crash-testing a MT-G2 on a copper Noctigon, graph done, mod done and repaired :-)

Thanks for coming through for us! You da man! (And I'm sure you girlfriend really was understanding of the need to satisfy flashaholic modsters with test data!)

-Garry

Thank you djozz. Your work is much appreciated. :slight_smile:

Excellent work, djozz! I'm glad the LED survived to live another day. :)

Real nice test! Again, great to know. My girlfriend understands but the wife doesn't...

Nope, not with an actual buck driver (i.e.: not a LD-29 or the like). As long as the input voltage is high enough (with some headroom) above the LED's forward voltage at the desired current, the driver should do the rated current no matter if it's a single 3.3v LED or 3 3.3v in series or a single 6v LED. The buck driver in the S1100 does 3.3A to a single XML and 3.3A to a single MTG2. IOS '5.5-12v 9A' driver does 9A to the MTG2 (I hope it's obvious that I haven't tested that one with a single XML, lol). The IOS '3-18v buck/boost'... see below

Wait, are you asking about that one, or stating what that combo actually does? Is that the '3-18v' one, that claims 5A to a single LED or 3.5A to multiples in series? If it is, the 3.5-3.6A you're seeing is correct... the 6v MTG2 appears to the driver as two 3v LEDs in series. That 5A number is for only a single 3.3v LED. I don't know why that driver does it that way when other non-boost buck-only drivers don't care how many LEDs there are/what the total forward voltage is, but that's what it is advertised to do, looks like it is doing what it says on the tin.

Whoohah… That 2nd video was a bit nerve wracking! :stuck_out_tongue:
I was almost holding my breath for the poor MT-G2… WOW 16A… You sure kept going far beyond the point where max output was reached! J)
Great stuff!
Thanks for torturing testing it. :slight_smile:

Very nice torture :slight_smile:
now, dedome it :bigsmile:

Running the IOS driver you mentioned will run a single XM-L U2 at 5 amps with a couple off 18650.
Running the same driver with an MTG-2 with 2 x 26650 will give you 2.88 amps. Exactly the same setup with three 26650 will give 3.6 amps at the led. Dont ask me why.
I think your measurements for the 9 driver were taken at the tailcap.

Not try to take this thread off topic but to answer questions as they arise here is the results for anyone that has not seen them.

No, my 9A number came from a meter inline with the LED. Same setup did 6.6A at the tailcap with 3x18650 (that's a lot - a lot of current, a lot of abuse for the cells & switch, a lot of wasted heat in the driver, and not a whole lot more light than with a 4A driver).

MTG2 running from only 2 cells is not enough voltage overhead, these drivers really need 3 or more cells in series to work properly with a Vf of 6.3-6.6v (depending on current).

I did the graph, and it comes with the following notes:

-I noted the data of all three runs, and it all appears to reproduce very well. I thought that at least the third run would show some damage (lower output, altered Vf) from the second run mayhem, but looking at the numbers now, that appears not to be the case, the led keeps performing equally well. The graph shows only the data of the second run.

-I can compare the measured Vf up to 3 amps to the official Cree data, and I measure 0.2 V higher, which is significant. For the higher currents there is no available data, also Match did not publish Vf data of the MT-G2. I don't know why the results differ from Cree (is it real, did I measure wrong, is my -not really trash-quality Amprobe- DMM off by 0.2V, is there a big influence from the Noctigon board?). Well, it is what I measure.

-At the end of the test run the though sizable heatsink was just too hot to touch, so temperature may have an influence on the higher current readings (but then, in your flashlight build that will only be worse, so the data just are more to the point as they are now )

Here we go:

no dedoming for me, dthrckt already proved that dedoming can be very unsuccesful. But the way I am going to use this led will make it look like it is dedomed, but does not throw a bit (*mysterious smile*). I'll get back to that when it is finished.

Thanks for the graph djozz. With all this work you have done how and what would you use for maximum realistic performance in a largish flashlight, driver and pill wise?

I am still not very good in driver electronics (only a general idea of what is going on), so no ideas about which one to use, but I do have idea's of how I think the MT-G2 could be driven well.

I will give a general idea of how I ideally use this kind of data to make my idea of a good useful flashlight, and I guess that is not so far from how some others in this forum think. (My game is usually not about setting records while being totally inefficient and burning through batteries in 5 minutes, that's another fun sport).

