Teardown and Mod Thread of Courui XML2 Aka "Big Head 3x18650 Side-Switch Thrower"

Yeah I know, cheating, it's not apples-to-apples unless it's on a XML board, but why go only halfway?

Another thing, semi-serious, stay away from copper MCPCBs if you want max amp draws. On aluminum the LED runs hotter and therefore pulls more current. Shouldn't it put out more light if the current is higher? No? :)

Thanks for the nice teardown Linus! This light will still be on my watchlist of lights very interested in.. (Im not pulling the trigger yet though)

Hopefully all info below will clarify some for those who have missed some of the info around XM-L vs XM-L2.

Having 3 cells to divide the current on, I would easily go with XM-L2. Which I even do in the majority of my single cell lights. It would be fairly easy to get 6A peak, or higher depending on cell. And you would also see 5A+ for a good/long amount of time depending on the cells... Having 2 or 3 cells in parallel helps limit sag a lot!

In general you either get higher output with XM-L2, or about similar output at lower current. But I had an example that was even worse (which I mentioned to Ouchy in another thread). So if you get an XM-L with lower than average Vf, and an XM-L2 with higher than average Vf, the XM-L could be brighter, in my extreme example with a partially drained battery it was 3,5A for the XM-L vs 2,2A to the XM-L2 with a drained battery. Its not common with such a large difference though..

As for lumen vs current:

(Direct copper mount, not copper MCPCB, but its probably quite similar)

6A to XM-L gives you around 1480 lumen.

In order to get similar output on XM-L2 you only need. 4,4A. Im not 100% sure what bins was used.

2,6A to XM-L2 gives you 1000lumen. 3A To XM-L gives you 1000 lumen..

And since the XML2's voltage is higher, which gives lower current, that also means the battery voltage stays higher for longer, which helps offset the difference in when the driver drops regulation and current starts to decline with cell voltage. I think it's pretty much a wash.

Boy. Am I ever having fun now.

Yep good sum up of your experiences thanks guys, it’s good to have a spread of information. What sparked my interest in this possible crossover performance situation was when I saw those XML U2-3C NW emitters on sale recently, was wondering if in a single cell light and highly driven it might actually result in a higher total output vs the newer equivalent available XM-L2 T6-3C emitters. Even if it’s only a slight difference or possibility I was curious if it was possible that the older tech could still outperform the newer stuff.
Also I feel safer knowing the current is being spread across 3 bond wires rather than just two but that may also be less relevant in reality.

I guess this is what it comes down to for me, knowing the output bin of an emitter can make all the difference when the differences are so small. Especially when you may have older emitters that are rated at a higher output bin and therefore gain an advantage over the newer stuff with a more efficient baseline.

I may just do this test myself out of curiosity with those particular emitter bins I mentioned above and see what happens. But as you say the variation between individual emitters even within a bin is probably going to make this stuff too muddy to draw any real conclusion from anyway.

-

Ok back to the big headed light! XM-L2s all the way with this one I agree! :slight_smile:
Speaking of which the tint of the one in my light is actually pretty nice, having used it a bit more I’d be pretty confident in saying it’s a 1C tint with a nice amount of green. Makes distant vegetation pop without being too obviously tinted. Surprisingly nice actually, particularly after fixing the focus. I guess the days of getting the horrendously blue XMLs in cheap lights are gone. :slight_smile:

I was a bit confused on the issue as well, but I think I understand a lot better now. Thanks for the explanation!

This flashlight looks amazing btw. :slight_smile:

It’s amazing and amazingly flawed all in one :stuck_out_tongue: I guess we kind of like them like the around here!
Wouldn’t be fun if there wasn’t anything to improve on though, would it… :wink:

The XM-L was T6. Match did the XM-L2 test when the emitter was new on the market. I believe it was T6 as well. Im just not 100% sure.. I assume its the same bin. But like you say, there are some difference within the same bins, and like I recently noticed in this thread (see the: "Why did I use a bunch of time modding this light for probably zero visible gain?" part). You can even risk getting a good difference in Vf on the same emitter. And on top of that, are you really 100% sure that the supplier are 100% trustworthy? There are many small factors can make results invalid..

But in this light, 3 cells in parallel will easily give you you 5-6A+. Even 3x protected NCR18650B should give you 5A+ on a typical XM-L2 based on what I have seen. And when you have an XM-L2 at those currents, XM-L just cant compete....

Are you planning on resistor modding the stock driver, or are you going directly to some DrJones stuff or a reprogrammed driver? Im curious what the stock driver are capable of. 0:)

I’ll be stripping the driver board eventually to go for a linear driver setup, so I’m happy to try and toast it and the emitter by doing some resistor mods first. :slight_smile:

:beer:

So far I have not modded lights but I’m eager to try my hand at it. At the same time I don’t want to mess up any good ones so to speak. This light has really appealed to me and from the sounds of it has lots of potential. Given some of its challenges would this be considered a good canidate for a first time or should I stick with something more mainstream like a C8 or something?

Convoy C8 host would be a good first 'Flashlights For Dummies' project, lol. There are tricks to learn even for a build that uses all off-the-shelf parts, it will serve you well to start with something rather more difficult to screw up. :)

... just ordered one from KD; I'll put in a linear driver.

Thanks for the tear down Linus.

