DIY leds Discussion Thread for all your how tos and doubts and anything related

[Rocket Soul]

Yea, it gets easier with practice but a single unit just has no economy of scale. To that day and a half add research/calculations to know what you want and then half a day of troubleshooting after it's done.

I got myself a GLA. Couldn't have done it better myself since good UV diodes are difficult to source.

Do you know where to source good 400 nm (ish) UV diodes (bare, not mounted)? I contemplate to DIY something for my fish tank.


At my scale I'm not sure if 3 % are really worth pursuing. Stop using seeds and "pheno hunt" myself a productive strain would be a better payoff. Or many other things I didn't optimize yet.

Theres stuff on alibaba and aliexpress that id rate as good enough. Talked about it with pcbuds the other day: https://www.icmag.com/threads/pcbuds-mini-grow.360815/post-18794714

Not allowed to give links to any of the alibubbers but if you look around aliexpress you almost find any nm, both mounted and unmounted. Modern uvs are mostly good enough.
The regular global distrubtors would work well: digikey, arrow electronics, mouser
I usually do alibaba but rhats like atleast 30$ in shipping so not worth it for a small order; hence aliexpress. TME in europe carries prolightopto who are very good aswell, and a bunch of cheapo brands.
GLA are excellent, strips or boards? We got the strips and been very happy with them. Have two other light fixtures with similar spectrum which also do well and me and my growbuddy never seem to agree which ones is best; sometimes it depends on the strain.

3% was just a number drawn thru the air, what you could get by optimizing spread in a ways you dont get with standard lighting.

 

[Orange's Greenhouse]

Theres stuff on alibaba and aliexpress that id rate as good enough. Talked about it with pcbuds the other day: https://www.icmag.com/threads/pcbuds-mini-grow.360815/post-18794714

Not allowed to give links to any of the alibubbers but if you look around aliexpress you almost find any nm, both mounted and unmounted. Modern uvs are mostly good enough.
The regular global distrubtors would work well: digikey, arrow electronics, mouser
I usually do alibaba but rhats like atleast 30$ in shipping so not worth it for a small order; hence aliexpress. TME in europe carries prolightopto who are very good aswell, and a bunch of cheapo brands.
GLA are excellent, strips or boards? We got the strips and been very happy with them. Have two other light fixtures with similar spectrum which also do well and me and my growbuddy never seem to agree which ones is best; sometimes it depends on the strain.

3% was just a number drawn thru the air, what you could get by optimizing spread in a ways you dont get with standard lighting.

I have the board. I don't see much benefit in the strips unless you need that specific form factor. Your build just spreads them even if I remember correctly? A board can get a similiar output but is much easier to handle. If you were to give yourself a realistic wage it would come to the same cost.

With the chinese diodes I'm not too confident that their wavelength binning is correct. Calibrating for luminous flux differences is simple to do at home (if wavelength is compareable) but I just don't have a spectrometer at home.

I just wanted to say, that I got something else that is easier to optimize than lighting. A proper strain gets me 20 % more productivity. A better light takes proper testing and might yield 3 % which is worth it at scale.

 

[Orange's Greenhouse]

But to get back on topic. @Prawn Connery proposed the idea of using strip light to build a vertical light setup.



I did some sketching and rough calculations. Given his numbers the idea is to light a 4x4 tent. The light is in the middle and 1 m high. The plants are 0,3 m from the lights, giving them 0,3 m of depth which should be sufficient for growth and ventilation. A traditionally lit 4x4 canopy is ~1.5 m². The vertical setup would yield (assuming 1,2 m is lit) 2.25 m² or 50 % more.

To light 2.25 m² at 900 µmol/(m²*s) you need a light with 2025 µmol/s. With a system efficiency of say 2.6 µmol/J that's 778 W.

You can use 6 strips per section and two sections high for a total of 12 strips. That's 64 W per strip. Which is a bit high but could be done if desired. More on that later.



Here is a mockup of how that could be assembled. The strips are roughly looking like the GLA branded ones and critical dimensions are correct. The idea is to have fittings that hold the strips in a hexagon shape. In the middle are hexagon brackets to provide additional support (only one is shown but multiple would be used). The bottom fitting provides an adaptor for an inline fan. This provides forced convection and is essential to remove heat. Above this module would be another set of strips to get to the desired height. Cable management is left as an exercise to the reader.

At the high wattage heat sinks would be necessary. The problem here is that heat sinks cost money and take time to assemble. I have seen some nice heat sinks that can be cut to size, are small (40 or 50 mm wide and 6 mm think) but cost >100 €/m, adding additional cost. Assembly gets more involved too.

