HiloReign
Well-Known Member
Super cool link Guod, veg/flower in one array.
PICO's DIY Thread - Advise, Ideas and Technology - NO PANEL REPS!
- Thread starter PICOGRAV
- Start date
PICOGRAV
Well-Known Member
Love it, I really think the 5000K is not going to be a wise choice, learn from the past 90 days of testing I have done...
PICOGRAV
Well-Known Member
As you can see the E, G and H refer to the CRI or how accurate they are at reproducing the normal daylight spectrum. The higher the better but you will lose efficiency the further you go. The 90s or "G" is what I would go with for this project.
Mechmike
Well-Known Member
The E designates the 80CRI, G=90CRI, H=97CRI. There higher CRI LEDs typically have a lower Lumen/watt output. I suggest going for the higher output. The discussion about 5000k/2700k 2:1 is valid for flower only for sure. I was trying to get you the most lumens with a good spectrum for both veg and flower. I think it would work just as well with a 1:2 ratio with less lumens, a bit more stretch and a slightly better flower response. Considering the size of the area you want to light you could probably do with 3 Veros only. I'm thinking that a PC case with 4000 lumens inside is like direct sunlight bright. There are some pretty impressive grows here on RIU done with white LEDs exclusively. Unfortunately I don't know of anyone selling the drivers and red/blue multichips in the US. There are some drivers available but I could never swallow the prices asked.
PICOGRAV
Well-Known Member
You see the problem is I once thought like you, I had to replicate the suns power, I thought that was king...
You might have the intensity of day light yes, but you are way off thinking that is what sunlight looks like, the sun is just a beautiful hydrogen gas explosion, with a smooth warm intense burn. The light travels through our atmosphere, sorta like a filter lighting us up.
Remember that the integrity of the CRI will always trump the intensity of the lumens, some times it just takes awhile too find out -
PetFlora
Well-Known Member
My view of simulating the sun has to do with what colors the sun provides during early veg (spring) growth (summer) flower (fall) + relative temps- cooler in early veg, warm/hot in summer, cooler in fall. Beyond that is optimal umols/m^, which I have yet to achieve, but still get vg results.
Neither do I have an environmentally controlled grow room. Summers here are long and often 95* heat and humidity. Even waiting until September's first new moon can be too hot/humid as well as late November, meaning my results (even though indoors) are often weather compromised
Neither do I have an environmentally controlled grow room. Summers here are long and often 95* heat and humidity. Even waiting until September's first new moon can be too hot/humid as well as late November, meaning my results (even though indoors) are often weather compromised
You see the problem is I once thought like you, I had to replicate the suns power, I thought that was king...
You might have the intensity of day light yes, but you are way off thinking that is what sunlight looks like, the sun is just a beautiful hydrogen gas explosion, with a smooth warm intense burn. The light travels through our atmosphere, sorta like a filter lighting us up.
Remember that the integrity of the CRI will always trump the intensity of the lumens, some times it just takes awhile too find out -
smokey the cat
Well-Known Member
So I've been sketching out some different lighting models with various top shelf emitters that are available these days - for those of you following along this is because a few pages back we established that Vero can't be run in parrallel with simple AC drivers, and the high forward voltage makes running Vero in series a real pain.
I started looking at Cree XP-g2/XM-L2 alternatives. Cutter has some $15 PCBs pre-loaded 3 XP-G2 warm white in series on them - they look great. I started sketching out some ideas of running 4-6 of these in series, trying to get around 50-100W of warm white output with dimmable drives.
After more than a few hours looking at data sheets I eventually came to the awful conclusion - this is stupid. Sure, I could build a nice array of relatively efficient LED modules, but why the hell am I trying to stitch together 12 to 18 small emitters? It just seems dumb to try to reinvent the wheel - Cree and Bridgelux have already stiched together a bunch of small emitters for us with the CXA and Vero. And they wont need me to farf around with my sub-standard soldering skillz.
Seriously: head-desk moment. But it's been useful and a nice learning experience.
So, back to the Vero. Particularly the Vero 18 - after a bunch of research I think the secret is to be a bit more conservative with my expectations and avoid running arrays of these things. The beautiful thing about the Vero is that it seems to match up really nicely with the $36 Meanwell LPF-60d-36. One driver per emitter. And it's a cheap yet reliable driver, from the budget end of the MeanWell range, offering simple resistance dimming making an easy build.
