Some IMR (and others) discharge testing...

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
Just some testing of a variety of readily available IMR cells. I gathered two of each sample that I had for testing. Testing was done at 1C to determine the manufacturer's rating then again at calculated maximum rated discharge (see the summary chart for explanation). Some of the cell didn't live up to their max ratings so they were retested to determine their (subjective) practical maximum. You should be able to extrapolate any current in between my test discharge currents.

Summary:


Junsi iCharger 306B w/3.00V cut off + LogView 2.7.4.485 + data tables exported to Excel 2010. I have the full data table available, but figured capacity and voltage were the primary concerns. Note: on the updated graphs where the plot line does not reach the 3.00V x-axis, this is due to the voltage dropping faster than the software's datalog interval of 2 seconds. See the summary chart for the final values as taken off of the iCharger's display.


aw14500.png



aw16340.png



aw18350.png



Out of four of these 18650 only one was able to perform near rated levels and able to withstand a 15C discharge. My three others performed similar to the blue, red and purple plots above.
aw18650.png



These 26500 are pretty touchy near their somewhat low discharge limit. While @10A their voltage holding looks good and capacity is adequate, @ 12A+ they were unable to hold any voltage and capacity was @10% or worse.
aw26500.png



sf14500.png



sf16340.png



sf17650.png



sfb18350.png



It looks like I had one bad Shao red wrapped 18350.
sfr18350.png



sf18500sgce.png



sfp18650.png



sfr18650.png



ms22430.png



...artificial cooling used to prevent iCharger cut off on over-temp (50°C)
th26650.png



And a popular LiCo for comparison…
rl18650.png



And a not so common LiCo...
awc.png



On deck: A123 systems 18650 and 26650 LiFePO4...



If I may generalize, there seems to be better power handling as you go longer vs. wider. I.E. 18650 handled higher current draw better than 26500; the same for 14500 vs. 16340. Additionally, if anyone thinks they will get these types of current draws in lights/devices with the batteries held by spring pressure alone is kidding themselves. I ran my tests multiple times with multiple test rigs and saw HUGE differences between testing with my modded 4Sevens battery cradle (with and without magnets on the cells for contact and 12awg wire connections) and my ratchet clamp setup (w/12awg wire connections).

-DF

EDIT: All charts up to date as of 11/2/11
 
Last edited:

Justin Case

Flashlight Enthusiast
Joined
Mar 19, 2008
Messages
3,797
Nice work.

For the graphs, it might be useful to use a common x-axis for cells with similar capacities. Makes it much easier to compare/contrast. Also, testing at the same current draws for the small cells and the large cells (e.g., 18350 and under for small cells, 17670 and larger for large cells) would also help to make 1-to-1 comparisons. The main question would be which current draws.
 

Battery Guy

Enlightened
Joined
Apr 28, 2010
Messages
807
Location
Portland, Oregon
DFiorentino

Really nice work. Thank you for doing this. I have been wanting to test and publish the line of AW cells for awhile now but have not had the time. You inspire me.

I am quite shocked at the results of the AW 26500 cells. Specifically, they seem to perform quite poorly at 10 A, which is a popular discharge current for many of the hotwire superbulbs that these cells commonly are used to power.

Might I ask how you made connection to the cells?

Cheers!
BG
 

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
Nice work.

For the graphs, it might be useful to use a common x-axis for cells with similar capacities. Makes it much easier to compare/contrast. Also, testing at the same current draws for the small cells and the large cells (e.g., 18350 and under for small cells, 17670 and larger for large cells) would also help to make 1-to-1 comparisons. The main question would be which current draws.

Yeah, I kind of thought about the x-axis after I finished posting them up. I can redo them in the next day of so to make for easier comparisons. I was also thinking of a single discharge test for all like size cells, but like you mentioned, what current. I was thinking maybe picking the lowest common denominator (lowest max discharge) for each cell size and using that current, seeing as each cell would only perform better from that point down.


DFiorentino

Really nice work. Thank you for doing this. I have been wanting to test and publish the line of AW cells for awhile now but have not had the time. You inspire me.

