Samsung LM561C two channel 100w PC Board

Recently I decided to design my own PCB for Samsung diodes. I really like the the Samsung LM561C diode as it can hit a maximum luminosity of 201umoles/w which is very impressive. In order to achieve this result the diodes have to be driven between 150mA - 165mA less than the maximum 200mA current of the diode.

There are a number of methods to achieve this result, one is thru current restriction, which I chose. To accomplish you must use parallel circuits with series circuits to achieve the result. Samsung LM561C diodes have a Forward Voltage of 2.9v each

My option is to string 32 diodes in series for a Forward Voltage of 92.8v. I placed six rows of these series circuits connected in parallel to create a circuit demand of 1200mA, which can not be achieved by circuit design; 1050/1200mA = 0.875 ratio allowing a maximum current flow of 175mA to each diode using a meanwell HLG-185H-C1050B driver constant current. Using a lesser driver will maintain he same ration, but lower the total current flow allowing softer use of the diodes with the same micro moles/watt capability.

I attached a picture of the finished board that is currently under fabrication at Shenzhen Hansion Technology Company on Alibaba. I have the dxf files for the pcb as well as svg files.
namaste

It seems more like 2*100W to me. That's a lot of heat to dissipate.
And soldering 384 leds to a single board. I hope you got that automated.

Actually it is 100w at 50w per channels because the load is established that way. I have a company fabricating 13 boards with 192 diodes on each channel. The board is only 100w total demand. It is not a direct volts x amps = watts because of parallel circuitry used; 32S6P 32 diodes in series connected in 6 parallel strings; to lower available amps below maximum. No real problem with heat as it is not that much and I mount these to a 6x18 heat sink with a small fan no problems no overheating. Diodes are running below 200mA the maximum current designed for the diode so they never get that hot more efficient that way.
namaste

To understand the design concept look at it in this manner. Connect 32 diodes in a row in series and you get 92.8v and 200mA, current is constant and volts are additive. So a single string of diodes is 92.8v @ 200mA. If I make two strings each of 92.8v @ 200mA and connect them in parallel, the volts remain hte same, 92.8v and the amps are now additive, so I have 92.8v @ 400mA. If I connect three strings in this manner I would have 92.8v @ 600mA.

Now, If I left the circuit as it is, three strings of diodes connected or 32S3P, I would have 92.8v and 600mA drive current. I could use a meanwell HLG-185H-C700B and I would be using 85% of the driver capacity making the driver less efficient than it's advertised efficiencies. The diodes would be at 200mA and deliver about 160umoles/w fairly low. Not a great design.

So, If I double the rows to six, I change the entire dynamic of the circuit. Now I have 1200mA of current demand. In order to meet this demand I would need a driver that delivers 185Vf or Forward Voltage at 1200mA or probably a HLG-185H-C1400B driver, but I would over drive the diodes making more heaters than photon emitters at 201umoles/watt available. By using a 1050 or lower driver I get just the opposite effect you would assume, I de;liver a constant current of 1050mA divided by six = 175mA less than 200mA design maximum current flow but much closer to maximum photon flow of 150-160mA.

The use of any driver from 1050mA to 700mA will deliver a lowered amount but the 1050mA driver dimmable is best suited for the board delivering the maximum photons.

I worked with the Engineers at the company I posted earlier, they have an Engineering department, I am also an Engineer, so we collaborated and came up with this scheme. I drew the initial gerber files, and the engineer designed the final gerber file and provided me with a dxf file. I am having 13 boards made, as the schema uses 384 diodes 192/3500k 192 5000k takes two reels of 2500 diodes each to make the boards. The manufacturer is more than willing and capable to do the work at hand.
hope this helps with understanding. The Samsung diode is the future as cree is entering the mid power diode market with horticulture in mind.
namaste

Let me try to use your numbers and see where I think there is a disconnect.

You are talking about 6 rows of 32 leds but your picture shows 2 * 6 rows of 32 leds.
For just 6 rows of 32 leds your calculation is correct but that is only half of the board.
Either this board runs both channels in parallel or in series. In series it acts like 6 rows of 64 leds and in parallel it is like 12 rows of 32 leds. In both cases either your voltage or your current is a factor 2 of, and consequently so is your power usage.

I normally look at the power consumption by starting at a single led. You say you use 175mA per led (instead of 200mA.) This means each led consumes just over 0.5W (2.9*0.175). Simply multiplying with the number of leds gives 384*0.5W = 192W.

On the other hand if you would run the leds with 87.5mA (=1050/12) then you would only consume about 0.25W per led. Multiplying with 384 gives a bit over 96W.

Phagoos said:

Let me try to use your numbers and see where I think there is a disconnect.

You are talking about 6 rows of 32 leds but your picture shows 2 * 6 rows of 32 leds.
For just 6 rows of 32 leds your calculation is correct but that is only half of the board.
Either this board runs both channels in parallel or in series. In series it acts like 6 rows of 64 leds and in parallel it is like 12 rows of 32 leds. In both cases either your voltage or your current is a factor 2 of, and consequently so is your power usage.

I normally look at the power consumption by starting at a single led. You say you use 175mA per led (instead of 200mA.) This means each led consumes just over 0.5W (2.9*0.175). Simply multiplying with the number of leds gives 384*0.5W = 192W.

On the other hand if you would run the leds with 87.5mA (=1050/12) then you would only consume about 0.25W per led. Multiplying with 384 gives a bit over 96W.

