Minimalistic multichip DIY LED build

Look att the datasheet again. When you run the driver in constant current mode you will have a voltage range in which the driver manage to retain the wanted current. The 120 W chip has a FV between 37 - 42 V (the actual FV depend of the current. If you run at 3,6 A - its the lower wattage and so on)

The HLG-320-48B has the current limit at 6,7 A and a Constant current region within 24-48V - this should work but I have not tested!

Probably no - they will give a lot of heat and I´m not sure they will give 100 watts i light radiation. I´m normaly think - a 100 watt in gives 100 wat heat out. But you can run them at a lower current - 5A givs 200 watt or rather close this

Sincerely Lasse

Lasse, Something I'm not understanding, if you run the HLG-320-48B at the constant current of 6.7A and 48V, you are under the chips rated 7A but the 48V will push the chip to 321W. The specks say the chips max wattage is 294W . Do you need to monitor the drivers voltage to only 42V ? Thanks---Rick
 
Lasse, Something I'm not understanding, if you run the HLG-320-48B at the constant current of 6.7A and 48V, you are under the chips rated 7A but the 48V will push the chip to 321W. The specks say the chips max wattage is 294W . Do you need to monitor the drivers voltage to only 42V ? Thanks---Rick

Do not hook up on 48 v. 48 V is if you run it under constant voltage mode. But if you chose constant current the driver will adjust the voltage in the region of 24 ~ 48V. If your FV on the chip is 36 V - the driver will give 6,7 A with 36 V. The B model run under constant current mode and you can adjust the current. If you adjust the current to 5 A the FV on the chip is somwhere arround 34 V and the driver will adjust the voltage to that. Read through this. However it looks like the drivers only are dimmable in some degree - but someone have to test and report

Sincerely Lasse
 
Any suggestion on a Meanwell driver that can push the EPISTAR EDI-EA1143 45*45mil 50W Super Actinic Blue Hybrid?

I have two of 50w multchips.
 
I personally have no experience of Mean Wells drivers, I just look at the specifications. In their HLG series is something strange with the specifications - only 240 and 320 has a constant current region. I think, stress think, that the others work the same but do not know. At worst you can, even if it is a constant current drive, make a parallel connection, and run them both at the same driver, for example, HLG-240-48B. The parallel connection allows each chip receive 2.5 A. But then the chips have to be as similar as possible. It is therefore wise not to feed each chip with its maximum current.

As usual - I do not know because I have not tested these drivers, but I think.


Sincerely Lasse
 
So I made the jump into this head first and ordered 4 of the AC-RC 50w Hybrid chips, 4 drivers and 4 sets of 60* lens/optics. This is all going over a 150xTall(48"x24"x30"tall) with the end goal being SPS dominated. Is this looking like it will be enough light/spread? I am considering ordering an additional 2 50w 10k's or possibly 20k's to tone the blue back down.
 
From what I have read you need active cooling on 50W + chips. I see on the AC-RC site they sell passive heatsinks for their 20W chips. Good luck.---Rick
 
I really would prefer to go with a heatsink along the lines of what folks are using for "traditional" 3w builds but have no clue if that would support it even with fans.

Why wouldn't it? The only difference between 50W of one chip and 51W of 3W chips is the concentration of the heat source. The heatsink is radiating the same amount of heat either way.

If you had a fan blowing directly above the LED, it would move the heat out to the fins of the "standard" heatsink. However, if you don't already have one, it would probably be easier to just get the CPU cooler. Just keep the junction below 65°C and you'll be good to go.
 
I hope it is ok to post this up. I'm looking at the heat sink options that may be available. Instead of having one heat sink for a 50w LED and then another for the rest of the LEDs I was hoping something like this may work:

http://www.heatsinkusa.com/10-080-wide-extruded-aluminum-heatsink/

This heat sink has 2.5" fins. They also have another that only has 1" fins. The one with 1" fins is a little cheaper, but not much.

Cut it to be a 10x10 size and add a 120mm fan on top in order to run one 50w 16k white and then 6 20w RB LEDs.

Thoughts?
 
I hate the fact that HeatsinkUSA doesn't list the surface area per inch of each of their profiles! (and the fact that none of these sellers list actual temperature rating curves!)

OK, if you have 2.5" fins, then that is 5" of surface area per fin. With 24 fins in that profile, that is 60 square inches of surface area, just in the fins. Add another 10" for the base and ignoring the valleys between the fins gets us with 70" per inch of heatsink.

Your ten inch hunk would have over 700 square inches of surface area.

The rule that I was told by AC-RC was that you need >32sq cm/Watt with no cooling assistance (fan). You have a little more than 4,516 square cm. Unassisted, in a perfect world, that would support 141 watts.

Figure your actual current draw (therefore, the actual heat you need to dissipate) by multiplying the voltage * current for each chip. (West, Virginia! W=VA)

I would add another two inches (or more, if you have the room). At 12", you have 843" sq of surface and that would almost support them without a fan. Put a fan on it and the heat dissipation increases dramatically.

Fans can fail. If you're on the bleeding edge with the fan running, you're likely to lose your chips if it/when fails.

That stuff is $4.59 per inch, so the more of it you can buy, the more insurance you have against toasting your LEDs.

Of course, run the numbers yourself to verify.
 
Wow, that is an excellent explanation Landsailor, thanks!

I have plenty of room to make the heat sink larger, I can to up to 20" but would prefer not to. I can see doing a 10x16 heat sink though and then putting the 120mm fan on top of it in order to help cool things.

I was also thinking of going with the 1 50w and 4 20w RB. This way I can run a 100w driver on the 4 20w LEDs and then a 50w driver on the single 50w LED. Both dimmable for the desired color. So each heat sink would have 130w of LEDs on it which would still be within the 32cq cm/watt and increasing the size to 10x16 would help even more with close to 7225.8 sq cm of area (and at the 32 sq cm/w would be about 225w) - all this if my math isn't too off.