A C8 is a good modding choice for the first time, very little chance of messing anything up and parts are readily available and don’t have to be made to fit.
The OP reflector is also very forgiving in terms of getting a good focus and alignment, this can be one of the more frustrating parts of modding a throwy light. At least it can be for me and ruin the whole experience if it’s difficult! HD2010 I’m looking at you! :stuck_out_tongue:

With this light you may have fun doing basic mods… focusing the emitter, upgraded wiring, emitter swap, maybe a resistor mod etc, but to really get the most out of it and drive it hard it needs a bit more custom work. Adding 7135 drivers will also require some special fitting/heatsinking and considerations regarding the electronic switch.

Have a stab at a C8 first and see how you go, then look at this light once you know some of the basic pitfalls of modding :slight_smile:

Is it possible to change the switch into mechanical switch and use one of those common Nanjg 105C drivers?

Yes you may be able to do something like this, there’s plenty of room in the driver cavity to mount a decent sized reverse clicky from the inside. Then add some kind of an extension onto the side switch plunger that goes all the way down to it. Or replace it with a small rubber boot and plunger. Probably hard to get it watertight but otherwise I don’t see why it wouldn’t be possible.

I’ve also been thinking of adding some kind of a lockout mechanism to the tailcap. I’m thinking along the lines of a simple contact plate or shunt attached to the tailcap that completes the circuit when the tailcap is screwed on tight, but can be locked out when the tailcap is unscrewed a couple of turns.
A momentary switch that can handle high currrent and can be depressed by the tailcap may also work and be neater…

Even easier would be to just replace the filament wire with a pair of bullet connectors on short wires that can be disconnected if the light is stored for longer periods.

There’d even be room in there for a tailcap clicky switch but you’d have to drill through the cap and install a switchboot to make that work.

Can the battery tube be reversed, so the anodized threads go at the front?

Not directly, the threads are the same on both ends but the tube isn’t completely symetrical. So for example the front of the battery tube has a deeper recess than the rear where they mount the spring pcb. Could probably be made to work with some standoffs though.

Also from what I can tell the front of the battery tube body only makes ground contact to the driver through the threads itself. It doesn’t press up against the grounding ring of the driver board directly as is usually the case in other lights of this setup.

It’s probably why I had the flickering problem at the start.
Poor grounding contact only through those solder blobs on the back of the contact board, relies on combined pressure from the press fit and the batteries to make contact. Another area for improvement :stuck_out_tongue:

So I don’t think lockout by unscrewing the battery tube a turn or two is going to be a possibility.


Small update:

Was tinkering with the light and taking some tailcap/emitter current readings tonight.

In stock form I was getting a stable 2.58A at the tailcap.
Emitter measurements went as follows.

Meter connected with long thin crocodile wires clipped to the original tiny emitter cables gave me 2.04A at the emitter.

Ditching the crocodile wires and soldering my thick meter leads directly to the thin emitter cables gave me an initial peak of 2.69A but then dropping to 2.58A.
The fact that reducing the wire resistance increased the current indicates that this isn’t a true constant current buck regulator right?
Or do I have that backwards?

Upgrading the emitter wires from stock to 20Awg raised the tailcap reading to 2.9A, I also managed to get a voltage reading of 3.17v across the emitter in this setup (that doesn’t seem right to me, going by the datasheet I should be seeing closer to 3.3v to get near 3A).

That’s the last measurement I took before I accidentally shorted the emitter positive across the reflector and blew something on the driver, at that point I only had a firefly mode high and low. All modes still worked but at severely reduced output…:frowning:

Looking closely at the driver I noticed the supposed sensing resistor (R200) had a burnt mark in the middle and measuring it’s impedance gave me in the region of 50 Ohm.
Ok so I figured if nothing else had been damaged I could just replace the burnt resistor and might as well do some resistor modding while I’m at it.
So I took some of those smaller R500 resistors I had around and stacked 4 of the buggers to replace the R200. Should give me a total value of 0.125 Ohm.

Tested the light and there was lots of light again, the driver was alive once more :slight_smile:
Maybe a bit brighter but hard to tell so I took some more tailcap measurements to check. These are a bit weird and I’m really not sure what’s going on anymore. I now get values that start at 2.8/2.9 and steadily rise to over 3 at the tailcap, max I measured was about 3.4A.
I guess the increase is somewhat consistent with the resistor modding but the rise has me baffled…before the mod it was very stable at 2.58A…

Anyway, I don’t really know what I’m doing with testing these kinds of drivers and this whole thing has left me more confused and disheartened than enlightened. The driver may not be working 100% after the short anyway so I would definitely take my values with whole lot of salt…

In any case I will soon be tearing the components off this board and getting a linear driver in this light instead.
So much easier to understand and control :slight_smile:
Cheers

Likely not a buck driver, that would be pretty pointless in a light with a max Vin of 4.2v. Try it with 8.4v and a XML, or 12.6v and a MTG2... doubt you'll get anything but fried parts.

'Constant Current' is a different kind of thing from a buck driver that outputs a constant current. 'Linear' and 'CC' are describing the same thing, somewhere along the line some buck drivers got the incorrect CC label attached to them.

Making it lock out at the front like a SRK is possible, in theory at least. Isolate the driver from the head, add a larger OD contact ring onto the driver's ground ring. That way it doesn't matter if the head is connected to BAT- through the threads at all times, as long as the connection between the battery tube and driver is broken it will be locked out.