Axiom's KiwiLED Division ends the Light Wars

[manuellynch]

Axiom's KiwiLED group in New Zealand has just introduced a T8 LED tube for use in countries with 230VAC power systems.

The KiwiLED tube is higher output than a 4 ft T8 fluorescent 36 Watt tube, operates at 50% less power at 18 Watts, and runs off the existing magnetic ballast and starters and requires no rewiring.

The AT12M18W50 is a 4 ft or 1200mm tube that has a viewing angle of 270 degrees and has built in safety measures to allow simple re-lamping of existing lamps. The product sells for $55 NZD or $42USD.

The product is only currently available for shipments to New Zealand, Australia, Fiji, Samoa, Tonga, New Caledonia, and Tahiti.

The product will be available in North America, Central America, South America, and Europe in 2012 and will work with ANY electronic ballast without rewiring - simply change the bulb. We expect pricing to be only $35.00 per tube at that time.

Learn more about the Smart Simple Technology from http://www.kiwiled.com/docs/imaje-smarttechnology.pdf

We will be posting several videos and stories of some of the largest companies in the world using this technology to save an average of 10-15% on their utility/electric bills per month.

[megavolt512]

Axiom's KiwiLED group in New Zealand has just introduced a T8 LED tube for use in countries with 230VAC power systems.

The KiwiLED tube is higher output than a 4 ft T8 fluorescent 36 Watt tube, operates at 50% less power at 18 Watts, and runs off the existing magnetic ballast and starters and requires no rewiring.

The AT12M18W50 is a 4 ft or 1200mm tube that has a viewing angle of 270 degrees and has built in safety measures to allow simple re-lamping of existing lamps. The product sells for $55 NZD or $42USD.

Wow, it produces more than 2800 lumens? For 18 watts? That would be an efficacy greater than 155 lumens/watt - nearly 3x anything the DOE Caliper program has measured in a T8 LED retrofit. Have you had any independent testing done that verifies this extraordinary performance?

http://apps1.eere.energy.gov/buildings/publications/pdfs/ssl/led-t8-flourescent-replacement.pdf

http://www.grainger.com/Grainger/GE-LIGHTING-Fluorescent-Lamp-4PL16

Edited January 18, 2011 by megavolt512

[manuellynch]

Axiom's KiwiLED group in New Zealand has just introduced a T8 LED tube for use in countries with 230VAC power systems.

The KiwiLED tube is higher output than a 4 ft T8 fluorescent 36 Watt tube, operates at 50% less power at 18 Watts, and runs off the existing magnetic ballast and starters and requires no rewiring.

The AT12M18W50 is a 4 ft or 1200mm tube that has a viewing angle of 270 degrees and has built in safety measures to allow simple re-lamping of existing lamps. The product sells for $55 NZD or $42USD.

Wow, it produces more than 2800 lumens? For 18 watts? That would be an efficacy greater than 155 lumens/watt - nearly 3x anything the DOE Caliper program has measured in a T8 LED retrofit. Have you had any independent testing done that verifies this extraordinary performance?

http://apps1.eere.en...replacement.pdf

http://www.grainger....cent-Lamp-4PL16

I love people like you - full of lots of data that doesnt mean anything. Do you know who started the Caliper Program? Yours truly with James Broderick of the DOE - my LED fixture was the first ever done.

A T8 Fluorescent in a recessed downlight fixture has an efficacy of about 60-70% of the 2800 lumens you claimed - or about 1960 lumens - as consistently shown for a T8 tube in a DOE approved LM79 optical performance test. Our T8 tube is about 2100 lumens...so yes it is greater.

We have run LM79 test or comparable in New Zealand and your product output is in compliance with all of photometric requirements.

So while I appreciate your ability to follow the great misinformation put out in these reports - I will remind you that they are very much aligned with CREE Lighting and not Nichia of Japan - so our tube is the first product on the market that uses Nichia LED....and yes are greater than 120 lumen/watt on our parts.

So thank you for challenging me - for those that know me - I will have fun running circles around you and your links. You will be seeing this tube in facilities all over the USA in 2011.

Find the first Caliper test and check with the DOE - it was run on a recessed downlight that I designed, developed, and introduced long before CREE and the DOE got involved...I patented it as well.

And you know - while all of those reports are great - do you know that the USA 2nd largest homebuilder had specified our product despite what the Caliper results eventually published.

