How to solve high power LED array heat issue?

[Dave_UK]

Re: LED array heat issue

This is a link to the LEDs:-

**broken link removed**

I cant find any info on the therninal resistance.

The array will be used on a small aquarium to provide light to coral. There is already alot of lighting available for this in the form of flourescent tubes and metal halide lamps but these have drawbacks suck as size (tubes) and heat. I was hoping toovercome these issues by using LEDs'. I selected the LEDs mainly due to their efficiency and because I am able to get the correct colour temperature (10000k).

 

[polymath]

Re: LED array heat issue

Hi Dave_UK
Have you considered a GU10 LED spot lamp? Runs simply from 240vac

spec:
Current per LED 13.8 mA
Total current 20mA
Total power consumption 2.1W
Surface illumination 29000lux
Luminous flux 70lm
Color temperature 10000K
LED luminous intensity 9000~14000mcd
LED’s viewing angle 15~20deg
Useful life 50000 hours

Looks like they have a UV filter on front but this is probably 'removeable'.

Find them here (Manchester UK):
**broken link removed**

regards ... Polymath

 

[PaulHolland]

Re: LED array heat issue

polymath, Your calculations make one fatal error and that is: emmitted light from an LED is NOT lineare with current !!!!.. two times the current does not give two times the light emmitted from the LED. If this was true they would simply use PWM to make superled's !!.

The second error you make is that you think your light emmission is 416% up but you forget to take the 1/10 emmision time into account. since 9/10 there is no light your total light emmision is = 7.5 * 1/10 = 0.75 about 50% of what you get with constant current.

Paul.

 

[Dave_UK]

Re: LED array heat issue

Thanks for the link Paul,

I looked at a few similar 240v lamps before I found the LEDs I am currently using. The problem with them is that their total output is only 70 lumens. The output of my 48 led array should be 864 lumes so I would have to use 12 lamps to get a similar output to the high power array and that would give me to large a footprint.

 

[polymath]

Re: LED array heat issue

polymath, Your calculations make one fatal error and that is: emmitted light from an LED is NOT lineare with current !!!!.. two times the current does not give two times the light emmitted from the LED. If this was true they would simply use PWM to make superled's !!.

The second error you make is that you think your light emmission is 416% up but you forget to take the 1/10 emmision time into account. since 9/10 there is no light your total light emmision is = 7.5 * 1/10 = 0.75 about 50% of what you get with constant current.

Paul.

Hi Paul
Dave_UK has shown that this is getting off-topic.
Thanks for your comments - can you refer me to any data on pulse output of LEDs?

Every data sheet I looked at on Kingbright shows a Linear function between Current and Luminous Intensity. Perhaps Kingbright have got it wrong.

I am aware that LED performance soon drops-off with temperature rise.

As the human eye does not respond to the average photon count but to peak values and slowly - the factoring of the Luminous Intensity - that you suggest is required - is not valid for human eyes - if it were then Multiplexing of LEDs or the CRT TV would not work!

I have designed many long distance IRED Tx and pulse driven these at values in excess of 2.0A! on a 5mm IRED and achieved remarkable distances in bright sunlight. No - don't tell me it is all in the Rx.

I look forward to reading your reference for LED Current/Luminous Intensity Function.

May I refer you to NICHIA App Note on 'Controlling the Luminous Intensity of LEDs:
http://www.nichia.com/specification/appli/controlling.pdf
Clearly states:

The intensity of NSPR346BS(Red) is a linear function of current

x10 current will produce x10 output - Linear
Because White LEDs are based on a blue die with a yellow overlay:

The intensity of NSCW215(White) is not a linear function of current

Dave_UK did not mention white LEDs in the start of the thread.
My calcs were based on a Red LED - consequently correct!
Nichia show that for a White LED:
x4 current will produce x3 output - not linear but still a massive increase in output.

Hi Dave_UK
Here is a link to thermal management of LEDs - might be too technical - have a look:
http://www.nichia.com/specification/appli/thermal.pdf

best regards ... Polymath

 

[PaulHolland]

Re: LED array heat issue

polymath,

A transistor also has a linear region and a saturated region when its relationship between input and output is nolonger linear. Power amplifiers for FM make use of this region for high efficiency (transistor looks like a switch in this mode). Your IR link makes the same use of this region by using frequency modulation, if you would use AM modulation you would not be able to get the same distance.

**broken link removed**

The human eye is integrating light it is seeing so a pulsed light source on/off for 50% will be half that of 100% etc..

to be short: a LED has an linear region followed by a non linear region close to the point of saturation.

Paul.

