XM-L dedomed by accident (really...)

I just love to tinker my Ultra-fire 502B that I recently purchased. You can see my review here.
A few hours ago I soldered a wire inside the springs to reduce the resistance path… Following this I got about 2.7A tail cap current. My DMM is adding about 100mOhms so I expect the real current with the tail switch to be around 3A. Am I right?

Anyway, I was about to put everything back together, I tested one last time that the aluminium foil inside the head was correctly placed by turning the light on high for about 4 minutes. After that time the head was at about 50°C and the reflector (I removed the front glass) was burning hot (so 60-70°C). At that point I stopped, thinking that everything looked good.
Before putting the front glass back on, I wanted to clean the reflector and LED dome with a tissue. I did that right after switching the light off after its 4min on.

To my surprise the dome fell off!

This is the best picture I could get. Most of the dome came off but there is still a thin layer on top of the emitter. Thank god the wire bonds are intact as well as the phosphor.

I have two questions:

1- Why did that happen? Is it normal for the dome to become soft when the LED runs for a few minutes? Or is there a problem somewhere in the thermal path causing the LED to be way too hot? I have a thermocouple but I can’t manage to get the temp of the die.

2- What should I do now? Leave it like that? There is some remains of the dome on the emitter, rough surface. A bit like the first picture of this thread but less.

One last note: If I kill my flashlight, I won’t be getting an other one any time soon… An it’s my only one…

Thanks for your input on the subject.

I just successfully placed my thermocouple on the silver pad in the corner of the LED.
It got to 140°C in 10 seconds. !!!

You can just leave it as it is. Try running it and check out how the beam looks now. Maybe you will like it more? I know I normally do on smaller lights. But then I just love a light that can throw :-)

I didn’t notice a warmer white. But I did notice a smaller spot.
I have to solve that temp issue though!

Apologies for the stupid question, but have you checked if there’s any thermal paste under the LED star?

I didn’t remove the LED as I don’t have any thermal adhesive (only CPU thermal paste).
But as you can see on this picture there is clearly some white glue under the LED.

I just ran some tests with the pill out to trouble shoot the problem.
I measured the temperature after 20 seconds at 3 different points and used thermal paste to get a good contact with my thermocouple.
Ambient: 20°C
On the pill: 55°C
On the aluminium board (“star”): 91°C
On the side of the LED die: 143°C


One of the techniques for dedoming is to get the LED really hot then slide the dome off. When it’s hot it’s much easier to remove. This can be done by running the light at max power until its hot, or by pushing the dome off immediately after reflowing.

In the future I suggest not touching the dome right after the light has been on for awhile. Wait for it to cool before cleaning.

Hi lagman. The dedoming is fine and will probably give you a nice beam profile in your 502B. The thermal paste on the other hand looks bad. It looks too thick and don’t seem to be under the entire emitter base. Thermal paste/adhesive is a relatively poor conductor of heat, but much better than air. It is only supposed to be thick enough to fill in the micro gaps. You would be better off with two flat/true surfaces and no thermal paste compared to what you were provided in you light. It appears that the emitter and pill are not touching other due to a fairly thick dollop of thermal adhesive under the center part of the emitter base.

A hot emitter will tend to appear cooler in tint. Fix that thermal transfer issue and you should see the a slightly warmer tint and slightly better CRI than the original led had.

You seem have a good head on your shoulders about this type of stuff. Best wishes on you modding.

Is it actually glue or just white thermal paste?

Anyway, it will definitely help to replace the white junk under the star with your CPU thermal paste in the manner Ima4wheelr suggested (super thin layer applied under the star).

I’ve had an emitter spin on the star (solder underneath was hot enough for it to dislodge when the bezel was loosened). It is best to avoid messing with the reflector/bezel/anything internal before the light has had ample time to cool.

3A is actually fairly hot for a P60. Having a MCPCB with a direct thermal path will help big time for P60 lights because of the relatively poor thermal path in these lights, even with foil wrapping. Also many P60 pills do not have a level pill top, so the thermal xfer from the MCPCB to the pill is poor in addition to the poor thermal issues there already. There are a couple of BLF threads on this topic with methods to fix it.

Besides these issues, not giving it a cool down period can really put the pressure on that poor over-heated LED. I tightened a bezel once on a direct drive (EAST-092 style) C8 with the power on, and the light went poof - the LED actually twisted on the MCPCB because it's solder softened up so much.