Alternatively it would be easier to increase the strip count to 10 per section, 20 total. This is 39 W per strip which is ok with forced convection and increases the diameter of the module from ~110 mm to ~180 mm. Price would be the same because no heat sinks are required.

 

[Rocket Soul]

But to get back on topic. @Prawn Connery proposed the idea of using strip light to build a vertical light setup.



I did some sketching and rough calculations. Given his numbers the idea is to light a 4x4 tent. The light is in the middle and 1 m high. The plants are 0,3 m from the lights, giving them 0,3 m of depth which should be sufficient for growth and ventilation. A traditionally lit 4x4 canopy is ~1.5 m². The vertical setup would yield (assuming 1,2 m is lit) 2.25 m² or 50 % more.

To light 2.25 m² at 900 µmol/(m²*s) you need a light with 2025 µmol/s. With a system efficiency of say 2.6 µmol/J that's 778 W.

You can use 6 strips per section and two sections high for a total of 12 strips. That's 64 W per strip. Which is a bit high but could be done if desired. More on that later.



Here is a mockup of how that could be assembled. The strips are roughly looking like the GLA branded ones and critical dimensions are correct. The idea is to have fittings that hold the strips in a hexagon shape. In the middle are hexagon brackets to provide additional support (only one is shown but multiple would be used). The bottom fitting provides an adaptor for an inline fan. This provides forced convection and is essential to remove heat. Above this module would be another set of strips to get to the desired height. Cable management is left as an exercise to the reader.

View attachment 19091266

View attachment 19091267



At the high wattage heat sinks would be necessary. The problem here is that heat sinks cost money and take time to assemble. I have seen some nice heat sinks that can be cut to size, are small (40 or 50 mm wide and 6 mm think) but cost >100 €/m, adding additional cost. Assembly gets more involved too.

Alternatively it would be easier to increase the strip count to 10 per section, 20 total. This is 39 W per strip which is ok with forced convection and increases the diameter of the module from ~110 mm to ~180 mm. Price would be the same because no heat sinks are required.

I think i saw @Scfarmer with similar ideas and ill admitt that ive also dabbled with the idea. I think i did some looking around and found it very hard to find any hexagonal tubing at anything but huge prices. Went with the idea to use square which i could find but one problem remains of access to the inside of the tube: whos skinny baby hands and arms reaches inside to attach the screws inside? Youd probably best of having some type of dummy test before getting the tubing. And youre likely needing one very anorectic girlfriend for this.
The benefit of a design like this is oblivious, you could extract air thru it making cooling the strips and renewing the air all in one go. But it would mean always extracting the bottom. Another similar approach would using 4x "lipped" heatsinks (i think the glas sinks come like that, and making holes in lip and finn and attach with screw and nuts. But seems a bit of a big project with lots of costs: 4 boards per per 40cm section.

My GLA light; im not sure i posted it here yet but it is up somewhere on RIU; yes it was more or less even distribution- a little closer on the last couple of heatsinks i think to get that even spread. I did mess that one up a bit, initially i had the strips the wrong way, lol.
The whole point of strips in that build was
1: my grow buddies preference
2 :access to fitted heatsinks locally thru t-slot
3 : boards need a bit higher hanging height, in openspace this can hamper you with light losses on walls.

Boards have a slightly different spectrum aswell, more fullcycle than the boards. They do great in our vegg aswell but should really replace that with a dedicated veg light: 8 of those strips to light one tray of vegg at 150w seems wasteful to me. It just came about that way during a refit and we went with it to try. Always try to put special cuts under it and you can see a slight better health under them than our standard 5k+ 660 veg light.

 

[Rocket Soul]

I have the board. I don't see much benefit in the strips unless you need that specific form factor. Your build just spreads them even if I remember correctly? A board can get a similiar output but is much easier to handle. If you were to give yourself a realistic wage it would come to the same cost.

With the chinese diodes I'm not too confident that their wavelength binning is correct. Calibrating for luminous flux differences is simple to do at home (if wavelength is compareable) but I just don't have a spectrometer at home.

I just wanted to say, that I got something else that is easier to optimize than lighting. A proper strain gets me 20 % more productivity. A better light takes proper testing and might yield 3 % which is worth it at scale.