Using the MT-G2 most factory lights go up to 5 amps, even the ones that use copper boards (Eagletac). That is a good idea, great output and good for efficiency and gives a managable amount of heat. But in this forum we want a bit more and we are not afraid to get our hands warmed up. Maximum output is reached at 11 amps, but the light will be hugely inefficient at that current, a real waste of energy and probably not good for the lifetime of the emitter. My personal rule of thumb with copper mounted leds is to aim at about 3/4 of maximum output: the led is used very well, the visual difference with max is negligable and it almost halves the current neccesary for the maximum output (the performance graphs of all other emitters have a similar profile so this is valid for all emitters). For the MT-G2 on copper and maximum heatsinked 3/4 of max is reached between 6 and 7 amps, so that would be my ideal. In a flashlight build heatsinking will be worse, but I'd still go for the 6/7 amps . The led alone will produce -apart from emitted light- about 30W of heat at that current, so the build needs a good thermal path and sufficient surface area on the outside of the light.

One remark about why it is great to use these copper boards as opposed to traditional boards in flashlights: because the heat can get away from the led so easily, the temperature difference between led and body is less (I believe it is even much less), so the body can be allowed to get hotter while the led is still performing well. This is advantageous in two ways that help each other: 1) obviously you are allowed to make more heat (the led can be driven harder), and 2) the difference between flashlight body temperature and ambient temperature will be larger so a same amount of energy can be dissipated with less body surface. In practice: I have not tested it yet, but my guess is that when a copper board is well mounted with a good heat path to the body, you can continuously keep running a flashlight that is too hot to touch. One design consequence is that the battery compartment will have to be thermally more isolated from the head of the light than usual (i.e. further away).

Sorry, this is a bit of a long answer, and it is perhaps not even an answer to your question .

Thanks for the graph!

May I suggest that once this/these threads get less popular. Start a new thread and just put all your emitter charts in it.
That should be a sticky! If you ever do future tests, everything could go into that once place. Would be nice to only check one post (and thread) for all your recent tests.
Your latest graphs look really nice btw! Great work! :slight_smile:

Thanks for the suggestion, a sticky might be a good idea and it might not be; the problem with graphs is that they tend to lead their own lives. In my opinion these graphs should be consulted together with the information on how the numbers are obtained because -at least in my measurements- I have lots of uncertainties about the measuring equipment used that I want the user of the data to understand. The raw graphs suggest an absoluteness that is just not there.

So I tend to think that if a separate stickied thread is made with performance data on these leds it should contain data obtained with calibrated meters and a proper integrating sphere, not this 'I have no idea how accurate the numbers are but you get the general picture' stuff.

(for that matter, I find my own thread back easily by searching 'djozz mt-g2' , and match's sinkpad graphs with 'match sinkpad' , I find that even easier than looking for stickies).

Please don’t apologise for answering my question. You have put in so much effort with this led I was curious what your thoughts were to sum up. My next question as I value your opinion would be what driver would you use if it is made.

At 7 ampere, the Vf of the MT-G2 is 7V, or at least close to 7V. It would have been great to have 7135 type lineair chips that could handle the 8.4V of two li-ions in series, in that case my no brainer ideal driver would be a lineair driver with enough chips to deliver 6 or 7A, running off 2S-2P li-ion batteries.

What is available is one of the available 2/3 leds-in-series buck drivers. Unfortunately they are big and comfychair found that for high currents they need 3 cells in series. But as said, for actual suggestions for drivers the MT-G2 driver threads gives more useful information than I can.

(slightly off topic: I actually think Eagletac has made the almost ultimate MT-G2 flashlight (SX25L3), the only weakness IMO is that they use 3 instead of 4 18650 batteries, I expect that it can only maintain that 2375 OTF lumens of the turbo-mode for a short period when the batteries are freshly loaded)

djozz, your numbers look great. I think they are perfect for relative output numbers. This is perfect for knowing how much is too much for the emitter. That part you have nailed right on.
Very few of us can measure absolute output of an emitter or light.
The equipment to do that truly proper is way above my pay grade. :beer:

djozz,
Perhaps using the ‘9V’ MT-G2 should somewhat alleviate the issue of achieving ideal overdrive current in certain cell configurations. It takes less current to drive it as hard as the ‘6V’ (e.g. Noctigon, 4S Li-ion, 3-18V IOS buck-boost driver: 9V vs. 6V). Ergo, driving the 9V at 5A is equivalent to driving the 6V one at 7.5A.

Thanks for the sweet graphs and your time invested in them.

The 6v part works pretty nice in direct drive off 2 cells, does around 5A. The 9v part would be interesting with 3 cells, either DD or a driver that does DD on high mode only.