Bare LED chip white light hot as hell, normal? And some LED questions

Hi guys!

Have 3 "228 lumens LED" bought from DX (yea maybe not the best ones but they work and ill guess they are not really that bad after all) and pulled of the lens to one, which suprised me was made of glass and not plastic, to get as wide beam as possible from them. The second suprise was when i noticed there was this soft silicon thing inside it as i thought maybe was used as a kind of heatsink or so to further transport away the heat because i couldnt really se any other use for it.

I fired the LED up at only 750mA and held a finger some mm from the chip to feel if the chip maybe got hotter or so without it and pretty much burned my finger of after a couple of seconds.


1. The heat i felt is that actually from the chip it self getting this insanely hot even with the heatsink underneath it?

2. Would the chip be totally ok for long-term use even at 1A without the silicone or will it overheat in the length or so?

3. The chip doesnt really have to be protected since it will be in an enclosed reflector but would it be possible to protect it by adding an minute amount of say clear epoxy over it or will there be a risk that it would react with the phosphor so it would be destroyed of maybe even dissolve in it? Water clear silicone is nowhere to be found.

Thanks!

Was the soft silicone thing a small dome? If so, it's part of the LED. Removing it changes the optical properties, and some people report the phosphor oxidizing some time after peeling it off. These LED domes are hard to stick back on properly.

What you are feeling is probably the photons hitting your finger and becoming heat. LEDs are about 1/3 efficient, and you're putting about 2-3 watts into the LED. The heat in an LED forms inside the diode and is spread through the bottom of the LED into the flashlight.

Clear epoxy is likely to degrade gradually, but should work well enough. I don't know the details of your device to suggest whether to leave things be or not.

You probably shouldn't stick your finger in a vapor-deposited reflector because it can damage the surface.

Dam that was a fast response.

It is exactly this one and the silicone is underneath the lens: DX SKU 2394

Aha so it could oxidize had no idea that it could, thought it was 100% inert. Dont want to stick it back on since i need it to emit a wide beam as possible due to the way the reflector and front lens is made to the original bulb in it.

Never thought about that it could have been the photons, really interesting and that made me much more calm.

Im just going to change the lightbulb in the front light to a vintage moped to get rid of the incandescent bulb and make it as a normal Hi-Low beam with heatpipe cooling. The reflector could be made totally airtight if needed with silicone for example but the whole enclosure is pretty much watertight. But if the LEDs would fail after some years i only buy new ones, they doesnt cost much as all and will most likely get down in price even more.

Of course i wouldnt touch the reflector but i need to mount the LEDs inside it to get a proper beam. And for an example so do i make vapor-deposit mirrors at home some times and i know that they should never been touched since it could be instant scratches on them.
[/SIZE]

Yep, that silicone dome serves a few purposes:

1. It extracts more light from the LED. If the photons hit a flat surface (the sheared-off silicone goo) they reflect back inside.
2. Protect the phosphor. Some people here peel the domes off to change the spread of light. Some (of that group) find brownish dimmer spots on their LEDs after many hours at high temperature. YMMV
3. Change output pattern. The dome sends the light out to the sides more, if I understand correctly. I might have that backwards, though - don't spend money based on this one.
4. Be extremely clear. The special optical silicone comes from Dow-Corning, about $50/kg, minimum order is a 200kg drum. It has about 99% transmittance of light, compared to about 90% for anti-reflective glass and maybe 70-85% for clear, non-bubbly epoxy.

I think the Custom forums has occasional posts about putting stuff in place of the optical silicone. I don't think you'll have problems with the LED too soon.

The difference between 750mA and 1A is not that much at all that it justify the extra heat generated so i think i will go for this but it give of more light then the orginal 15W bulb did. And at 750mA the baseplate doesnt really get that hot so mounted directly on a heatpipe and a litte passive heatsink on the outside and i think it will be held quite cool, so i guess it will be ok even without the dome at least heatwise.

And btw its just insane how much light you could get from a small chip that is not bigger then around 1x1 mm..


Thanks a lot for the help in this, interesting!

Henry_A said:

And btw its just insane how much light you could get from a small chip that is not bigger then around 1x1 mm..

Click to expand...

Once I killed an LED by accident. I had soldered one like yours directly to a copper plate, then accidentally put 18V through it from a high-amperage source. The flash was really impressive, but it didn't last very long. You should see what the Customs people here do with overdriven LEDs on massive chunks of metal. I have a pinky-size flashlight that pours out over 500 lumens.