The thing is that the cheap chinese Ebay drivers arent that much cheaper. Around $20 for a non-dimming unit, with completely unknown quality, probable awful EMI and over-specced in the auction description, and will probably run much hotter. I just can't justify saving the $30-$40 for two drivers that I want to grow with for 5 years.
The LPF-60D-36 covers the operating voltage of the Vero 18 beautifully. According to the data sheet the V18 runs from 22-35V at absolute extremes, the LPF-60D-36 covers 21-36V.
The driver wont push the V-18 to max though - LPF-60D-36 gives us 1.67A (max given for the V-18 is 2.1A, nominal test current is 1.05A). At the max current the driver can feed the Vero should run at 52.8W, producing somewhere in the region of 140%-145% of nominal illumination.
MrFlux's brilliant CXA thread - https://www.rollitup.org/led-other-lighting/743645-cree-cxa-analysis.html - has a look at the Vero, and picks the 3000k tint as the sweet spot in the range. Vero 18 3000K are only $17 on Newark for the lowCRI/high-output version. Add a couple of dollars for the wiring harness and you have an emitter that can plug straight into the Meanwell without even firing up the soldering iron, let alone the multimeter. The hardest thing looks like it'll be finding the right heatsink for my cab and attaching the Vero - (two m2.5 self-tapping screws FYI). I will need the soldering iron to hook up a 100kohm trimpot to the dimming input of the driver mind.
The only negatives to this that I can see is that the Mean Well is PWM dimming. Is this really a thing? Efficiency at very low levels wont be that great, and it can't dim to zero. Neither of these seem like a genuine negative to me - especially considering how cheap the drivers are and I don't need super-low or zero dimming.
I'm planning two 3000k V-18s for a 2.2sq foot micro cab - because I only have 90cm of vertical space I am expecting to be running these closer to 50% of max power (~55W total) than 100% of max (~105W total). With adjustable dimming output I should be able to veg and flower with these, even with only the 3000k spectrum, they look like they have enough blue in the output.
Does this plan look like it will succeed? Anything I've missed here?
If this looks right to you it means that if you're US based and source a cheap Ebay heatsink you can strap to it two Vero-18s powered by two LPF-60D-36. You'll have built a dimmable warm white panel with 105W of output for about $100 bucks US. With a 5 year guarantee for the drivers. Without having to solder any parts more critical than a couple of $2 trimpot. This is truly insane.
Driver link - http://www.meanwell.com/search/LPF-60D/LPF-60D-spec.pdf
I started looking at Cree XP-g2/XM-L2 alternatives. Cutter has some $15 PCBs pre-loaded 3 XP-G2 warm white in series on them - they look great. I started sketching out some ideas of running 4-6 of these in series, trying to get around 50-100W of warm white output with dimmable drives.
After more than a few hours looking at data sheets I eventually came to the awful conclusion - this is stupid. Sure, I could build a nice array of relatively efficient LED modules, but why the hell am I trying to stitch together 12 to 18 small emitters? It just seems dumb to try to reinvent the wheel - Cree and Bridgelux have already stiched together a bunch of small emitters for us with the CXA and Vero. And they wont need me to farf around with my sub-standard soldering skillz.
Seriously: head-desk moment. But it's been useful and a nice learning experience.
So, back to the Vero. Particularly the Vero 18 - after a bunch of research I think the secret is to be a bit more conservative with my expectations and avoid running arrays of these things. The beautiful thing about the Vero is that it seems to match up really nicely with the $36 Meanwell LPF-60d-36. One driver per emitter. And it's a cheap yet reliable driver, from the budget end of the MeanWell range, offering simple resistance dimming making an easy build.
The thing is that the cheap chinese Ebay drivers arent that much cheaper. Around $20 for a non-dimming unit, with completely unknown quality, probable awful EMI and over-specced in the auction description, and will probably run much hotter. I just can't justify saving the $30-$40 for two drivers that I want to grow with for 5 years.
The LPF-60D-36 covers the operating voltage of the Vero 18 beautifully. According to the data sheet the V18 runs from 22-35V at absolute extremes, the LPF-60D-36 covers 21-36V.
The driver wont push the V-18 to max though - LPF-60D-36 gives us 1.67A (max given for the V-18 is 2.1A, nominal test current is 1.05A). At the max current the driver can feed the Vero should run at 52.8W, producing somewhere in the region of 140%-145% of nominal illumination.
