DIY Bridgelux Vero 29 + 3.95" profile heatsink usa

[SupraSPL]

LER lets you convert from lumens/W to W/W (radiant power/dissipation power) = % conversion efficiency

The 555nm LER is 683lm/W. A 660nm LED could never have an LER of 300 because it has a set LER based on its wavelength/curve (~50 lm/W). Luckily 660nm LEDs dont normally bother with lumens they give us mW so we don't have to bother with LER in that case.

Long story short, I am not pulling your leg my friend, the Vero 29 at Tj100C will give you about 10% less light/W than it will at Tj50C. The temp droop at Tj 50C is about 96% and at Tj 100C it is about 87%. I got this info from the Vero 29 PDF from this chart

 

[churchhaze]

It's not that I don't believe you about the temperature and efficiency. I've read both the vero 29 and cxa 3070 datasheets very thoroughly and have a good understanding of what all the charts mean. It's that I don't think choosing vero over cxa was a death decision as you make it out to be. You make it seem like AB chips get 52% efficiency while vero 29 get 34.5%, and that's where I question your numbers. That couldn't possibly be true. Both of these companies use almost exactly the same die and phosphor technology.

My defensive stance has nothing to do with the junction temperatures and how that relates to efficiency under ideal conditions. I'd stop this project right now and get CXA AB chips if I really thought the efficiency gain would be that great, so naturally I'm second guessing your numbers..

35% efficiency vs 52% ideal efficiency makes this project a non-starter regardless of other details. Of course I'm going to question your math before scrapping the whole project.

Long story short, I am not pulling your leg my friend, the Vero 29 at Tj100C will give you about 10% less light/W than it will at Tj50C. The temp droop at Tj 50C is about 96% and at Tj 100C it is about 87%. I got this info from the Vero 29 PDF from this chart
View attachment 3196639

 

[SupraSPL]

Gotcha, I encourage anyone to question my math and methods. I make mistakes all the time. But I wouldn't stand behind the numbers if they couldn't stand up to scrutiny.

My main reason for piping up here is just to warn you that it takes a huge amount of heatsink to use passive cooling and get a good Tj. Any breeze from your circulation fan will help a lot especially with those wide gaps between the fins.

As far as CXA vs Vero, since the ABs are somewhat scarce, in practice we are dealing with CXA3070 Z2 vs Vero29. Since the 3000K Vero has less blue than the 3000K CXA, it is expected to have slightly lower efficiency. A more fair comparison might be CXA3070 3000K Z2 vs Vero29 4000K. At 1.4A the CXA3070 is 36.8% and cost $2.10. At 1.4A the Vero29 4000K is 35.8% and cost $2.12. If you use the numbers on their chart rather than their vF curve, it is 37.1%. They are so close that with margin of error might as well call them equal unless you are running soft (700mA), at which point the CXA3070 gets the upper hand. I do run soft and I was lucky to get the AB bins, and those are the main reasons I prefer the CXA3070 for my projects.

Mr Flux has pointed out that since the Vero is not binned and the CXAs are, that my spreadsheet may be giving a slight advantage to the Vero, because I am using the minimum numbers rather than the typical numbers. I agree with that point, but I also view that as a strike against the Vero because I like to know what I am buying.

 

[churchhaze]

I've gotta be honest, I'm kinda glad your opinion on the cooling is so different. I was thinking I bought too much heatsink for 50W power usage. I really will have to do some tests to see how temperature correlates with airflow to size up how much "blow" I'll need. I won't get the power supply until tuesday, so until then, i can only test with these 9V batteries, and 1.5V batteries to bump up the voltage, which isn't enough time to heat up the heatsink.

For example, if the difference between a light breeze and a 6000RPM 90mm fan is only 4-5C, I might consider somewhere around 2500rpm.

Gotcha, I encourage anyone to question my math and methods. I make mistakes all the time. But I wouldn't stand behind the numbers if they couldn't stand up to scrutiny.

My main reason for piping up here is just to warn you that it takes a huge amount of heatsink to use passive cooling and get a good Tj. Any breeze from your circulation fan will help a lot especially with those wide gaps between the fins.

