Looking for a specific DC CC driver for LEDs

[karimwassef]

So.. why multichips at 100W?

The multichips deliver 100W at 40mm = 2.5W/mm linear density (on a strip). At 430nm, they're about $20 each. At 395nm, they're about $40 each. At 365nm, they're about $150 each. Taking 395nm as a base: 800W ~ $320

The Luxeon looks to be 3.2W at 1.7mm and two could fit side by side on a 1" strip. So ~ 6.4W every 1.7mm = 3.8W/mm. But at $14 each (3.2W each), then 800W ~ $3500

So the Luxeons would be 10x or $3200 more each side or $6400 total cost increase...

That's just looking at power density ... I realize it's not light density (that's what the exercise in this thread is about). I'm open to looking at the light to power differences and compare. But I think the price here is an order of magnitude higher, but the benefits may not be an order of magnitude higher.

This is a big project, and I need to optimize. If the multichips are inferior in total value, then they'd be the more expensive option and these Luxeons may be better. Let me know if you see any errors above.

 

[karimwassef]

Also, the Luxeons are soldered? The multichips are designed to be easily replaceable.

I don't share all this stuff since it's all still in deep design, but I've been looking at this for a year or more - from multiple angles. Lighting is one key vector. You should check out the greenhouse structure and the flow threads

 

[karimwassef]

I love how the Luxeon datasheet rates radiated power in mW and efficiency%. Also confirms what we expected that radiated output is a function of current and the voltage difference indicates the power lost.

Here's where the Luxeons stack up at 395nm:

1000mA at 3.1V = 3.1W. The radiated power is 1000mW, so 1W/3.1W efficiency = 32%.

Based on the LED chips tested so far, if the 100W wastes 80W as heat of 20% efficiency, then the Luxeons are 1.5 x more radiated output per Watt (I know these assumptions still needed to be tested).

Even the worst LEDs are ~14% efficiency. Even assuming 10% efficiency, the Luxeons would be 3x the output.

At 3x light output and 10x price increase, the math doesn't line up?

I'm not accounting for the cost and complexity of assembling and powering 250 Luxeons per side vs. 8 multichips to get to 800W. Assuming that I only need 1/3 the Luxeons, that's 80 Luxeons to 8 multichips. Still doesn't look like a better fit.

Does anyone assemble Luxeons into subassemblies?

 

[karimwassef]

Mouser is better, but 395nm is still very expensive

http://www.mouser.com/Optoelectroni.../N-8usfn?P=1yzsl7j&Keyword=Lumiled+UV&FS=True

It's $8.19 each for 100pcs. Assuming 80 needed per side (3 x better than multichips) would leave some unused. So let's take 100pcs per side = $820x2 = $1640 for the total LED lighting.

The multichips are $320x2 = $640 for the equivalent light output. That's $1000 less.

I still need to look at the cost of wasted power over a year's worth of consumption and the front end cost.

 

[zachts]

So.. why multichips at 100W?



That's just looking at power density ... I realize it's not light density (that's what the exercise in this thread is about). I'm open to looking at the light to power differences and compare. But I think the price here is an order of magnitude higher, but the benefits may not be an order of magnitude higher.

This is what I'm very curious about, how do the multichip violets actually perform in terms of light output/watt. mW/Watt is how I've allways compared violets and royal blues, but without labratory equipment, or reliable manufacturer data sheets how do you go about determining this? at least short of measuring PAR on a violet multichip and an equally dense array of luxeon z for example.

I agree the cost/benifit may not be there for you at this point in time, but in a year? I anticipate the luxeon violets, or even semiLED violets will start to become available in dense arrays just as the Cree "X" chips did all over web a couple years ago.

As you mentioned the cost of power consumption over time, allong with having to replace multichips every six months? could add up pretty fast in favor of a higher upfront cost solution.

 

[karimwassef]

This is what I'm very curious about, how do the multichip violets actually perform in terms of light output/watt. mW/Watt is how I've allways compared violets and royal blues, but without labratory equipment, or reliable manufacturer data sheets how do you go about determining this? at least short of measuring PAR on a violet multichip and an equally dense array of luxeon z for example.

