1W LED's

[Overclocked]

What do Brits call H2O?
Woe-tahh!
We call it water.

DiHydrogen mono Oxide.

 

[Hero999]

Is "worter" Cockney?

No it's my attempt at spelling water how it sounds.

I say water the same as you do except that I pronounce the T, you probably say D instead of T.

Many people say wa?er - they miss the T off completely.

The Cockneys say woo?er, but it's isn't woo-er either the war sound is in between woo and war.

How do you say new?

I say n-you, you probably say noo - you don't pronounce the y sound like I do.

 

[audioguru]

Americans say Wahdder, I say Water.
A nyew is almost a female deer. I say noo like a cow would say it.

 

[logto]

there is various of led www.atgelectronics.com

 

[audioguru]

That Chinese LED company has a "fluorescent tube" 4 feet long with 276 LEDs inside. They have another one with 342 LEDs but don't say how long it is and they don't say how bright they are.

 

[Overclocked]

1500mA - no way! Not sure which datasheet you're looking at. 350mA at 2.79V (the minimum voltage specified) is the absolute maximum that you'd want to run it at. The 1W rating is the current x voltage. This is the datasheet for your LED **broken link removed** . It looks to be the lambertain.
I did a test at work to test how well different thermally conducting compounds work. At the end I cranked up the current in stages see how much current these LEDs can draw before blowing up. Around 800mA for a few seconds.

Audioguru - its a torch! The flash light flashes no more than a ceiling light. (Do you call ceilings something different over there too?)

I was looking at data for the K2 Star, which can go up to 1500mA. But Im going to wait till they come and hope there is a part number for them.

 

[logto]

about fluorescent tube

this fluorescent tube is designed by mine .
do you want some informations about it i can give you /

 

[logto]

what is k2 star

 

[fingers]

logto - https://www.lumileds.com/pdfs/DS51.pdf
K2 is the name of the LED and star is the shape of the PCB which it is attached to.

Overclocked - I haven't thoroughly read the datasheet before <41 pages!> - see what you mean. 1500mA is a serious amount of current! Is the heatsink a portable air-con unit? We only ever use them up to 350mA, seems like a waste!

 

[Overclocked]

logto - https://www.lumileds.com/pdfs/DS51.pdf
K2 is the name of the LED and star is the shape of the PCB which it is attached to.

Overclocked - I haven't thoroughly read the datasheet before <41 pages!> - see what you mean. 1500mA is a serious amount of current! Is the heatsink a portable air-con unit? We only ever use them up to 350mA, seems like a waste!

The heatsink is pretty big, as long as a florscent bulb, and as wide as 4 or them! I beleive its made of steel with white paint, Id have to sand a bit off of in order to conduct heat properly.

If the LED ran at a maximum of 1500mA, The power would be 5.25W. The datasheet calls for 100 Lumens @1A @3.5V. Those are probably what the 3W LEDs produce..100 Lumens, and 50 lumens @ 350mA @3.5V...I only read the first couple of pages, the rest is physical measurments of the LED package.

 

[logto]

LEDs which is 3 W the natural current is 750MA, and 1 W leds current is 350 mA
The VF of all white leds either 3w or 1w is 3.2-3.4V

 

[logto]

do someone know AMC7150 this IC which is used for drive LEDs

 

[Overclocked]

do someone know AMC7150 this IC which is used for drive LEDs

There is no clear cut way to say which Controller to use, it all depends on your end Application. Are you going to run LEDs in serial or parallel? What are your output requirements (ie, Voltage and current)? What are your input Voltages? Are you running from a battery? etc etc.

If your running from a source that has a large output voltage, you'll have a easier time looking for a controller than if you were running from a battery.

If your just going to experiment with LEDs a old computer PSU will suffice. Be sure to use the proper resistor for the LEDs.

 

[Hero999]

For experimental purposes the LM317 is probably the best thing to use as it's pretty easy to build a constant current source that ranges from 10mA all the way up to 1.5A.

Builing a switch mode constant current controller isn't too different from building a contanst voltage supply except you monitor the output current rather than the voltage. However the input and output voltages are still important factors. If the input voltage is higher than the LEDs' maximum voltage drop then use a buck, if the minimum volt drop is always higher than the battery voltage then use a boost, if it's about the same as the battey voltage then use a SEPIC of a buck boost (the negitive output voltage is seldom a problem for LEDs).

 

[Overclocked]

I got the LEDs in today,So I looked around for a part number on the LED. Unfortunately, there was none to be found, So I went to testing. Included is a excel file of the data I took. Some of the data I took is questionable (ie, current- in some cases, the current went down instead of going up when resistance went down. Data my physics teacher calls "fubar" data, or simply put, data that has error) . Resistances, and voltages were taken with my meter. The only thing calculated is Current.

