The Planted Tank Forum banner
1 - 20 of 43 Posts

·
Registered
Joined
·
140 Posts
Discussion Starter · #1 ·
First, let me say I’ve read a TON of threads here on DIY LED stuff, but I haven’t found anything matching my goals.

I have a 48x18x18 tank (64gal) currently running a Fluval 3.0, which I have decided to supplement with a pair of DIY LED pendants. Depending on how I like the pendants, they may totally replace the Fluval. Growth is great with the Fluval, but I’d like a bit more intensity at the substrate to get stuff to grow bushier. I’m also looking for better coverage; my Fluval leaves a big dark spot at the surface. My primary goal is to make the plants/tank look nicer. I find the Fluval doesn’t have enough punch in the red channel for my taste, so I’d specifically like more power there. I’m also looking for a bit more shimmer/caustics.

I want the setup to be per-channel dimmable/programmable (likely via Arduino), but I should be able to figure that out on my own. Seems like there's a lot of good info here that I just haven't explored yet.

I played around with Spectra and worked out a bunch of overcomplicated setups before deciding to simplify by using high-CRI whites (which are sadly not programmed into Spectra). Particularly, I came across the Bridgelux Vesta Tunable Gen 2, which is a single COB with cool and warm white channels. The lower CRI offerings are not terribly remarkable, but the 95-CRI “Thrive” version seems unparalleled.
1031390


This option is appealing to me because of its simplicity and small package size (which I imagine leads to nice color mixing). I plan to get the BXRV-TR-2765S-20A0-A-23-S2, which can put out about 1700 lumens between both channels. My Fluval produces 4250 lumens, so a pair of the Vesta chips would fall a bit short of being able to totally replace it. Maybe I’d add a third pendant if I decided to replace rather than supplement it.

Anyway, my question is what to round out the tunable chip with. I’m sure the chip on its own would be plenty of light, look nice, grow plants well, etc., but I want to be able to fine-tune color more than cool/warm. The Vesta chip is a 20mm square, so a nice arrangement would be one 2cm star board on each side. I’m drawn to the triple chip star like these (same site has two others that are a bit more expensive). I’m thinking violet, blue (specifically 460nm to fill the dip in the 6500K), deep red, and far red. Or perhaps I should trade the far red for blue or a royal blue – I’m wary of a red-heavy build causing leggy plant growth, and most LED systems have a pretty sharp royal blue spike I’ll be lacking.

I’m also curious if you have any recommendations regarding optics and/or reducing glare, etc. One big pet peeve is glare from lighting. It’s a rimless tank, and I sit with my eyes just at the water level, so glare will be super hard to avoid. I’ve thought about tilting the pendants, creating a shroud around their perimeter, etc., but I haven’t landed on a good solution. The star board chips are ~165° viewing angle, so I’m certain I’d need some optics to avoid atrocious glare. However, the widest optic I could find was this one at 72° viewing angle. I’m not certain how well that would pair with the 120° of the Vesta chip.

Lastly, color mixing is a concern. The colored chips would be over an inch apart, and I don’t have the experience to say whether that would cause noticeable “disco” effect.

Any thoughts on 1) supplementary chip choices, 2) optics, and 3) color mixing?
 

·
Registered
Joined
·
12,619 Posts
First, let me say I’ve read a TON of threads here on DIY LED stuff, but I haven’t found anything matching my goals.