As far as CXA vs Vero, since the ABs are somewhat scarce, in practice we are dealing with CXA3070 Z2 vs Vero29. Since the 3000K Vero has less blue than the 3000K CXA, it is expected to have slightly lower efficiency. A more fair comparison might be CXA3070 3000K Z2 vs Vero29 4000K. At 1.4A the CXA3070 is 36.8% and cost $2.10. At 1.4A the Vero29 4000K is 35.8% and cost $2.12. If you use the numbers on their chart rather than their vF curve, it is 37.1%. They are so close that with margin of error might as well call them equal unless you are running soft (700mA), at which point the CXA3070 gets the upper hand. I do run soft and I was lucky to get the AB bins, and those are the main reasons I prefer the CXA3070 for my projects.

Mr Flux has pointed out that since the Vero is not binned and the CXAs are, that my spreadsheet may be giving a slight advantage to the Vero, because I am using the minimum numbers rather than the typical numbers. I agree with that point, but I also view that as a strike against the Vero because I like to know what I am buying.

 

[SupraSPL]

I will be very interested in your results. For the longest time I have been running passive only. I did some testing with COBs on CPU coolers and was amazed how cool they can run with a minimal amount of fan power. Since then I have been promoting active cooling for new builds. The main reason I still run passive cooling is because I have a huge amount of heatsinks on hand, 132 pounds of the 10.08" profile (6 lbs each, good for ~50W):


If I were to do it all over again I would probably go with this design, 200W per heatsink with a 2W fan

 

[foreverflyhi]

Haha another epic picture from supra!

Worm bins and heat sinks.

 

[churchhaze]

I've got the mean well driver now and 3 of these wired up for testing. I've got Arctic silver 5 as TIM, and supra is correct. Within 5 minutes, they're all very hot. I will continue to run this test unless it gets above 100C

These pictures really don't do these lights justice. At 1.4A, when I put my hand right next to the light, the radiant power burns my hand. Just Wow...

One thing to note... If these were put flat on the table facing down, I think there'd be a risk of setting the table on fire from radiant heat.

 

[churchhaze]

Another thing to note is that the mean well driver I'm using really does not like being current limited. When I turn the pot down to limit it, it squeals. Let's not do that....

 

[SupraSPL]

I agree, when I am testing COBs, if one of them flopped over so it was beaming into something, I can see how it would be a fire hazard just from the sheer heat of the photons converting. The light is so bright it will cook flesh in a matter of seconds. But on the other hand I run some of them only 6" from the canopy (700mA) with no increase in canopy temp.

 

[epicfail]

Another thing to note is that the mean well driver I'm using really does not like being current limited. When I turn the pot down to limit it, it squeals. Let's not do that....

Are you saying you can't dim that driver with out it squealing?
Is it only quiet on full? that doesn't seem normal to me. I would say try a different pot or resistor on the dim wires but yours is the C1400A version with the internal pot. Also is it only dim-able to 700mA? (50%) B version dims to 10% .

I wonder if the C1400B version makes noise when dimmed. I know JavaCo has the HLG-120H-C1400B set up on a dimmer, maybe he will chime in if it does.

Here is the 3590 i slapped together today cranked all the way up (117watts).

Here it is at 1.2 amps Cree test current for the cxa3590

I have the HLN-60H-1400B just dimmed with resistors now for testing (I don't have the right pots), it runs silent and cool. Sorry for all the questions, I'm still undecided on which drivers I want to buy and noisy drivers are not something I'm looking for.

 

[smokey the cat]

Hey Church - awesome looking setup. Fantastic bright lights you've got there. Can't wait to see what you grow with em.


I've got a small scale version of your build with 2 Vero 18s on a single heatsink.

I added a 120mm fan - it just had to be done as passive heatsink performed poorly in my situation. Would love to see someone set up a huge ducted heatsink style setup like a modern graphics card cooler.



My meanwells dim without any noise - I reckon you should check with the supplier and ask if they wanna replace it for you because it shouldn't do it.


On Vero spectrum - I I have found the Vero 3000k too lacking in blue. The 3000 looks real yellow to my eye, like a HID though which looks awesome like hot yellow sun pouring over the plants. But I don't think it's a great light - much rather switch it to a 4000 I think. Too much stretch.

Maybe it'd be an OK supplementary if paired with bluer lights.