I agree the cost/benifit may not be there for you at this point in time, but in a year? I anticipate the luxeon violets, or even semiLED violets will start to become available in dense arrays just as the Cree "X" chips did all over web a couple years ago.

As you mentioned the cost of power consumption over time, allong with having to replace multichips every six months? could add up pretty fast in favor of a higher upfront cost solution.

I followed up on that in the next couple of posts. Basically, if you assume that the Luxeons are 2x - 3x the output per Watt, that would justify a 2x - 3x price premium. But the Luxeons require a lot more effort to assemble too and are not as easily replaceable. That's why I asked if there were any Luxeon Multichips or other subassemblies.

I don't think I would need to swap out the multichips every 6 months. Monitoring will determine the real replacement rate.

The last cost element remaining in the cost of wasted power as heat.

 

[karimwassef]

By the way, I do plan on measuring PAR and relating it to current and power.

For low wavelength UV, I can't do a true PAR, but I do plan on doing a comparative analysis of similar wavelength chips... The idea is to use fluorescent powder to represent the irradiance of the UV ... once it's converted to a lower frequency (higher wavelength) like green or yellow, then the PAR meter can capture it.

Measuring two chips at the same wavelength should provide a comparative "measure" of radiation efficiency. I can't measure true radiation efficiency (also called optical wall plug efficiency) but I don't really need to.

 

[zachts]

Aside from RapidLED's Borealis multichip (which I'm suspect didn't provide adequate thermal management) I've only seen 4up arrays. But I haven't really been looking. I'm sure one wouldn't be hard to source but I'm not sure the cost would be worth it to have them fabricated. RapidLED's fabricator might be worth looking into, and possibly Steve's LEDs whom I know has done custom work for people using the luxeon Z in the past.

On a side note, for your night time viewing White LEDs I'd highly recommend the High CRI Bridgelux Vero, it wouldn't take very many of their large array to give you more than enough light for evening or early morning viewing and would look pretty good with some of the violets/UVs running.

 

[karimwassef]

Actually, I go very simple for viewing lights. I like the 7020 chip strip configuration

https://www.amazon.com/gp/product/B00FMQOUB2

They're incredibly bright and very simple to apply.

 

[karimwassef]

Ok. I have PAR data and it's interesting. Some of it should have been expected, but some was a little surprising.

I also got several more chips that need power

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/100D1FBF-C8D7-4887-9F05-F62674847725_zpseoukesoo.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/100D1FBF-C8D7-4887-9F05-F62674847725_zpseoukesoo.jpg" border="0" alt=" photo 100D1FBF-C8D7-4887-9F05-F62674847725_zpseoukesoo.jpg"/></a>

I'm amazed at the variability in the packaging... I'm going to match these up to the prices and then look at their performance metrics. I wonder if it's possible to just toss a chip in the trash based on the package it arrives in...

 

[karimwassef]

So - first, the experiment process:

I don't think it needs to be this controlled, but until I know what variables matter and what don't, I'll control EVERYTHING I can.

I started with a hole drilled at the top of a plastic drawer frame (was about to toss it). I have a large heavy aluminum block that I use as a weight for some projects. I'm using that as my sensor base with different size clamps to control location.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/D5EA2BFE-F1DD-4AC6-97CF-C4FAF7823813_zpsk5bzwbwt.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/D5EA2BFE-F1DD-4AC6-97CF-C4FAF7823813_zpsk5bzwbwt.jpg" border="0" alt=" photo D5EA2BFE-F1DD-4AC6-97CF-C4FAF7823813_zpsk5bzwbwt.jpg"/></a>

I use a level to make sure the source and sensor are lined up. and that they're both level. I wasn't planning on making any measurements with lenses, but then decided to grab some data. I love me some data.