[saneon]

Wow , I think you touched a nerve John. How dare you challenge his inflated numbers.

Don't you know that everyone on this board is suppose to accept Manuels BS without question or face public humiliation by an Axiom NZ

beat down. f-ing rediculous

[megavolt512]

A T8 Fluorescent in a recessed downlight fixture has an efficacy of about 60-70% of the 2800 lumens you claimed - or about 1960 lumens - as consistently shown for a T8 tube in a DOE approved LM79 optical performance test. Our T8 tube is about 2100 lumens...so yes it is greater.

We have run LM79 test or comparable in New Zealand and your product output is in compliance with all of photometric requirements.

68% is actually the typical T8 luminare efficiency. What is the luminare efficiency with your T8 retrofit? (please don't say 100 percent). If you've done an LM79 test sequence (which tests the entire luminare, not just the chip), you already know electrical characteristics, light output, luminous intensity distribution, and color characteristics. So what are they?? Just saying "higher output than a T8" doesn't mean much.

36K Hour T8 has 90+ percent lumen-maint at end of life. Cost to relamp: $3.50 ($2.50 for the bulb, $1.00 to completely recycle it)

Data for your T8 LED product here?

What about lumen-maintenance? A relamping of your retrofit product will cost 10x that of a fluorescent relamping, and it is critical to know when your product is expected to reach 70% output, and require replacement. Will the product show a similar lumen-degradation to your signage product in the GWH project? You've shown some LM80 lumen testing data on your web site, but it appears to be Nichia's own test data for the chips themselves, not your completed product. These chips were mounted on a test bench with water-cooled heatsinks if I read it correctly?

Sorry to sound skeptical Manual, but other T8 LED products have not measured up to the claimed performance in ANY category. The one's operating off the existing ballasts are particularly bad, with ballast losses even higher than when a florescent tube was used. You've announced a product here that claims to have solved ALL these problems, and actually performs twice as well than the fluorescent lamp it replaces. Great!!! However, it should be no surprise folks will ask to see more data with such a bold claim.

And btw, I'm not disputing your performance, I'd just like to see your data. Preferably from a 3rd party, but if you don't have that... how about posting the LM79 report that you have done? That would be a great start.

You also said the Caliper program is "closely aligned with CREE." Please explain that. The program tests completed LED luminares, purchased through normal distribution channels. Individual chip mfgs are not identified. Are you saying the really good products that perform 2x... 3x better are being excluded to promote a specific mfg.. CREE? If this is true, than by all means let's sink the credibility of this DOE program. You'd have lots of support here.

I never said the DOE Caliper program was flawless, but it's the largest independent testing program in the world for SSL luminares right now. If you know of better one, please post it here. I wish my industry (cold cathode... neon) had such a taxpayer-funded advocacy group testing and rating our products for performance. If I made LED luminares I thought far-outperformed the competition, I'd embrace the Caliper program.

Edited January 19, 2011 by megavolt512

[manuellynch]

Wow , I think you touched a nerve John. How dare you challenge his inflated numbers.

Don't you know that everyone on this board is suppose to accept Manuels BS without question or face public humiliation by an Axiom NZ

beat down. f-ing rediculous

Now thats not true - just some hot buttons I have at times. Thanks for posting this though Saneon - made me relook at what I read and realize I was a bit "harsh" - thanks for drawing that to my attention.

[Brian]

Waiting to see what comes next!

[manuellynch]

your questions

1) You said "68% is actually the typical T8 luminare efficiency - so that means it is only 68% efficient in getting out fluorescent light"

those numbers are usually based on open fixtures. On our LM79 testing we did on our list retrofit kits our light output was higher - these LM79 test results are on our website. Our new system has higher performance than our last kits and will have higher output - LM79 testing in process - will post in 30-60 days. So I will make sure you get a copy.

2) You say a "36K Hour T8 has 90+ percent lumen-maint at end of life. Cost to relamp: $3.50 ($2.50 for the bulb, $1.00 to completely recycle it)"

- oh we know all that - we have to have the same or better performance and we do. in fact in New Zealand we arent competing with T8 or T12 we are competing with companies deciding on whether or not to move to T5 or our LED solutions. Most of the decision...honestly...is a commitment to companies being "more green" - and what that means....no one really knows. It is just what they want to do. Fonterra Dairy (largest milk producer in the world) wants to move away from fluorescent due to end of life and relamping "sparks" - in high methane areas it has caused fires. LEDs help fix a safety problem. The largest property management company wants to do the change out to LED because it makes their properties more "modern and attractive to customers". So everyone has their "real reason" for making the change.

