660nm 680nm 690nm 700nm diode build

I'm wanting to try to build a red light in strip lighting but I could use a little help from people.

I'm not buying any premade strip as I want to craft it.

I will probably need a better soldering iron, so before I order something off Amazon, I will wait for recommendations on a kit or a workstation.

I also plan on using some higher than average nm. I see a lot of reports on action spectrum and the absorption maximas of chlorophyll a + b, which seem to vary somewhat depending if they come from reviewed scientific sources or from websites, who by coincidence or not, also sell LED lights. My only raison d'etre is that the P680 + P700 are called that because those are maxima absorption frequencies. I want to look a little more into it... let's say for photomorphogenetic reasons.

I'll need some help throughout this project. For one thing, I could use some recommendations on some deep red 660nm diodes. I will go back and read some threads (far red, etc), but 78 pages :'( along the 37 webpages + 12 pdf files + ugh...

I plan on using three similar diodes made by Ushio, including the SMBB680D-1100, SMBB690D-1100, and SMBB700-1100.

here are some stats from their spec sheet


Now no information is provided regarding PAR or anything. I even asked the salesman before I realized ... why would he know that? Instead we have total radiated power and radiant intensity (RI) reported. Its too bad I can't find an RI on other brands of diodes, like LM301Bs. If someone knows, let me know.

Orders done, diodes on the way.

Also list any ideas you may have for heatsinks, starboards, solder types, and related threads.

just do it.diy panels its way beyond my skill ambition

.use 561 who cares its a great diode....the plants don't really care..they just want lots of good photons..

anyway after starting with burples..and we are now at full circle.. red panels are interesting..

please get back to us with the results/build..

Will do, I ordered the diodes. They are on the way.

I have an HLG-240H-48A driver.
I am planning on running 5 boards in parallel allowing 5x 1A per circuit.
Three of the boards are going to be Samsung LM09 HinFlux boards rated at 48V/1A.

I was planning on using five of each diode (680, 690, 700) on a strip; so 15 diodes per circuit.
That's 3.1V per diode at 1A = 48V. There would be two strips.

If I do it this way, I shouldn't need to add any resistors, right?

trojanvirus said:

I have an HLG-240H-48A driver.
I am planning on running 5 boards in parallel allowing 5x 1A per circuit.
Three of the boards are going to be Samsung LM09 HinFlux boards rated at 48V/1A.

I was planning on using five of each diode (680, 690, 700) on a strip; so 15 diodes per circuit.
That's 3.1V per diode at 1A = 48V. There would be two strips.

If I do it this way, I shouldn't need to add any resistors, right?

Click to expand...

Oh-oh! I see a risk to get thermal runaways, bro.
Boards and red strips would need to have exact the same voltage and resistance. But red and white diodes heat up and behave much different and therefor the voltage would be not the same when they are warmed up.
The one with lower resistance would get more current, would heat up even more, get more current, heat up even more and bang!!..it's kaputt.
Better use an additional HLG-40H-48 for the red strips. These diodes are too expensive to blow them up and an HLG-40 you can already get for a little over 30 bucks.

Randomblame said:

Oh-oh! I see a risk to get thermal runaways, bro.
Boards and red strips would need to have exact the same voltage and resistance. But red and white diodes heat up and behave much different and therefor the voltage would be not the same when they are warmed up.
The one with lower resistance would get more current, would heat up even more, get more current, heat up even more and bang!!..it's kaputt.
Better use an additional HLG-40H-48 for the red strips. These diodes are too expensive to blow them up and an HLG-40 you can already get for a little over 30 bucks.

Click to expand...

Or get a CC driver and run them in series. Btw are there any special issues or concerns about running long single diode strings?

trojanvirus said:

I'm wanting to try to build a red light in strip lighting but I could use a little help from people.

I'm not buying any premade strip as I want to craft it.

I will probably need a better soldering iron, so before I order something off Amazon, I will wait for recommendations on a kit or a workstation.

I also plan on using some higher than average nm. I see a lot of reports on action spectrum and the absorption maximas of chlorophyll a + b, which seem to vary somewhat depending if they come from reviewed scientific sources or from websites, who by coincidence or not, also sell LED lights. My only raison d'etre is that the P680 + P700 are called that because those are maxima absorption frequencies. I want to look a little more into it... let's say for photomorphogenetic reasons.