This is the basic pedestal concept. Rotating the clamps to alter their height to position the sensor and control its level.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/2034B8C2-AF17-4457-A08A-266917309F45_zpsfj6ojfda.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/2034B8C2-AF17-4457-A08A-266917309F45_zpsfj6ojfda.jpg" border="0" alt=" photo 2034B8C2-AF17-4457-A08A-266917309F45_zpsfj6ojfda.jpg"/></a>

The hole is large enough for a lens, so a littl bigger than the chip opening. Maybe I'll go 200W at some point, and I'd like to keep things the same to avoid having to retake data.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/654E377D-C8AD-4EFC-841F-5721F379FA9B_zpsd1wssxc3.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/654E377D-C8AD-4EFC-841F-5721F379FA9B_zpsd1wssxc3.jpg" border="0" alt=" photo 654E377D-C8AD-4EFC-841F-5721F379FA9B_zpsd1wssxc3.jpg"/></a>

Used black garbage bags and black duct tape to seal out all external light

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/66CC26B7-F025-4A96-A643-64CEE31B7586_zpsizidktt2.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/66CC26B7-F025-4A96-A643-64CEE31B7586_zpsizidktt2.jpg" border="0" alt=" photo 66CC26B7-F025-4A96-A643-64CEE31B7586_zpsizidktt2.jpg"/></a>

LED, heatsink and fan sit on top with a circular hole and duct tape cylinder to allow access.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/0C166BFD-D5D5-435B-AB49-AF548FC047F2_zpsuf21kq0j.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/0C166BFD-D5D5-435B-AB49-AF548FC047F2_zpsuf21kq0j.jpg" border="0" alt=" photo 0C166BFD-D5D5-435B-AB49-AF548FC047F2_zpsuf21kq0j.jpg"/></a>

It allow enough air to go through, so should be fine

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/54121623-E4C9-4296-844F-5F3E99C8F442_zpsbhxmlzb1.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/54121623-E4C9-4296-844F-5F3E99C8F442_zpsbhxmlzb1.jpg" border="0" alt=" photo 54121623-E4C9-4296-844F-5F3E99C8F442_zpsbhxmlzb1.jpg"/></a>

The probe looks ok, but the Styrofoam plate didn't look like a controlled setup.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/F40A20F9-21CE-492E-82B0-E951488A19FE_zpsngjubfo9.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/F40A20F9-21CE-492E-82B0-E951488A19FE_zpsngjubfo9.jpg" border="0" alt=" photo F40A20F9-21CE-492E-82B0-E951488A19FE_zpsngjubfo9.jpg"/></a>

Decided to use double clamps and then a 12" metal ruler with a level to control distance and level

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/ED4FDE74-DF6B-4525-BB9A-15B90F59F207_zpsvxwyhhwl.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/ED4FDE74-DF6B-4525-BB9A-15B90F59F207_zpsvxwyhhwl.jpg" border="0" alt=" photo ED4FDE74-DF6B-4525-BB9A-15B90F59F207_zpsvxwyhhwl.jpg"/></a>

 

[karimwassef]

The LED is slightly recessed into the heatsink assembly so this allowed me to take that into account too.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/D3DAAFED-B578-435F-88EA-4A12CBB0077B_zpskniq4ayt.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/D3DAAFED-B578-435F-88EA-4A12CBB0077B_zpskniq4ayt.jpg" border="0" alt=" photo D3DAAFED-B578-435F-88EA-4A12CBB0077B_zpskniq4ayt.jpg"/></a>

This view isn't very clear, but I'm using a black magnet (old glass cleaner) to hold the metal ruler and level up against the opening.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/0053D19D-CC40-40B7-ACAA-07D941B6AA77_zpsflqbctsn.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/0053D19D-CC40-40B7-ACAA-07D941B6AA77_zpsflqbctsn.jpg" border="0" alt=" photo 0053D19D-CC40-40B7-ACAA-07D941B6AA77_zpsflqbctsn.jpg"/></a>



This should be a little clearer

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/05072853-BE88-424D-993F-21B3C1A33046_zpsslubvahg.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/05072853-BE88-424D-993F-21B3C1A33046_zpsslubvahg.jpg" border="0" alt=" photo 05072853-BE88-424D-993F-21B3C1A33046_zpsslubvahg.jpg"/></a>

and that's the stage ready for the testing... almost

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/468A3061-375C-482C-AA35-EFAD1C2A07B1_zpsflq8ob04.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/468A3061-375C-482C-AA35-EFAD1C2A07B1_zpsflq8ob04.jpg" border="0" alt=" photo 468A3061-375C-482C-AA35-EFAD1C2A07B1_zpsflq8ob04.jpg"/></a>