I have a 48x18x18 tank (64gal) currently running a Fluval 3.0, which I have decided to supplement with a pair of DIY LED pendants. Depending on how I like the pendants, they may totally replace the Fluval. Growth is great with the Fluval, but I’d like a bit more intensity at the substrate to get stuff to grow bushier. I’m also looking for better coverage; my Fluval leaves a big dark spot at the surface. My primary goal is to make the plants/tank look nicer. I find the Fluval doesn’t have enough punch in the red channel for my taste, so I’d specifically like more power there. I’m also looking for a bit more shimmer/caustics.
first keep in mind recommended max for the entire chip is 700mA
I'd venture to guess that for viewing you will use, at most, 1/2 of the warm channel, BUT is is personal taste.
So balance drivers accordingly.
For example 300mA on warm, 500mA on cool will exceed the 700mA recommended max rating..
Maximum Drive Current is maximum combined drive currents between both 2700K and 6500K channels. For example, if 700mA is applied to one channel, no current may be applied to the other channel of the array. If 300mA is applied to one channel, then a maximum of 400mA can be applied to the other channel
Best would be current adjustable drivers..
see Bluacro as an example:
EXCEPT they don't really go low enough.
Ideally you would like like 200/500 ww/cw but again personal preference.
25W a chip at combined 700mA 36V
110L/watt approx (all broad estimates)
2750 lumens for full power on the chip.

Tuneable chips are a good idea on paper but personally I find them a bit cumbersome.
Would LOVE to recommend them but it gets difficult. Form factor is great, specs are a bit difficult to design around without a bunch of waste.
Like say 350mA per channel but you only run one channel at 50% for looks..19W total in real life 2000 lumens each.

Point is you sort of have to design it baed on what you "think" you will like the best or over-build (more chips) than you think you need.

As to color additions.. 660nm red is really the only necessity w/ good high K high CRI whites, In my "opinion"
 

·
Registered
Joined
·
12,619 Posts
Seperate topic needs separation.
If you want a pair of COBs I'd recommend "Freah fish" and "crisp white" by Lumileds..
There is some violet diodes in the crisp white.

Or go any high CRI high K white and add some 660nm red and warm white "stars"..

1031392

4:1 power ratio of ff:cw
* MIXING LIST
----------------------------------------
myData fresh-fish-v2.csv x1
myData luxeon-crispwhite-1203-3000k.csv x1
----------------------------------------

* SIMULATION DATA
----------------------------------------
Luminous flux : 4,699 lm
Radiant flux : 16,030 mW
PPF : 71.4 umol/s
TCP : 5710 K
CRI : 95
λp : 453 nm
Color : #FFD4CA
----------------------------------------

* PERFORMANCE @ 60cm & 120° (compulsory)
----------------------------------------
Irradiance : 4.7 W/m²/s
Illuminance : 1,385 lx
PPFD : 21 umol/m²/s
----------------------------------------
Did "the math" on the 6500k channel on the Vesta Thrive
* MIXING LIST
----------------------------------------
myData bridgeluxtune6500.txt x1
----------------------------------------

* SIMULATION DATA
----------------------------------------
Luminous flux : 1,900 lm
Radiant flux : 6,467 mW
PPF : 28.6 umol/s
TCP : 6250 K
CRI : 95
λp : 472 nm
Color : #F8FFE7
----------------------------------------

* PERFORMANCE @ 60cm & 120° (compulsory)
----------------------------------------
Irradiance : 1.9 W/m²/s
Illuminance : 559 lx
PPFD : 8.4 umol/m²/s
----------------------------------------

by SPECTRA 1.0β @ 1.023world
1031393
 
  • Like
Reactions: jake21

·
Registered
Joined
·
12,619 Posts
For fun.. The (almost) perfect daylight LED array..

* MIXING LIST
----------------------------------------
myData luxz405.txt x1
myData viosys420(2).csv x1
myData luxeonfr.txt x1
myData bridgeluxtune6500.txt x1
----------------------------------------

* SIMULATION DATA
----------------------------------------
Luminous flux : 1,906 lm
Radiant flux : 7,838 mW
PPF : 30.8 umol/s
TCP : 6620 K
CRI : 99
λp : 472 nm
Color : #FDFDFF
----------------------------------------

