LED Current Limiting thermal protection

Thread Starter

trader007

Joined Feb 27, 2010
249
Hey there everyone.

I have a need to protect some LED's from overheating, and I have some thermal switches that work well but I would like to dim the LED automatically rather then them shutting off completely.

What type of circuit do manufacturers use to do this? I was thinking maybe just a simple thermistor pushing on a FET to drive the current?

The powers supply source is 12VDC running about 1a of lights. Its those string lites so the current limiting resistors are built into the strip.

thanks for any help
 

Kermit2

Joined Feb 5, 2010
4,162
A simple 555 timer could do this. Instead of the variable pot. to adjust duty cycle, you could substitute a fixed resistor and a negative coefficient thermistor. The duty cycle would decrease as the thermistor heated up. Or even have both sides of the pot subbed with identical thermistors to account for ambient temp levels. then the duty cycle would decrease by a given amount per degree temp change between ambient and the temp of the LED strip.

I'm to tired to draw it up right now, but I know lots of folks here could whip one out in no time if you need a circuit to examine. If not I'll try to do it in the 'morrow.
 

#12

Joined Nov 30, 2010
18,224
Just noodling:

Twelve volts at about an amp? You could easily get that out of a cigar lighter socket and an amp worth of LED string would distribute the light rather evenly, but "the manufacturers" don't provide temperature compensation for any LEDs or LED strings (as far as I know). Pin 5 of the LM555 chip is designed for use as a modifier if you did a ramp and a square wave to make a PWM controller. I can't think of a thermistor that would have enough range of variation to do this to a mosfet or a 555, unassisted. A bjt would have a better chance as a thermistor interpreter because 10 millivolts makes a lot of difference to a base-emitter junction. I would try to do the single transistor method just to avoid all the brain work of the PWM method, if it's feasible.

Kermit's idea is valid because the difference between ambient temperature and the active parts is pretty much a constant. The only difference in performance is that the LED string would start up dim on a hot day instead of waiting for the LEDs to heat up before turning down the power. And...it's fairly difficult to measure the temperature of one LED without the measuring device becoming a significant part of the thermal equation.

Edit: Then, there is room to consider that an LED string is already designed well enough to not need thermal protection.

Let's start with specifications. Which temperatures?
 

Attachments

Last edited:

Thread Starter

trader007

Joined Feb 27, 2010
249
Thanks for the input!

I am not sure if it will be feasible though. The LED's in my situation are wrapped around a steel 1/2" pipe, which is why they can overheat. the 555 timer and its components would have to fit inside the pipe, and even if it does there will be a lot of heat in there... all of the components would normally be around 130F degrees or so, because thats just how this gadget is running during tests.

The only real reason I want thermal protection is because if something is left covering the lights they wont have any airflow and they can overheat.. or quite possibly if its 110F outside they will simply overheat from the ambient temperature being too high.

I have already lowered my power input to 11v, which helped the heat output a lot with barely sacrificing any lumen output, so that was nice. But I am starting to think maybe my 50 cent poly switches that easily fit inside the pipe might be the best bet. I just dont like the fact that when the led's hit the thermal cutoff point at 80C, they just shut off completely and wont come back on for at least 15 minutes or so.
 

inwo

Joined Nov 7, 2013
2,419
How about a current limit circuit in parallel with a thermo-disc limit switch?

When switch opens limited current will still flow.
 

Alec_t

Joined Sep 17, 2013
14,280
I don't think having the heat sensor inside the pipe would work optimally. A sudden rise in LED temperature would not be detected soon enough to prevent LED damage/fire, because of the large thermal capacity of the pipe. The sensor would have to be in close thermal contact with the LEDs, but the rest of the control circuit could be remote from the LEDs.
 

#12

Joined Nov 30, 2010
18,224
Commenting on inwo, Thermodisks like Klixon brand usually have a hysteresis of 10 to 20 degrees F.

Commenting on Alec, this idea elaborates on my suggestion (post #4, paragraph 2, last sentence) that trying to measure the temperature of one LED will probably add enough mass to ruin the accuracy of the measurement.

"Wrapped around a steel pipe" is a good way to make them NOT overheat.

PTC thermistors designed as resettable fuses might be available such that they don't just shut off, but rather modulate.
 

ronv

Joined Nov 12, 2008
3,770
If your happy with your thermal switch, why not use Inwo's idea and just put a resistor across the switch so the LEDs go to 1/2 power or so?
 