I have a 90g tank and with the prior calculations of (48x18)/14 *3 = 185 w total for the tank I would be looking at 130w per side or 260w for the tank.
 
What is the spread of the multi-chip LED without optics? Is it narrow or very wide? Also can anyone comment on AC-RC's drivers? I want to drive twenty 20w lamps on 4-5 dimable circuits. I would rather have the nicer fit and finish of a meanwell setup if there is an appropriate fit.

http://www.ebay.com/itm/85-265VAC-5...986?pt=LH_DefaultDomain_0&hash=item3371803a1a

Sorry, I missed your question.

Normally, they spread 120 degrees, but look at the data sheet - which usually bear the beam angle.

I have one of their drivers (100 watt) and it seems to work - have not driven it for very long. However, calling it dimmable, it's not the whole truth. It is possible to reduce the intensity, but not quite down to zero and the function is "jerky".

See my answer in # 568th I think it is in HLG series you should look.

Sincerely Lasse
 
Lasse, Can you confirm my math @ your post 568. Say I want to run 4x 50W chips in parallel. Will a MeanWell HLG320H-36B driver work. The driver specks say 18-36V and a current of 8.9A. The 50W chip specks say 32-36V and a max current of 3.5 A. Therefor 8.9 / 4 = 2.22A each chip. Soooo 36V x 2.22A would run the chips at 80W each? Is this the right way to figure a driver? Thanks---Rick
 
Lasse, Can you confirm my math @ your post 568. Say I want to run 4x 50W chips in parallel. Will a MeanWell HLG320H-36B driver work. The driver specks say 18-36V and a current of 8.9A. The 50W chip specks say 32-36V and a max current of 3.5 A. Therefor 8.9 / 4 = 2.22A each chip. Soooo 36V x 2.22A would run the chips at 80W each? Is this the right way to figure a driver? Thanks---Rick

Thanks for asking that 007...I'm a complete idiot when it comes to electricity!

I'm no electronics expert but I have explained to me that - yes - you can calculate in this way if you work with a driver that provides constant current and chip´s are as similar as possible. In fact, multi-chip design relay of this particular way of calculate.. A 50-watt chip consists of five parallel strings of 10 in serie connected LEDs.

There are trap doors, if the chips are too different there are not an exactly equal current split and if a chip fails, the others share the remaining current.

Example. If a chip breaks in your exemple each of the remaining chips share the same 8.9 A, which gives about 3 A to each - no problem. If two broken - it will be around 4.45 A to each and thats not god, they will burn

Its also important to stress that a chip`s FV is determined to partly by the current. In your case when you calculates the effect so you should probably only use about 34 V. The effect will then be 34 * 2:22 = 75.48W.

When doing this kind of design you should not use the maximum current without having a small safety margin. You can also have a fuse prior each chip. The fuse must then be the same or slightly higher than the chip's maximum current.

Once again, the above requires a driver that provides a constant current and with the total FV in the regulatory area. FV in a parallel configuration is the same as FV of one chip or string of serial chip. FV of a series connected string is the sum of the individual chip `s FV.
FV = Forward Voltage = the minimum voltage needed to obtain the LED to function.

With a driver that works with a constant voltage everything is different.

Sincerely Lasse
 
Example. If a chip breaks in your exemple each of the remaining chips share the same 8.9 A, which gives about 3 A to each - no problem. If two broken - it will be around 4.45 A to each and thats not god, they will burn

Its also important to stress that a chip`s FV is determined to partly by the current. In your case when you calculates the effect so you should probably only use about 34 V. The effect will then be 34 * 2:22 = 75.48W.

When doing this kind of design you should not use the maximum current without having a small safety margin. You can also have a fuse prior each chip. The fuse must then be the same or slightly higher than the chip's maximum current.

Once again, the above requires a driver that provides a constant current and with the total FV in the regulatory area. FV in a parallel configuration is the same as FV of one chip or string of serial chip. FV of a series connected string is the sum of the individual chip `s FV.
FV = Forward Voltage = the minimum voltage needed to obtain the LED to function.

With a driver that works with a constant voltage everything is different.

Sincerely Lasse

Lasse, I need to ask a couple questions to your answers. When using a HLG320-36B how do you regulate the voltage to 34V? Is it with a dimming signal? So in a Parallel circuit, the voltage output from the driver is the same to all the chips, and you divide the number of chips into the drivers max current to find the total current to each chip?? Yes, I would add a fuse at each chip as suggested in the 3W builds. Would you figure the fuse at the 2.22A drive current to each chip? Thanks---Rick
 
When using the driver in constant current mode, the current is maintained by that the voltage is varied automatically dependin on load. Voltage will be automatically set to the actual FV. It is therefore important that the chip's FV is in the driver's field of work. HLG-320-36B have the constant current region of between 18 and 36 V. Epistar 50W Super Actinic Blue Hybrid 45mil LED panel has its FV from 31.5 to 36V. Thus, FV is within the driver's work area. If you feed with 2.22 A, then will the driver control the voltage to the current FV - which I think will be around 33-34V if you use 2,22 A. If yuo use the maximal 3,5 A - the FV will surly be arround 36V.

So in a Parallel circuit, the voltage output from the driver is the same to all the chips, and you divide the number of chips into the drivers max current to find the total current to each chip??
Yea - for a constant current driver and similar chip´s - its true - at least - I have learned so

I should use a 3,5 fuse beacause the chip manage 3,5 A as maximum.

Sincerely Lasse

Edit:
drivers max current
Wrong question - its not the drivers max current - its the current of the driver. Its a constant current driver!!!!
 
Last edited:
Back
Top