18650 & 21700 battery internal resistance cold testing

maukka · 2019-01-22T17:00:59.000Z

edit 23.1.2019: added 18650 results at the end of the post

Many people report that lights can’t sustain their turbo modes in freezing temperatures. I tested four 21700 batteries and one 18650 after they had been outside in (–12°C) for an hour. Their surface temperature was at –1 to –3 degrees Celsius when I started the internal resistance (Ri) measurement. The test was done inside. I did a similar test previously on some 16340s.

For example, estimating that a high power light draws about 10 amps, the voltage would drop from 4.2 volts to about 3.3 volts on a battery with an Ri of 0.09 ohms. This is such a low voltage that the light would most probably assume that the battery doesn’t have enough charge to sustain the turbo output.

A Samsung 30T with an Ri of 0.05 ohms would probably perform better in cold temperatures on high modes. Samsung 30Q acted as an 18650 reference, and it was significantly worse than any of the 21700s. None of the cells were brand new and have several discharge cycled under their belt.

Using a lower high mode a while would warm up the battery, after which the turbo max mode might work.

Testing was done at 5 amps DC at room temperature (22°C) and at about –1°C.

Acebeam 21700 5100mAh IMR21700NP-510A
Warm 0.039 ohm
Cold 0.088 ohm

Liitokala 21700 4000mAh Lii-40A (Manufactured by Lishen)
Warm 0.025 ohm
Cold 0.089 ohm

Wuben 21700 4800mAh ABD4800
Warm 0.043 ohm
Cold 0.097 ohm

Samsung 30T 21700 3000mAh (Vapcell branded)
Warm 0.011 ohm
Cold 0.053 ohm

Samsung 30Q 18650 3000mAh
Warm 0.051 ohm
Cold 0.159 ohm

edit:

Data for a couple of 18650 cells. They were outside in –4°C for 3 hours. Surface temperature was 0°C before the discharge test started so not quite as cold as in the 21700 test. The same 30Q performed better this time around.

Very similar results to the previous tests. The lower the internal resistance warm, the lower it is when cold. When at 0 degrees, there’s consistently about 60 milliohms more resistance compared to room temp of 23°C. Sanyo GA increased 80 milliohms.

Samsung 30Q
Warm 0.052 ohm
Cold 0.112 ohm

Samsung 25R
Warm 0.033 ohm
Cold 0.090 ohm

Sanyo GA
Warm 0.069 ohm
Cold 0.149 ohm

Sony VTC5A
Warm 0.027 ohm
Cold 0.083 ohm

Sony VTC6
Warm 0.037 ohm
Cold 0.102 ohm

RobertB · 2019-01-22T17:23:41.000Z

It was –14ºF (–26C) here yesterday. I wonder if a 30Q would work at all at that temp.

maukka · 2019-01-22T17:39:17.000Z

I used a 30Q in a Skilhunt H03 yesterday at –25°C for half an hour and it worked fine. Didn’t use maximum output though (did once cycle through the modes and passing by turbo, nothing of note happened).

teacher · 2019-01-22T18:12:32.000Z

Interesting information maukka, thank you for sharing it. .

BlueSwordM · 2019-01-22T18:15:27.000Z

Are you using a 4 wire measurement?

My NCR20700A 20700 has a DC internal resistance of 15mOhms at 20°C.

A Samsung 30T should be even lower than that.

Are you counting lead/contact resistance too?

maukka · 2019-01-22T18:33:10.000Z

Yes, I use a 4-wire setup. 15mOhms sounds low for NCR20700A, HKJ measured 0.03 ohms.

On the other hand, I just checked one of my old measurements which show an Ri of 0.012 ohms for the 30T. Other warm measurements do match up though.

edit: thanks for the tip, checked the data again and seems I did the calculation wrong originally, actual Ri is 0.011 ohms.

