My ACEBEAM EC65 XHP35 HI is dead (Now fixed!)

I purchased this flashlight a few days ago, and started to measure its run-time curves since then.

I tested this flashlight with the following batteries:

  • ACEBEAM 21700 5100mAh: A protected battery that came with EC65. I suspect this should be a Samsung INR21700-50E. The internal resistance measured is 30.7mΩ (with YR1030). 5017mAh/17.922Wh is obtained after 1A discharged down to 2.5V.

  • Samsung INR21700-48G: The internal resistance measured is 15.32mΩ. 4758mAh/16.863Wh is obtained after 1A discharged down to 2.5V. HKJ's review can be found here.

  • LISHEN LR2170SA: LISHEN is a known li-ion manufacturer from China, and this cell is a good one. LiitoKala also sells the re-wrap of this cell. The internal resistance measured is 14.61mΩ. 4056mAh/14.858Wh is obtained after 1A discharged down to 2.5V. HKJ's review can be found here

  • ACEBEAM 18650 3100mAh: A protected battery that came with my ACEBEAM H15. I suspect this should be a Samsung INR18650-30Q. The internal resistance measured is 23.4mΩ. 2972mAh/10.677Wh is obtained after 1A discharged down to 2.5V.

NOTE: the lumen numbers mentioned below are rough estimations, and by no means should be considered as accurate.

First, I tested the EC65 Turbo mode with these batteries.

The LISHEN battery can sustain the ~3700 lumen Turbo until the thermal step-down kicks-in, while the other three cannot. All the other three batteries can only sustain ~3700 lumens for less than 25 seconds then the output will drop to ~3100 lumens. Then the maximum output you can get at Turbo mode will be ~3100 lumens or less, even after cooling down the flashlight.

Then I conducted more extended experiments on Turbo mode. I repeated the following process to measure the approximated output curve of Turbo.

  • Turn on EC65 at Turbo.

  • Turn off EC65 when the thermal stepdown kicks-in.

  • Apply cooling to EC65 (with two fans, for at least 5 minutes).

Some findings from this result:

  • The Turbo mode of EC65 will decrease its output in accordance with the battery voltage. ~3700 lumens → ~3100 lumens → ~2700 lumens → ~2300 lumens.

  • The three batteries (ACEBEAM 21700 5100mAh, Samsung INR21700-48G, ACEBEAM 18650 3100mAh) can only sustain ~3700 lumens for a very short period. So with these batteries you'd better consider EC65 as a 3000~3100 lumen flashlight.

  • When conducting this experiment with the LISHEN LR2170SA battery, EC65 stopped working after several seconds and turned out to be dead after I cooled EC65 for the first time then started the 2nd round of Turbo.

It looks like something within EC65 (some circuit component?) cannot afford the ~3700 lumen load for a longer period, so when a more powerful (high-drain) battery is given, this vulnerability is exposed. I guess this problem is more easily to be found if tested with a even more powerful 21700 battery, for example, Samsung INR21700-40T and INR21700-30T.

I plan to report this problem to ACEBEAM. Before ACEBEAM solve this problem, you'd better not use a true high drain battery with EC65.

EDIT: I received a new EC65 from ACEBEAM. The new test results are as follows.

This new EC65 is able to sustain the extended Turbo test and performs noticeably better than the dead one. The highest output is now fairly close to 4000 lumens (~4100 lumens @0s, ~3900 lumens @30s). This number is no longer a single burst, it can last for ~3.5 minutes with the included 21700 battery. And, the new EC65 is more efficient as well.

It's good to see ACEBEAM fix this problem quickly, I'm impressed!

Wow. Thanks a lot for sharing.

1) Interesting that output decreases with depleting cell voltage. Unlike the Emisar D4 with FET driver, the EC65 uses a boost driver which could have been designed for constant output, and this would have been an advantage over the D4.
2) It seems it’s a win for the D4 with the FET driver when it comes to reliability.

