Mountain Bike Reviews Forum banner

Triple XM-L light review

4K views 19 replies 8 participants last post by  -Archie- 
#1 ·
The lights similar to this one seems to still be attractive nowadays, so I'll share my personal experience with one of them:

3600-Lumen 3T6 LED High Power Bicycle Light For 3*Cree XM-L T6 4-Mode LED bike light Kit - LED Bicycle light wholesale online Worldwide Free Shipping!!!



Long story short, it is good light - but as usually with cheap Chinese unbranded stuffs, it requires a bit of work to become trustworthy device. The level of engineering and the quality of parts are quite good, but it assembled terribly.

One interesting fact is, that LEDs aren't soldered to the single baseplate (as in many multy-LED lights), but are individually mounted on more or less standard 14-mm round bases, with one edge grinded out. The pill have 3 recessed places for each LED base, with very fine polished (almost mirror-grade) surfaces. Mounting pressure is provided not by separate screws, but by reflector of quite complicated shape: each hole is surrounded by protruding circular skirting on the inner side, with cable channels milled between them in the rest of surface. The LED bases are covered with stiff adhesive sheets of teflon-like plastic with square holes for LED dome. That way, back side of reflector provide equal pressure to the LED bases in circular areas around each LED, leaving enough free space for the wires and soldering points. Reflector itself is fixed to the pill by single M3 screw in the center.





Upon inspection, I've found that there's no thermal paste in the light: all the LEDs were glued to the pill by thick layer of silicone-like substance (obviously, it started hardening long before final assembly), which wasn't even covered the contact surfaces fully!



The wires were not aligned to the respective channels in the reflector, and prevented it from applying the correct force to the LEDs. In addition, the reflector's screw wasn't tightened at all.

Inside of pill, I've found several metal particles from drilling, milling and soldering: fortunately, they not managed to cause electrical shortage during first switching of light on. Thermal contact between the pill and outer case was merely absent: no any traces of thermal paste, and relatively narrow matching surfaces.



I've completely disassembled the light, desoldered the LEDs, cleaned all the residues of silicone, and applied good Arctic Silver thermal paste to them. After soldering and aligning the parts, the reflector was mounted back and fully tightened, with Locktite applied to the thread. For thermal conductivity, I've applied layer of viscous thermal paste (Russian-made KPT-8) between pill and the case.

Also, I've replaced weak original power cable with good MagicShine one, and installed rubber boot into cable hole in the outer case (initially, there was a wide gap, compromising the watertightness of light). Internal wires between LEDs and driver are good ones, with silicone insulation, although a bit too long IMHO.

Driver (marked TR-0192) is of unusual design. There's essentially two independent drivers: for one lower LED and for two upper ones connected in series: either of them or both can be switched on, resulting 3-mode arrangement.



The manufacturer marking on some components was unfortunately sanded out at the factory. Initial current consumption of the light was as follow: Low mode takes 0.58 A from battery, medium - 1.70, and high - 1.85.



As I've managed to provide normal heat dissipation path in the light, I've decided to increase its output: after little bit of reverse-engineering, I've located the current-sense resistors (they're R5/0.1 Ohm and R10/0.15 Ohm, respectively) and reduced the values by about 20 %. This allowed me to unleash some of power, increasing current consumption to the 0.82 A, 2.45 A and 2.27A, and light output to 3950, 11120, and 11070, respectively - making the light to be a 20-watt one.



There's interesting issue with difference between mid and high modes: as you see from the table, actual current is less in high mode than in middle: most likely, this caused by some kind of protection limiting the overall current for the light, or by parasitic current flow across the driver's PCB on "unexpectedly-high" modes. I've neither the time nor desire to investigate this further, so let it be. As for the eye perception, the change between mid and max is more interesting: the brightness of beam essentially does not change (as lower current is compensated by more LEDs in action), but flood is increased, and cone of light is more consistent.

One more pretty thing: lhe light is still operational on voltage down to the 4 volt. This is especially good for winter use, when it's powered by LiFePO4 battery with slightly shifted voltage range (if compared to the "standard" lithium-cobalt ones).

Indication of battery discharge doesn't work. There's separate green & red LEDs on the driver, but the latter is never lit: the button illuminated by green LED all the way down to the complete discharge of battery and its disconnection by protective circuit.

