Problems with M@g mods

[Firecop]

Hey guys, I was hoping someone could shed some light on what happened today to one of my M@g mods. I built a 3xCree Q2 direct drive off 9AA Eneloops for my partner, which he has been using for a year with no problems. The Eneloops were new, and had been charged on my MAHA C-808M 2 days before the incident.

We were in the basement bedroom where a fire had just occurred. The atmosphere was hot (~110F), but our meters said that we should be okay without breathing apparatus. We had been digging for about 10-15 minutes when, in the space of about 5 seconds, his light dimmed dramatically then quit. When he touched the head of the light, it was so hot he almost burned himself!

I have not had any issues like this in the past while direct driving, and my meter measures ~1-1.2 amps when used in place of the tailcap (between the wall and the battery holder's - terminal)

Just now, I pulled everything off and found that 2 of the 3 emitters were dead. Both emitters close to the + and - coming from the switch didnt' work. Here's a closeup:

Please ignore the shoddy workmanship; this was my first attempt at a M@g mod. The blue RTV is there to help hold the optics on the emitter.

What do you guys think? How did this happen, and how might I prevent this in the future?

Edit: I just checked the batteries. In their cheap 3AA-1D adapters, all 9 Eneloops ran a brand new 3xQ5 DD M@g mod at .8A and all the cells, individually, had 1.44v.

 

[Illum]

the heatsink heated to the point where solder begins to melt, once solder melts on both sides it bypasses the LED and allow unrestricted current flow between the terminals, which then yields a thermal runaway that throughly cooks your batteries:shakehead

the black wire looks okay but I would suggest potting some arctic alumina thermal epoxy under it, it will act as a thermal conductor and an electric insulator. The green, presumed positive, looks like it was pulled out by wire tension when the solder melted. Since your project was a multi-die light, one CREE may have been bypassed because of this and the reduction in load cooked the other CREE's.

Note that such failures don't usually happen instantaneously as solder could melt by increments until a drop hits the sink.

Advice on your next mod::thumbsup:

• don't use the stuff in hot glue guns, they are insulators in both thermally and electrically, meaning it will isolate your CREE from the heatsink to effectively reduce its ability to heatsink.
• Use arctic Alumina on the bottom of CREE to insure that there is no static air/gap between the CREE and the heatsink
• Use an anodized sink, the ano will reduce the sinks ability to act as a conductor.
• Depend on current consumption, proper sizing of wire gauges could mean better performance due to reduced linear resistance
• If your planning to use this for extended periods of time [the duration of the battery's capacity] consider investing in higher temperature solder

 

[jag808808]

It sounds like a short. Check your switch and LEDs to locate the short. Good luck!

Aloha,

jag

 

[datiLED]

That emitter has seen some serious trauma. The silicone under the dome is disturbed. Was it dropped?

Also, the black wire looks like it is soldered onto the ring. That could be bad news, too.

 

[Firecop]

Thanks for everyone's help! I should have made it clear that each emitter was Arctic Silver epoxied to an aluminum disc, which was in turn Arctic Alumina epoxied to the inside of the mag lip. The wires are 22 ga. solid-I thought they'd handle this wattage...

Where's a diode on the emitter?

Jag: I did a quick check of each emitter by touching each pad with wires hooked up to a 3AAA stack. Only one lit; would that indicate a short (and if so, where?), or a dead Cree?

DatiLED: of course it's been dropped, we're firemen! Just be thankful it hasn't seen use as an impact weapon during an arrest oo:

 

[Illum]

well, an LED is a diode
sorry if that confused you

 

[LuxLuthor]

It's dead, Jim.

 

[allburger]

What may have helped would have been to electrically insulate the heatsink from electrical flow. You can do this a number of ways whether it is anodizing it, or simply putting some rustoleum paint on it. I know the heatsink could get hot enough to make the paint release some gas, but there must be some way to insulate that.

 

[TigerhawkT3]

the heatsink heated to the point where solder begins to melt, once solder melts on both sides it bypasses the LED and allow unrestricted current flow between the terminals, which then yields a thermal runaway that throughly cooks your batteries:shakehead

the black wire looks okay but I would suggest potting some arctic alumina thermal epoxy under it, it will act as a thermal conductor and an electric insulator. The green, presumed positive, looks like it was pulled out by wire tension when the solder melted. Since your project was a multi-die light, one CREE may have been bypassed because of this and the reduction in load cooked the other CREE's.