3) You asked about the LM80 data.

Well you are correct - this is Nichia data running the parts at 80C ambient and 80% humidity - but I am pretty sure they are running on thin FR4 board with no heatsink - they are doing the test to show it working for a TV, Mobile Phone, or Car dash - not a lot of metal around. They all show the data running the parts at full power. We dont run our LEDs at full power - we run 15-25% below what they recommend. Keep in mind that the high temp for an office in New Zealand is 22C... a bit less than 80C. I am not sure you know but I came from a world of military, satellite and medical electronics - so lifetime testing and performance is engrained in what we do.

4) You said "Sorry to sound skeptical Manual, but other T8 LED products have not measured up to the claimed performance in ANY category. The one's operating off the existing ballasts are particularly bad, with ballast losses even higher than when a florescent tube was used. You've announced a product here that claims to have solved ALL these problems, and actually performs twice as well than the fluorescent lamp it replaces. Great!!!"

None of the LED tubes that were on the market used Nichia LEDs - they used some pretty crappy Chinese LEDs. Please note that we have not yet finalized the tube running on electronic ballast and I agree with you - there are quite a few losses there...for now. On the magnetic ballast we have shown in repeated test that we are running them at 97.5% efficient.... this was tested by a power company and confirmed. So what we really did is make it work with magnetic ballast found outside the USA. So relamping is easy and quick and requires no modification of the fixture...this is the biggest selling point...because in New Zealand it is AGAINST THE LAW to rewire or change a fixture without a hell of a lot of supporting data from safety organizations...they are very very very careful down here. Safety to a new level.

5) You said "And btw, I'm not disputing your performance, I'd just like to see your data. Preferably from a 3rd party, but if you don't have that... how about posting the LM79 report that you have done? That would be a great start."

We will be posting this in 30-60 days - we have found that most of the interest is in a 5 foot solution and not in 4 foot - so we are modifying a few things to finalize all and then do the data. But we will do the data along with a lot of customer testimonials and stories. Hell - what we are doing will be on the world stage in Sept 2011 as the Rugby World Cup is in New Zealand and our lights will be everywhere!

6) You said "You also said the Caliper program is "closely aligned with CREE." Please explain that."

CREE is a military electronics company - they get a lot of money from the DOE and DARPA to help develop technologies - I think they have received $60mil+ plus. It is known that CREE sits very heavily on these government officials to ensure systems are put in place to help make "Cree Shine" - this is politics guys. Cree is a public company with a stock that people make a lot of money on. Cree has moved to China and is now focused on the Chinese market - good for them. They realized the USA market for LED lighting was not what they thought it was going to be - hell, I am the one who personally started going after home builders for LED lighting - and we know what happened to home building. DOE ignores Nichia - who by all analyst accounts is the world leader in both size, product performance, and in low cost. If you use a CREE product you get helped along in USA government programs. Good on them for building the parts in China now and getting around the fact that USA military cannot use Chinese made products - political friends clearly helped.

7) You said "I never said the DOE Caliper program was flawless, but it's the largest independent testing program in the world for SSL luminares right now. If you know of better one, please post it here. I wish my industry (cold cathode... neon) had such a taxpayer-funded advocacy group testing and rating our products for performance. If I made LED luminares I thought far-outperformed the competition, I'd embrace the Caliper program."

Well we should talk about CCFL - I am a huge fan of it. Not sure if you know but I was VP of Biz Development for Microsemi - the company that made the chips to drive CCFL more efficiently...so I appreciate your technology. As far as a better program than Caliper....well I think that program is about done since Barack Obama but the reigns back into EPA Energy Star hands... they are developing a means an way to measure LED fixtures that is going to make it even more complicated. What is very funny is that there is no Caliper program for incandescent, CFL, etc.

I want to thank you for challenging me - hopefully you are coming to ISA in LAs Vegas - would love to have a drink or two and talk about this more. Apologies for coming off so harsh before - that was pretty crappy of me.

[Brian]

Manuel

#2 You really did not answer the question at all

#4 Most retrofits done in the US will have electronic ballasts in the sign/fixture not magnetic

#5 Has nothing to do with the world cup. I understand you are proud of that but kinda tired of hearing about it in most posts LOL!