Im surprised these worked up to this much current, The most I tried is 1.3 Amps! 1W LEDs arent supposed to work up to this much, but the LED didnt work after that, But I cant say it was exactly 1.3A. Strange, I didnt hear a "pop" or anything, even the LED doesnt look burned. Surprisingly, these "supposed" 1W LEDs work up to 1A of current (and still work afterwards!). I still do wonder what part # these are...

Or I could have gotten lucky and these could have been mixed up with 5W LEDs...

 

[audioguru]

A LED that has had too much current doesn't make a "pop" and it doesn't look burnt. It is just melted inside. Sometimes its bonding wires melt like a fuse.

 

[Overclocked]

It seems they wont survive at 1A of current, I just had another one blow, So looks like They have to stay at 350mA. Im down to 3 LEDs, which should still be enough.

<strike>But that gives me an idea, I can probably pulse them 1A current pulses. I wonder if I could use a boost converter to power a 555 timer to provide high current pulses.</strike>

ADD: Even better idea-Boost converter without Output capacitor! That way the LED will be powered by high power pulses from the rectifier!

 

[justDIY]

pulsing 'next gen' leds doesn't yield any more brightness, perceived or average. the best way to get more light from most power leds is to lower the junction temperature. if you can keep the junction at 25C or less, you can keep upping the current until the internal bond structure can't take it anymore.

on the other hand, some new leds produce more light at higher temperatures, but you sacrifice longevity for brightness. The luxeon rebel works like this. you can run them at 350ma for years and years, or run them at 700+ma for months, at nearly double the brightness. The luxeon PWT is an even more extreme example of this. The pwt is rated to last a mere 2000 hrs, but cranks out 26 lumens from roughly a 0805 package

 

[Overclocked]

pulsing 'next gen' leds doesn't yield any more brightness, perceived or average. the best way to get more light from most power leds is to lower the junction temperature. if you can keep the junction at 25C or less, you can keep upping the current until the internal bond structure can't take it anymore.

on the other hand, some new leds produce more light at higher temperatures, but you sacrifice longevity for brightness. The luxeon rebel works like this. you can run them at 350ma for years and years, or run them at 700+ma for months, at nearly double the brightness. The luxeon PWT is an even more extreme example of this. The pwt is rated to last a mere 2000 hrs, but cranks out 26 lumens from roughly a 0805 package

I did have it on a heatsink, and the thing didnt even feel hot or warm.

Heres a thought, A 100W Energy efficientbulb emits 1100 Lumens (according to the packaging anyway...) and use only 22W. Luxeon Rebel (tri-star) emit 285 Lumens @ 2.45W (assuming Vf=3.5V and Iled= 700mA). To match up with up a energy efficient bulb, you would need 4 of these things. If each LED uses 2.45W, then that would bring up the total to 9.8W!

I thought LEDs were poop poo when compared to Those energy efficient bulbs? If my math is correct, then the new Rebel LEDs are more efficient than those energy effcient bulbs, BUT cost 14X more (the 100W pack of bulbs had 4 bulbs and were $4, bringing the cost per bulb to $1 each). However, If the LEDs produce 1140 Lumens @ 9.8W, then you can still double the amount of LEDs and still be energy efficient.

But I wonder if the cost of the LEDs are worth it?

 

[justDIY]

wait, 100w equivilant cfl's are only $1 ea in Connecticut, is that some sort of government rebate program or social welfare? Damn, what store is this - I need to grab some!

the 23 watt cfl I have (claims to replace a 100w bulb) outputs 1600 lumens, that's roughly 70 lm/w. The best rated cool white rebel is 180 lm @ 2.38w (700ma * 3.4v) which is roughly 76 lm/w. To match the output of my $8 CFL, I'd need 9 rebels, at $6.60 ea that's $60. I think I'll stick with compact fluorescent for now. It is good to see the numbers are getting better, but lumens per watt isn't the metric the public will judge new lighting technology by. It's lumens per dollar:

rebel = 3.6¢ per lm
cfl = 0.5¢ per lm

The reason Edison is still rich with his hot-wire bulb = 0.016¢ per lumen ... that's right, $0.00016 per lumen (figuring 25 cents per bulb)

edit:

oh, don't forget to include energy used by the driving circuitry - cfl's have very efficient offline smps powering them. the same technology is just now becoming available for leds. running leds using resistors or linear regulators puts their efficiency numbers back in the stone age with the hot wire bulb.

edit2:

if your heatsink didn't even feel warm, then you have a problem with thermal coupling. most "star" type leds have a rather high thermal resistance, although it's better than the olden days. new ceramic composite leds have a much lower resistance (rebel, ostar, lamina) - they're the ones you can really push. you need a laser IR thermometer or a well placed thermocouple to really measure the junction temperature - lumileds has some data about this in their thermal management white paper.