How to solve high power LED array heat issue?

[Dave_UK]

Hi,

I am new to electronics and need some help with a project I am building so any advice would be greatly appreciated.

I have built an array of 48 high power LED's (3.5v, 125ma) running on a 12v power supply. The array was built using stripboard and is housed in an aluminium instrument case. After about 30 seconds of power the array starts to get hot and the board starts to smoke.

I am planning to add a couple of fans to the case and I assume some sort of heatsink is in order but I dont know where to start with this.


Many thanks in advance.

Dave.

 

[IanP]

410 nm led array

Stripboards are not very good as far as real copper resistance is concerned ..
And you have 68x0.125mA of currents flowing around this board ..
I would start with soldering plated copper wire of certain crossection - for example 1mm², or more - thus increasing condactivity of existing tracks ..
Then you can think about adding fans ..

Regards,
IanP

 

[Dave_UK]

led array 10000k

Thanks for the reply.

I will eventually be using a PCB for the array so Im not too worried about the limitations of the stripboard. I am more concerned about how to keep the LEDs running at a reasonable temperature.

Any suggestions about which type of heatsink to use and how I could mount it?

Cheers

Dave.

 

[PaulHolland]

cooling an led array

Hi, The current through the PCB tracks is killing you .. 48 * 0.125 = 6 Amps. This needs tracks that have a minimum of 5 mm with copper !. You also put a lot of heat in the resistors: ( 12 - 3.5 ) * 0.125 * 48 = 51 Watt (1 Watt per resistor) Do you use 1 Watt resistors ???????... Do not think you do !.

Paul.

 

[Dave_UK]

heating issue led

The LEDs are connected in groups of 3 in series so the current is 2 Amps. I am using 0.5 Watt resistors. Would using higher rated resistors lower the temperature?

 

[polymath]

led tubes

Hi Dave_UK

I agree with IanP - avoid stripboard if you can or solder Bus strips on it.

Have you considered PWM driving the LEDs?
The idea being that you turn the LEDs ON for a short very high current period (heating) then OFF for a relativly long time (cooling). eg. ON= 1ms OFF=9ms
The ON current being far greater than the continous current rating for the LEDs.

This method usually makes for a more power efficient design.
The repetition rate and Make/Space Ratio (on/off ratio) are such that the human eye perceives a continuous brightness from the LEDs.

hope this helps ... Polymath

 

[xxargs]

led array cooling

The LEDs are connected in groups of 3 in series so the current is 2 Amps. I am using 0.5 Watt resistors. Would using higher rated resistors lower the temperature?

Is exactly 12 Volt or 'car' 12-volt (ie between 11.5 Volt - 14.5 Volt)?

if you designs using exactly 12 volt => 12 - 10.5 (3x diodes) = 1.5 Volt and make resistors for .125 mA on 1.5 Volt = 12 Ohm (0.18 Watt heating) to feed 3 diodes each chain, each diode give 0.438 Watt heat. and totaly (48 * 0.438) + (16 * 0.18 (16 resistors)) = 23.8 Watt for matrix (if very tigth mounting you need fan to coling this)


If using above matrix on cars 12- volt (~14.00 volt), give 14 - 10.5 (3xLED) = 3.5 Volt over 12 Ohm resistor and give 0.292 A current -> going hot (1.021 Watt) and all LED in chain overloaded 2.33 times (0.292 A) current and make heat of 1.02 Watt each LED.

Total power in this case is (1.02 * 48 ) + (16 * 1.02) = 65.28 Watt and matrice possible begin to smoke - For only for 2 Volt higher input voltage, power going near 3 times higher...

Other parameter is LED:s 3.5 Volt in cold state going lower in heat state, and current incrase, more power dissepated, heating incrase, voltage going down, more power dissepated... and current and power run away.


In reallity, Voltage incrase over LED diode with higher current, so current incrase not so much as decribe above - but power consumtion still incrase very quickly with few volt higher value compared to designed voltage if using simple low valued serial resistor as current limit of LED-chain.

If you have varying voltage, is possible better to using current source to make controlled 125 mA (possible LM317 or simular transistor coupling) compare to using serial resistors each 3 LED chain. Or using externa regulators to be sure this matrix feeds exactly 12 volt and current limit (2 amp) to protect from heating runaway.

