LED Dome Modification

[Flashlight Aficionado]

Someone had an LED that the dome fell off. It had better throw because of the smaller apparent area. (the dome wasn't there to magnify it)

The problem is the die dies without the dome protecting it. naughty: pun intended) Not sure why it dies. Can a flat piece of glass be put in the dome's place to keep the die alive?

I hope to achieve two objectives. One, make better throwers. Two allow the use of multi-die LEDs to be substituted for single-die LEDs. This should produce more light but with no loss in throw/focus.

So, can it be done?

 

[LukeA]

You will experience a bunch of loss, especially if you don't use AR-coated glass.

 

[saabluster]

The problem is the die dies without the dome protecting it. naughty: pun intended) Not sure why it dies.

Do you have any evidence that the die dies? There is this rumor going around that it does but I have yet to see any evidence. Quite the contrary my experience has been that it does just fine without it and that after the initial lumens loss there are no problems.

 

[Flashlight Aficionado]

LukeA - I am more concerned with keeping the LED alive. phosphor and other coatings are another matter.

saabluster - I just read it here on CPF. I assumed it was true, because no one argued with that statement. Here is a Link to that post.

 

[saabluster]

saabluster - I just read it here on CPF. I assumed it was true, because no one argued with that statement. Here is a Link to that post.

I had not seen that post but I did post an objection in a thread here which you were a part of. There is zero evidence, that I have seen, that moisture kills the dies in anything other than reflow soldering.

I hope to achieve two objectives. One, make better throwers. Two allow the use of multi-die LEDs to be substituted for single-die LEDs. This should produce more light but with no loss in throw/focus.

I had this same thought myself and I plan to test it out when I get some more MC-Es in.

 

[2xTrinity]

...allow the use of multi-die LEDs to be substituted for single-die LEDs. This should produce more light but with no loss in throw/focus.

You will experience a bunch of loss, especially if you don't use AR-coated glass.

I read this not as no loss compared to leaving the dome on, but no loss underdriving an MC-E with the dome scraped off, compared to a conventional XR-E with the dome intact, running higher current. Removing the dome will result in less lumens making it out, part of the functionality to having that dome there is to aid in extraction of light from the emitter that might otherwise be lost inside the packaging, through the use of index matching the dome to the LED directly with the silicone gel. Removing the dome will certainly worsen efficiency, but in a reflectored setup, it might still slightly enhance throw.

 

[Gryloc]

IIRC, the only thing that you should notice if you remove the dome from the LED (assuming you are talking about the XR-E) is a slight reduction of brightness (different beam angle, but a true reduction in total lumens?), and a slight color shift (usually to the warmer side). Back in the day, when the XR-E was released, this was a practice done to make the XR-E focus better with the lambertian Luxeon reflectors and optics. At first, the XR-E had such a strange beam pattern, that some resorted to this practice to get rid of ugly rings and get the beam to focus. Nowadays, there is an amazing offering of XR-E reflectors and optics out there.

I have ran a P4 XR-E for lengthy period of time in a Maglite at over 1.5A with no problem. Actually, not only did I remove the glass dome, but I also removed the metal ring surrounding the die. It took careful prying and preparation so you leave an intact soft silicone blob around the die and fragile bond wires. I did it to improve it, optically, for my particular task (focus good in a large Mag reflector). I was not able to measure the before and after lumen amounts. The tint did shift slightly warmer as well (a plus in my eyes).

I have done the same with a white 0100 Rebel (it uses no bond wires and the phosphor stays intact if you are careful and precise with your tools). This was done by accident after mounting the emitter when I hit it with the reflector's bezel. I am using the small McR-10R reflector (a 10mm dia reflector designed for the Rebel) in this light. After I lost the silicone dome, the beam actually got tighter and brighter for me (the apparent die image shrunk considerably). I purposely did the same to another Rebel 0090 that I used as a test platform for reflectors. It appears to be healthy still.

I am curious to see if you can remove the dome on a MC-E without harming the many bond wires. The apparent image of the dies on the MC-E are slightly smaller than the XR-E, but I hope that removing the dome will shrink it further. See post #92 for my experimental findings:
http://www.candlepowerforums.com/vb/showthread.php?t=204439&page=4

My MC-Es have not shipped yet, but there is a chance I may remove the dome on one of mine in the name of science to see how it affects the reflected beam (hopefully it makes it smaller and more intense). According to cmacclel, the MC-E is not a nice emitter to focus tightly. There is a noticeable donut that is difficult to minimize. I hope changing the apparent die size will help the matter.