 

[polymath]

Re: LED array heat issue

polymath,

A transistor also has a linear region and a saturated region when its relationship between input and output is nolonger linear. Power amplifiers for FM make use of this region for high efficiency (transistor looks like a switch in this mode). Your IR link makes the same use of this region by using frequency modulation, if you would use AM modulation you would not be able to get the same distance.

h**p://www.automotivedesignline.com/howto/190303172;jsessionid=K2L0WO2ZVI35AQSNDLQCKH0CJUNN2JVN

The human eye is integrating light it is seeing so a pulsed light source on/off for 50% will be half that of 100% etc..

to be short: a LED has an linear region followed by a non linear region close to the point of saturation.

Paul.

Thanks Paul

I am aware of the workings of a P-N junction down to the FERMI Level analysis.

Your 'technical' comments on an IR system on which you have NO technical data and have NOT seen are pure fantasy.
I suggest you replace your cloudy crystal ball!

You comment:

The human eye is integrating light it is seeing so a pulsed light source on/off for 50% will be half that of 100% etc..

confirms the fact that if the human eye is provided with a short 500% 'bright' pulse for 50% time will integrate this pulse over the repetition period and conclude it is 250% 'bright' - just what I was saying - thanks.

Paul you wrote earlier:

polymath, Your calculations make one fatal error and that is: emmitted light from an LED is NOT lineare with current !!!!..

I notice you have avoided making any comment about:
a LED has a LINEAR OUTPUT with CURRENT
So you know better than NICHIA and Kingbright - good for you.


best regards ... Polymath

 

[PaulHolland]

Re: LED array heat issue

I must be dealing with a child on the otherside of this line. End discussion !.

Paul.

PS: Emmitted white light fro LED is NOT lineare with current input. You should try before you think (dream) something. Nothing in NATURE is lineare not even your eyes, ears etc..

 

[polymath]

Re: LED array heat issue

Thanks Paul

Posting insults I will leave to you - with the lost debate!

LED OUTPUT INTENSITY is LINEAR with CURRENT
as stated by NICHIA and Kingbright - read it here:
http://www.nichia.com/specification/appli/controlling.pdf

I do not know more than Major Multi-National LED manufacturers like NICHIA and Kingbright - like you think you do.
I am still open minded and continually learning.


best regards ... Polymath

 

[PaulHolland]

Re: LED array heat issue

From the largest LED and opto maker in Taiwan:

This was the original high efficiency red LED, which was available as far back as 1976. The chemistry is gallium phosphide doped with zinc oxide. These LEDs are usually impressively efficient (1 to maybe 2 lumens/watt) at low currents of a few mA or less, but are only about 2-3 times as efficient as the original formula red ones at 20 mA. The color varies with current, and is nearly enough pure red at .5-1 mA but more orange at higher currents. If the lamp is not tinted red, the emitted light is usually orange to sometimes slightly yellowish orange at 30 mA. The spectral output is a broad band, nominally peaking at 697 nm and maybe only peaking that far out in the red at really low currents. There is a secondary spectral band in the green peaking around 545-550 nm which is more noticeable at higher currents. Maximum drive current is usually 30 mA, but these LEDs have noticeably nonlinear light output that increases less than proportionately with current above a few mA. Voltage drop is around 1.9 volts.

(5) Super High Brightness Red, GaAlAsP –
These things became available in the mid 1980's. Earlier models with opaque substrates were impressive back then with efficiencies of 1-2 lumens/watt. These have since been improved, with transparant substrates and other refinements. Agilent has a similar chemistry which they call "T.S. AlGaAsP". The overall luminous efficacy of the best models is about 9 lumens/watt. The spectrum usually peaks between 650 and 670 nm, but some Agilent models peak slightly shorter. The color is pure red to He-Ne laser red, with a dominant wavelength (wavelength of monochromatic light of matching color) usually in the 640's but anywhere from about 635 to about 650 nm. Efficiency is usually maximized at currents near 20 or 25 mA. Efficiency at low currents of around a mA to a few mA is not impaired as badly as it is in many other types that have efficiency peaking near or over 20 mA. Typical voltage drop at 20 mA is 1.8 to 1.9 volts. Maximum rated drive current is usually 30 mA but sometimes 50 mA. These usually glow dimly at 1.5 volts.

(6) High Efficiency Red, Orange Red, and Orange, GaAsP on GaP substrate –
This was the first non-low-current high-efficiency red LED. The working chemistry is gallium arsenide phosphide, with an arsenic/phosphorus ratio around 40-60 on a gallium phosphide substrate. The GaP substrate is transparant to the emitted light and GaAs is not. This is one reason why these LEDs are more efficient than the original formula red ones. The color is normally orange-red with a dominant wavelength around 620 nm. A slight variation of this has a dominant wavelength usually around 610-615 nm and is considered orange. The peak wavelength is usually around 630 nm. Typical drive current is 5 to 20 mA and maximum current is usually 30 mA. Typical voltage drop at 20 mA is around 1.9 volts.