 

[Ave_Rapina]

taiwan led manufacturer 5mm white

For an led array you need to use some kind of pwm on the alimentation to reduce the power.
Has you know 100hz is very fast to our eye feels the blink.
You can use in the alimentation and pulse modulation to reduce the power.
You need too to see the response time of the led and the down ramp to calculate the best duty cycle to use in the circuit.
You see that the dissipate power will reduce.

 

[Dave_UK]

420 nm led

Thanks for the replies.

I am using a regulated 12v(variable) power supply and I have measured the voltage to be 11.99v.

What is a PWM? The light output from the LEDs is critical for this application.. would using a pwm reduce the intensity?

Thanks

Dave.

 

[Ave_Rapina]

420nm led

You can have almost the maximum bright with an modulation with a frequenci that your eye can´t notest
If you do that you reduce a lot the current consumed.
I wil send you an propose of schematic In this moment I m a little busy

 

[PaulHolland]

maplan heat pdf

Hi Dave,

Do not let you fool yourself by people telling you that PWM will give more light and uses less power. It is NOT. The only think that PWM is giving you is the ability to control brightness, like a light dimmer. But then one that is very efficient for LED's.

In general, LEDs have a nonlinear I-V (current/voltage) behavior and thus current limitation is required to control two important issues, power dissipation and light output. Since both dissipation and output are current-dependent, the ideal source for driving an LED is a constant current source.

regards,

Paul.

REF: **broken link removed**

 

[Ave_Rapina]

led actinic blue 410nm

Hi Dave,

Do not let you fool yourself by people telling you that PWM will give more light and uses less power. It is NOT. The only think that PWM is giving you is the ability to control brightness, like a light dimmer. But then one that is very efficient for LED's.

In general, LEDs have a nonlinear I-V (current/voltage) behavior and thus current limitation is required to control two important issues, power dissipation and light output. Since both dissipation and output are current-dependent, the ideal source for driving an LED is a constant current source.

regards,

Paul.

REF: h**p://www.maxim-ic.com/appnotes.cfm/appnote_number/3668

You have all reason but he can regulate the bright for example at night to be less intensive and day more intensive.
That was the sugestion that I want to give.
But of course an constant courrent you can pull maximum bright from the leds.
The curve of the led is similar to an exponencial without control .
If you intend to to give a little more bright you need to pull more amps from the power

Figure 4 shows the LED current pulsed at 100Hz in response to a squarewave control waveform ranging from 0V to approximately 3.9V. For such low-frequency PWM dimming, the typical efficiency is higher than for the linear-current approach shown in Figure 2.

The site that you give explain all vantages disvantages and is the more indicated.

That depends the nature of the problem.
Regards

 

[PaulHolland]

Re: LED array heat issue

The efficiency of the complete system is higher since you do not have the losses through the resistor that provides current limiting to the diode (LED) but efficiency in terms of more light for less current is not true.

The circuit from Maxim is nothing more than a step-down regulator with PWM to control output voltage and current to the LED. This circuit provides a constant current through the LED's and NOT a pulsed one.

Paul.

 

[Dave_UK]

Re: LED array heat issue

ok... Im confused now. Im a newbie to electronics so this is all a little over my head. lol

I am happy with the light output I am currently getting from the LEDs and I am looking for a solution to keep them cool enough.

If I used a PCB instead of stripboard and mounted the front of the leds in a sheet of aluminium then added a case fan would this be sufficient?

Would adding higher rated resistors reduce their heat output?

 

[polymath]

Re: LED array heat issue

Calculations based on Kingbright super bright 5mm RED LED WP7113SRSGW - a common Red LED.
h**p://www.us.kingbright.com/product.asp?catalog%5Fname=LED&category%5Fname=Through+Hole+LED%
2F5%2E0mm%2F35+degree+Viewing+Angle&product%5Fid=WP7113SRSGW

Assumptions: 12v psu, perfect loss-free driver.