Click to expand...

actually it is just as I said there are six rows each on each channel as the picture shows. two channels 12 rows total. math is still the same. There are 192 diode on each channel, 6 rows of 32, please look again more closely. thank-you
namaste

Subd!

I have been experimenting with various configurations and I thyink the 3000k 90cri and 3500k 80cri seem fairly close id plant development. Adding the 5000k with plenty of blue and some more red seems to be a great mix. With strips it is hard to get a volume of diodes in a smaller area, so I am following Vitally of chilledled in an approach to get there. I purchased three of his 90w boards, mounted them on my style heat sink and placed them in a 2x4 and they are killing it. he uses 3000k 90cri and additional blues, etc. SO I decided to go with the two favorites. You can take two boards and wire the channels in series and control 3500k on one driver and 5000k on one channel. A great light for both veg and bloom. I am doing a DIY style of fabrication and if you hit me up in PM and I can share the details, no sales, just group ordering. A minimum of 13 boards is required for an order as it takes 2500 diodes of each 3500K and 5000K, two reels, to make 13 boards with 384 diodes per board and no wasted leftovers. I already have my order in but others can do the same in a group and order 13 boards from the same supplier and they will give you credit for the design services already completed, just ask for the GrowGreen 100w LM561C board.
namaste

Nice thread, I would to make my own led pannel soon, so subbed!
Big up

Nico Farmer said:

Nice thread, I would to make my own led pannel soon, so subbed!
Big up

Click to expand...

thanks
You can contact the same supplier I used,as he has an engineering team available to assist with design parameters. I read an article from ledgardner and I saw what i wanted to do which is to drive the diodes more softly and bring out the micromoles for the plants. I experimented with flexible Constant Current strips to figure out what I wanted.

I vacillated between using 3000k 90cri which I have in strips or use 3500k 80cri. I decided to go with the 3500k because I think tyhe two together will fill out the light spectrum better.

I already have a light with some 3000k 90cri but it has additional royal blue for balance so I figure a little less red in the 90cri than the 80cri.

If you search on Alibaba you can find plenty, I have given names to two that I have worked with with very high success in quality, reliability, and trust.

thanks

what was the cost per board if you dont mind my asking? im perched on the edge of investing 10k into LED lighting and wanna keep my options open

Desert Hydro said:

what was the cost per board if you dont mind my asking? im perched on the edge of investing 10k into LED lighting and wanna keep my options open

Click to expand...

YO, BRO, whats UP? Here is the deal. You can go direct to the manufacturer. A nice guy named Jeff is easy to work with. If you go with a mixed spectrum like I did, you will use two reels; $150/ea to make 13 boards; $100 fab; with 384 diodes per board. I designed two channels for flexibility. They will provide design engineering for $100 per project or you can use my design if you like. The final cost is $800 for 13 boards that can deliver 1300 true wall watts of power.

https://tophspcb.en.alibaba.com/?spm=a2700.8443308.0.0.iZMiFz

these are the guys I used, ask for Jeff.

I was reading your thread, I had hermies too. I retired cob's just a gut and got a few chilled boards but I decided to do a few thin gs myself too. I saw your grow, I think a pcb style will do better, no heat to dissipate in the room. I have retired ALL cobs forever now. I have 12 in retirement But now I have DIY QB304 in veg and clone and I plan on using these for both veg and clone.

I think this is your best option. You can get any design you want. If you like the QB geometry you can use that one too. All you need is the formula for engineers; 38S8P - 58 series diodes in a string with 8 strings in parallel, will yield the 304, 160w or you can do like my DIY model and squeeze a little more out with 42S8P for 343 diodes and 180w. I mounted on a cookie sheet to make a point NO HEAT.
namaste

short answer $60/board
namaste

Update GrowGreen336 on a 2x2 frame in a 2x2 tent. I would rather NOT talk about other folks work. So here is a 42s8p with mixed spectrum on 1/8" thick x 1" wide aluminum flat stock NO HEAT NO FANS, just lots of good micromoles.
namaste

Fabrication is complete.
For a single board I recommend using a HLG-185H-C1050B driver for 100w. For two boards I recommend using a HLG-185H-C500B driver for 200w. The boards are designed to use lower amps to generate higher watts, so lower is better. The board will run the Samsung LM5661C diode at 167mA very close to the 200 lumens/watt output capacity.
namaste

It would be nice to see the wall power used by this board with both channels in series and in parallel on the HLG-185H-C1050B.

Phagoos said:

It would be nice to see the wall power used by this board with both channels in series and in parallel on the HLG-185H-C1050B.

Click to expand...

First, the Vf per channel is 92v. So you can only hook up two channels in series to use a 185H-C1050B/A driver that will yield 90-100w. If you run the channels in parallel you get 92v @ 2100mA so you would need a 185H-C2100B/A driver if available. You can use a 240H-C1050B/A but very inefficient.

The most efficient driver is the HLG-185H-C1050B/A driver for 100w. When the boards come in you will see the two board version, as that is what I am building with channels on board to board wired in parallel, and the two boards wire in series to a HLG-185H-C500B driver for 180-200w at low amperage which is what the board is designed to perform, not higher wattage.

If you are interested in participating in the CO-OP the time is running out in days, otherwise you will have to purchase one at a higher price and a later date.
namaste

Phagoos said:

It would be nice to see the wall power used by this board with both channels in series and in parallel on the HLG-185H-C1050B.

Click to expand...

I reviewed your total posts to date, and it seems you are critical of everyone else but have never started a thread about anything. Why is that?
namaste

those look legit