If you optimize both you get 23 i just meant that for that perfect even spread which in the end can give you some decimals diy is the way to go. And getting it right WILL make you feel awesome- building and repairing is just such a kick. My proudest moment of this year was first crashing and breaking my projector, damning myself, only to figure out it was optic related and opening that sucker up and to my own great surprise managing to fix it! In one fell swoop i scored every last man point and beamed like a little sun for a week. Getting your own light perfect is the growers version of this: every single crop and record you break will only renew that feeling. Its that "smells like victory"-feeling from apocalypse now, lol, beating the world at its own game by just bending it to your will, just using a drill, saw and a screwdriver

Inexact binning in china would not surprise me; my guess is that a lot of diodes rejected for being somewhat off spec in cct would go to lights in the past. In my latest project it seems like the supplier is sourcing thru the globals, digikey/arrow etc rather than some fly by night china operation.
The amount of cooperation and patience ive had recently may be indicative on a slowdown in the industry and in general in china: alibaba negotiations used to be very to the point before: "you buy now?"-like rather than really trying to get you what you want in your heart-of-custom-heart.

 

[Rocket Soul]

It would be really cool to get an idea what people would do for a 1m2 tent with diy; finding the right solution seems hard. Some would say strips, some may say boards. What do you diy minded guys say? Candidates for even spread seems to preclude 2 foot strips unless you want to put a bunch in the middle or make something with both, middlestrips and strips around a square frame. Who has a go to design here which both gives even spread and doesnt eat up too much vertical space? I got some suggestions but its nice to hear from everybody...

 

[led05]

Any build pics & finished results?

Here’s a 500w one I did, I’ve done so many now I’ve lost count (25+) & even started selling some off, it’s addicting building them

 

[Rocket Soul]

Any build pics & finished results?

Here’s a 500w one I did, I’ve done so many now I’ve lost count (25+) & even started selling some off, it’s addicting building them

View attachment 19091284

This is really nice! The other thread is dedicated to builds and buds results, this one is open for anything diy discussion jibber jabber; want to keep the other on subject, with this thread to divert any discussion going beyond the actual builds/buds. But maybe some info on components, what space your running it in? Cool to see builds with cobs still nowadays
Build and bud thread: https://www.icmag.com/threads/diy-led-growlights-builds-and-results.18131610/

And check out the poll aswell here in this thread

 

[led05]

Cool to see builds with cobs still nowadays

Ty & Yup, plants crave more than just low wattage high count underdiven LED IMO, cobs driven hard are very bright AND put out some radiant which is a requirement of all plants, I won't get into Phytochromes, Emerson Effect etc here, I also use LED side by side with HID HPS and Fluro pure UVB, LED's regardless of what some say are horrible at producing real / true UV, they simply burn themselves out or are cost prohibitive

after decades of using LED and HID etc, the perfect combo is LED, DE HPS & Fluro for UVB

Here's a 880w LED I built, this thing is bright !

Peace

 

[Prawn Connery]

But to get back on topic. @Prawn Connery proposed the idea of using strip light to build a vertical light setup.



I did some sketching and rough calculations. Given his numbers the idea is to light a 4x4 tent. The light is in the middle and 1 m high. The plants are 0,3 m from the lights, giving them 0,3 m of depth which should be sufficient for growth and ventilation. A traditionally lit 4x4 canopy is ~1.5 m². The vertical setup would yield (assuming 1,2 m is lit) 2.25 m² or 50 % more.

To light 2.25 m² at 900 µmol/(m²*s) you need a light with 2025 µmol/s. With a system efficiency of say 2.6 µmol/J that's 778 W.

You can use 6 strips per section and two sections high for a total of 12 strips. That's 64 W per strip. Which is a bit high but could be done if desired. More on that later.



Here is a mockup of how that could be assembled. The strips are roughly looking like the GLA branded ones and critical dimensions are correct. The idea is to have fittings that hold the strips in a hexagon shape. In the middle are hexagon brackets to provide additional support (only one is shown but multiple would be used). The bottom fitting provides an adaptor for an inline fan. This provides forced convection and is essential to remove heat. Above this module would be another set of strips to get to the desired height. Cable management is left as an exercise to the reader.

View attachment 19091266

View attachment 19091267



At the high wattage heat sinks would be necessary. The problem here is that heat sinks cost money and take time to assemble. I have seen some nice heat sinks that can be cut to size, are small (40 or 50 mm wide and 6 mm think) but cost >100 €/m, adding additional cost. Assembly gets more involved too.

Alternatively it would be easier to increase the strip count to 10 per section, 20 total. This is 39 W per strip which is ok with forced convection and increases the diameter of the module from ~110 mm to ~180 mm. Price would be the same because no heat sinks are required.

That's almost exactly what I had in mind – nice graphic!

And yes, I was also thinking of using a total of 12 strips and your maths is almost identical to what I had calculated. A simple way of looking at it would be to compare how many watts of LED you would need to replace 1200W of HPS – which I have used in the past to very good effect – and the answer is around 800W or less.