I have a 3500k too which seems better, but not ideal on its own.

Only a few month's experience with Vero 18s here, so this is all first steps here and not the definitive account.

 

[Observe & Report]

All three of my LPF-90D's make a squeal when they are dimmed. It's barely audible and I think it is normal. It is possible that yours are squealing but the sound is too soft and/or high pitched to be heard in your space.

 

[SupraSPL]

Good info SmokeyC!

I will continue to run this test unless it gets above 100C

Sorry I probably should have mentioned this earlier but at 1400mA, a junction temp (Tj) of 100C is implied by a case temp (Tc) of about 75C or a heatsink temp of about 60C.

 

[JavaCo]

Mine squeals but i have to pretty much have my ear right next to it . I dont hear it over the fan when adjusting the Current from a arms length. But i also filter my electricity at the panel with caps. If it is loud enough to hear From a few feet then I would say it is dirty juice. When neighbors start cranking up washing machines , dryers , AC and what ever else it can cause noise In your current. If you have one of those audio video power strips that filters power to your TV and equipment you can try running it off that see if helps make the squeal less audible. Or if it is just a minor squeal then that is normal don't worry about it.

 

[churchhaze]

I wired up the 4th unit to see just how much current the series combination can draw at 143V.

When it's first turned on, current is 0.96A. It quickly rises to 0.99A, within 10 seconds. After 5 minutes or so it gets to 1.12A. (It probably won't get this high when there's airflow)


So here are my choices..

1) Run all 4 on the this mean well power supply at 1A (maybe 1.1A if warm)
2) Run 3 on this supply at the max current of 1.4A and get 2 more vero 29 and another supply for 3 and 3

Which would you do?

 

[churchhaze]

By the way, at ~1.1A, the heatsinks are warm to the touch whereas at 1.4A, they are too hot to handle.

39.325W average electric power at 1.1A vs 50.05W at 1.4A.

The configuration of 4 uses 157.3W of 200.2W of mean wells max power

The configuration of 3 uses 150.15W of 200.2W of the max power

So they're pretty close in terms of making the best out of the expensive power supply, configuration of 4 having the slight advantage.

 

[smokey the cat]

wow, interesting reports on the squeal. Wonder if this means I'm getting old? I'm mid thirties and have always been pretty sensitive to high pitched noises, but this a sign...

 

[smokey the cat]

Mr Haze either in a pinch will work, but prefer a larger numbers of soft emitters over fewer harder emitters. 4 Vero per driver is best - you'll be running in the most efficient part of each led's curve so you'll get more lumens for those watts.

More $ upfront, but these things are so cheap and long lasting.

Supra might do the maths for the outputs if he has the figures for a V29???

 

[churchhaze]

I'm thinking 4 will work better too. It will allow me to use everything I have now without having to get a separate supply and 2 more vero 29. Would that be 5 of them, like in mr flux's build?

If I ever want to extract the remaining 300mA out of the power supply, I could put 143V of vero 10 in parallel with the vero 29 chain.

 

[churchhaze]

I realize that the majority of the experts here recommend 4000k vero or 3000k cxa, but then you see a lot of people supplementing with 660nm. Also, like you say, HPS like color is what I was going for. I want a ton of yellow, amber and red, and don't really care about blue as long as there's enough to see a blue lighter. These are the 80CRI btw.

I'll just have to see for myself if there's really too much stretch. I doubt there will be with the R:FR ratio being so high. The R:FR ratio of HPS is about 2:1 while for the vero 3000k is about 6:1. The gradient for the HPS is far less steep dropping off with higher wavelengths than the white leds.

I may add UV-A and violet LEDs later, but right now I want to replace a bunch of HPS light.

On Vero spectrum - I I have found the Vero 3000k too lacking in blue. The 3000 looks real yellow to my eye, like a HID though which looks awesome like hot yellow sun pouring over the plants. But I don't think it's a great light - much rather switch it to a 4000 I think. Too much stretch.

Maybe it'd be an OK supplementary if paired with bluer lights.

I have a 3500k too which seems better, but not ideal on its own.

I had intended this thread to be more of a build DIY and potential discussion rather than a journal, so in theory anyone could make a similar design with whichever color temperature.