* PERFORMANCE @ 60cm & 120° (compulsory)
----------------------------------------
Irradiance : 2.3 W/m²/s
Illuminance : 561 lx
PPFD : 9.1 umol/m²/s
----------------------------------------

by SPECTRA 1.0β @ 1.023world
1031394


Drop the 405 to .7 would tame the peak in violet
#FDFFFE
 

·
Registered
Joined
·
140 Posts
Discussion Starter · #5 ·
Jeff, thanks. You're a wealth of knowledge.

first keep in mind recommended max for the entire chip is 700mA
I'd venture to guess that for viewing you will use, at most, 1/2 of the warm channel, BUT is is personal taste.
So balance drivers accordingly.
For example 300mA on warm, 500mA on cool will exceed the 700mA recommended max rating..

Best would be current adjustable drivers..
see Bluacro as an example:
EXCEPT they don't really go low enough.
Ideally you would like like 200/500 ww/cw but again personal preference.
25W a chip at combined 700mA 36V
110L/watt approx (all broad estimates)
2750 lumens for full power on the chip.

Tuneable chips are a good idea on paper but personally I find them a bit cumbersome.
Would LOVE to recommend them but it gets difficult. Form factor is great, specs are a bit difficult to design around without a bunch of waste.
Like say 350mA per channel but you only run one channel at 50% for looks..19W total in real life 2000 lumens each.

Point is you sort of have to design it baed on what you "think" you will like the best or over-build (more chips) than you think you need.

As to color additions.. 660nm red is really the only necessity w/ good high K high CRI whites, In my "opinion"
Thanks for pointing that out - I only recently noticed that there's an overall limit of 700mA. It hadn't occurred to me though the driver issues that raises. It looks like Mean Well's ELG-100-C can reduce down to 175mA and could drive the two chips' warm white channels in series, and the ELG-150-C reduces to 250mA and has enough punch for the cool whites.
Why couldn't I run a higher current and just dim it? I'm assuming the 700mA limit is from thermals, so 2x350mA @ 100% DC should be equivalent to 2x700mA @ 50% DC. I guess then there's just the risk of me accidentally giving it too much juice when I'm programming it?
Speaking of dimming, I vaguely recall that DC dimming is better than PWM - is this true?

I think you're right that I'd run the warm white channel at much lower power than the cool white - even with my Fluval's weak warm white channel I find myself dialing it back at times.
I agree that the tunable chips are not optimum. For me, I think it will be worth it to allow me to dial in the look I want and readjust accordingly without swapping chips. (For example, I just changed my tank's sand from black to white and found myself wanting 2-3x as much power to the warm whites.) Of course, I could do that with a discrete combo of WW/CW, but then I face the same issue of "waste" (having more chip than I will use at one time, or maybe I misunderstand what you mean by waste).

Why 660 red and not far red? I assumed it would give plants nicer reds to have a broader spectrum of red light. I guess I know that far red is in the visible spectrum, but I don't know much else (like what it actually looks like). Is it like violet where it's visible but our eyes aren't sensitive to it?

Seperate topic needs separation.
If you want a pair of COBs I'd recommend "Freah fish" and "crisp white" by Lumileds..
There is some violet diodes in the crisp white.

Or go any high CRI high K white and add some 660nm red and warm white "stars"..

4:1 power ratio of ff:cw


Did "the math" on the 6500k channel on the Vesta Thrive
I read your previous posts recommending the FreshFocus Fish, but I wasn't sure how I felt about its huge blue spike and cyan dip. Can you explain why that's preferable to the Bridgelux 6500K (whether in the tunable chip or in a discrete COB like the Gen 7 V10)?
Also, would you mind sharing the 6500K's CSV so I could play around with it in Spectra? (I think that's how it works at least.)

I had also come across the Crisp White, but I didn't connect the dots that it has substantial violet output. That does seem nice to mix the violet into the white channel so you don't see purple LEDs. What about mixing Crisp White with Bridgelux 6500K?
1031400


For fun.. The (almost) perfect daylight LED array..

