reason my current drivers look so small is because it's 4x 9V batteries in series just to test the setup.

Hey church..wouldn't adjusting the internal pot down to maybe 1.2a help fit the 4... Plus might run cooler at that current and sink combo.

I know looking at inventronics spec sheets as current drops voltage available goes up.

Wonder what the max wattage rating on that driver is also? I bet it's not over 200w...would probably need about 220w for the driver inefficiencys running 50w each led

 

[churchhaze]

Thanks for reply!

Your math makes sense, but my internal reality check just isn't accepting it. The 19W cree bulb I got from home depot is passively cooled with a white painted piece of crap heatsink with few and short fins. Somehow my brain rejects that this beast of a heatsink can only support 14W passively.

I guess a lot of it does come down to what is an acceptable temperature, because those cree bulbs are hot as hell. I'm sure if i put the leds from the 19W cree bulb onto these sinks, they'd run way cooler.

Another problem I have with the math is that in general, heatsinks with thinner, but more fins, while they do have a surface area advantage, each fin has a lower thermal conductivity, as conductivity is proportional to width and inversely proportional to length. My point is that thin finned, active, sinks have a very hard time spreading the heat to the top of the fins, while thick, passive, fins conduct heat to the top much better. The thin fins win when there's no lint and the blowers work, but it's not as simple as finding the surface area on each sink. Fins can also be thick or thin.

I'm very curious now what the temperatures on this sink/cob would be running at 1.4A for an hour with only a passive breeze.

Hey CH, very nice install work! I had that profile in the spreadsheet each 4" piece has 1513 in² of surface area. So for passive cooling they are good for about 14Watts each. So if you run the Vero29 at 1.4A that is about 53W and the heatsink be very hot. If you actively cool it you could fit a 92mm fan that really cranks some airflow. That should put you in a decent temp range, each heatsink would be good for about 40-50W.

You are correct about the Meanwell driver, it would be good for 3 Vero29s at 1.4A. If you dimmed it down to just under 700mA, it could run all 4.

PG, the meanwell driver is 228*68*38.8mm (L*W*H), sort of like the size of a brick but smaller.

 

[churchhaze]

I don't think it would need to be adjusted down to 1.2A, it would just only reach 1.2A if it was going to get that high. Putting it at 1.2A would only limit it at 1.2A. If i ran it that way, I'd just leave it at 1.4A and hope VF lowered as Tj warmed up.

Yeah, this supply is rated for 200W. 1.4A at 143V out.

Hey church..wouldn't adjusting the internal pot down to maybe 1.2a help fit the 4... Plus might run cooler at that current and sink combo.

I know looking at inventronics spec sheets as current drops voltage available goes up.

Wonder what the max wattage rating on that driver is also? I bet it's not over 200w...would probably need about 220w for the driver inefficiencys running 50w each led

 

[SupraSPL]

I guess a lot of it does come down to what is an acceptable temperature, because those cree bulbs are hot as hell.

That is correct, the COB will not be destroyed by running on that heatsink and it will still grow pot. But the consequence may be that loses much of its advantage over HPS. You are also correct that thick fins have a better thermal conductivity, they move heat faster and further with a lower temperature gradient, but without sufficient surface area they cannot get rid of the heat. The surface area allows the heatsink to take advantage of the temp difference between itself and the surrounding air. So the best thing is a combination of thick base plate+ lots of surface area (or lots of air movement which accomplishes the same thing).

Running at 1.4A and Tj50C they are 34.5% efficient. Running at Tj 100C they are 31.4% efficient. That is 10% less light/W. So for each 53W module, if you spent less than 5W cooling it, you would see a worthwhile gain.

Regarding vF, based on the "typical" numbers from the vF curve in the PDF, you would need to run at .7A in order to fit 4 Vero 29s on the HLG-185H-C1400A. In practice you might be able to go a bit higher, especially if the Veros are running hot. If you hook up 4 Vero29s without dimming it first, the driver will go into "limping" mode. As soon as you go even slightly above 143-144V, current will drop off very steeply (less than 700mA). As a side note, these drivers are very impressive. They deliver the specs they promise, awesome efficiency, very competitive price, power factor corrected and you never know when dimming will come in handy.