Just 3 things that hit me right away

[manuellynch]

Manuel

#2 You really did not answer the question at all

#4 Most retrofits done in the US will have electronic ballasts in the sign/fixture not magnetic

#5 Has nothing to do with the world cup. I understand you are proud of that but kinda tired of hearing about it in most posts LOL!

Just 3 things that hit me right away

Bryan -

First - I will answer ALL questions when our product is introduced in the USA in late 2011. So if you really do read these post - than please do. We are not introducing in the USA yet because there is no consistency in what anyone uses - so we are debugging how to do this "effectively and efficiently".

And secondly - do you ever chime into conversations without just being an argumentative prick or is this just an outlet for you? I know I can be a prick on issues but I also try to add some value on other topics so people can learn and grow. In-fighting and hostility between differences of opinion really leaves nobody in a great place - hell, look at what happened in Tucson.

And on the World Cup - ok - understood, I will not bother you with stuff from New Zealand - we would have liked to have you at our ISA Party - but since there will be a lot naked models from New Zealand I dont want to offend you. So instead - watch the Super Bowl - we are involved heavily in the lighting in the stadium and in the halftime show.

Manuel

[Brian]

Wow someone got up on the wrong side of the bed this morning! I did not attack you at all just needed to know if you could answer those questions. I really don't care if I go to your party or not Manuel so that is not a threat to me. Man don't be a baby

And if you are comparing me to the freak in AZ F you!

Edited January 19, 2011 by chubbygumby

[megavolt512]

[/color]

I want to thank you for challenging me - hopefully you are coming to ISA in LAs Vegas - would love to have a drink or two and talk about this more. Apologies for coming off so harsh before - that was pretty crappy of me.

no appologies necessary! I love talking about LED-fluorescent performance, design, and efficacy.

I'll eagerly await your LM79 and LM80 reports.

[YYZ]

I won't get into any of Manuel's products, but can comment on a couple of things as another LED developer.

You've shown some LM80 lumen testing data on your web site, but it appears to be Nichia's own test data for the chips themselves, not your completed product.

You are correct - there is a difference between chip performance and luminaire performance - the latter is what matters. Knowing the LEDs that Manuel us using I have no doubt that it will be very bright, but will reserve praise or criticism until the LM79 results are shared. LM79 has nothing to do with Caliper and is essentially the defacto measuring stick for any lighting products using LEDs.

These chips were mounted on a test bench with water-cooled heatsinks if I read it correctly?

They do not use water cooled heat sinks for their testing. Most of the time the heat sink is minimal or non-existent, actually, and is generally a PCB with the recommended copper design for thermal transfer away from the mounting pads.

You also said the Caliper program is "closely aligned with CREE." Please explain that. The program tests completed LED luminares, purchased through normal distribution channels. Individual chip mfgs are not identified. Are you saying the really good products that perform 2x... 3x better are being excluded to promote a specific mfg.. CREE? If this is true, than by all means let's sink the credibility of this DOE program. You'd have lots of support here.

Cree, being an American company, has been a big player in trying to steer the DOE in directions that they would like to go. It's not unlike GE trying to write standards for power transmission, Philips for trying to limit PFC to just 70%, or 3M trying to get the DOT specs for reflective sheet to be require a whole lot of their latest materials to be used.

That said - products are not excluded based on LED or chip type and if Nichia product is better it will be proven in the testing that Caliper does. It has been proven in some cases, if you know the products they have tested, but they don't release specific product details. You are only left to make educated guesses which results are for which products, which isn't too hard if you follow things very closely.

Beyond Caliper, though, the DOE and other domestic "industry groups" certainly do favor Cree as the LED of choice. To win certain bids or grants, you have to use the sponsor's product and Cree has been a sponsor of many of these things from the outset. It's a smart strategy on their part, but I generally have a hard time choosing their product over Nichia or Toyota when the performance, price and ability to work together is better with the latter parties. They listen and respond better with their product offerings IMO. I'll never forget asking one of their product development guys if a particular LED might be possible or become available and being told very clearly "no, why would we do that?". Toyota and Nichia both made this LED (LEDs with 3 or 6 chips in series) and next thing you know - Cree has it too.

[manuellynch]

I won't get into any of Manuel's products, but can comment on a couple of things as another LED developer.

Marko - you are one of the best listeners - you always blow me away! And what you said about your feet be cold and Canada on the phone the other day- that was funny. And I think Dubrovnik is the place.

