Testing my modded lights

Hi everyone

I managed to get what I think were pretty good results with my modded F13 & F15 lights a while back with the help of this awesome community.

I was looking for a good bike light and the F15 with all the mods has been rock solid for over a year: LD-25 driver @ 2.7A, no PWM in mid & low modes, XM-L2 U3 3C on noctigon, filed down bezel, braided springs, OP reflector and whatnot.

I was bored today and thought I’d run a quick test, so here’s my question: with a Samsung ICR18650-26F battery at 3.85V idle I measured about 3.45V and 2A through the battery on load. This doesn’t seem right to me because according to Test of 17mm CC 2600mA LED Driver 2 Group Mode, 3-4.2v, 2.6A
the driver drops regulation at 3.5V, while at 3.45V the current is significantly lower than 2.7A ( 2A). Also the battery drops from 3.85V to 3.45V under 2A load. That seems steep. Is this all due to extra resistance in the circuit somewhere? Can you please give me a few pointers to determine where I might be losing efficiency if that’s the case?

Or perhaps I’m missing something obvious here?

Are you measuring the voltage right at the cell when measuring under load? It sounds to me like the 26F cell internal resistance has increased and that is causing the large voltage drop you are seeing. Also other resistance in the circuit would cause voltage drops such as the springs, if they are not bypassed.

If you want to increase the time the light stays in regulation get a cell with lower IR, like the sanyo GA, and minimize other resistance like springs by bypassing them.

The springs are braided. That’s not to say that I did it right, LOL. And yes, I’m measuring the voltage at the cell. My Liitokala charger measures some 100mR for the cell resistance.
I should have probably mentioned that my DMM is dirt cheap, the leads are 22awg and about 12 inches long.

Based on this chart it looks like the cell should sag about .2 volts at this load and level of depletion.

EDIT: well I got that by comparing the 2 amp line to the .2 amp line. Sag is probably more than that compared to resting voltage.

Dogecoin supporter!? Wow, hehe

Thanks Angler

Did you measure 2A or just estimate that from the driver test? If you measured, your long DMM leads would affect the measurement. However, they shouldn’t affect voltage measurements.

The Lii 500 IR measurements are not accurate, based on HKJ’s test. So it is possible your cell IR has increased from 0.1 ohms (when new) to 0.2 ohms, which would explain your measurement of the voltage drop.

The 2A were measured at the tailcap.

Makes sense, thanks. I’ll try different batteries, leads, etc and see if I can determine where the extra resistance is. I would have thought the DMM and leads are not the problem because they read 2.7A with the battery fully charged

ETA: The above is obviously nonsense, right? The added resistance in the leads may not be enough to get the driver to drop regulation with the battery fully charged but may still not be negligible

This is correct. Now, the question is: when you measured the voltage drop of the battery, are you sure the current draw was 2A? If you weren’t using the DMM to measure the current at this time, the circuit resistance would be less, and the resulting current would be higher. So, maybe the driver was in regulation after all.

Another thing to note is that since this is a buck driver, the current from the battery won’t equal the current to the emitter in general.
Edit: my mistake, it is a linear driver, not a buck driver.

I said before the 2A were measured at the tailcap, but it wasn’t. My bad. I had the 2 DMM setup with one in series and one in parallel to measure both I & V simultaneously. So I just tried again, this time used some very short leads I made myself and found that the current dropped significantly when I connected one alligator clip to the threads of the flashlight body. So my former measurements were most likely skewed.
This time around, with the same battery, fully charged (4.22V) I got 2.9A and 3.85V at the battery without the alligator clip. Interestingly, with the alligator clip presumably producing some nasty contact resistance, the current dropped to 2.56A but the battery voltage remained at 3.85V. Does this make any sense? The driver is not supposed to produce more than 2.7A to begin with. LOL, I have no idea what I’m doing

I tried a different approach: ceiling test. So I measured the lux output pointing the light to the ceiling, noting the results every 5 minutes:

Time Lux
0:00 220
0:05 220
0:10 219
0:15 209
0:20 198
0:25 190
0:30 182
0:35 176
0:40 168
0:45 148
0:50 143
0:55 136
1:00 127
1:05 115
1:10 94
1:15 66

So the driver drops regulation after just 10 minutes. That doesn’t seem right

ETA: Gonna try a NCR18650B

That does seem rather quick to drop out of regulation, but it may not be too out of the ordinary. For example, the XPL tested here has a VF of about 3.5V at 2.9A. Your cell, after about 15min of 3A draw is at 3.65V. Adding in the the ~0.15V drop of the linear regulator, the light would go out of regulation at this point. If your cell has gained a bit in IR, or if your LED has a bit higher-than-normal Vf, it would go out of regulation even sooner.

Using a cell with a lower IR, such as the samsung 30Q, and using a lower Vf LED, such as the XPL2, would dramatically increase regulation time.

Great stuff, thanks for the info and for the excuse to go back to tweaking my lights haha. As for the battery, and since these F13/F15 are 26650 lights the LiitoKala INR26650 seems like a good option for a low resistance cell (0.04 Ohms), right? Test of LiitoKala INR26650-50A 5000mAh (Cyan)

Yep, that liitolala cell is a top choice. To be more accurate, the IR as well as the shape of the discharge curve determine how long the light will stay in regulation. Look at the discharge curve in the test with time on the x-axis. With your setup, the time it takes for the cell to reach ~3.65V at 3A is the approximate regulation time. So for the liitokala cell that would be about 36 minutes.

So this would be the XP-L I’m looking for right? http://intl-outdoor.com/noctigon-xp20-cree-xpl-high-intensity-v3-3c-p-902.html
I’ve done some googling and some suggest it’s more of a thrower compared to XM-L2, some others say there’s no difference. I don’t want my bike lights to be too throwy. I’m gonna get a couple of them anyway to experiment a bit.

As for the batteries, the protection board adds to the IR, right? Would it make a significant difference if I removed it?

One more question please. The 0.15V drop of the linear regulator you mentioned, is that something specific of the LD-25? Trying to figure out from here where that 0.15V comes from

OK, so I tried the ceiling test again with a Panasonic NCR18650B. It dropped out of regulation after some 17 minutes. That’s exactly when it would be at 3.65V according to HKJ’s graphs!
So I’m going to conclude that there’s nothing wrong with the build itself. Apparently I’ll need to keep an eye on those Samsung 26F’s

No, the XPL2 is the one with the lower Vf. Like this one. It should give a similar beam to the XML2. The XPL HI would give a more narrow beam with more throw.

The protection circuit would add resistance, I’m not sure if it would be significant.

The 0.15V drop also exists in 7135-based linear drivers. Not sure exactly where it comes from, but in one of the graphs in HKJ’s review he has “voltage change in driver”.

One thing to consider with the XPL2 is that the driver will be dissipating more heat because of the lower LED voltage. In some cases the driver components could get too hot. I don’t know what would happen with your driver, but just so you know of the possibility.

Thanks again mate. I’ll use the F13 as a “test bench” for that XP-L2, but I’m in Europe and mtnelectronics charges $13 for shipping. Time to shop around

This is the one I want, right?


You might be able to switch to a better FET. Rds-on which is the internal resistance of the FET at a given gate voltage varies significantly and is the main criteria for choosing a particular FET. Cheap drivers use cheaper fets which start showing higher resistance sooner on the discharge curve. One of led4powers drivers with the sense resistor set to 2.5A should perform better.

No problem, glad the circuit seems to be working and making sense.

That LED you linked is still an XPL version 1. Int-outdoors doesn’t seem to have XPL2s. Here is one:

It’s probably time for me to go read about drivers because that sounds like chinese to me LOL. Thanks for the tip, I’ll try to learn a thing or two about this stuff and make an informed decision, but for now I’m a bit reluctant to swap drivers. What I like the most about the LD-25 is that it has no PWM. All modes are regulated. I don’t want potentially distracting PWMs in my bike lights