I'll need some help throughout this project. For one thing, I could use some recommendations on some deep red 660nm diodes. I will go back and read some threads (far red, etc), but 78 pages :'( along the 37 webpages + 12 pdf files + ugh...

I plan on using three similar diodes made by Ushio, including the SMBB680D-1100, SMBB690D-1100, and SMBB700-1100.

here are some stats from their spec sheet
View attachment 4233671

Now no information is provided regarding PAR or anything. I even asked the salesman before I realized ... why would he know that? Instead we have total radiated power and radiant intensity (RI) reported. Its too bad I can't find an RI on other brands of diodes, like LM301Bs. If someone knows, let me know.

Orders done, diodes on the way.

Also list any ideas you may have for heatsinks, starboards, solder types, and related threads.

Click to expand...

I like what youre doing but it would be even more interesting if each wavelength diode was run individually, so you can investigate them 1 by 1.
One issue with this build ia that with only a little power dissipation per strip its going to be hard to get something out of the results.

Just want to say that I received the diodes, but unfortunately I learned after reading docs that they will require a specially designed board. I thought 5250s were all standard... damn learning curve beating me down.


I'm thinking common Cree XML pads would work, but figure I should just get some boards crafted.
Here's a shot of XML board specs:

Design a mpcb board? Send this to R2T? ugh... this project is complicated!
I will probably look for someone that can just process the whole thing.

trojanvirus said:

Just want to say that I received the diodes, but unfortunately I learned after reading docs that they will require a specially designed board. I thought 5250s were all standard... damn learning curve beating me down.
View attachment 4242943

I'm thinking common Cree XML pads would work, but figure I should just get some boards crafted.
Here's a shot of XML board specs:
View attachment 4242948
Design a mpcb board? Send this to R2T? ugh... this project is complicated!
I will probably look for someone that can just process the whole thing.

Click to expand...

Check out cutter they will customize strips or boards

https://www.cutter.com.au/product/solskin/

Here's a link

trojanvirus said:

Just want to say that I received the diodes, but unfortunately I learned after reading docs that they will require a specially designed board. I thought 5250s were all standard... damn learning curve beating me down.
View attachment 4242943

I'm thinking common Cree XML pads would work, but figure I should just get some boards crafted.
Here's a shot of XML board specs:
View attachment 4242948
Design a mpcb board? Send this to R2T? ugh... this project is complicated!
I will probably look for someone that can just process the whole thing.

Click to expand...

You would be better off looking at RapidLED stars. Not the cheapest way, but by the time you purchase a reflow oven you'll have a chunk of change just to get set-up.

trojanvirus said:

Just want to say that I received the diodes, but unfortunately I learned after reading docs that they will require a specially designed board. I thought 5250s were all standard... damn learning curve beating me down.
View attachment 4242943

I'm thinking common Cree XML pads would work, but figure I should just get some boards crafted.
Here's a shot of XML board specs:
View attachment 4242948
Design a mpcb board? Send this to R2T? ugh... this project is complicated!
I will probably look for someone that can just process the whole thing.

Click to expand...

I don't know it for sure with R2T but Hanison and Meizu do also produce own PCB's. In fact PCB's are their main business.
I would ask them for creating the PCB's too! Tell them which diode you wanna use and how much diodes you want per strip/board/PCB and they can also create the Gerbel files. When they have used them diodes in the past you could use an exsisting board layout to keep the costs as small as possible.

Rocket Soul said:

Or get a CC driver and run them in series. Btw are there any special issues or concerns about running long single diode strings?

I like what youre doing but it would be even more interesting if each wavelength diode was run individually, so you can investigate them 1 by 1.
One issue with this build ia that with only a little power dissipation per strip its going to be hard to get something out of the results.

Click to expand...

Depends on the length of the string. Citizen is using 36 diodes in series on their 3618 COB's and HLG has used 38 diodes in series on QB304's. But they don't use wires between each diode. With 4" diode spacings and 38 diodes the wire length total would be 152" and with AWG20 or 22 thats already enough to measure/feel it.