Nichia doesnt spend resources on marketing that CREE does - for that matter Axiom doesnt either - so I guess there is a reason why our LED solutions are high value and low cost.

Thanks for clarifying the Cree - DOE connection Marko

[megavolt512]

I won't get into any of Manuel's products, but can comment on a couple of things as another LED developer.

These chips were mounted on a test bench with water-cooled heatsinks if I read it correctly?

They do not use water cooled heat sinks for their testing. Most of the time the heat sink is minimal or non-existent, actually, and is generally a PCB with the recommended copper design for thermal transfer away from the mounting pads.

Yes they do.

From the KiwiLED site posted that began this thread, on their home page where it says "Our Lifetime Data (LM80)" - http://www.kiwiled.com/docs/LM80Data-AoTuroaLED.pdf

"LM80 Test Data

Description of LED light sources

Part Number: NS3L183-H3

Part Name: Nichia Chip Type Warm White LED

Description of auxiliary equipment

Active cooling life test system

Consisting of small enclosed boxes for devices under test and water-cooled

heat sinks to control device temperature."

It should be no surprise the chips deliver stellar lumen-maintenance in a setup like this? Especially running only 6000 hours. Quite a far cry from the T8 retrofit described, where high power chips are put on a narrow strip and enclosed in a plastic tube with no forced-air cooling? Certainly you guys don't expect this type of test data to transfer to the finished product?

Once the light output has dropped to 70 percent, the product life for an illumination product is over, and it needs to be replaced. Folks are going to look much further down the road than 6000 hours for a retrofit that costs 10x the fluorescent lamps it replaces.

Edited January 20, 2011 by megavolt512

[Brian]

Thanks for that info megavolt. I have not seen a lot of your posts on this site but you seem very well informed.

[YYZ]

Yes they do.

From the KiwiLED site posted that began this thread, on their home page where it says "Our Lifetime Data (LM80)" - http://www.kiwiled.com/docs/LM80Data-AoTuroaLED.pdf

You're right, they did, and it appears that it may be the new way they're doing things.

Looking more closely, though, there are details we need to pay attention to. They didn't use the water to create only the most perfect conditions possible as you are making it sound. They used the water to regulate the case temperature of the product and they did this for 3 different temperatures:

55C / 131F (you really think they were cooling the water?)

80C / 176F (which would certainly not be cooled and probably heated)

105C / 221F (!) (which would certainly require heating of the water)

Taking only the best-case results would be a mistake, but weighing the worst case results at 105 degrees - hot enough to burn you on contact and a probably junction temperature of 115 degrees - still suggests the product performs favorably.

Edited January 20, 2011 by YYZ

[megavolt512]

The LED is the heat source here, the water need not be heated.

These LED's can operate at 105C. Without a heatsink the junction (and case temps) will soar above this, and ultimately ruin the chip. In other words, you still have to cool it to keep it at 105. You can't operate a 350ma LED with no heatsink at all (at least not for an extended period).

105C is more than enough to boil water all on it's own. To regulate the temp at a certain level, they need only restrict/increase the flow of water.

Almost certainly, most of the GWH entrants are dropping off lumens faster than their chip datasheets because the thermal management is less than provided on the test bench during evaluation.

Edited January 20, 2011 by megavolt512

[YYZ]

Almost certainly, most of the GWH entrants are dropping off lumens faster than their chip datasheets because the thermal management is less than provided on the test bench during evaluation.

Yes, of course. These LEDs are generally not the grade being used in most channel letter modules, and said modules have little to no heat dissipation and use simple fibreglas PCBs and not metalclad PCBs. (I don't know what Manuel is using in his fluorescent type product). In fact, the way Erik has them all clustered in neighboring compartments probably heats the rear of this cabinet even more than a channel letter would, with added surface>air area to help dissipate heat.

Channel letters can often create more heat than the LEDs if used in warmer climates, so the test end up being pretty accurate when it's all said and done, but the depreciation is also due to some products being driven at or closer to their limits than other products. We're talking about a wide range of items here with varying levels of ability to withstand fluctuations in heat, incoming voltage and drive current.

The LED is the heat source here, the water need not be heated.

These LED's can operate at 105C. Without a heatsink the junction (and case temps) will soar above this, and ultimately ruin the chip. In other words, you still have to cool it to keep it at 105. You can't operate a 350ma LED with no heatsink at all (at least not for an extended period).