Note that such failures don't usually happen instantaneously as solder could melt by increments until a drop hits the sink.


Advice on your next mod::thumbsup:

• don't use the stuff in hot glue guns, they are insulators in both thermally and electrically, meaning it will isolate your CREE from the heatsink to effectively reduce its ability to heatsink.
• Use arctic Alumina on the bottom of CREE to insure that there is no static air/gap between the CREE and the heatsink
• Use an anodized sink, the ano will reduce the sinks ability to act as a conductor.
• Depend on current consumption, proper sizing of wire gauges could mean better performance due to reduced linear resistance
• If your planning to use this for extended periods of time [the duration of the battery's capacity] consider investing in higher temperature solder

That shouldn't be happening in an LED build unless something else, such as a short, was already there to heat the heatsink enough to melt solder. The emitters would be dead long before the solder got close to its melting point.

 

[jag808808]

Check each LED with a voltmeter for resistance and continuity. The drop may have been just enough to jar something within the LED itself from the pics.

Thanks for everyone's help! I should have made it clear that each emitter was Arctic Silver epoxied to an aluminum disc, which was in turn Arctic Alumina epoxied to the inside of the mag lip. The wires are 22 ga. solid-I thought they'd handle this wattage...

Where's a diode on the emitter?

Jag: I did a quick check of each emitter by touching each pad with wires hooked up to a 3AAA stack. Only one lit; would that indicate a short (and if so, where?), or a dead Cree?

DatiLED: of course it's been dropped, we're firemen! Just be thankful it hasn't seen use as an impact weapon during an arrest oo:

 

[Mr_Light]

I make all of my Tri-LED Mags running 3D cells wired parallel. I'm wondering if this is safer for isolating failed LEDS since I know each individual LED can handle 3D cells by themselves?

 

[Illum]

I built a 3xCree Q2 direct drive off 9AA Eneloops

Direct drive...I'm assuming with a resistor involved?

That shouldn't be happening in an LED build unless something else, such as a short, was already there to heat the heatsink enough to melt solder. The emitters would be dead long before the solder got close to its melting point.

well, from experience werking with 5mm LEDs in a linear circuit...the LED will die at a certain temperature but it still allows current to pass through it like a resistor, which bears the possibility of a remaining short. :shakehead But yeah, I would agree that there has to be other factors...like an existing resistor.

I am convinced that a solder melting temperature was attained because of the green wire in the pic. notice there is solder on the CREE's contact pad? the green wire should look the same as the black if thats how he soldered it

FYI, My project was making a 16 LED cluster in series [18V@30ma] and when a few of the [DX] LEDs begin to burn out, hey...the string still works and all the LEDs down the line is still getting 30ma.

when an LED dies, it goes from a Light emitting diode to just a diode...and both types have some sort of voltage drop over it...Vdrop = heat.

 

[LukeA]

For 3D/3s NiMH, I like 3 separate AMC7135 drivers each running a single LED.

Also, Eneloops fit 3s4p in a 3D Mag.

 

[kenster]

Howdy! :wave:

I shouldn`t be posting but I am!

First, I enjoy goofing with flashlights and I have modified more lights than I can count but this is just a hobby to me so take everything I say with a grain of salt, please!

Where do I start??? It is obvious the LED in the picture was smokin hot! I have Cree`s with air under the dome but this was caused by flashlights being dropped by friends lets say "DROPPED HARD". I have not overheated a Cree to the point of melting solder but I can see this as a possible cause of your emitters air under the dome.
Oh, many questions like from reading this.......

"Edit: I just checked the batteries. In their cheap 3AA-1D adapters, all 9 Eneloops........"

..... I assume this means the LED`s wired in series running DD 9 eneloops 3s3p. Then I see this.......

".......... all 9 Eneloops ran a brand new 3xQ5 DD M@g mod at .8A........... "

......then this...........

"I have not had any issues like this in the past while direct driving, and my meter measures ~1-1.2 amps when used in place of the tailcap."