With active cooling (computer or floor fan on the bottom blowing up through the middle), 60-65W per strip shouldn't be an issue. Although admittedly I was thinking around 600W (50W per strip) would be a good start.

Alternatively, an octagon would also be ideal. That would be 16x strips running at no more than 50W each for 800W total.

Th wiring would run through the middle to hide it for a neat solution, and the power cable would be at the top, out of the way.

 

[Rocket Soul]

That's almost exactly what I had in mind – nice graphic!

And yes, I was also thinking of using a total of 12 strips and your maths is almost identical to what I had calculated. A simple way of looking at it would be to compare how many watts of LED you would need to replace 1200W of HPS – which I have used in the past to very good effect – and the answer is around 800W or less.

With active cooling (computer or floor fan on the bottom blowing up through the middle), 60-65W per strip shouldn't be an issue. Although admittedly I was thinking around 600W (50W per strip) would be a good start.

Alternatively, an octagon would also be ideal. That would be 16x strips running at no more than 50W each for 800W total.

Th wiring would run through the middle to hide it for a neat solution, and the power cable would be at the top, out of the way.

I wanna see this done badly. But i think getting the hectagonal/octagon alu will be quite hard and expensive.
Please remember that any screw and nut solution needs to come with access to the inside of the pipe with nimble little hands or alternative is to tap hole to screw in directly without nuts. Hope your doing well Prawn and always a delight to see you around

 

[liam788fh]

If you optimize both you get 23 i just meant that for that perfect even spread which in the end can give you some decimals diy is the way to go. And getting it right WILL make you feel awesome- building and repairing is just such a kick. My proudest moment of this year was first crashing and breaking my projector, damning myself, only to figure out it was optic related and opening that sucker up and to my own great surprise managing to fix it! In one fell swoop i scored every last man point and beamed like a little sun for a week. Getting your own light perfect is the growers version of this: every single crop and record you break will only renew that feeling. Its that "smells like victory"-feeling from apocalypse now, lol, beating the world at its own game by just bending it to your will, just using a drill, saw and a screwdriver

Inexact binning in china would not surprise me; my guess is that a lot of diodes rejected for being somewhat off spec in cct would go to lights in the past. In my latest project it seems like the supplier is sourcing thru the globals, digikey/arrow etc rather than some fly by night china operation.
The amount of cooperation and patience ive had recently may be indicative on a slowdown in the industry and in general in china: alibaba negotiations used to be very to the point before: "you buy now?"-like rather than really trying to get you what you want in your heart-of-custom-heart.

After decades of experience with LED, HID, and other lighting technologies, the ideal combination has proven to be LED, double-ended HPS, and fluorescent lighting for UVB.

 

[led05]

After decades of experience with LED, HID, and other lighting technologies, the ideal combination has proven to be LED, double-ended HPS, and fluorescent lighting for UVB.

Sounds like an echo in here

after decades of using LED and HID etc, the perfect combo is LED, DE HPS & Fluro for UVB

Here's a 880w LED I built, this thing is bright !

View attachment 19091519


Peace

 

[Orange's Greenhouse]

UVB Diodes are efficient and cheap nowadays. And why add the fire risk of HID lighting when you can get the same results by controlling the environment?

 

[led05]

UVB Diodes are efficient and cheap nowadays. And why add the fire risk of HID lighting when you can get the same results by controlling the environment?

You’re being tricked I’d bet, what range nm are your UVB diodes?

Control of environment doesn’t equal radiant energy blasted onto plants like the SUN or say DE HID does, plants crave it. HID is no more dangerous than LED or a space heater etc, it’s the people operating things & overloading power on lines that is….

I’ve been testing led for more than two decades to grow, few can say this it’s not like I’m against them, I just know you need that radiant energy & DE HPS is the best & cheapest way to do that - what do you use for radiant energy?

 

[Rocket Soul]

UVB Diodes are efficient and cheap nowadays. And why add the fire risk of HID lighting when you can get the same results by controlling the environment?

Indeed, if you source them from the right place. Digikey still lists them around 20e a pop.
They also have the advantage of being dimmable, better possibility to spread them, and no problems with turning them on and off every hour if one prefers the uvb pulse rather than keeping them on all the time.

 

[led05]

Indeed, if you source them from the right place. Digikey still lists them around 20e a pop.
They also have the advantage of being dimmable, better possibility to spread them, and no problems with turning them on and off every hour if one prefers the uvb pulse rather than keeping them on all the time.