105C is more than enough to boil water all on it's own. To regulate the temp at a certain level, they need only restrict/increase the flow of water.

With respect, the test details clearly state that there was an aluminum heat sink. We use them, so I can tell you with absolute certainty that there is no way that those LEDs attached to anything more than their own PCB would ever reach anything at or even near 105C. You'd have to add heat to the heat sink externally, which in this case means heating the water. Further, it's often drive components that generate as much or more heat than LEDs, believe it or not. Ask Erik to touch the Permlight el Plato's in the GWH and he'll tell you that the resistors are a whole lot hotter than the LEDs. Efficient product design is critical and some are better than others in this respect.

For example, we use 2 versions of that LED. One is driven at 285mA (the 350mA rated one in Manuel's report) and the other is driven at 560mA (rated for 700mA). On the 'hottest' light engine, we have 9 such LEDs clustered together on a PCB that is about 3.5" square. We've done measurements with and without heat sinks and drive currents up to 700mA (just to test our boundaries a little) and we never reached anything close to 100C at any point in time. Using a FLIR system to measure at the desired 560 mA and without any heat sink, the max junction temp did not exceed 80C, and with a heat sink (no liquid cooling or anything) it stayed in the upper 60s, which is well within spec.

Our on-board switching driver is between 92~96% efficient, so we're not using measly resistors to sink off 40% of the energy of a product that might only be 60% efficient. Heat generation really is limited to the LEDs with very little coming from ICs and such. Again - this is our product and not Manuel's - but it's entirely possible to design products that are efficient and stay within all recommended design guidelines. We're expecting the LM79 data shortly, while the LM80 takes a lot longer.

[megavolt512]

Agree with almost everything you said Marco. Good post.

However I'd clarify the PCB **is** the heat sink (or at least part of it). So is whatever you secure it down to. A 3.5 in square clad PCB is 12.25 square inches. If my calculations are correct, you are dissipating about 1 watt per chip at 285ma. 9 watts. A cooler board would almost certainly provide more lumen-maintenance (and more flexibility) but yes it sounds like as you've tested it things are below the max junction temp. Not a place I'd want to hang too close too for too long. (I don't think you were saying you were running the Nichia chips with NO heatsink (no pcb, nothing) at 350ma and staying well below max junction temp?)

Agree the drive components are additional heat sources. Yet more heat-sinking needed.

This has been a great thread - I wish there was more like it. I'll go back into my usual listen-mode now. I'm glad to see we're all passionate about our products and industry.

Edited January 21, 2011 by megavolt512

[YYZ]

Agree with almost everything you said Marco. Good post.

However I'd clarify the PCB **is** the heat sink (or at least part of it). So is whatever you secure it down to. A 3.5 in square clad PCB is 12.25 square inches. If my calculations are correct, you are dissipating about 1 watt per chip at 285ma. 9 watts. A cooler board would almost certainly provide more lumen-maintenance (and more flexibility) but yes it sounds like as you've tested it things are below the max junction temp. Not a place I'd want to hang out too close for too long.

Agree the drive components are additional heat sources. Yet more heat-sinking needed.

This has been a great thread - I wish there was more like it. I'll go back into my usual listen-mode now. I'm glad to see we're all passionate about our products and industry.

Just to be clear, that 9 LED board is always attached to a heat sink in the fixture(s) we supply it for. It's one of our OEM clients and we make about a dozen "engines" for them. This one is a recessed downlight with about 1250 source lumens @ 18W power consumption (including PSU losses, etc), so a luminaire efficiency of almost 70 lumens per watt.

Channel letter modules should be able to operate without being attached to anything at all - so you're right - the PCB should essentially be the only heat sink. The GWH cabinet is likely warmer than most signs out there and I'm unsure if our EasyStroke being at the top of the TW cabinet is bad, by being where all the heat goes, or good, by being where the heat also exits - but it is what it is. Our testing showed it to increase over the first bit and then decrease a little over 1% per 1000 hours - so a projection of 50% brightness at about 55 hours. Erik's testing in SoCal shows about 12.5% drop after 11.3K hours, but the upswing due to the colder ambient temps will bring this average back in line with our projections. There are peaks and valleys every year, so you have to track over multiple years to plot out a more linear 'average', which can be done a bit better in the GWH that we're not actually a part of.