Drop the 405 to .7 would tame the peak in violet
#FDFFFE
Hmm, so that's a combo of 405nm, 420nm, fresh fish, and Bridgelux 6500K? That looks pretty nice - you think it would be good in reality or just on paper? I.e., I've read people suggest replicating the sun doesn't necessarily give the nicest look.
 

·
Registered
Joined
·
12,619 Posts
Nicest is relative . Natural isn't as punchy as say RGB arrays or RGBAMBER or RGBALIME.
Depends on personal preference. My RGBLA lights have a crisp white w/ a property that is hard to quantify over simple high CRI ( even w/ accent color additions) leds.
To be 100% honest as great as the RGBLA look is ( def not 90 + CRI) it has differences that I can' t say is better or less than "whites".
Wish I could explain it better but the punch of non- white can be err....... unnerving?
Color theory states blue plus yellow ( normal white led) is " color dulling" to people.



8000k/660nm ala Finnex and old BML DISCUS pattern (6500k/660) and own tests is a simple "crisp white w/ red punch"

Adding 740 would be a supplement
to 660nm not a replacement in my opinion
The Emerson effect is the increase in the rate of photosynthesis after chloroplasts are exposed to light of wavelength 680 nm and more than 680 nm. When simultaneously exposed to light of both wavelengths, the rate of photosynthesis is far higher than the sum of the red light and far red light photosynthesis rates. Wikipedia
There is no shortage of cyan in ff.
Actually would prefer Bridgelux 6500k over ff If one can find lone 6500 cobs so yes Bridgelux/Crisp white is fine and probably cheaper choice Like I said sometimes what bridgelux has on paper is not what mere mortals can buy. Now eqiv in say 5000/5500 k were always easier to actually buy
On a micro tint level guess Bridgelux tints green, ff tints blue. As to the spike...ehh good and bad





.
 

·
Registered
Joined
·
12,619 Posts
Tough love post ..

10v analog dimming ( btw analog dim protocol does not imply analog output ) is garbage and only popular because it is an unfortunate " industry standard".
Even there ( control circuitry, not lighting) current is "preferred" over voltage. Well by some. :)


RARELY smoothly dims from 100 -0% and usually needs a separate full off circuit via timer or switch ..
Funding "affordable" controllers off the shelf is ny impossible. Apex, Bluefish full ect
One can convert pwm 5v (3.3v) to 10 v pwn, 10v analog but adds a layer of complexity .

As to OUTPUT pwm vs current both have plusses and minuses. Pwm w/ err strobing.
Analog with larger color shifts as current is decreased

5v pwm is simple, cheap, dims to zero w/ minimal color shifts.

Opinions are always arguable ;)
 

·
Registered
Joined
·
140 Posts
Discussion Starter · #8 ·
Thanks Jeff. Since so much comes down to preference, I think I'll use the tunable chips to get a feel for what I like. The 6500K gives me a nice flat spectrum where I can add specific colors as desired, and even if I only run warm white at 10%, the chips aren't too expensive. Arguably, a "better" solution might be a single channel 5000K Thrive chip, which is uber-flat at ~90% spectral power from 440nm-630nm, but I want the warm/cool tunability.

Here is a list of all the Thrive chips Digikey stocks (there are a few, like a higher-power Vesta, that nowhere seems to sell). And here are the 6500K chips. $3-$4 for 700-1450 lumens at the test current. The V4, V6, and V8 chips have marginally higher CRI (98 vs 95), but the spectrum is only a touch smoother than the Gen 7 V10. It's notable that the V4 and V6 chips are the "HD" (high density?) models, which have about 20% less lumens/Watt. I'd probably go for the V10C if I were running one of these. $3.65, 1450 lumens (up to 2610 at 700mA), 95 CRI, and just 13.5mm square.


Hmm, thanks for the info on dimming. So wait - "10v analog" vs "10v PWM" vs "5v PWM" is just specifying the input to the dimming circuit? In that case, 5v PWM seems like a no-brainer, as I'll (tentatively) be using an Arduino to control this.
If I understand correctly, a driver takes that input signal and dims the LED either by PWM or by reducing drive current? Assuming that's right, would the output PWM be the same frequency as the input PWM?
While I don't notice a flicker, my Fluval puts out the most obnoxious high-frequency whine (maybe 15kHz?) when it's dimmed. Is this a result of PWM dimming?
 

·
Registered
Joined
·
12,619 Posts
Side note:

875 total current for the large cobs.
500/350 or 500/300 drive currentz possibke soo easy Ldd choices.

Still need good coiling though
Bridgelux still holds the minor spectrum lead so to speak.
Would be a tradeoff.
Sooo just an fyi.

You asked about the not running at full, only rhing I can add is it isn't t a problem unless you tweak it ir a controller failed. Sime ldd boards have a pull down resistor on board which means lights will go off in the event of a controller failure.
Still leaves hunan error.
Technically I rarely run anything near nax outout anyways. Just SOP for passive cooled.

IN order to attach it here I renamed it .txt. Should import fine. If not just rename it. .csv. Ignore windows warning.
 

Attachments

·
Registered
Joined
·
140 Posts
Discussion Starter · #11 · (Edited)
I came across those Luminus COBs, but I was turned off by: 1) the lack of spectral data, and 2) the form of an array of independent chips rather than a single large COB (mostly from an aesthetic standpoint, but I think I also recall that a single COB is somehow better than a bunch of discrete chips).

Thanks for the CSV. I got the datapoints for the 2700K, which I have attached. (Not sure what you use, but this site allowed me to easily pull data from the graph, then I used Excel to interpolate the points I needed. I could have grabbed those exact points from the graph, but it would have been a lot more time consuming.)

This wasn't immediately apparent to me, but I've found Spectra allows decimal values; e.g., I can simulate 10% duty cycle dimming by specifying 0.1 of a certain LED. Anyway, I think I settled on a combo that gives me the controllability and spectrum I desire.
* MIXING LIST
----------------------------------------
myData 6500kbridgelux.txt (Thrive) [120°] x1
myData 2700kbridgelux.txt (Thrive) [120°] x0.1
LED Violet (420nm) [120°] x1
LED Blue (460nm) [120°] x1
LED Amber (590nm) [120°] x1
LED DeepRed (660nm) [120°] x1
LED DeepRed (680nm) [120°] x1
----------------------------------------

* SIMULATION DATA
----------------------------------------
Luminous flux : 2,186 lm
Radiant flux : 9,250 mW
PPF : 41.2 umol/s
TCP : 6800 K
CRI : 95
λp : 468 nm
Color : #FFD9EF
----------------------------------------

* PERFORMANCE @ 30cm
----------------------------------------
Irradiance : 10.9 W/m²/s
Illuminance : 2,577 lx
PPFD : 48.5 umol/m²/s
----------------------------------------

by SPECTRA 1.0β @ 1.023world
1031413

1031414
I'm not set on the amber, but it might help with a nice sunrise/sunset effect. Any suggestions on specific chips? Star boards are easy, but I don't like that using them would set the colored LEDs an inch away from the white chip. Am I stressing about color mixing too much?


Edit: Here's an idea. This post discusses using an acrylic diffuser for mixing colors. I wonder if I'd still get nice caustics.
 

Attachments

·
Registered
Joined
·
12,619 Posts
I came across those Luminus COBs, but I was turned off by: 1) the lack of spectral data, and 2) the form of an array of independent chips rather than a single large COB (mostly from an aesthetic standpoint, but I think I also recall that a single COB is somehow better than a bunch of discrete chips).

Thanks for the CSV. I got the datapoints for the 2700K, which I have attached. (Not sure what you use, but this site allowed me to easily pull data from the graph, then I used Excel to interpolate the points I needed. I could have grabbed those exact points from the graph, but it would have been a lot more time consuming.)

This wasn't immediately apparent to me, but I've found Spectra allows decimal values; e.g., I can simulate 10% duty cycle dimming by specifying 0.1 of a certain LED. Anyway, I think I settled on a combo that gives me the controllability and spectrum I desire.
* MIXING LIST
----------------------------------------
myData 6500kbridgelux.txt (Thrive) [120°] x1
myData 2700kbridgelux.txt (Thrive) [120°] x0.1
LED Violet (420nm) [120°] x1
LED Blue (460nm) [120°] x1
LED Amber (590nm) [120°] x1
LED DeepRed (660nm) [120°] x1
LED DeepRed (680nm) [120°] x1
----------------------------------------

* SIMULATION DATA
----------------------------------------
Luminous flux : 2,186 lm
Radiant flux : 9,250 mW
PPF : 41.2 umol/s
TCP : 6800 K
CRI : 95
λp : 468 nm
Color : #FFD9EF
----------------------------------------

* PERFORMANCE @ 30cm
----------------------------------------
Irradiance : 10.9 W/m²/s
Illuminance : 2,577 lx
PPFD : 48.5 umol/m²/s
----------------------------------------

by SPECTRA 1.0β @ 1.023world
View attachment 1031413
View attachment 1031414
I'm not set on the amber, but it might help with a nice sunrise/sunset effect. Any suggestions on specific chips? Star boards are easy, but I don't like that using them would set the colored LEDs an inch away from the white chip. Am I stressing about color mixing too much?


Edit: Here's an idea. This post discusses using an acrylic diffuser for mixing colors. I wonder if I'd still get nice caustics.
Like this.. ;)
1031423

Doesn't take long to hand place points every 5nm.. I tried the "automatic extraction" and just found it cludgy.
5digit fixed.. format Sure there might be better but my workflow is adequate..

Import into notepad ++
1031424


Manually clean making sure to start at 380nm
1031425

Save and import..Deleting a comma or a point error like 504 instead of 505 will cause errors in import to Spectra
Thought I mentioned the fraction part ..somewhere..
 

·
Registered
Joined
·
12,619 Posts
As to time..don't think the whole from import chart to import into Spectra takes me more than 15 minutes.

as to Luminous devices spectrums are available or were..
Old "studio" COB, some of the first readily available high CRI COB's but not 6500k
 

Attachments

·
Registered
Joined
·
140 Posts
Discussion Starter · #14 · (Edited)
Yeah, auto extraction is great for less cluttered plots but was not at all useful for the plot with the 6500K data.
You must have more patience and/or be faster at clicking than me. If I can ever use scripting/Excel formulas to avoid monotony, I will (which is quite often in my job, lol). My method took about 4 minutes from start to finish, and I'm actually surprised how close the results were to yours. To each their own though! (As an aside, it looks like you accidentally zeroed out 760nm and up.)
1031428


I found spectra for a bunch of the Luminus chips, but not for those particular ones that are on the tunable array. I may well have overlooked something.

Speaking of Luminus, I'm thinking I'll supplement the Bridgelux one of their RGBW chips. I'll probably never use the green or white channels, but I think the 6.5W of red and blue alone are worth $12. Sure, I could probably beat that with fleabay chips, but I'd rather spend a couple extra bucks to buy from a reputable manufacturer. I also like the idea of having red and blue commingled on the same chip. Although I'm a little stumped how to drive it at just 250mA per channel... Edit: The Mean Well LDH-25-250W could drive two of the Luminus chips in series at 250mA. Only 10v PWM dimming, but that's not tough to make work.

I think I'll ditch the violet and deep red chips. Sure, those would round out the spectrum, but they're so dim to human perception I don't think their minimal visual addition would be worth the extra PAR. I expect I will be having to dial the lights way back to avoid algae, so PAR without visual brightness is a disadvantage.
 