(7) Ultrabright Orange-Red, Orange, Yellow, and Green, InGaAlP –
The indium gallium aluminum phosphide LEDs came out in the early 1990's. These are usually either red-orange with a dominant wavelength around 610-617 nm, or yellow or "amber" with a dominant wavelength around 590 nm. Dominant wavelength can be as low as the 550's of nm ("pure green" or "emerald green", yellowish green but less yellow than usual for LEDs) and as high as the low 630's (He-Ne laser red). The most efficient ones are usually orange to orange-red. Overall luminous efficacy for orange and orange-red ones is anywhere from 7 to 24 lumens/watt, with most made after 2000 achieving at least 12 lumens/watt. Yellow ones are mostly a little less efficient, 4 to 15 lumens/watt with most made after 2000 achieving at least 8 lumens/watt. Green ones are less efficient still (mostly 3-4 to about 8 lumens/watt) but more efficient than GaP and GaAlP green LEDs. With the exception of some more-red models, these LEDs have significantly reduced efficiency at low currents of a few mA or less. Typical drive current is 5 to 20 mA. Maximum drive current is usually 30 mA but sometimes optimistically 60 mA. Typical voltage drop at 20 mA is around 1.9 to 2.3 volts depending on the specific variation.

 

[polymath]

Re: LED array heat issue

Hi Paul

Thanks for the LED history - I do not use obsolete components.

Can you post the link to the Taiwan LED Manufacturer?
Always willing to learn.

regards ... Polymath

 

[Dave_UK]

Re: LED array heat issue

Now Im completely confused...lol

can someone suggest a solution for cooling the led array?

I will eventually be using a PCB. how do I attach a heatsink to this?

Thanks

Dave

 

[PaulHolland]

Re: LED array heat issue

Hi, Why confused. Did you check the pdf file I uploaded ?.

One way of cooling your LED's is mounting them close to the PCB as possible and use copper planes (top/bottom PCB) to get rid of the heat. Other ways are small CPU fan's.

Can you tell me what wavelength your light source needs to be maybe LED's are not the best choice.

Paul.

 

[Dave_UK]

Re: LED array heat issue

The white light needs to have a colour temp of 10,000k or higher. I will also be using blue LEDS which are 465nm. Ideally the blue light source should be 420nm but I couldnt find any LEDs of that wavelength.

 

[polymath]

Re: LED array heat issue

Hi Dave_UK

Maplin have got 420nm 5mm LEDs - I don't know if they are what you want:

**broken link removed**

regards ... Polymath

 

[PaulHolland]

Re: LED array heat issue

Hi, Best use TL based light source. Minimal heat and maximal light. TL based light sources can provide 5000K per tube.

**broken link removed**

Blacklight (use in disco's) is about 400nm maybe they even have 420nm.

I also would advice you to look at: https://www.lighting.philips.com contact them direct, If you need help I can provide contact within Philips (Happens to have good contact there from previous contracting work).

regards,

Paul.

 

[polymath]

Re: LED array heat issue

Hi Paul
I agree - TL seems to be the best solution - perhaps Dave_UK has a maximum size limit.

I have just read your LED-lighting.pdf posted after my controlling.pdf
An extract from your LED-lighting.pdf:

Contrast the steep voltage/intensity curve with the linear
relationship between current and intensity in Insert 1.

This is what I understand to be the case - re. previous postings.
Perhaps you believe your posting to be incorrect?

Is it little wonder Dave_UK is confused! I going that way as well - must be the heat!

best regards ... Polymath

 

[Dave_UK]

Re: LED array heat issue

Thanks for the suggestions.

The LED from maplin is 410nm so wouldnt be quite right. Im after actinic light to suppliment the white light and I think it has a specific wavelength of 420nm. Maplin also seem to be excessivley expensive for LEDs

There are a couple of reasons I cant use tubes. Firstly because of their size. Also, LED lighting gives a desirable "shimmer" effect similar to that of MH. Tubes dont give this effect as they are not a single point light source.

Im realy going for something new here. Once ive finished getting the basics right Id love to addnsome sort of programming so I can dimm and turn of LEDs in sequence to simulate sunrise/sunset and clouds etc. the If the final product is any good I may even think about making a load to sell them.

 

[PaulHolland]

Re: LED array heat issue

polymath you do not want to give up do you ..

Please read the rest of the document:

Insert 1:

The responce is nearly linear and the output can be....

The right hand graph also clearly shows what I was telling you that at high currents saturation effect start. Beginning is somewhat linear and later it start to saturate.

Dave_UK

Have a look at www.farnell.com maybe they are cheaper. Use the information from MAPLIN to find the part number and search for a cheaper source using google ..

Paul.

Maplin has a 420nm LED: **broken link removed**
check datasheet and find other source.