LED positioned

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/59FD2937-E612-4CB8-8C6F-9172D4A6291D_zpsfiinh3ah.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/59FD2937-E612-4CB8-8C6F-9172D4A6291D_zpsfiinh3ah.jpg" border="0" alt=" photo 59FD2937-E612-4CB8-8C6F-9172D4A6291D_zpsfiinh3ah.jpg"/></a>

and a little fine tuning on the LED chip to center it

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/28D4503F-7856-4B64-81B6-060E508478C2_zpsgqeds7qf.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/28D4503F-7856-4B64-81B6-060E508478C2_zpsgqeds7qf.jpg" border="0" alt=" photo 28D4503F-7856-4B64-81B6-060E508478C2_zpsgqeds7qf.jpg"/></a>

You can see the duct tape cylinder in the bottom right hand side here. Almost ready to test!

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/1E52C596-78C0-4A4A-86FA-2349116BD0F9_zpswtusfbea.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/1E52C596-78C0-4A4A-86FA-2349116BD0F9_zpswtusfbea.jpg" border="0" alt=" photo 1E52C596-78C0-4A4A-86FA-2349116BD0F9_zpswtusfbea.jpg"/></a>

 

[karimwassef]

Ok. TESTING IN PROGRESS

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/A2BD2D6E-C155-4EDB-9EF4-70B6B32031A3_zpsevghbglf.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/A2BD2D6E-C155-4EDB-9EF4-70B6B32031A3_zpsevghbglf.jpg" border="0" alt=" photo A2BD2D6E-C155-4EDB-9EF4-70B6B32031A3_zpsevghbglf.jpg"/></a>

The whole thing lights up and the trash bags inflate like a funhouse, but there's no light escaping anywhere, so I'm assuming there's no outside light getting in.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/E444378E-1E5E-42B2-927C-2A348DE239C6_zpsnceqm2uc.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/E444378E-1E5E-42B2-927C-2A348DE239C6_zpsnceqm2uc.jpg" border="0" alt=" photo E444378E-1E5E-42B2-927C-2A348DE239C6_zpsnceqm2uc.jpg"/></a>

Just in case though, I run the test at night in the dark. After all the tests are done, I opened it up to see what it looks like in there:

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/109D5CCC-A7F9-486B-9B5F-0A3CBE96AC5E_zpsfc7rjvlg.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/109D5CCC-A7F9-486B-9B5F-0A3CBE96AC5E_zpsfc7rjvlg.jpg" border="0" alt=" photo 109D5CCC-A7F9-486B-9B5F-0A3CBE96AC5E_zpsfc7rjvlg.jpg"/></a>

Pretty intense

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/25A5DCC2-9333-43B3-A863-709C8A46D9C8_zpsjzapxvlz.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/25A5DCC2-9333-43B3-A863-709C8A46D9C8_zpsjzapxvlz.jpg" border="0" alt=" photo 25A5DCC2-9333-43B3-A863-709C8A46D9C8_zpsjzapxvlz.jpg"/></a>

The camera clearly can't work with this light, but it's the best it can do

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/C0004EE6-CD13-4A54-AF02-1B38DD91B9A8_zpsnno4xg9c.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/C0004EE6-CD13-4A54-AF02-1B38DD91B9A8_zpsnno4xg9c.jpg" border="0" alt=" photo C0004EE6-CD13-4A54-AF02-1B38DD91B9A8_zpsnno4xg9c.jpg"/></a>

I reran the testing on one chip with the lens.

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/A143BDE7-3236-4DA6-9675-0DC1D0829A8F_zpslyrhkaey.jpg.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/A143BDE7-3236-4DA6-9675-0DC1D0829A8F_zpslyrhkaey.jpg" border="0" alt=" photo A143BDE7-3236-4DA6-9675-0DC1D0829A8F_zpslyrhkaey.jpg"/></a>

I have collected the data and the analysis is pretty straightforward, but I need to make it more presentable.

Out of curiosity - what would you think the lens would do to the PAR reading? Go up? Go down? by how much?

