LED circuit

[dombanks]

for the electrical engineer guys amongst us. can anyone point me in the right direction for a LED power circuit that will compensate for the effects of ambient temperature on the LED.

 

at the mo im using a MCP1702 and 330R resistor to provide a stable 15.2mA which is great but as the LED heats (its a bivar 405nm 3mmLED) the light output is dropping. The drop is not so bad on its own as its time is much less than my reaction time so i could cope with it if it was in a stable condition but it wont be. just holding the LED assembly with your finger is enough to make it change.

 

i was thinking of replacing the  330R or adding in parallel some sort of NCT R-t themistor that could be situated by the LED which would increase the current through the LED as temp increased to keep a constant luminosity.

 

can any onepoint me i the right direction?

[tightscot]

If I understood any of that I'd have been glad to help..

[bhouse]

A tangential answer perhaps but could the LED be faulty? Is there a change in voltage drop across the LED as it heats up? That could indicate a fault. I've never experienced a power drop with temperature in an LED...

 

 

Assuming it's not a fault, then I think an NTC thermistor in parallel with the 300 ohm resistor should work but would need a fair degree of tinkering.

[Paul Morcom]

Assuming your electrical query is a serious one then I suggest you try posting your question in the "Tech Talk" forum - clever people view that forum and might be able to help.

The clue is in this forum's title (Stuff & Nonsense) for the type of posts and responses you may get.

[pete g]

sounds like faulty led or resistor

[pistonbroke]

Asfair  the LED is a current operated device so why not use a constant current source , plenty of suitable chips available just "do a search"  

[dombanks]

i wasnt sure for either forum tbh and i wasnt really expecting many answers if any but you never know someone might give me an opinion etc that may nudge me into a different thought process. electronics are not my thing so this stuff is usually done for me but hes away so i htought i would see if there was a generic circuit floatign about.

 

no the led is not faulty. all leds will have an output difference due to ambient temps and as the material itself heats up. its not a visible effect. to surmise the light passes through a medium that changes opacity and its this change i am detecting. the light is collected by a photodiode and amplified. so small changes in output of the are a bit of an issue and show up as a big change in the reading.

 

if you read the post above the LED is already powered by a constant current souce. its fixed at 15.2mA. the current source is not the problem its the LED changing its output with temperature which it will do. we are not talking by much either. it is defuinalty not visible by the naked eye

[pistonbroke]

Might be talking b*****k here but if light intensity changes then surely the current through the devise must change , more brighter = more current draw, any temperature change being a result of increased current flow .

 

perhaps your supply cc isn't constant enough 

[Wile E. Coyote]

for the electrical engineer guys amongst us. can anyone point me in the right direction for a LED power circuit that will compensate for the effects of ambient temperature on the LED.

 

at the mo im using a MCP1702 and 330R resistor to provide a stable 15.2mA which is great but as the LED heats (its a bivar 405nm 3mmLED) the light output is dropping. The drop is not so bad on its own as its time is much less than my reaction time so i could cope with it if it was in a stable condition but it wont be. just holding the LED assembly with your finger is enough to make it change.

 

I'm a bit confused Dom.  Is it ambient temperature correction you're looking for, or to offset the self-heating effects of the LED itself?

 

If the former, a NTC should do the trick.  A Zener diode arrangement might also be worth a look, as these can be used for temperature compensating circuits.

If it's the latter, then I suspect that what you'll need to do is look at:

1) Better heatsinking; and/or

2) Setting <100% output as your baseline.

 

The reason I say this is that what you're proposing to do otherwise is to increasingly over-drive the LED (pumping more and more power through it as heats up).  This current will, I rather suspect, further increase the rate of temperature rise... which will result in you wanting to pump even more through it... which in turn will... you get the picture! 

 

Caveat: it's a good few years since I did this sort of stuff seriously, so may be off target with current tech!

[pistonbroke]

A few ideas here ?

 

http://www.google.co.uk/search?q=constant+current+regulator+circuit&rlz=1C2CHFX_en-GBGB543GB545&tbm=isch&tbo=u&source=univ&sa=X&ei=nYcUUuOfBaLU0QXXrYDYDA&ved=0CEcQsAQ&biw=1920&bih=955

[dombanks]

PB, your not wrong but there is a also a heating effect of the LED itself so as i have it a the mo the current into the LED is constant using something similar to the LM components in amny of the diagrams. basically as it heats the semiconductor properties change chaging the amount of emission

 

WC

its a combo of the two i need. i was thnking of mounting the NTC onto the LED housing (led and lens are in an ali cylinder). i found that by chaging the ambient temp around the housing by either heating/cooling was having the effect. a couple of drops of ether on the side of the housing made it go mental! likewise if i blew hot air onto it it wold change. so long as i kept the enviroment constant it was ok as in there was some change but it was acceptable. but if the enviromnet changed i.e. i.e. held a soldering iron next to the fan it would change quickly.

 

so my thought was to introduce a NTC on the side of the housing. in an arrngement with the 330R resistor and linear regulator. the regulator will carry on providing a constant output, the 330R/ntc will provide the base line current and the NTC would alow the total resistance to change with temp thus varying the current supplied into the LED.  thing is thats about as far as i can go and i know its not that simple. i was hoping i could find a simple circuit that would do this i could rig up and test but a couple of days searching is just finding stuff that beyond me.

 

edt i should say better heatsinking will deffo help

[Wile E. Coyote]

Occurs to me that getting a reliable temperature reading (to drive the compensation circuit) might be a problem.

 

Have you looked at some of the torch forums?  I was researching a new torch over the weekend and there seems to be a lot of knowledge around LEDs and thermal effects on "candle power forum".

 

As a tangential idea, if you don't find anything specific to LEDs, it might be worth having a look at temperature stabilisation/compensation techniques used for TCXOs, OCXOs (which involve controlling the temperature around the component in question), or MCXOs.

[Hammy]

I just googled this - based on using a resistor and thermistor in parallel to increase the power supply voltage  as the temperature increases, linky:

 

http://www.google.com/patents/US6693394

 

 

There are a few others out there , but this looks pretty simple. Most look fairly crude as the liight output verses temperature is not that linear, but its a start and may be good enough

 

 

I've just  come up with this too - You could also use an "extra  reference LED" inside a tube aimed at a LDR . A simple feedback circuit  with an op amp could then be used to drive the VDR  to achive a constant voltage across it by varying the supply voltage to all the LEDS's and hence keeping the light output constant

 

      

 

 

 

Dunno if LDR's  have a significant temp co-efficient tho, you colud use other devices with the same idea !

[dombanks]

cool thanks for the suggestions,   hammy that patent is pretty much where we started with athat sort of idea. watty (the guy who does this for me and is called mr Watt   ) was thinking putting the NCT onto the side of the LED housing but concerns we raised for regarding the ability of the NTC to change quick enough.

 

anyway thanks for the ideas its given me somethign to look for.