My EC65 died with the supplied battery on turbo after being on for about 30 seconds. Acebeam said they have fixed the problem since and I’m getting another one next week. I wonder if yours is the newer batch since it was able to take much more punishment than mine.

Yes, ACEBEAM products seem to regulate its output on Turbo when possible, and step down a bit when the battery voltage is too low. I found similar characteristics when testing my EC50 II, EC50 III, EC60, L30, etc. I personally prefer this behavior.

Well, maybe you can torture your new EC65 with Samsung 30T/40T and see what heppens. I got no chance to do so before its death :smiley:

Yes, just ordered some 30T and Lishens just for this purpose :slight_smile:


Thanks for sharing.

Still good since it died peacefully and nothing dangerous happened.

That’s a very powerful boost driver, but that means the driver electronics have to dissipate a lot of heat, probably 5W or more, which is a new challenge in the confines of a small flashlight cavity.

Yeah pretty impressive for a non-fet driver to take 10A+ on the input

Thanks for the review and info. Does this use the same noctigon pcb and carclo optic as the D4?

Boost drivers lose efficiency the more difference there is between input and output voltages.
If the output voltage remains constant and the input voltage drops, the amount of current it can deliver decreases significantly, so it steps down to a lower mode.
I assume you’re talking about the large steep drops in output.

If you’re talking about the slightly inclined output line which isn’t perfectly horizontal, that’s because of heat.

I know this is the case in theory, but I don’t know to what extent this is the case in practice. The M43 (3S4P) is able to produce max output with cells as low as 3.5V, and the Imalent DN70, although with a 6V emitter, can produce flat turbo output until cell is depleted as well. Imalent DN70 - Review / Vorstellung / Test + Passaround | Taschenlampen Forum
Of course neither these two lights have a boost voltage of 12V, but if the latter turns out to be a hurdle, then I’d preferred something like a 2S2P setup for the Nichia leds, and instead of XHP35 use XP-L in 2S2P. Why complicate it with a 12V XHP35?
I understand there is a qualitative complexity with increasing boost voltage, but I don’t see this should result in a quantitative behaviour that output cannot remain flat.

It’s indeed using a Carclo optic, however the MCPCB used doesn’t seem to be a NOCTIGON.

The possible reasons I can think about:

The bundled battery is not a true high drain one, and ~10A should be the maximum load it can sustain.

EC65 will report a warning (triple flashes for every 5 seconds) when the battery voltage is too low. The lower the battery voltage, the higher the input current is needed for the boost driver to maintain a constant output. And the higher the input current, the more battery voltage sag. That means if the driver keeps the constant output until the low voltage warning, the battery voltage sag will be far too high. And then even though the warning is given on Turbo, there is still much energy remaining unused in the battery.

Yeah, the ones I have are 38 and 41mOhm (DC IR). So they’re a bit worse than a VTC5A (30mOhm), on the same level as a decent sample of a 30Q (mine vary a lot, from 30 to 60mOhm).

Thanks. With YR1030, my measured IRs are: 13~14mΩ (unprotected 30Q/VTC6), ~10mΩ (unprotected VTC5A), 14~15mΩ (unprotected PLB 26650-55A), and 30~31mΩ (ACEBEAM bundled 21700).

Maybe you need a similar 4 terminal device for resistance measurement, so as to eliminate the effects like contact resistance?

Reading this thread… yikes!

I’m intrigued by this light. 21700 support, a nice looking quad, and Acebeam. I only own one other Acebeam (UC15), but was impressed with its quality, output and UI.

However, for $122 I don’t think I want to buy a light that goes poof after just 30 seconds on turbo.

Don’t worry, it’s not a $122 light that goes poof after 30 seconds.

… It’s actually a $139.90 light that goes poof after 30 seconds… :person_facepalming:

Does it look replaceable with a noctigon mcpcb?