I have been practicing adding 7135s to an IS 3.04A driver, and I’ve now added 2 more 380 mA 7135s, so emitter current should theoretically be ~3800mA or 3.8 amps, so I’ve been testing the driver with an XM-L2 emitter.
I have an emitter that I bought from FT awhile ago that should be an XM-L2 T6:
The emitter is mounted on a copper plate from a computer heat sink.
I am using a 30V/5 amp bench power supply to test the driver, and I noticed that I can get to ~3.78 amps (both at the driver + contact (display from power supply) and at the emitter leads (measured using a clamp meter), but only by increasing voltage to 4.4V (in general, I’m finding the driver + current to be about the same as the emitter current in my setup).
So, I was checking against Match’s testing:
and, looking at the Vf vs. current, it looks like I should be able to get to 3800 mA at 3.6V.
Is this difference that I’m seeing (Match: 3800 mA at 3.6V vs. my setup: 3780 mA at 4.4V) due to only resistance (leads, etc.)?
EDIT: I forgot to say that what I’m really worried about is that if my test results are ok, there would be no way to get to the 3800 mA with a single Li-ion battery (since it wouldn’t go to 4.4V).
Try using a second multimeter to measure Vf across the LED/MCPCB to make sure you aren't losing significant electrical potential across other components.
According to the XM-L2 datasheet Match's results are correct. Even if you linearly extrapolate the Current vs Voltage graph conservatively, 4.4V should get you at least 5 amps.
What do you get with the PS direct, without the driver?
That way you can see what the led will do without the driver in the way and see if your numbers aren't closer. The driver is going to have some affect and there will be the rule of diminishing returns when adding chips to it. A single 18650 probably will not do it. That's why I used to use 4 NiMHs when I added chips to a 7135 driver, to keep the voltage up over 4.5v
380mA chips are almost always over-rated, they won't do a true 380mA each. They're all different - in fact all the 7135s are churned out of the same line, and then they're checked and the better ones over a certain break point get the '380mA' label (and a higher price tag) and the rest get the '350mA' label (binning - this is the same as how computer CPUs can have the same physical die structure & features but some are set to run at 3400mHz and some at only 2800mHz, same goes for mulit-core CPUs, the ones with defects get some of the cores disabled and turn into 2- or 3-core CPUs even though they all came from the same batch of silicon and started as 4-cores).
Whatever current that particular driver outputs with that specific set of 7135s is what it's gonna output, as long as the current isn't limited by the LEDs Vf. The low current you're measuring could be either the LED's Vf, or you may have '380mA' 7135s that only do an average of 370mA each.
Test a bunch of 7135s individually, one at a time on the same LED, and make a chart of the current each one will do. Just doing '380 x (number of 7135s)' you'll always get less current output than that theoretical number.
Keep adding more 7135s until the output stops increasing... then you'll know the limit is the LED, and not the driver.
And as for the 350mA binned chips... that 350mA is the minimum, if they aimed for an exact 350 due to the peculiarities of semiconductor manufacturing there would be significant numbers of chips that were below that minimum and would have to be tossed. So they aim for a safe target that will give the highest yield (number of usable parts from a single wafer, not the current output of each individual die), say 355, or 360. From a single silicon wafer some will have defects and won't work at all, some will work but be under spec and get tossed, some will land just above the 350 minimum, the majority will be safely above that minimum and close to the target, and a few will be exceptional and get the '380mA' label slapped on em. Doesn't mean they do exactly 380 or exactly 350, just that they fall close enough to those two ranges to be labeled as such.
see also: http://en.wikipedia.org/wiki/Wafer_testing
I just put an XM-L2 T6 on a Nitro BLF SinkPAD last night, running it on an QLite driver with 2 380 binned chips stacked, running off a single Samsung 20R cell, for 3.59A. It outshines my M3 with de-domed XM-L U2 emitter at 4.28A.
I’m figuring the binned chips not being true 380, and some inefficiency from the driver is why I’m getting 3.59A vs the mathematical 3.8A. 200mA? No biggie. Love it, love the beam and the tint.
For the record, a single Efest IMR18650 V2 pulled 3.4A.
Edit: with a 200mA spread and 10 chips on board, each chip is off less than 20ma…the driver cannot be 100% efficient. So with driver loss counted the chips are coming very close to the binned 380mA.