Measuring Temperature of LEDs

[VanIsleDSM]

So what's the best way to try and approximate the die temperature of LEDs for home hobbyists?

I know the wavelength shift or the Vf drop is a good.. but we need a system that we can all use to get an idea of what's going on.

I was just about to order an IR thermometer, even though I know they aren't that great, at least I could try and get some form of reference to work from.. but I'm not sure if it'll just be a total waste of money?

Please discuss, there must be something we can figure out.

 

[dat2zip]

If the LED is mounted to a good heat sink the LED takes time to heat up you use the Vf of the LED with a constant current source to calculate the temperature of the LED. Since the Vf of most white LEDs has a negative temperature coefficient you can apply a constant current to the LED and measure the Vf over time. The Vf at start and the Vf at the steady state can then give you a fairly accurate temperature based on the LEDs data sheet number for the negative coefficient value.

Say the data sheet shows a -2mV/C change. If you apply a constant current to the LED and it starts off at 3.5V and then levels out 10 minutes later to 3.4V then it has dropped (3.5 - 3.4) / 2mV = 50C or the LED has risen 50C from ambient.

Measuring the true die temperature with a IR temperature meter is not accurate.

Wayne

 

[LukeA]

If the LED is mounted to a good heat sink the LED takes time to heat up you use the Vf of the LED with a constant current source to calculate the temperature of the LED. Since the Vf of most white LEDs has a negative temperature coefficient you can apply a constant current to the LED and measure the Vf over time. The Vf at start and the Vf at the steady state can then give you a fairly accurate temperature based on the LEDs data sheet number for the negative coefficient value.

Say the data sheet shows a -2mV/C change. If you apply a constant current to the LED and it starts off at 3.5V and then levels out 10 minutes later to 3.4V then it has dropped (3.5 - 3.4) / 2mV = 50C or the LED has risen 50C from ambient.

Measuring the true die temperature with a IR temperature meter is not accurate.

Wayne

What about mounting the LED to a thermoelectric junction and measuring the current produced? Or make a setup with a thermocouple probe in it so that each LED is in the same place in relation to the probe.

 

[frenzee]

I have often thought about this question and have not found a good answer. I think LED manufacturers use the current vs. Vf curve to guestimate the junction T, but how do they plot the curve in the first place? I don't know.

In one of his experiments, Newbie used one of those really, really tiny thermocuples and Arctic Silvered it to the base of the slug. This will give you pretty good readings, but to get to the junction T you have to make some assumptions about the Q of several layers of metal, ceramic, thermal glue, etc. Good enough? Probably. Accurate? No, not in my opinion.

I almost odered one of those IR thermometers too, but the problem with those things is that the viewing angle of the sensor is typically 5-10 degrees. I think no matter how close you try to hold the thing, inevitably what you will end up with is an average of the temperature of the die itself plus the temperature of the reflector, the casing, the slug, the heatsink, and the solder mask, not to mention the optical dome.

Another technique Newbie used was thermographic imaging. If you can afford one of these cameras, this will give you the most accurate readings of the die temprature in a production environment. I wish I had the dough to spare for one, but it is definitely outside my meager hobbyist budget.

 

[snarfer]

Interesting point about the thermographic imaging. Do you think a digital SLR modified for IR photography by removing the glass over the sensor could actually be used for this purpose and generate fairly accurate results? Or what about just taking pictures with IR film?

 

[frenzee]

Interesting point about the thermographic imaging. Do you think a digital SLR modified for IR photography by removing the glass over the sensor could actually be used for this purpose and generate fairly accurate results? Or what about just taking pictures with IR film?

No, unfortunately not. A modified digcam or even an IR cam will only record the intensity of the IR radiation. It has no idea what the wevelength - and therefore the temperature - of the object is. I wish it were that simple. A thermographic camera shows the temperature of the object in false colors and then you can compare the color to the reference bargraph and get an idea of what temperature range you are looking at.

 

[VanIsleDSM]

The only thing is you still can't get a die temp with IR imaging, because the die temp won't show through the dome...

Maybe the best way is to pluck the dome off and use an IR temp gun.. I know they come off the rebels easy enough..

but I'd like to find a non destructive way.. I think the Vf drop is probably most viable.. But if it was as easy as a micro controller with something to measure the Vf, plug in your constant current and voltage/temp coefficient and have the thing tell you the temp of the LED... wouldn't there be one on the market?