So, there is no proof either way how the LED will die without the dome. If the emitter uses a soft silicone dome, then the dome may aid in removing heat from the die (which glass and plastic may not help). This is something I heard of, but I cannot be certain on. The light transmittance is affected, however, and that may reduce the total output. Actually, I never understood why the XR-E would not increase in total lumens without the die and retaining ring, since that ring does block light up to a certain angle. That light may not be totally wasted, but it is not focused by the main dome/optic as well. Hmmm... :thinking:

-Tony

 

[Flashlight Aficionado]

I had not seen that post but I did post an objection in a thread here which you were a part of. There is zero evidence, that I have seen, that moisture kills the dies in anything other than reflow soldering.

I did read that. I blame my selective bad memory. Thanks, that was the thread that sparked my curiosity!

I had this same thought myself and I plan to test it out when I get some more MC-Es in.

:twothumbs I can't wait. I forgot it was you who was getting the broken MC-Es. I am glad we are having the same crazy ideas.

I have done the same with a white 0100 Rebel (it uses no bond wires and the phosphor stays intact if you are careful and precise with your tools). After I lost the silicone dome, the beam actually got tighter and brighter for me (the apparent die image shrunk considerably). I purposely did the same to another Rebel 0090 that I used as a test platform for reflectors. It appears to be healthy still.

Excellent! Brighter and brighter? I assume it leveled off? Did the the LED that you purposely removed the dome get brighter too? Side Note: Let us all use the term Die Image. It is the perfect description. My apologies if I am behind the curve and it is already the accepted term.

saabluster - can you check the lumens or lux in your tests? In comparison to a unmodified LED. If you do, put the lux meter point blank at the undomed die and the domed die.

Between saabluster and Gryloc we can make huge improvements in LEDs!

 

[saabluster]

saabluster - can you check the lumens or lux in your tests? In comparison to a unmodified LED. If you do, put the lux meter point blank at the undomed die and the domed die.

I did do a rough test to see what the lumens loss was after removing the dome and it was around a 30% reduction in output. I would have to do the test again to be sure though.

 

[Flashlight Aficionado]

That's interesting. You measured a 30% decrease in lumen output, but Gryloc observed an increase. I wonder if it is a roll of the die (intentional again) whether it increases or decreases.

 

[Gryloc]

Flashlight Aficionado,

Hold on there. I guess that I did not say that right. When I said that the light got tighter and brighter, that was with a tiny reflector (where die image really counts). The beam diameter actually was cut nearly in half, and as a result the light was more concentrated (more intense). I did not mean that total light output increased (I mentioned that I have no way of testing that). I am sorry for the confusion.

I did make a small light box to strictly make comparisons of lumen output, but it was made way after I knocked both domes off, and took off the dome on the XR-E. With a light meter (alone) directly in front of the emitter, I can see how the flux decreased because of a widening of the radiation pattern. Saabluster, can you confirm that test with the setup in some sort of small white room, milk carton, or white-lined box? If I have time tonight, I may sacrifice a spare XR-E or XR-C LED I got from DX with an unknown flux, and test the differences in light output in my make-shift light box. Again, sorry for the unclear wording...

-Tony

 

[Flashlight Aficionado]

Gryloc - No problem. It made no sense to me, unless there was runaway heat issues. Thanks for clarifying that the hotspot got brighter, but the spill got dimmer. But the overall output was less. Now that makes sense.

 

[Gryloc]

Oh, shoot. I tested a few LEDs days ago, but then forgot to post my findings.

I had two LED emitter to test. One is a Cree XR-E, and the other is a Cree XR-C. I do not know what the flux bins were on either LED, because both were from DX (removed from the flashlight during previous mods), but that will not matter. I do not have an integrating shpere, but I do have a white paper lined box, which gives me great comparative measurements between LEDs. This is what I use to compare different flashlights and LED emitters at various currents.

For my test setup and process, I secured an emitter to an old PIII/P4 heatsink (no fan necessary), then applied 350mA to the LED via a CC/CV labratory power supply. I set the heatsink against my custom light box (positioned the same way every time thanks to a little jig I created), let the LED warm up for 2 minutes at 350mA, then took my lux readings. I would take a reading, turn the PS off, for a few seconds, then turn it back on for a few seconds to take anther reading. I did this three times for every test, although this redundancy did not show any irregularities a single time (but I did it to be safe).

