how hot is hot?

[lc-150]

Hello there!

I have a question about HS. How hot is hot? You know, i bought 5 types of heatsinks for my bike light (xml), the thing is that all of them perform different, some are big enought to not even get warm, some get pretty hot in a couple of minutes. My question will be, what temperature is "safe" for a xml?

Thanks a lot, happy new year for all of you!

 

[jtr1962]

I'm also in the process of making a bike light. I added a thermistor to drop the drive current so the heat sink stays at ~130°F (~55°C) in case there's little or no air flow. A good rule of thumb I use is if the heat sink is too hot to keep your hands on it for at least a few seconds then it's too hot. The idea is to keep the junction temperatures under about 80°C. If the heat sink is no hotter than about 50° to 55°C then junction temperatures should be OK.

 

[SemiMan]

I'm also in the process of making a bike light. I added a thermistor to drop the drive current so the heat sink stays at ~130°F (~55°C) in case there's little or no air flow. A good rule of thumb I use is if the heat sink is too hot to keep your hands on it for at least a few seconds then it's too hot. The idea is to keep the junction temperatures under about 80°C. If the heat sink is no hotter than about 50° to 55°C then junction temperatures should be OK.

That strikes me as overly conservative for a bicycle light. Realistically you may use a bicycle light thousands of hours over its whole life. With most modern LEDs, even at 125C junction temp, there will be little degradation over that period of time. There is certainly a benefit to keeping the LEDs cooler to maximize output, but for a bicycle, form factor is important, so I will take the unnoticeable 5 or 10% decrease in output to keep the form factor small which is generally more important.

Even for a commercial light, this is for the most part overly conservative. Sure your LEDs will be still have over 70% output at 100K hours, but you may have a tough time selling them if they are too expensive.

LC-150. For the XM-L, (or any other LED), it is all about the junction temperature and drive current. If you are driving it really hard, then you need to keep it cooler as it will degrade quicker at high drive currents and high temperatures. The junction temp will also be correspondingly w.r.t. the heatsink temperature. At lower drive currents, you can run them quite hot and they will still last a very long time. The data sheet and application notes on heat management are quite informative.

 

[Steve K]

The ultimate goal is to avoid killing the LED (or degrading it) before its intended lifetime. For my dynamo light, which runs constantly, I'd like it to stay 50C or less in order to get a longer lifetime. It has an advantage in that it only operates while the bike is moving, so it has airflow over it.

For a LED conversion of a battery powered bike light that I did a few years back, it has a smaller heatsink, and it does get pretty warm when stationary. It doesn't get used much, and the Lux III LED is pretty obsolete now, so the heatsinking has proven to be sufficient.

I'm currently doing a LED conversion of a Cateye Micro II bike light, and I'm roughly doubling the size of the heatsink relative to the light I mentioned above. The design easily permits this larger heatsink, so it's not a sacrifice.

You might want to find a way to measure the temperature close to the LED and see how hot the LED is really getting, and then make a decision on the heatsink based on that. I use a LM35 temperature sensor, which is just a TO-92 semiconductor device. The TO-92 version is about $5 now, but the SOIC version is under $2 (at Digikey).

I agree with JTR's rule of thumb (and fingers) that if you can touch the heatsink at the LED for 10 seconds or so, then it's probably at 50C or less, and should be okay (assuming that you have a good heat path from the LED).

regards,
Steve K.

 

[eebowler]

A good rule of thumb I use is if the heat sink is too hot to keep your hands on it for at least a few seconds then it's too hot.

I practice this as well. If space permits you to use a large heatsink compared to a small one, use the large IMHO.

 

[Illum]

Anything that would result in a junction temperature exceeding 145C is too hot [LED limit is approx 150C]. Anything that would result in a junction temperature below 15C is too cold. [Below this the forward voltage of the LED will soar, an XRE may require up to 6V as opposed to the conventional 3.3v at 350ma]

 

[Steve K]

..... Anything that would result in a junction temperature below 15C is too cold. [Below this the forward voltage of the LED will soar, an XRE may require up to 6V as opposed to the conventional 3.3v at 350ma]

Can you provide a source for this lower temperature limit?

The XR-E datasheet that I have lists a maximum junction temperature of 150C, but no minimum. For the "Temperature coefficient of voltage - white, royal blue, blue, green", it lists it as -4mV/deg C typical, with no indication of any change while still in the range of acceptable junction temperatures. Is there some other app note from Cree that the warning about low temperatures appears in?

regards,
Steve K.

 

[hotdrop]

IT really depends on the LED itself. The luxeon K2 for example actually gets 110% efficient at -15*C JT but in the deep blue spectrum it is really flat. Id attach the spec sheet but I dont have upload priveleges so you will have to find them yourself.