Thread Starter

trader007

Joined Feb 27, 2010
249
How about a current limit circuit in parallel with a thermo-disc limit switch?

When switch opens limited current will still flow.
i have no idea why i didnt think of this. i wouldnt even need that big of a resistor... probably 1000 ohms / 1w rated resistor would work

WILL TRY it and let you know!
 

burger2227

Joined Feb 3, 2014
194
When LED's get hot, you are in trouble already! I've seen them pop their tops off in milliseconds just giving them too much current. Physically, the top blown off would create enough space to negate the thermal sensor's accuracy.

If it is big enough to have a heat sink then give it a bigger one!
 

Thread Starter

trader007

Joined Feb 27, 2010
249
The diodes don't get too hot under normal use. Its for when someone throws a towel over the light, or the ambient air is unusually hot and the light just cant cool itself quite enough.

Even with the light fully insulated by cloth, it takes about a half hour to build up 65c, and then the poly switch trips. So the heat buildup is very slow. The chances of the light overheating in normal use is about zero, but I never can be too safe.

So I tried the bypass resistor- it works! 150ohms is all i need for a nice low glow. I thought for sure it would be more then 1/2w, but thats the largest resistor I had on hand and its working fine... doesnt even feel warm to the touch.

This is pretty sweet... very simple/durable safety measure that doesnt totally stop working when in default mode!
 

Thread Starter

trader007

Joined Feb 27, 2010
249
I had originally thought that, but I had a hard time trying to figure out which model poly fuse would work..

Then I found these KSD9700 thermal switches that have simple temperature rating and thats it... and they work fine. I mean they're a little pricey at $0.60/piece I think, but thats still affordable.

Of couse I had to buy a set of them with an NO configuration! I swear it was NC when I bought them... whatever, I also was shipped the wrong color lights so I have to wait for a new batch to come in anyway.

Just setup my test light under some clothes, it overheated and went to dim mode!! Looks like it's taking a bit more time to cool off and reset now, but at least this thing won't leave people in total dark. Of course too, my thermal switches are 65c rated, which is too low. I have 80c on order!
 

Thread Starter

trader007

Joined Feb 27, 2010
249
MOTHER$*@()$*@)(!

ok, so someone explain to me how a 65c switch cuts off when the light is just getting warm, but an 80c switch will let the light get so hot it can hurt to touch it.

Now what do I do? go with 75c or 70c? Those are my only two options left... Maybe this time I'll actually learn and just order both! I just hate that it takes 2 weeks to get these things in... but I don't want to buy them from america because they cost 3x the price.

On one hand, I am glad we have ebay and they allow china to sell directly to us, on the other hand I still wish we still just made stuff here in the first place :(
 

Thread Starter

trader007

Joined Feb 27, 2010
249
Ok i am back. I dont like the KDS9700 thermal switches. They are those bi-metal things that just open up when they reach a certain temperature. The reason I dislike them is they seem to build up heat in the switch itself... so if the lights run warm for awhile the switch will eventually trip even if the LED's really aren't overheating.

My specs are as follows-

LED's- Epistar 6350 string, 1 meter (60 diodes). These are the tape strings with the resistors built in for current limiting... so i just give it voltage and it works fine.

Voltage- 11v. I run them under spec for lower heat. In my application, there is very little heat dissipation so I need to run them at 11v.

Working current- about 1 to 1.4 amps... depending on how much i crank up the power supply voltage. 12v makes it about 1.4amps

I want to go with a PTC thermristor, but I don't want to use a 555 circuit or anything like that. I need simplicity (mass producible). If I can find a PTC polyswitch that gives full current to about 80c and then quickly starts raising resistance above that I would be set. It would need to pass about 1.5 amps of current until about 80c and then drop off.

I would REALLY appreciate if someone could point me toward a model range that would suit this. I have been researching a lot and I feel I am just running around wild...
 

Thread Starter

trader007

Joined Feb 27, 2010
249
Thanks... those look better for my application than the chinese ones im using

But they are still switches... i would rather have the LED's dim a little when they get too hot, instead of just shutting off. Limiting the current just a little provides a massive heat reduction, and the purpose of this project is to always have light that is reliable.
 
Top