I repeated this process for each test. I made three tests per LED emitter.
-First, I measured the lux from the light box with the original, unharmed LED emitter.
-Next, I carefully pried off the dome by taking the tip of a new standard Exacto knife blade and placing it under the edge of the dome. I carefully pried at the dome and worked my way around. With the dome removed, I checked to make sure that the emitter was still well seated on the heatsink. I placed the modified LED against the lightbox and took my second lux reading.
-Finally, I pried the ring off of the Cree emitter substrate, careful not to destroy the soft silicone material around the die and the bond wires. I took the same Exacto blade and scored the jelly-like silicone around the die (slid it around the inside perimeter of the metal ring). Then I used side-snipping wire cutters to pry the ring off of the emitter substrate. I was less patient here than I used to be. It is better to cut a notch on the ring with a Dremel cutting disk, and use a flat-head screwdriver to pry it off (lower mortality rate this way). I just moved, so I did not want to dig out my Dremel. With the ring removed, I again checked to make sure that the emitter was still well seated on the heatsink. I set the LED back against the lightbox for the last reading. I repeated the three steps for each LED.


XR-C:
Inital lux reading:
109 lux
Removed dome optic:
74 lux (-28.2% from initial)
Removed dome support ring:
55 lux (-49.5% from initial)

XR-E:
Inital lux reading:
142 lux
Removed dome optic:
10 lux (-25.4% from initial)
Removed dome support ring:
98 lux (-30.9% from initial)

Again, these lux readings are not exactly like "lumens". I never had a chance to calibrate my box to find a coefficient to match it with a lumen value.

So, you can see that the XR-C suffered more from the modification than the XR-E. I am concerned that I damaged the gummy silicone on the XR-C when I removed the ring, so I could have disturbed the phosphor layer. I am very confident with the results of the XR-E, however. The important part of this experiment (and the topic of this thread) is the light reduction after removing the dome, so removing the ring is not as relevent. The die size could have been a contributing factor to all of this as well.


So, someday, I will test this on a Lumileds Rebel. I am searching around for a cheap Earth globe (which uses two cardboard half-spheres) so I can paint the inside the proper glossy white color, then install a better baffle. Then, with this setup, I will then find the coefficient to convert my lux values to lumen values.

Cheers,
-Tony


NOTE: Something odd is going on with the date and time. I think this was why the CPF was temporarily shut back down. It sure is confusing. Oh, I predict that someone with the name of peterthomson will post about his 2D Maglite with Q5 in a apheric lens setup in a post tomorrow (the 13th) at some time between 4pm and 4:30pm....

 

[Flashlight Aficionado]

I'm using a Q5 aspheric 2D Mag with Kai buck/boost converter. I've removed the dome and am able to focus the beam even tighter than before, not sure about lumens as I don't have a light meter, but it is running fine without any adverse affects...

Noticed any difference at all? Perceived loss of lumens, tint change anything?

How happy are you with the new focus? What is covering the die? Silicone or is it bare to the elements?

 

[saabluster]

Great work Gryloc. It would be nice to have this test done in an integrating sphere but this data is better than nothing. Good to see our results were so close to one another.

Removed dome support ring:
98 lux (-30.9% from initial)

I said I had a reduction of 30%. What I did not say was that it was without the support ring. That means we are within .9%. I like that. Thanks for your hard work.

 

[Flashlight Aficionado]

Secondary question. Can we paint on phosphor? This way we can either repair LEDs that were damaged when the dome was removed and also possibly change the tint.

 

[Hammer Train]

I'm using a Q5 aspheric 2D Mag with Kai buck/boost converter. I've removed the dome and am able to focus the beam even tighter than before, not sure about lumens as I don't have a light meter, but it is running fine without any adverse affects...

 

[Calina]

NOTE: Something odd is going on with the date and time. I think this was why the CPF was temporarily shut back down. It sure is confusing. Oh, I predict that someone with the name of peterthomson will post about his 2D Maglite with Q5 in a apheric lens setup in a post tomorrow (the 13th) at some time between 4pm and 4:30pm....

Woo ouh! Halloween's magic is already in the air.
Your prediction is (sorry! was) amazingly accurate.

Would you please PM me next week lottery results? :devil: