Mag XM-L Mod

LilKevin715

Enlightened
Joined
May 25, 2010
Messages
712
Location
San Diego, CA USA
Here is a Mag XM-L mod that was pretty straight-forward. The main goal of the project was bang for the buck. I hope this mini-guide will be helpful to others as I myself have learned a lot from others here on CPF; just my way of contributing back to the forums.

Specs:
Mag 2D Gray (Incan version)
Cree XM-L T6 1D tint on 14mm round board
H22A Flattop Bare Heatsink
Shiningbeam 2.8A 3-Mode PWM Linear Regulator Driver (8 X AMC7135)
20 AWG Silver coated Teflon jacket wire
3 x C Accupower 4500mAh NiMH LSD Batteries
52.1mm Hardcoat Acrylic Lens



The first pic shows 20 gauge Teflon wire soldered to the emitter board. I then attached the emitter board to the heatsink with Arctic Alumina thermal epoxy. The emitter wires were then bent into shape for preparation for the driver as seen in the next set of pictures.



The battery wires were then soldered to the driver as well as selecting the desired mode (3-mode) by bridging the correct star to the outer ground connection. The emitter wires were then soldered to the driver. I then attached an anodized Akasa BGA ram heatsink to the LED side of the driver with Fujik thermal glue. The main purpose of the BGA heatsink is to prevent any possible shorting of the driver on the main heatsink.



The stock switch was then modified (cam cut off) and the battery wires from the driver were soldered to the switch. The switch assembly was then placed back inside the battery tube and the snap ring was reinstalled. Ceramique thermal paste was then used on the inside of the battery tube and on the heatsink accordingly.



The heatsink was then pushed into the battery tube and any excess thermal paste cleaned off. Also pictured is the modified stock reflector with the focusing cam cut off. The best focus was obtained when the top of the emitter pedestal base was at the same level as the bottom of the reflector lip.



The use of 3 C cells in a 2-D host requires a battery sleeve (1 inch ID/1.25"OD PVC pipe). The tailcap spring was cut down and the anodizing was removed from the inside of the tailcap with Drano Max Gel.

Beamshots will follow shortly…
 

LilKevin715

Enlightened
Joined
May 25, 2010
Messages
712
Location
San Diego, CA USA
Beamshots have been uploaded. Not the best quality but good enough to get a general idea of the beam profile. Pictures are slightly underexposed to show better detail of the beam. Shots were taken about 0.5m away from the wall.

Low/Medium/High


XM-L Mag vs. SST-50 (4500k) deep SMO Mag, both use same driver.
 
Last edited:

LilKevin715

Enlightened
Joined
May 25, 2010
Messages
712
Location
San Diego, CA USA
Before soldering the wires to the emitter I did check the height of the wire & teflon insulation. If light from the emitter was emitted at a 180 degree angle then yes it would probably affect the beam profile. However the XM-L has a light emitting angle of 125 degrees.

Here is a picture at close to eye level


I did press down on the wire as hard as I could with tweezers while soldering the wires in place. You can see some indentations from the tip of the tweezers on the wire insulation in the pics.
 

StefanFS

Flashlight Enthusiast
Joined
Oct 12, 2006
Messages
1,262
Location
Silicon Road 1, Sweden
That's how I would do it.
You could even add a generous amount of glow powder/epoxy mix around the emitter with no ill effects to the beam. That's a nice effect.
 

Essexman

Flashlight Enthusiast
Joined
Oct 9, 2006
Messages
1,192
Location
U.K.
I'm building a maglite just like that, thanks for the pics and text.

When running on full power what current draw are you seeing at the tailcap?
 

LilKevin715

Enlightened
Joined
May 25, 2010
Messages
712
Location
San Diego, CA USA
When running on full power what current draw are you seeing at the tailcap?
Tailcap current draw measurements are 0.13/0.86/2.79 amps. If I assume 25% optical losses then OTF lumens are in the neighborhood of 42/252/655 (according to Cree data sheet).
 
Last edited:

don2k9

Newly Enlightened
Joined
Jun 6, 2011
Messages
5
This mag mod is so beautifully done. I did my mod on my 4D with a P7 on the same heatsink from H22A two years ago. Powered by 3x18650, the thing is the heat generated from the driver is so out of control it will shut itself down in 15 minutes during the summer days but works fine in the winter:).
I've quit myself from flashlight for nearly 2 years, even lost my old account here in CPF. I decide to try something new on my Mags this time, since there are so many new things out there on the market.
 