MrFlux's brilliant CXA thread - https://www.rollitup.org/led-other-lighting/743645-cree-cxa-analysis.html - has a look at the Vero, and picks the 3000k tint as the sweet spot in the range. Vero 18 3000K are only $17 on Newark for the lowCRI/high-output version. Add a couple of dollars for the wiring harness and you have an emitter that can plug straight into the Meanwell without even firing up the soldering iron, let alone the multimeter. The hardest thing looks like it'll be finding the right heatsink for my cab and attaching the Vero - (two m2.5 self-tapping screws FYI). I will need the soldering iron to hook up a 100kohm trimpot to the dimming input of the driver mind.
The only negatives to this that I can see is that the Mean Well is PWM dimming. Is this really a thing? Efficiency at very low levels wont be that great, and it can't dim to zero. Neither of these seem like a genuine negative to me - especially considering how cheap the drivers are and I don't need super-low or zero dimming.
I'm planning two 3000k V-18s for a 2.2sq foot micro cab - because I only have 90cm of vertical space I am expecting to be running these closer to 50% of max power (~55W total) than 100% of max (~105W total). With adjustable dimming output I should be able to veg and flower with these, even with only the 3000k spectrum, they look like they have enough blue in the output.
Does this plan look like it will succeed? Anything I've missed here?
If this looks right to you it means that if you're US based and source a cheap Ebay heatsink you can strap to it two Vero-18s powered by two LPF-60D-36. You'll have built a dimmable warm white panel with 105W of output for about $100 bucks US. With a 5 year guarantee for the drivers. Without having to solder any parts more critical than a couple of $2 trimpot. This is truly insane.
Driver link - http://www.meanwell.com/search/LPF-60D/LPF-60D-spec.pdf
NewGrower2011
Well-Known Member
What type of cable lengths can be achieved between driver(s) and the actual light fixture if someone wanted to try and design a DIY with a 'remote ballast' type of approach?
I've seen the DC-DC parts on various sites where there's a big metal cage (AC to DC transformer) that can then drive 3 or was it 4 output circuits with 48v DC? Then there were little black sealed boxes with pigtails in/out that actually went to the LED(s) (basically just a DC driven LED driver instead of AC based right?)... I'm new to this so I'm seeing this as opposed to these other meanwell drivers/etc where it's a single sealed plastic box with an AC pigtail on the one end and then the LED output on the other... So, what's the difference here and why would you pick one camp over the other? Also would the other AC/DC transformer need an enclosure to be 'waterproofed' or IP65 or whatever the spec is? It's obviously not waterproof with the open mesh for ventilation...
I'm looking to see if that approach buys me some additional design options and whether it would change the parts I would need...
i.e. Voltage drop/loss over distance? Would I need some magic formula to figure out my loss and then add that to the demand calculated for a given driver?
i.e. Don't do it!!! Here's why...
i.e. It's plain inefficient and a waste...
I've seen the DC-DC parts on various sites where there's a big metal cage (AC to DC transformer) that can then drive 3 or was it 4 output circuits with 48v DC? Then there were little black sealed boxes with pigtails in/out that actually went to the LED(s) (basically just a DC driven LED driver instead of AC based right?)... I'm new to this so I'm seeing this as opposed to these other meanwell drivers/etc where it's a single sealed plastic box with an AC pigtail on the one end and then the LED output on the other... So, what's the difference here and why would you pick one camp over the other? Also would the other AC/DC transformer need an enclosure to be 'waterproofed' or IP65 or whatever the spec is? It's obviously not waterproof with the open mesh for ventilation...
I'm looking to see if that approach buys me some additional design options and whether it would change the parts I would need...
i.e. Voltage drop/loss over distance? Would I need some magic formula to figure out my loss and then add that to the demand calculated for a given driver?
i.e. Don't do it!!! Here's why...
i.e. It's plain inefficient and a waste...
idk whats better, i went for a lot of ac-dc Drivers (11 pieces lol). If you search voltage drop calculator on Google, u can put in your specs and get the result....I think putting the Drivers at a distance is a good idea, i dont want to be near the mains...i suggest you think about some sort of connector for your Drivers. If you want to take your light out for repairs or changes you dont want to unscrew 100 wires and think about which wire belongs where. So i went with a Mainboard atx connector cable, which only fits/connects in one way...just as a Suggestion. So when the Panel is done, its just plug and Play.
Attachments
NewGrower2011
Well-Known Member
Right. I figure if I do go down that road I'll look at something like a Foxconn catalog and look for something with the right gauge of wire and a 1-way-to-fit industrial/rugged connector... waterproof/etc... I know something out there exists... just depends on # of channels....
or take a look at IDE power connectors (those which you put in your cdrom drive or harddrive), im using them too,since i got 11drivers and that motherboardconnector only gives 10 channels, and i Need to connect my heatsink fans aswell. Gives you "2 channels" per Piece (2x +, 2x-). What do you Need "waterproof" for? Dont overcomplicate things 
guod
Well-Known Member
looong thread...
[URL]http://www.reefcentral.com/forums/showthread.php?t=2222702
PCBs for the Meanwell LDD ..
http://coralux.net/?wpsc-product=ldd-5-driver-board[/URL]
http://www.youtube.com/watch?v=jbbnzNmOg9c
NewGrower2011
Well-Known Member
Guod - awesome vid... thanks!
smokey the cat
Well-Known Member
Thanks for the response mate - definitely couldn't have got this far without you guys giving input.
So I just spent the last couple of hours ordering the parts for this grow light online. One 3000k Vero-18, one in 3500k. I've coped pretty well with stretch so far, so here's hoping the blue from one 3500k light is enough to keep things under control. Probably don't need 100% of output, so I'm prepared to run one light harder than the other depending if I want to dial down blue or red for veg or flower.
The heatsink I got has plenty of room on it, so I can always add a couple of 3W blue emitters if it gets too leggy, or 660nm or 740nm if I want to experiment with deep reds.
And, if anyone else is like me, and doesn't want to drill a heatsink, I found this lovely looking stuff at StevesLEDs - http://shop.stevesleds.com/Thermal-Adhesive-1-Part-Glue-TSE3941M-Thermal-Adhesive.htm Thermal glue - not epoxy. Supposedly as conductive as paste, and LEDs are easy to remove and remount. I haven't mounted a CPU in years so have no idea where my thermal compound is anymore, so I had to buy something anyway and this looked like a lazy man's way of avoiding mounting mishaps and broken drill bits.
not really, just look at some arctic silver and compare...those r like 10times more conductive. In their defense they dont really compare it to thermal compound, just adhesive. Other than that it sounds really nice. Have you found Reviews for that stuff? If it works -nice, but a cob is not a small 3W led, make sure to give a Feedback on how it worked out
Good luck.
smokey the cat
Well-Known Member
Here's the source I think: http://www.momentive.com/Products/home.aspx?id=20992&tid=1139\\
With thermal conductivty listed as 0.84W/m*K it seems very close to Arctic Silver's claimed 0.87W/m*k. White silicon paste is supposed to be mid-70sW/m*k iirc, so it seems well rated.
It'll need a good tight fit to be able to compete with screwing down though. Bridgelux recommend tightening till the pastic surround deforms
With thermal conductivty listed as 0.84W/m*K it seems very close to Arctic Silver's claimed 0.87W/m*k. White silicon paste is supposed to be mid-70sW/m*k iirc, so it seems well rated.
It'll need a good tight fit to be able to compete with screwing down though. Bridgelux recommend tightening till the pastic surround deforms
smokey the cat
Well-Known Member
It's not Arctic Silver epoxy - its some weird one-part hygroscopic stuff from Momentive. At 8.4w/mk its closer to AS5, rather than the epoxy.
Gotta be honest I really have no idea what the performance difference is between any of this stuff in real world applications. A decade ago when when I was a thermal compound obssessive this goop review was considered compulsory reading - the guy tests a bunch of thermal goops alongside a couple of unusual suspects. http://www.dansdata.com/goop.htm (note the heat resistance figures he provides reflect his testing setup only).
Can always bust out some toothpaste to compare, lol
Gotta be honest I really have no idea what the performance difference is between any of this stuff in real world applications. A decade ago when when I was a thermal compound obssessive this goop review was considered compulsory reading - the guy tests a bunch of thermal goops alongside a couple of unusual suspects. http://www.dansdata.com/goop.htm (note the heat resistance figures he provides reflect his testing setup only).
I had a think about it - the thermal patch on the back of the Vero is pretty similar to a CPU die and at most should be generating around 30-35W of heat the way I'm gonna use it. Providing I can get a decent contact I'm pretty confident in any product from white goop up to the AS5. I really think clamp strength is gonna be the big issue, not compound conductivity.
Can always bust out some toothpaste to compare, lol