I've had LED domes pop off on me, as has others - pretty much all in relatively high amp lights with poor cooling.

These guys are on the money. Think about your thermal path. From the outside in (cool to hot):

  • Aluminum body
  • Aluminum foil
  • Aluminum Reflector
  • Brass pill (not great thermal conductivity)
  • Thick (and low quality) thermal adhesive (terrible thermal conductivity)
  • Aluminum MCPCB base
  • dielectric substrate (terrible thermal conductivity)
  • LED soldered to copper trace

The interface from each material from the next is non-ideal, there are air gaps, inadequate pressure, etc. Furthermore some items are just bad conductors of heat!

The most critical thing is to reduce the thermal resistance right at the LED. That should be your priority, then work your way out. A direct thermal path MCPCB such as those from Noctigon or SinkPAD is a great first step, followed by a thin layer of a higher quality thermal interface material (thermal grease or thermal adhesive) instead of that thick layer of unknown adhesive.

i used to dedome mine running the led hot so you can pull of the dome. have done it that way on most of my XML (2)’s. you might wanna clean the old thermal past and apply your own. make sure there isnt to much, this can be a problem. a small drop and press the board down with some pressure. and 3A on a Alu board for a P60 is a lot. you might wanna look for a copper board?

I thank you all for your advices.
I realize now that a good light with a big piece of metal in the head and less thermal interfaces is much better than a P60 drop in.
The white paste is a glue. I haven’t been successful yet to remove the mcpcb because of that, even with a screwdriver.

Here is what I’ll do for now:
*Remove the mcpcb and the white glue.
*Put a thin layer of thermal paste the same way I would do on my CPU.
*Reduce the current to below 2A by adding a serial resistor.

Yes I know, it’s a shame efficiency wise, but I want that light to keep working as I don’t have any other one. At some point I’ll remove that resistor and put a proper 4*7135 driver in place of the current one.

I was a fool thinking I could drive a P60 to 3A. Thanks for bringing me back to reality. I’m glad I learned that lesson without much damage to the light.

Well the emitter is only about 4$

and drivers are cheap

Is this a single mode light ?

Personally I don't think I would have been trying to do any resistance mods on a p-60 light in the first place . And secondly I wouldn't be terribly concerned with running it just as it is .

If it's a single mode I'd drop in a 2$ driver and turn it down if it gets too hot .

I'm not sure what too hot on an led actually is ....?? Unless you're experiencing what TomE is talking about .

My guess is if you only have one light , you probably don't have a battery capable of melting solder .

Ok, everything is back together. I soldered a 0.35ohms resistor in series to lower the current to 2A.
I had to use a lot of force to remove the star because of that white glue. I placed a small amount of silver thermal paste and put everything back together.
By using my technique of placing a thermocouple on the little square in the corner of the LED die the temp rises to about 90°C very rapidly and then stabilizes at about 100°C.
Sorry, I didn’t take any picture of the inside but here is a picture of the beam now that it’s de-domed.

It’s not about the cost of emitters/driver. It’s about the time it takes to come to me (usually one month).
I could buy it locally but then the cost becomes an issue.
The light was a 3 modes but I shorted the driver so now it’s one mode.

You are right about the part that I shouldn’t have tried to mess with the light…
“too hot” means that after I reduced the resistance path and got 3A to the LED, it would start smoking after 20 seconds. That’s probably why the dome fell off. You don’t need a good battery to do that, just a bad thermal management. :slight_smile:

Good work, that was a big improvement. With decent thermal management 3A in a P60 should be fine. You might want modes though ;-).

Please do post your temperature results if you happen to move over to a direct-thermal-path MCPCB.

What kind of thermometer / thermal probe are you using?

The cheapest K-Type thermocouple I could find on ebay. It’s accurate enough for me and goes up to 400°C, which is what I wanted the most. I’m quite happy with it! Here is a review
To really appreciate the effect of changing the thermal paste I should have kept the same conditions as my previous measurements (3A), but I did everything at the same time so…

Ah, I didn’t notice that you’d done the temperature measurements after adding the series resistor. Still, a lower die temp is a good thing regardless of how it’s achieved.

Thanks for the info on the thermocouple setup. I only have an IR thermometer right now, so I cannot accurately measure the temperature of a small hot object. This may be the trick!