Justin Case

Flashlight Enthusiast
Joined
Mar 19, 2008
Messages
3,797
IMO, using 3xC NiMH might not provide enough Vbatt to reach full regulation when using a 2.8A 7135 driver.

At that drive current, an XM-L Vf may be about 3.3V based on the datasheet. Add 0.12V for the 7135 overhead. Thus, you need at least 3.42V to reach/stay in regulation. If the Mag mod has 0.1 ohms of parasitic resistance, that is another 0.28V, for a total of about 3.7V. Unfortunately, based on Silverfox's NiMH shootout data, Accupower C NiMH cells will sag below 1.25V at 3A draw within about 10 min of run time. Most of the Accupower C cell's voltage plateau looks to be at around 1.2V. Out of about 5Ah/2.8A = 1.8hrs battery capacity, only 10 min of it is in full regulation. The rest of it is out of regulation.

If the actual Vf is lower, say 3.2V, then that helps. If the Accupower C NiMH cells can hold a higher voltage (let's say that the plateau is 1.25V at 3A draw), that helps. If your Mag has only 0.05 ohms parasitic resistance, that helps. I wouldn't count on better performance from the Accupowers. But a lower Vf certainly can happen. And you can implement various resistance fixes for the Mag. Then, you need about 3.2V + 0.12V + 0.14V = 3.46V to reach/stay in regulation, while Vbatt could be at least 3.6V.

it all depends on what you have.
 

moderator007

Enlightened
Joined
Jan 1, 2010
Messages
990
Justin Case, I agree with your statement above. Do you know if anyone has actually tested the amc7135 at 2.8A and a XML with 3 D Nimh's in series? It would be interesting to see the results. I know we can mathematically figure if it will stay in regulation or guess on when it falls out of regulation. But I have never seen the actual numbers. And I do realize there would be alot of variables but at least we would have some guide to go by.
 

Justin Case

Flashlight Enthusiast
Joined
Mar 19, 2008
Messages
3,797
I have a triple XP-E powered by 6x7135 (2.1A) and it requires about 3.66V to reach full regulation when powered from my bench supply. There is probably some parasitic resistance in making contact to the center spring and ground trace of the driver. The connecting wires themselves are 18 gauge stranded copper, 1 foot long. I measure about 3.2V for Vf at 2.1A nominal drive. So the overhead is already about 0.46V, which seems like a lot since I'm simply using a bench supply to power the triple drop-in. Of course, I didn't put a lot of effort into optimizing the test setup.

Edit: The XP-E datasheet says that Vf is 3.4V at 700mA drive. So my sample is 0.2V under that.
 
Last edited:

Techjunkie

Enlightened
Joined
Nov 16, 2007
Messages
943
Location
in the brightly lit suburbs of NYC (Long Island)
I have a triple XP-E powered by 6x7135 (2.1A) and it requires about 3.66V to reach full regulation when powered from my bench supply. There is probably some parasitic resistance in making contact to the center spring and ground trace of the driver. The connecting wires themselves are 18 gauge stranded copper, 1 foot long. I measure about 3.2V for Vf at 2.1A nominal drive. So the overhead is already about 0.46V, which seems like a lot since I'm simply using a bench supply to power the triple drop-in. Of course, I didn't put a lot of effort into optimizing the test setup.

Edit: The XP-E datasheet says that Vf is 3.4V at 700mA drive. So my sample is 0.2V under that.

As long as the light is a side-switch light, where Batt+ is interrupted by the switch, like in a Maglite, you can escape the switch resistance to the LED by hardwiring the Batt+ contact at the bottom of the switch directly to the LED+. LED- still has to flow through the regulator to get to ground (Batt-), and can't do that while the regulator is powered-off. See here.



I wonder if a burn in of a XML would help lower its forward voltage Like the sst-90. Could help stay in regulation longer if it worked.

Hey, I know that thread ;)
 

Justin Case

Flashlight Enthusiast
Joined
Mar 19, 2008
Messages
3,797
As long as the light is a side-switch light, where Batt+ is interrupted by the switch, like in a Maglite, you can escape the switch resistance to the LED by hardwiring the Batt+ contact at the bottom of the switch directly to the LED+. LED- still has to flow through the regulator to get to ground (Batt-), and can't do that while the regulator is powered-off. See here.

It's a P60 triple in a SureFire 6P. No side switch. The Maglite has lots of resistance fixes that can be implemented.
 
Last edited:
Top