 

[zachts]

One would assume a lens would make the PAR increase, however that has much to do with the radiation pattern of the lens, wide angle could very likely have a dark spot in the center which is what you are measuring, and it could go way, way down in this test setup..........just for the sake of guessing lets go with 38% decrease measured at center beam.....course if that's the case I'd expect a substantial increase at some point if you move the sensor horizontal off center.....

 

[karimwassef]

One would assume a lens would make the PAR increase, however that has much to do with the radiation pattern of the lens, wide angle could very likely have a dark spot in the center which is what you are measuring, and it could go way, way down in this test setup..........just for the sake of guessing lets go with 38% decrease measured at center beam.....course if that's the case I'd expect a substantial increase at some point if you move the sensor horizontal off center.....

Dude! You basically guessed every possible outcome

Not sure that counts ... but ok.

So the PAR went WAY UP. I decided to take a measure of the radiation angle with and without the lens. Without the lens, it's about 80 degrees. With the lens, it's about 55 degrees.

I worked out the ratio of the radiated surface area and it turns out to be 2.6x (80 degrees is 2.6x the 55 degrees).

The PAR went up by.... 2.5x

I'm attributing the difference to the lens glass absorbing a little of the light.

These are just experimental. I'm getting 90-120 degree lenses

as well as 60x160 degrees asymmetrical lenses to test.

 

[karimwassef]

Ok. It took a while to complete the analysis and put it in a usable format.

First, I compared the old and new I x V data for the same LEDs. I could tell that LED #4 was degrading because it took higher and higher voltage to maintain the same current. By last weekend, I couldn't get over 2A without crossing the 48V threshold where I expected rapid degradation.

Apparently, even 48V was too high. After a month of running, the voltage at 2A was over 48V and over 3A was nearly 60V and would have failed catastrophically. This is where a buck converter can keep the runaway under control whereas a boost wouldn't - at least not without protections.

So there were four LEDs: # 1 (blue), 3 (blue), 4 (degraded blue), 5 (white) as well as T (theatrus's blue LED).

I didn't get around to mounting and testing # 2 (blue), or any of the new LEDs # 6-14.

Here's what they look like on an I x V

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/1_zpscactmhuo.png.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/1_zpscactmhuo.png" border="0" alt=" photo 1_zpscactmhuo.png"/></a>

I put the old data on the same graph for continuity also

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/2_zpspqh3mulz.png.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/2_zpspqh3mulz.png" border="0" alt=" photo 2_zpspqh3mulz.png"/></a>

 

[karimwassef]

The convention for a dataset is LED # . TEST #, so 3.2 is the second test using LED # 3.

So LEDs 1 and 3 remained in about the same general place. 4 degraded. 5 (white) looks like it got better?

I went back to my current shunt and I found that it wandered by 0.5A. I'm not sure why... I decided to bypass it and use my handheld multimeter for the current measurement. I have two so I checked both and got the same current (to two digits), so I'm using that.

So, with the more accurate data captured, I looked at PAR:

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/6_zpskywgvmkb.png.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/6_zpskywgvmkb.png" border="0" alt=" photo 6_zpskywgvmkb.png"/></a>

 

[karimwassef]

So, the white #5 was highest, then 1 and 3 were comparable. The degraded blue #4 was incapable of getting over 110... This is after only a month or so of operating at ~2A.

 

[karimwassef]

I used the coverage area to convert PAR to light radiated power (W)

<a href="http://s1062.photobucket.com/user/karimwassef/media/LED%20experiments/3_zps1ndq5jnd.png.html" target="_blank"><img src="http://i1062.photobucket.com/albums/t496/karimwassef/LED%20experiments/3_zps1ndq5jnd.png" border="0" alt=" photo 3_zps1ndq5jnd.png"/></a>

That's not exactly right because there's radiated light that's not photosynthetically available, but for the purposes of this analysis - anything not in PAR is considered heat.

 

[karimwassef]

It's interesting that the PAR has a quadratic relationship to current (PAR~I^2), as does light power (directly proportional to PAR by area). But the curvature is downward vs. upward as with the rest of the power...