·
Registered
Joined
·
12,619 Posts
90 + mp-1616 xnova cubes are in the tunable arrays. Sadly not a lot there either.
Phone hates me today. Lost my whole response .

One thing is you can split the current between 2 equal strings. 500mA over 2 parallel branches is 250 to each.
Not recommended and at the least like a 300mA fuse on each leg in case one fails open shunting all 500mA to the remaining string. There are fancier ways.

As to missing points, value judgement and depends on my mood.
 

·
Registered
Joined
·
140 Posts
Discussion Starter · #16 ·
Thanks! Yeah, their datasheet doesn't say if the spectrum provided is for 80 or 90 CRI. I assume it must be 80, as it doesn't look that special.

Man, I hate when that happens... and that's on PC. I periodically copy my post to the clipboard because I'm paranoid. If I lost a post typing on my phone I'd lose my mind. Some forums auto-save drafts of your posts and it's a godsend.

I thought about putting them in parallel, but that somewhat defeats the purpose of a constant current driver. I hadn't thought of using fuses, though. Easy to overlook the simple solutions.

I just finished reading through all of Mean Well's constant-current driver datasheets. It was about as fun as it sounds. Anyway, I compiled a list of PWM-dimmable drivers that would work for:
  • 1 Vesta channel (350mA @ 38v),
  • 2 Vesta channels in series (350mA @ 76v),
  • 1 Luminus channel (250mA @ 26v), or
  • 2 Luminus channels in series (250mA @ 52v)
It would probably be a better idea to unevenly split the Vesta with, eg., 500mA CW and 200mA WW, but my brain is wearing out and that's too complicated.

Probably unsurprisingly to anyone in the know, the LD- drivers took the cake for 1x Vesta. Particularly, the LDD-350HW was second cheapest, supports 5v PWM, and had the highest efficiency. There was an adjustable driver that would work, but it cost 3x as much with lower efficiency. Granted, the 8% drop in efficiency comes out to like a dollar per year, but the engineer in me loves efficiency.

I'm surprised that there's nothing to be saved by putting the two chips in series - the cheapest option was over twice the price of the LDD-350HW while having lower efficiency. While I'd like to keep the number of drivers down, I'm a little sketched out at the thought of a DIY 80v DC system over water.

Options were slim for the Luminus, but I found two. The LDH-25-250W was the obvious choice; it was the same price and efficiency as the LDDS-250HW while being able to drive both in series instead of just one. They're both 10v PWM, but I can make do.

So I think that leaves me with 4x LDD-350HW and 2x LDH-25-250W. With the cost of the chips, drivers, and heatsinks, I'm at $150. Adds up quick! I was thinking I'd run the Kessil gooseneck mounts, which would have brought me to $200, but I realized they're not long enough to get the lights where I want them. I'll probably hang them to save money and for the clean look.

Are there any safety precautions I need to take here with 50v DC hanging over water? I could conformally coat everything, but that would make tweaks/upgrades a pain. I guess a GFCI outlet and/or running a ground to the heatsink?

Mean Well modelDrivesPWM voltagePriceEfficiencyPower [W]Adjustable?
LDH-25-3501x Vesta
10​
$9​
95%​
25​
No
LDD-350HW1x Vesta
5​
$8​
97%​
?No
LDD-350H-WDA1x Vesta
5​
$14​
95%​
16​
No
NLDD-350HW1x Vesta
5​
$7​
96%​
?No
XLG-251x Vesta
10​
$24​
88%​
25​
Yes
LPF-16D-481x Vesta
10​
$20​
85%​
16​
No
HVGC-65-3502x Vesta
10​
$51​
90%​
65​
Yes
LDH-45B-3502x Vesta
2-8​
$19​
93%​
44​
No
ODLC-45-3502x Vesta
10​
$24​
86%​
33​
No
LCM-402x Vesta
10​
$31​
91%​
42​
No
LDDS-250HW1x Luminus
10​
$9​
94%​
?No
LDH-25-250W2x Luminus
10​
$9​
94%​
21​
No
 

·
Registered
Joined
·
12,619 Posts
Actually it does auto save here ..unless you do some voodoo.