....... Well, just a tailcap reading which isn`t perfect but .8 amps, 1 amp or 1.2 amps??? Total forward voltage??? Use a resistor??? One thing I have found is the forward voltage of an LED drops after some use so DD at the same voltage means a little higher drive current down the road. The drive current level makes a big difference in the heat produced in a multi LED modified Mag and 3 Cree`s driven at around 1 amp creates a lot of heat. Especially for long runs and in a room with a temperature of 110 degrees. YIKES! From the picture I don`t recognize the heatsink. I do see what looks like more than just a thin layer of epoxy between it and the head. This could be OK as long as the other end of the heatsink has a sufficient amount of tighter metal to metal fit with the head and/or body of the flashlight to keep things cool. Even though AA Epoxy is thermally conductive having the thinnest layer possible is better so a heatsink that is a loose fit is not too great. CPF member modamag has made several different tri/quad LED heatsinks for mags. The first versions had very little metal and very little metal to metal contact with the head which caused heat issues when folks drove the LED`s at high current levels and/or long runs so he improved his heatsinks by making them larger with more contact surface to the body of the light for a better thermal relief path. You mentioned ya`ll were digging which is hard to do while holding a light so was it set down and left on??? Hand versus air to keep a flashlight cool especially if the air is 110 degrees, HAND! Not the same if wearing a glove which insulates your hand from absorbing the heat. Anyway, just my worthless .02 on how I would begin to work through what might have caused your problem. Others mentioned a short which may be but did a short cause the heat or did the heat cause the solder to melt and then a short??? :shrug:

OK, OK I`ll stop rambling my speculating worthless .02 of nonsense! Best of luck solving the mystery.

Ken :tinfoil:

 

[TigerhawkT3]

...
well, from experience werking with 5mm LEDs in a linear circuit...the LED will die at a certain temperature but it still allows current to pass through it like a resistor, which bears the possibility of a remaining short. :shakehead But yeah, I would agree that there has to be other factors...like an existing resistor.

I am convinced that a solder melting temperature was attained because of the green wire in the pic. notice there is solder on the CREE's contact pad? the green wire should look the same as the black if thats how he soldered it

FYI, My project was making a 16 LED cluster in series [18V@30ma] and when a few of the [DX] LEDs begin to burn out, hey...the string still works and all the LEDs down the line is still getting 30ma.

when an LED dies, it goes from a Light emitting diode to just a diode...and both types have some sort of voltage drop over it...Vdrop = heat.

Huh. I was under the impression that when an emitter goes , it generally goes OC and makes the whole string lose power. I wonder which failure mode (OC or dead short) is more common, and by how much.

Firecop, could you check for continuity between the solder points and the heatsink?

 

[Illum]

Huh. I was under the impression that when an emitter goes , it generally goes OC and makes the whole string lose power. I wonder which failure mode (OC or dead short) is more common, and by how much.

Firecop, could you check for continuity between the solder points and the heatsink?

well, :shrug:
5mm LEDs is one thing, but I haven't had any crees short out [and I hope none ever do] to know whether the end result is absolute, or just relative to lower current loads

 

[Firecop]

I think I see a problem...the sleeving on the wire was always off the solder point like that; the new heat didn't do it. It's hard to see in the photo, but the bare wire is actually raised off the 'sink ~1mm, and doesn't move very easily (thus shorting out to the 'sink). Please remember that this was very early in my flashaholism-my handiwork has improved greatly since this sorry specimen, which is why I gave it to my partner

Illum: there is no resistor involved. These are pure direct drive from 3s3p Eneloops.

Kenster: the batteries that were in the defunct light were switched into a different light. I measured .8 amp draw with those particular batteries. When I swapped new Eneloops into the new light, I got a 1.2 amp draw with 3xCree Q5 in series.

I use "digging" as a term for slowly unearthing the debris in our area of fire origin. We normally hold the light with one hand and scape off the debris with the other. I've run several of these mods for extended periods of time with little noticable heat build-up. You wouldn't recognize the 'sink because cut it myself out of 1/8" aluminum plate.

Tiger: there's no continuity between any of my solder points and the 'sink.

I just tested the new light now by running it continuously for 30 minutes at 79F ambient, then testing the amp draw to 1A. The head did get quite warm, but not outrageously so, and it stabilized after about 10 minutes.

Upon further examination of the fried LED (I was *really* tired last night after a 12 hour shift), it seems like the two "cracks" in the photo are actually the wires of the emitter that have come off the yellow coating under the lens!

Thanks again for all the thoughts!