At home, the light quickly become hot on the mid/max modes (too hot to touch by bare hand). On open air, during my typical 3-hour night ride, it's just warm: thermal dissipation seems to be good enough. The beam gives quite good impression: low mode (that is, single LED) is comparable to the MagicShine light and totally sufficient for road, while mid and high provide a plenty of light for the forest trails and so on.

Some more details: I've lubricated all the threads with silicone grease, fixed the mount's screw with Locktite, and added a layer of rubber to the mount surface contacting the handlebar.

The last thing: the battery. It was clear in advance (giving the price of whole set) that it's not the best one - and tests confirmed that. Measured capacity is miserable 2508 mAh, so the only use for it is, perhaps, to be a spare battery for MagicShine rear light, or something like that. Obviously, the light is used with one of my high-capacity MagicShine (and compatible) batteries.



As for now, I'm extensively using this light for about one year. It works without any problems, with single exception: the handlebar mount broke after about half-year use (including winter), so I've replaced it with good-quality MagicShine one.
 
See less See more
9
#4 ·
steelhmr, thank you!

Vancbiker, yes, you're absolutely right. Of course I've fixed the inductor: I do that on almost any Chinese light or torch. Not actually glued it, though: I use double-sided adhesive tape between the coil and PCB, and add piece of some elastic material (like pen eraser) between driver board and the aluminium pill it is inserted into.

Another thing probably worth mentioning, is the power cord preparation. To protect it from being pulled out, I use plastic cable tie placed inside the case before assembling. Also, the cable ends should be properly prepared before soldering to the driver: pieces of thin thermal-shrinking tube on the red and black wires will prevent insulation melting and possible shortage of exposed inner wires.
 
#6 ·
I use double-sided adhesive tape between the coil and PCB, and add piece of some elastic material (like pen eraser) between driver board and the aluminium pill it is inserted into.
What do you think about using thermal rubber for holding it in place & providing some extra cooling for inductor and driver components?
https://www.fasttech.com/p/1285500 - like this one

For example Fenix BT20 uses 2 pieces of such rubber on both sides of driver's pcb:
 
#5 ·
Nice write up.
Props on the tech break down and possible improvements.

I have one of this light and it has been great. Used in the cold of winter and then rain of spring. Bounced from a moist basement for a while to bouncing around under the seat of the car ( and the heat of a closed up car) .
Broke it out for the first night ride last week and it still works fine.

One thing on mine: the red battery warning LED does come on.....about 3 minutes before the battery cuts out.

I'm thinking of purchasing the Hi_Max U3
version ( from the other thread) just in case this one dies
 
#7 ·
I really appreciate your effort to review and mod the lights :) But a bit of math reveals something interesting about this light's driver eficiency:
you pumped 30% more current into 3XML (15W vs 20W) and got 13% more light (9800 vs 11070).

Initial current consumption of the light was as follow: Low mode takes 0.58 A from battery, medium - 1.70, and high - 1.85.



As I've managed to provide normal heat dissipation path in the light, I've decided to increase its output: after little bit of reverse-engineering, I've located the current-sense resistors (they're R5/0.1 Ohm and R10/0.15 Ohm, respectively) and reduced the values by about 20 %. This allowed me to unleash some of power, increasing current consumption to the 0.82 A, 2.45 A and 2.27A, and light output to 3950, 11120, and 11070, respectively - making the light to be a 20-watt one.

OT: celebrating post 100 :)
 
#18 ·
I'll take a stab at answering this: This lamp apparently has two circuits. One that operates a single LED and one that operates two at once. When the sensing resistors were tampered with this changed the available current. When both circuits are operating at the same time ( on high ) a parallel circuit is created. Current will always take the path of least resistance. Somehow changing the resistors changed the ratio of resistance to the parallel set-up. When operating in mid-mode all the available current goes only through the two upper resistors which is likely now able to deliver more current ( in mid mode )...Anyway, that's my guess not knowing exactly how these drivers are set-up. Odd that it would do this but what can you say, it's brighter. :p
 
#16 ·
There's interesting issue with difference between mid and high modes: as you see from the table, actual current is less in high mode than in middle: most likely, this caused by some kind of protection limiting the overall current for the light, or by parasitic current flow across the driver's PCB on "unexpectedly-high" modes. I've neither the time nor desire to investigate this further, so let it be.
... :)
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top