People hang 110/220 over tanks all the time i.e ballasts.

That js one thing about cobs.. the voltage/driver/power supply issue.
There are some huge meanwells and those chinese black box reef lights have drivers inside w 80-100v dc 550mA and line voltage to them...There is precident.
Not going to get into ac is safer than dc issue.

Sadly just fried a 56v 250w n2 power air cooled power supply. Had a 12 v dc and 5v dc rail in addition to the 56v side. Instead of shutting it off at my little home made box I went to the power strip. Had a weird switch and apparently I short cycled the ac.
Blew up "resistor 32" with a loud bang.
It was aging and the 2 large power caps were starting to bulge a bit ( don't get me started on caps). Other than that all the rest looks brand new. Company said basically toss it and wouldn't tell me the value of r32.

I digress....
I've only recently started putting the drivers on the light head itself so mostly low voltage/low amps over the water but still had the potential from the ps I guess. Finding like 8-10 conductor wire was a chore, at least at reasonable gauge.
After 5 drivers thought it gets ridiculous sort of.
GFCI's though...always.

Yea funny thing diodes are about the least of the cost really.
 

·
Registered
Joined
·
140 Posts
Discussion Starter · #18 · (Edited)
That's true. I think my issue is more that there will be exposed wires here with the pendant style. I guess I'll just have to hot glue any exposed solder joints. The fact that it's DC also gives me pause - I've lit myself up with 120 AC and been fine, but I've never experienced a DC shock (excluding some high voltage, low current stuff). The outlet is right behind the tank, but I think I can put a GFCI anywhere on the circuit and have the same result.

I just had an idea - what if I used a few MOSFETs to switch a single 700mA power supply between both channels of the Vesta chip? Rather than two $8 drivers, I could use one $8 driver (LDD-700HW) and a few $1 MOSFETs and gain the ability to drive either channel anywhere from zero to full juice. Only thing I could think is that maybe the driver wouldn't like being switched between two chips?

Or, I think I could use some similar circuitry to just invert the PWM signal between two 700mA drivers (e.g. one gets 80% DC, the other gets 20%; one gets 30%, the other gets 70%).

Here are the diagrams. I'm probably missing some resistors/grounds/etc.
1031440


Edit: Wow I over-complicated that. Updated the schematic. With the second option I just need a couple resistors and a single MOSFET. In addition to the simplicity, I like the idea of under-driving a pair of 700mA drivers rather than having one at full load, and doing switching on the signal side would keep heat down.
 

Attachments

·
Registered
Joined
·
140 Posts
Discussion Starter · #19 ·
I just realized my "solution" would force the chip to constantly run at 700mA split between the channels - I couldn't dim it to, say, 100mA per channel. I guess I'll ditch the circuitry and just code a safeguard into my routine that won't allow the two channels' duty cycles to sum over 100%.
 

·
Registered
Joined
·
140 Posts
Discussion Starter · #20 ·
@jeffkrol
Any experience putting lenses on LEDs? I'd like to tighten the beam angle up a bit and make it so the bare chips aren't visible. The heatsink I'm using is pre-drilled for Ledil Stella lenses. The lens is for 30mm LES chips, whereas I have a 19mm and a 22mm chip. It looks like they'd still fit, but I'm not sure how the lens would handle having two slightly off-center chips. I can't find any lenses specifically for the Vesta 13mm or for the Luminus RGBW, so something hacky like this is pretty much my only option. I guess $20 wouldn't be a terrible gamble, but I figured I might as well ask.
 
1 - 20 of 43 Posts
Top