I’ve been buying from Digi-Key, Mouser, Cutter etc for decades now; talk to me specifics please about these UVB diodes, what are there specs, what is the NM range and what are the life of them?

There’s a reason industry uses Fluros for UVB and not LED, many many many industries

 

[Orange's Greenhouse]

You’re being tricked I’d bet, what range nm are your UVB diodes?

Control of environment doesn’t equal radiant energy blasted onto plants like the SUN or say DE HID does, plants crave it. HID is no more dangerous than LED or a space heater etc, it’s the people operating things & overloading power on lines that is….

I’ve been testing led for more than two decades to grow, few can say this it’s not like I’m against them, I just know you need that radiant energy & DE HPS is the best & cheapest way to do that - what do you use for radiant energy?

Why do you need radiant energy? PAR is required for photosynthesis and you it them most efficiently from LEDs. If you need heat, then use a damn heater. HID radiation to heat leaves is indistinguishable from convection or black body radiation from the space around it. The cheapest way to heat are fossil fuels or heat pumps. Fuel is cheap and Heat pumps have an efficiency > 100 %.
The last CMH I owned was designed to explode at the end of its lifetime. High temperatures are never safe.

There’s a reason industry uses Fluros for UVB and not LED, many many many industries

You mean they don't throw everything they have out the moment a new technology is available? The reasons you're looking at are inertia and path decisions.

 

[Rocket Soul]

I’ve been buying from Digi-Key, Mouser, Cutter etc for decades now; talk to me specifics please about these UVB diodes, what are there specs, what is the NM range and what are the life of them?

There’s a reason industry uses Fluros for UVB and not LED, many many many industries

Alibaba diodes, not sure what life span you get from them tbh. But the diodes ive bought and tested (640/660/680) were all on spec.
Id go for 285nm for best coverage of uvr8 gene actionspectrum, with minimum of uvc. 280nm is very common but half of the output is uvc, best to avoid

 

[led05]

Alibaba diodes, not sure what life span you get from them tbh. But the diodes ive bought and tested (640/660/680) were all on spec.
Id go for 285nm for best coverage of uvr8 gene actionspectrum, with minimum of uvc. 280nm is very common but half of the output is uvc, best to avoid

What are you using to test nm specific wavelengths…?

That said LED UVB Tech still isn’t there; I’ve been involved in the lighting space for decades; UVB diodes are cost prohibitive & burn themselves out fast - ushio makes a Fluoro that peaks at 306nm & spectrum covers 280-320nm, it follows vitamin D synthesis curve too which is exactly what UVB erythema does & is an ideal pure UVB bulb, there’s a lot of science behind this tech and it’s used for various industries including the medical fields -

there's a reason you’ve gotta go fishing on alibaba for them “UVB” diodes, think about it -

Why do you need radiant energy? PAR is required for photosynthesis and you it them most efficiently from LEDs. If you need heat, then use a damn heater. HID radiation to heat leaves is indistinguishable from convection or black body radiation from the space around it. The cheapest way to heat are fossil fuels or heat pumps. Fuel is cheap and Heat pumps have an efficiency > 100 %.
The last CMH I owned was designed to explode at the end of its lifetime. High temperatures are never safe.


You mean they don't throw everything they have out the moment a new technology is available? The reasons you're looking at are inertia and path decisions.

Heat pumps are > than 100% efficient, right next to perpetual motion eh? Think you’re gonna loose most arguments stating nonsense like that in your posts - lol !

If you don’t understand plants crave radiant energy being blasted on them you need to head back to AG class & the fields; anytime you wanna have a grow off of any species of plants lmk - friendly challenge

lol on new tech, led’s are replacing nearly everything, there’s zero replacement in the UV medical, health & industrial fields and for good cause, Fluoro tech dominates LED in the space, nature of the beast - do a bit of research & learn a few things before you speak condescendingly to someone whom knows a great deal of what their talking about…

I was the guy pushing LED back when everyone was calling me crazy (have probably 50 builds under my belt, spanning decades), now all those folk are pushing LED calling me crazy for still liking HID - it’s a trip

My only bias & agenda is to use what works best; I have more lights sitting @ here of all tech than I can count, the luxury to use exactly what I want ($$$ not a factor) and yet I use exactly as I’ve stated above for the very simple reason - it works best for now,

Id love me some new cutting edge tech though, please just point me in the actual direction(s) it exists and I’m game to test out myself…

99/100 times I hear people talking LED UVB and then they link to purple 380nm diodes, at least it seems you guys understand what nm range UVB Actually is

I’ve got expensive meters to test for UVB too, yet to find a diode actually putting out ANY !