Basically, at 12 hours a day we're projecting 10+ years of usable life, depending on geographical location - but our modules are sort of a lower priority right now. We have some key clients using them and a mix of occasional buyers, but we're really not pushing them like other companies who do nothing but this sort of thing. We have other sign and display stuff that does much better for us and we have little to no competition on some of the products - but again - with key clients only.

I guess I should shut up and stop hijacking Manuel's thread. We started with general LED talk, but I've gone a little too much into our stuff... though with the number of times he has jacked other threads, he has a little coming back to him ;)

Speaking of which...

Dubrovnik? Are you sure? I've been there and even being from a Croatian background, I have to say I'd have a hard time making that choice. It'd be a kick teaching you some of the language, if you want... but you really need to look at Split if the coast is what you have in mind.

[Erik Sine]

Agree with almost everything you said Marco. Good post.

However I'd clarify the PCB **is** the heat sink (or at least part of it). So is whatever you secure it down to. A 3.5 in square clad PCB is 12.25 square inches. If my calculations are correct, you are dissipating about 1 watt per chip at 285ma. 9 watts. A cooler board would almost certainly provide more lumen-maintenance (and more flexibility) but yes it sounds like as you've tested it things are below the max junction temp. Not a place I'd want to hang out too close for too long.

Agree the drive components are additional heat sources. Yet more heat-sinking needed.

This has been a great thread - I wish there was more like it. I'll go back into my usual listen-mode now. I'm glad to see we're all passionate about our products and industry.

Just to be clear, that 9 LED board is always attached to a heat sink in the fixture(s) we supply it for. It's one of our OEM clients and we make about a dozen "engines" for them. This one is a recessed downlight with about 1250 source lumens @ 18W power consumption (including PSU losses, etc), so a luminaire efficiency of almost 70 lumens per watt.

Channel letter modules should be able to operate without being attached to anything at all - so you're right - the PCB should essentially be the only heat sink. The GWH cabinet is likely warmer than most signs out there and I'm unsure if our EasyStroke being at the top of the TW cabinet is bad, by being where all the heat goes, or good, by being where the heat also exits - but it is what it is. Our testing showed it to increase over the first bit and then decrease a little over 1% per 1000 hours - so a projection of 50% brightness at about 55 hours. Erik's testing in SoCal shows about 12.5% drop after 11.3K hours, but the upswing due to the colder ambient temps will bring this average back in line with our projections. There are peaks and valleys every year, so you have to track over multiple years to plot out a more linear 'average', which can be done a bit better in the GWH that we're not actually a part of.

Basically, at 12 hours a day we're projecting 10+ years of usable life, depending on geographical location - but our modules are sort of a lower priority right now. We have some key clients using them and a mix of occasional buyers, but we're really not pushing them like other companies who do nothing but this sort of thing. We have other sign and display stuff that does much better for us and we have little to no competition on some of the products - but again - with key clients only.

I guess I should shut up and stop hijacking Manuel's thread. We started with general LED talk, but I've gone a little too much into our stuff... though with the number of times he has jacked other threads, he has a little coming back to him ;)

Speaking of which...

Dubrovnik? Are you sure? I've been there and even being from a Croatian background, I have to say I'd have a hard time making that choice. It'd be a kick teaching you some of the language, if you want... but you really need to look at Split if the coast is what you have in mind.

Remember the PTW which YYZ is in are individual channels fastened to wood (rears exposed to open air), so there is no contact with any other light source or channel. GWH is two cabinets (for now maybe 3) 5" deep to simulate large channel letters divided into separate compartments. The rear of these cabinets are exposed to open air, so there is cooling there unlike a channel letter fasted to a wall, a wall of which might generate and trap heat and passed onto the CL, and the cabinet is indoor so the sunlight will not bake generate heat and possibly bake components. This is also why I take compartment temp readings in every monthly evaluation, The LED compartments in the GWH are usually 10 to 20 deg coller than CCFL compartments. But some LEDs put out a lot of heat, more so than others.

This is a damn good thread, wish I could separate it.

[manuellynch]

concrete building sidings make good heatsinks - hell, feel the sidewalk in Las Vegas in the summer - there is a reason why the heat last all night long at times.

Mounted to wood with air will be actually worse than the side of a building. In your shop you dont have a lot of air moving. Letters on the side of a building have air flow over them all the time. Air is a coolant.

So I think to solve all of this Erik should put the signs all over the outside of his house!!! hahaha

Edited January 21, 2011 by Manuel Lynch

[LiteGirl]

thanks for the wealth of information guys