Cereal_killer · 2019-01-22T20:59:08.000Z

Wow I’m surprised the 30Q did so bad (relatively)! Was it in the same condition, age & cycle count wise as the 21700’s?

moderator007 · 2019-01-23T14:03:47.000Z

Even measuring the Ir right off the charger lowers the reading. Charging or using the cell heats up the internal cell temp.
I have noticed when running a discharge test with a cooler than room temp cell say 60 degrees F, the voltage sags and lowers until it plato’s then after a minute or so after heating up from the load, the voltage climbs up a little then plato’s again where it starts to decrease again until depleted. A warm cell can give you higher lumens than a cool one.
.
Last night I installed a blue spring in the tail cap of a UltraTac A1 modded with a mtn boost driver and a xhp50.2. In my basement where most all my hobby junk is kept, the temp probably stays around 60 F. The first try with a 14500 Efest I got about 1600 lumens. Battery should have been around 60 F. I charged the cell back to 4.2v and stuck it in front of a portable electric heater for a minute. Picked up the cell and held it in my hand tight for about another minute, letting the heat absorb in the cell. Ran the test again for 1850 lumens OTF of a A1. With a few more mods maybe I can reach 2000 OTF.
I would guess to say that the cell temp on the second test was around 80 to 90 F, just a guess.
.
Thanks Maukka for this valuable test data that shows cell temp affects IR readings.

Freedom · 2019-01-22T23:39:40.000Z

I just thawed out an 18,000 mah jump starter (300A/600A peak) from –16F. Fully charged/like new condition previously…dead…no lights…nothing. After it was thawed out it started charging somewhere between the 4th and 5th indicator lights out of 5 like normal. I’m guessing that it tripped a LVP circuit.
I wouldn’t have kept it out there if I hadn’t misplaced it yet deep freezing these cells at all is just a hassle in my opinion.

Agro · 2019-01-23T08:50:48.000Z

One of things that may affect 30Q performance:
it was most likely colder than the other cells, simply because it’s smaller.

Other than this….it’s surprising because I’ve seen multiple reports of 30Q doing very well in the cold.

Satan · 2019-01-23T09:34:57.000Z

Look at that Lii 40A IR ,it’s the best for me cheap price (12 usd for 4 shiped) decent capacity and easy to buy(to where I live) at AE that make me pefer my new flashlight to be 21700 than 18650.

maukka · 2019-01-23T12:10:47.000Z

That sample of 30Q was picked randomly. It has tens, probably a hundred cycles and definitely isn’t the best with regard to Ri even when warm. I would probably find a better one. I’ve measured some 30Qs as low as 0.023 ohm.

maukka · 2019-01-23T15:13:52.000Z

I tested three 30Qs from different batches. All of them have been used and none are newer than about a year. Incidentally, the newest one also has the lowest internal resistance (the same one in the original post). They measured 0.079, 0.066 and 0.052 ohms tested at 5A fully charged straight from the charger. Relatively new they were in the 0.023-0.040 ohm range. Seems they age quite badly.

I’ll do another post with a similar test on worn 18650s such as the Samsung 25R, Sanyo GA, Sony VTC5A and VTC6. I don’t have any brand new cells to compare them to so YMMV.

maukka · 2019-01-23T18:36:08.000Z

Data for a couple of 18650 cells. They were outside in –4°C for 3 hours. Surface temperature was 0°C before the discharge test started so not quite as cold as in the 21700 test. The same 30Q performed better this time around.

Very similar results to the previous tests. The lower the internal resistance warm, the lower it is when cold. When at 0 degrees, there’s consistently about 60 milliohms more resistance compared to room temp of 23°C. Sanyo GA increased 80 milliohms.

Samsung 30Q
Warm 0.052 ohm
Cold 0.112 ohm

Samsung 25R
Warm 0.033 ohm
Cold 0.090 ohm

Sanyo GA
Warm 0.069 ohm
Cold 0.149 ohm

Sony VTC5A
Warm 0.027 ohm
Cold 0.083 ohm

Sony VTC6
Warm 0.037 ohm
Cold 0.102 ohm

Phlogiston · 2019-01-24T00:33:08.000Z

Good data, maukka - thank you!

Your Sanyo GA results are pretty close to some datasheet-bashing I did in an old CPF post. Just in case anyone finds the cold-weather discussion in that thread useful, here it is:

Candle Power Flashlight Forum

Cold temperature performance rechargeables: 18650 vs RCR123a?

I know a CR123a primary battery is the optimal choice in cold weather, but if I for some reason want to stick to rechargeables, is there a general difference (ceteris paribus) between a 1x18650 and 2xRCR123a configuration? All thoughts welcome.

THX · 2019-01-25T16:04:29.000Z

Molicel’s datesheet can be written to a minimum of –40 degrees discharge.

I heard that their batteries are designated for use in outer space technology.