LED Spec:
Max DC current 30mA  Vf = 1.8v+((I-10/20)*0.1v)  Pdmax = 100mW

Pulsed Imax = 155mA 1/10 duty cycle 100uS pulse

Luminous Intensity is linear with current. Normalised = 1.0 @ 20mA

*******************************************

LED with 30mA

Luminous Intensity = 30/20 = 1.5

Resistor dissipation = (12v - (1.8v+(((I-10)/20)*0.1v)))*30mA = (12v-1v9)*30mA = 10v1*30mA = 303mW 

LED dissipation = 1.8v+(((I-10)/20)*0.1v) * I = 1.9V*30mA = 57mW

Total Dissipation = 303mW + 57mW = 360mW

*******************************************

LED with 155mA,  100uS pulse, 1/10 Duty Cycle

Luminous Intensity = 155/20 = 7.75

Continuous Resistor Dissipation = 12v-1.8v-(((I-10)/20)*0.1v) * I
                               = (12v-1.8v-(((155-10)/20)*0.1v)*155mA
                               = (10.2v-(72.5*0.1v))*155mA
                               = (10.2v-0.725v)*155mA = 9.475*155mW
                               = 1470mW

Resistor Dissipation (1/10) = 1470/10 = 147mW

Continous LED Dissipation = 1.8v+(((I-10)/20)*0.1v) * I = (1.8v+0.725v)*155mA
                          = 392mW

LED Dissipation (1/10) = 392mW/10 = 39.2mW

Total Dissipation = 147mW + 39.2mW
                 = 186.2mW

*******************************************

Summary:              Continous (MaxI)   PWM (MaxI 100us 1/10)       PWM Performance

Luminous Intensity       1.5                  7.75                    416% more

LED Dissipation          57mW                 39.2mW                   31% less

Res Dissipation         303mW                 147mW                    51% less 

Total Dissipation       360mW                 186.2mW                  48% less

So by using PWM, a pulsed current of 5*Idcmax gives 416% more light using 48% less total power.

Your chioce? ... Polymath

 

[IanP]

Re: LED array heat issue

155mA @ 1/10 duty cycle => 15mA on average and .. bla bla bla .. boolshit mate ..

12V supply, packs of 3 LEDs and 12Ω/0.5W resitors are perfectly fine ..
Just mount everything on a properly designed PCB (as much copper left as possible - copper works as a heatsink as well) then add a fan and that would be it ..

Regards,
IanP

 

[polymath]

Re: LED array heat issue

155mA @ 1/10 duty cycle => 15mA on average and .. bla bla bla .. boolshit mate ..

12V supply, packs of 3 LEDs and 12Ω/0.5W resitors are perfectly fine ..
Just mount everything on a properly designed PCB (as much copper left as possible - copper works as a heatsink as well) then add a fan and that would be it ..

Regards,
IanP

It is sad that you revert to insult if you are incapable of posting a valid rational engineering response.

LED forward drop at 155mA is 2v525 == 392mW - CONTINOUS
but LED is only ON for 1/10 of time. Disp = 39.2mW

Likewise resistor Disp = 147mW

To my knowledge LEDs and Resistors only disipate power when they are passing current ie.ON.

If you can reasonably and rationally correct my maths - please do.

Yes - the average LED current is 15.5mA but the human eye does not respond to light in a linear or average fashion. Perhaps you did not know this.

regards ... Polymath

 

[Dave_UK]

Re: LED array heat issue

oh dear... it looks like Ive started an argument.

If it helps.. the array will be used to provide light to photosynthetic organisms... so I need a constat source of light as Im not sure how they will respond to a flashing light no matter how fast.

I have checked on the retailers website and the LEDs have a maximum operating temp of 85 degrees. Do you think that this will be acheivable without a seperate heatsink?

The reason I ask is because other high power led's such as luxeons come with an integrated heatsink.

 

[PaulHolland]

Re: LED array heat issue

Hi, Please check the therninal resistance in K/W or C/W ( junction to free air) this enables you to calculate the maximum temperature. One question ?. Why not using TL lights or much better Xenon lights used in modern cars they generate a very large about of IR. Also used in the old days is carbon wire lamps (used normally for chickken egs). IR is heat radiation so any heat source would do in your case.

Paul.

 

[polymath]

Re: LED array heat issue

Hi Dave_UK

I was unaware that you needed a constant output light source - I did not see it in the thread.
Apologies to all for digressing the thread - unknowingly.

w.r.t. photosynthetic organisms - I know of Plant Breeders using standard flourescent light fittings as an artificial light source. Their light output does fluctuate at 100 Hz - by how much I am not sure.

Perhaps you could investigate - once you have your current design working.

regards ... Polymath