The longer the string and the smaller the wire diameter the more voltage droop will occur (higher resistance) and the driver has to compensate that. At least if the driver is working in constant current mode.
Light lutput would be the same but power draw would increase and the wires can get warm because of the droop. 1v voltage droop with 5,25A drive current(HLG-240H-48 driver) would be already 5,25w watts and this watts are dissipated by ~12ft of small AWG20 wire. That means, the wires would heat up with the time and in the worst case they could melt down at some point. I would keep the strings short and rather use a few srings in parallel. 12-18 diodes are most common. F-strips for instance use 8 in series and 9 strings in parallel on their 2ft strips and 16 diodes in series 9 times parallel on the double row and 4ft strips. This way the voltage stay in a safe range(<60v) and you will easily find a siutable driver.

Hi guys,
exponential learning phase in progress....
@Ryante55 & boilingoil

I contacted Cutter; they recommended XML stars, which I will consider. The reason I am hesitating is that its not a 100% fit. Also, if its just stars, they are cheaper from different sources. I am in Canada, so duty, shipping, etc. adds up fast (as I just learned ordering the diodes in the first place). Rapidled would be another similar alternative.

I should have said that I can get the 'stars', or rather boards, from Ushio, but they are pricey for what they are (I think ~$175 CAD before shipping and duty) which is just a bit more than it would be for stars.
Ushio diagram:


For that price I can just use a MCPCB design engine (or sketchup) and have some fabricated for maybe cheaper... I think (as Randomblame just suggested). I asked for a quote from 4MCPCB, which have just asked me for a Gerber file or detailed specs, which I am about to make. This maybe a better alternative; just take me a little while tonight. I think AliB guys will say ok for 50+ (I know one AliB seller did from me before). I can try them afterwards.

As far as reflow oven... I won't need one. I've been exploring several options since I started this post regarding soldering with a soldering iron, hot air, and (no joke) even a clothes iron. Here's a LED education playlist I been following. Maybe a bad idea idk... lol. I may reconsider...

Lastly, the drivers. refer to my original post for specs.

The test voltage... 600mA @ 3.0Vmax. I'm not doing pulsed voltage, so forget IFP.
Should I go over this rating? Or should I just find a constant Amp driver of 600mA that will provide up to ~30V for each separate wavelength? No meanwell! sorry I'm buying some cheapos if its just 600mA.

Alternatively I can just buy a few stars and a power supply to see how much we can take these babies to... but I believe the wavelength increases as the power rating increases, so maybe not a good idea.



Now if there is a tech wizzy (like Bigclivedotcom) who is lurking and wants to design me a plug'n'play board, have at her. Also, anyone who wants to 'chip in' in any way, you are welcome to it.

I've actually rigged a heatgun on a frame to replace some old diodes on some of my old boards.
I'd look into the XLM2 diodes,10% increase in efficiency over XLM's.

boilingoil said:

I've actually rigged a heatgun on a frame to replace some old diodes on some of my old boards.
I'd look into the XLM2 diodes,10% increase in efficiency over XLM's.

Click to expand...

You can also use Cree XPL's to increase efficiency by another 20%. You can find them for less than 3bucks per diode at ebay or alibaba. They use they same footprint like XML and XML2 so you can change the without issues. Around 400-500lm at 1050mA, +200lm/w at 350mA and at 10w you'll get 120-130lm/w depending on CCT.
I've added the datasheet below, brother...

Randomblame said:

You can also use Cree XPL's to increase efficiency by another 20%. You can find them for less than 3bucks per diode at ebay or alibaba. They use they same footprint like XML and XML2 so you can change the without issues. Around 400-500lm at 1050mA, +200lm/w at 350mA and at 10w you'll get 120-130lm/w depending on CCT.
I've added the datasheet below, brother...

Click to expand...

whoAH XPL DOES NOT USE THE same footprint as XML or XML 2, XPL is a 3.5mm sq pack same as XPG
Cheers
Mark

Oh-oh! Ash on my head... seems I'm wrong!
I thought with its 10w its the successor of XML2. Thanks for correcting me!

welight said:

whoAH XPL DOES NOT USE THE same footprint as XML or XML 2, XPL is a 3.5mm sq pack same as XPG
Cheers
Mark

Click to expand...