18V trough a single LED?! Poor one, but i guess that would been equivalent to take down the sun for a sec no matter what LED..

500 lumen from a small flashlight that is not even what im going to get from 2 of the leds in the link but 456 lumens is never the less way more then i guess the 15W original bulb would ever give.

Yea i have seen some few ppl here some years ago (havent searched that much) that got ridiculous amount of lumen from small chips and it just insane the amount you could from them, im really suprised every time.
If i should try that my self i would solder it off from the alu baseplate and then directly cool the LED with an compressor system so would it constant be held at say below -40C no matter what. Then i guess you could get some lumens out from even the smallest LEDs and probably not even destroying them in the process. The only thing then (as i would guess) that could break the LED is the bonding wires that would break due to the current and/or the deterioration of the semiconductor material or maybe even breakdown the phosphor perhaps. And i guess there also has to be a certain border where the phosphor cant convert more blue light to white.

Could be fun to test one day but i would guess that someone here already has.

AnAppleSnail said:

Once I killed an LED by accident. I had soldered one like yours directly to a copper plate, then accidentally put 18V through it from a high-amperage source. The flash was really impressive, but it didn't last very long. You should see what the Customs people here do with overdriven LEDs on massive chunks of metal. I have a pinky-size flashlight that pours out over 500 lumens.

Click to expand...

I had such experience. 12V directly to XP-G. but it instantly killed. I think bond wire blow like fuse.

And I over driven Luxeon K2 up to 2.5A. works couple of hours. then it dim when current over 1A, After few weeks, I noticed phosphor coating faded.

I guess its the heat that destroy the coating?

It could have be fun to chill the chip it self by compressor cooling in some way to get it way below freezing if you prevented the oil in the system to get on to it and if the phosphor was inert to the gas that is. By this way the bonding threads would have been chilled as well and it should then be possible to drive it way more then if it only was chilled by the baseplate and then the phosphor would have been chillied as well.

But due to boiling refrigerant on top and around the chip so would the light pattern be all over the place so any lumen measurement would be hard to measure to say the least. But it could maybe be interpolated in some way by the current or so.

Another way would be water instead, once again if the phosphor wouldnt be affected, and a heatexchanger then it would be easier and no boiling liquid but not under freezing since adding any antifreeze to the water could effect the phosphor.

Would have been fun to do that to an 100W led..

Henry_A said:

I guess its the heat that destroy the coating?

It could have be fun to chill the chip it self by compressor cooling in some way to get it way below freezing if you prevented the oil in the system to get on to it and if the phosphor was inert to the gas that is. By this way the bonding threads would have been chilled as well and it should then be possible to drive it way more then if it only was chilled by the baseplate and then the phosphor would have been chillied as well.


But due to boiling refrigerant on top and around the chip so would the light pattern be all over the place so any lumen measurement would be hard to measure to say the least. But it could maybe be interpolated in some way by the current or so.

Another way would be water instead, once again if the phosphor wouldnt be affected, and a heatexchanger then it would be easier and no boiling liquid but not under freezing since adding any antifreeze to the water could effect the phosphor.

Would have been fun to do that to an 100W led..

Click to expand...

why not just dip in into some liq N2?

Henry_A said:

I guess its the heat that destroy the coating?

It could have be fun to chill the chip it self by compressor cooling in some way to get it way below freezing if you prevented the oil in the system to get on to it and if the phosphor was inert to the gas that is. By this way the bonding threads would have been chilled as well and it should then be possible to drive it way more then if it only was chilled by the baseplate and then the phosphor would have been chillied as well.

But due to boiling refrigerant on top and around the chip so would the light pattern be all over the place so any lumen measurement would be hard to measure to say the least. But it could maybe be interpolated in some way by the current or so.

Another way would be water instead, once again if the phosphor wouldnt be affected, and a heatexchanger then it would be easier and no boiling liquid but not under freezing since adding any antifreeze to the water could effect the phosphor.

Would have been fun to do that to an 100W led..

Click to expand...

It destroy coating because I noticed lamp colour tend to blue. & found tiny silver color spots in die.

forced cooling possible with direct soldering LED to copper. but I have no idea about LED die/junction to bottom heat pad thermal resistance.

pavithra_uk: So the phosphor tends to be sensitive then to little of everything, good to know. The juncion resistance would i guess be quite all right but you have to go as low in temp as possible not to fry the chip if going to drive it hard.

FRITZHID: Sure if you have some handy so why not, but ill guess it could be a little costly in the length and i guess it would boil off quite fast as well if you dont isolate it very well.