It's all in the re-reading

Thanks for the props man, this community helps each other out in so many ways, we have a great group of LED growers here

 

[CaliWorthington]

Maybe you could get a bigger driver and run all 4 at 1.05 mA or 1.1 mA?

 

[churchhaze]

Efficiency is not a measure of light/W, it's always without units. Efficiency is always W/W. I know there is a chart somewhere showing the efficiency of the CXA3070 at various currents, but I haven't been able to find the same for vero, only LER, which imo is a bit misleading. It is not the same thing as efficiency.

That being said, vero 29 3000k may be less efficient than cxa3070 3000k mid-range bin, but I think the difference is very exaggerated.

Either way, like you said, the biggest risk is that the leds would not be running as efficiently, so I think I'd rather do a test run without cooling to see just where the temperature converges. Then I can see how adding circulation brings the temperature down and design a solution based on that.

Maybe I'll end up just using a small oscillating fan. I'm not sure yet.

Running at 1.4A and Tj50C they are 34.5% efficient. Running at Tj 100C they are 31.4% efficient. That is 10% less light/W. So for each 53W module, if you spent less than 5W cooling it, you would see a worthwhile gain.

 

[SupraSPL]

These numbers are based on the minimum flux values in the PDF, so we know we are getting at least these figure. Of course they are not perfect but it give us a general idea where they stand. Every value is based on a Tj of 50C, although in practice, COBs running at low currents can have a Tj as low as 40C and the ones running hard maybe not be able to get below 60C. So this chart give a slight advantage to the higher currents and slightly penalizes the lower currents.

The reason there are 2 charts for the Vero 29 is because the vF numbers in the PDF chart do not agree with the Vf numbers in the PDF graph/curve. Hopefully some Vero29 owners can clear that up for us.

 

[churchhaze]

I'm pretty sure the bridgelux and cree datasheets only provide luminous flux values (lumens), not radiant flux (watts).

This is why it's very hard to compare the luminous flux and LER of varying spectral distributions. It requires a bunch of calculus to actually make any sense of.

For example, 5000k vero will always have a higher lumen/watt efficacy than 3000k vero, but it's very hard to figure out whether that means the 5000k is actually more efficient (Watts out / watts in) just by looking at their datasheets.

Obviously if you were a marketing department and you knew all your customers rated your products by lumens, in order to get high lumens, you'd be driven to design spectral distribution more focused around 555nm. That makes it extremely hard to find true efficiency.

Taken one step further, it's even more difficult to know how the plant will respond morphologically to the spectrum, and if 1W of red might cause different growth characteristics than 1W of green or blue for example.

These numbers are based on the minimum flux values in the PDF, so we know we are getting at least these figure. Of course they are not perfect but it give us a general idea where they stand. Every value is based on a Tj of 50C, although in practice, COBs running at low currents can have a Tj as low as 40C and the ones running hard maybe not be able to get below 60C. So this chart give a slight advantage to the higher currents and slightly penalizes the lower currents.

 

[SupraSPL]

You are exactly right, that is why we need the LER, which represents the calculus to convert from lumens to Watts. Cree provided us with the CXA 3000K LER, it is 325. Mr Flux and SDS have attempted to calculate the LER from the curve in the PDF and arrived at 328 which closely agrees. Bridgelux provides good curves in their PDF and Mr Flux has calculated the LER for 3000K 4000K and 5000K.

Using those numbers, the spreadsheet I posted takes LER into account and converts from lumens/W into actual electrical efficiency, which is a great starting point for comparisons. Mr Flux and SDS have taken it one step further and attempted to incorporate the photosynthetic efficiency, which is not represented in my spreadsheets. But I can verify the CXA3000K (14% blue) is a very suitable balance of blue for cannabis hybrids in terms of photomorphogenesis.

The spreadsheet also takes into account $/PAR W, which is critical to make a fair comparison.

 

[churchhaze]

LER lets you convert lumens to power input (W), not radiant power output (also W).

A 50W 660nm led with a LER of 300 would probably burn the plants down with radiant power, while a 50W 555nm led with a LER of 300 would be cool to the touch.