Randomblame said:

Oh-oh! Ash on my head... seems I'm wrong!
I thought with its 10w its the successor of XML2. Thanks for correcting me!

Click to expand...

Hi Random
yes your correct the XPL is recommended as the replacement for XML as it uses a slightly smaller cut down die of the XML that allows it to fit into the 3.5mm package, so same spec just the smaller footprint
Cheers
Mark

I'm still working on this... I know its been long, but I finish what I start. Just take my sweet azz time about it is all.

Cost me too much money but I kinda see it like a college class fee. Its the diodes that cost way too much for a poor guy like me; just grabbing 660s ffs; oh well, school cost money. The solder kit I purchased is dumb. It has no constant current, so testing is out of the question. If you increase the voltage, you increase the amps, which lights up the diodes. Its not that dumb actually, cuz if I went too high it beeped and turned off. But really, I should have just purchased a solder iron. HIndSight 100%, but the kit is actually ok, it came with a lot of accessories, and I will use the iron still. so whatever.



The soldering to XM-L chips was accomplished by applying a minimal amount of soldering paste onto the diode footprint on the chip, placing the diode on the chip, and cooking them on a frying pan (set at 4.5/10; ir temperature ~210C and stable) until it seated. If I used too much solder, I had to press a little on the diode (with forceps) and the extra solder squished out the side. I lost (the first) two 700nm diodes trying to solder them with the iron. Don't do this; buy a cheap kit if you want to practice. Oh well, college admin fees I suppose.


Turns out, once you get a hang of it, its a breeze. Started doing two at a time.


I QA'd each diode by visual inspection and testing with the power supply (that I bitched about heh heh).

I'm still waiting on a couple more parts to complete everything, namely the driver and some extra components...

OK project is taking too long, so I am just going to use the official numbers...
question is... '2 strips in parallel of 5 in series' or just '10 in series'?


The first set of numbers is for 10x series (s), the other is for two 5-series in parallel (p). Totals are shown (t). Note that 700 diode doesn't match (grrrr), so... going to do something weird on top of this weirdness.

Cool little heat sink:



It can easily fit 10 stars, so I prepped the diodes with thermoconductive tape and was thinking to stick 5 stars of 680 next to 5 stars of 690 on two heatsinks and have pin connectors to daisy chain the two heatsinks. This results in 10 in series of either 680 or 690 or 680/690 combo. The eight 700 diodes can just go on its own heatsink. The fan is also 12V x 60mA so it would be nice to power everything together...

I purchased this driver:



and also these cool little buck-boosters:

which have a narrower output voltage range (5-30V) than the Meanwell LDB-L (2-40V), but its so adaptable I just had to.

Now the driver can power all the heatsink fans AND the diodes at their specific voltages and amperages... I think.

And thank you to everyone who has been helping me, including LedExpert over at GrassCity.

I will receive the parts soon enough; sorry for the lag in my progress reports... I still need to 'build' a spectrometer.

And please, critical feedback is desired (except for the part that says this costs too much, just buy a prefab 660).

When I get this all done, I will compare with some other 660s (probably, just so busy all the time).

I kinda learned the hard way I suppose, but I'm glad I did.
Learned a lot about diodes and about 'what's it really worth'.

The diodes work. The buckboosters work. The driver, the mounts and fans, all work.
Cost a fair bit, and probably not worth the expense, but hey, I have them and can test them out when I get a spec (coming).

BUt what I really learned is that sometimes its ok to use the next best thing. Why not just use 660nm? or even better, why not just use pre-soldered 730nm diodes that are available for 50c a piece? Quality of diodes? give it a rest. At least for the money I have the specialized diodes that can be used for some research.

and I like HLG products but the new red strip with the 6 diodes for $70 bucks.. idk folks. After doing the soldering and drivers and buckboosters, for the same price, I would have 200 diodes (not 6) that are the next best thing.

I'm going to test a bunch a cheap diodes now too. If I can get 200W of diodes (already mounted/driverless/etc) for $15 and if they grow good weed, then that's the way to go.