I am quite shocked at the results of the AW 26500 cells. Specifically, they seem to perform quite poorly at 10 A, which is a popular discharge current for many of the hotwire superbulbs that these cells commonly are used to power.

Might I ask how you made connection to the cells?

Thanks. It's curiosity about my own mods and powering them that got me to finally do this; the 26500's in particular. I have a couple lights, LED and hotwire, using these 26500's and when you strike the match they are great, but sag quickly as voltage drops. I thought it may have been a heat issue, but alas, it's just the cells themselves. I've got 5 of them and the two tested here are the best (most consistent) of the batch. The others performed even worse, like 1600mAh @2.3A worse. It's like for a given length, going up in diameter doesn't increase power handling/capacity proportionally. Taking brand/manufacturer out of the equation the 18500's I tested performed better than the 26500's and the 18650's performed better than the 26650s. When you look at the C rating performance that is. I'd be really curious to test out an IMR 32650 to see how bad in does in retrospect. My connection for these tests was via your typical clamp:

img1853ng.jpg


Stranded 12awg wire soldered (more like welded) to heavy brass contacts.

-DF
 

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
I wish I did...or at least I used to wish. Seeing these results makes me feel more comfortable going with a 2P 18650 setup. Though it does require a modded host more times then not. Even a 3P 17650 setup should be able to handle ~30A.

-DF
 

xul

Enlightened
Joined
Aug 28, 2011
Messages
336
Location
MD
incredibly, y plotted as a linear function of x.

aw imr 26500
2300 2300 mAh

2.3A 1950
10A 1650
19.1A 1380
X Y
I % of rated capacity
2.3 0.847826087
10 0.717391304
19.1 0.6



-0.014695968 slope, a


0.875556926 intercept, b


Y = aX + b

Here's some predicted values.
X ......Y
2.3 0.8417562
10 0.728597249
19.1 0.594863942
6 0.78738112
15 0.65511741
 
Last edited:

ampdude

Flashlight Enthusiast
Joined
Apr 7, 2007
Messages
4,585
Location
USA
DFiorentino, can you come up with any reason why the AW IMR14500 performed so much better at higher discharge rates than AW's IMR16340's and IMR18350's? This doesn't make any sense to me.

In theory the IMR16340's should have about the same capacity and the IMR18350's roughly the same or more by volume, don't they?

Perhaps the IMR14500's are a whole different cell made by someone else. Or newer.. or different chemistry/mix?
 
Last edited:

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
No scientific reason. I've known for a while from a practical standpoint that this would be the case, however. When you look at a lot of the reviews on here, similar lights always seem to fare better from a performance and/or runtime standpoint with 14500s as opposed to 16340s. The 18350s did shock me a bit; AWs especially. They underperformed at all discharge levels. Like i hinted to previously, it seems the "fatter" cells perform adequately when discharged near 1C. Crank up the juice and the story is different. It may just boil down to a manufacturing standpoint in that maybe the quality or quality control is better on say 18650s as they likely sell more than say 26500s. Or, I just don't know...

-DF
 

ampdude

Flashlight Enthusiast
Joined
Apr 7, 2007
Messages
4,585
Location
USA
That's wild. Doesn't make any sense to me. Well I'm glad you tested them, very good information here, thanks.
 

xul

Enlightened
Joined
Aug 28, 2011
Messages
336
Location
MD
???

None of my 26500's were able to hold 19.1A without tripping the 3.00V cutoff first.

-DF
BTW, the percentages should be multiplied by 100.

I thought the graph showed this battery crossing the X axis at 1380 and 1650. It must have been the 10A line. Here's revision A.


aw imr 26500
2300 mAh


2.3A 1950
10A 1515


X Y
I % of rated capacity
2.3 85
10 66

-2.456239413 slope, a


90.43195935 intercept, b


Y = aX + b
X Y
2.3 85
10 66
6 76 predicted, so you should get 1750 mAh at 6A. This third value will tell if the relationship is linear.
 
Last edited:

jasonck08

Flashlight Enthusiast
Joined
Sep 16, 2007
Messages
1,516
Location
Redding, CA
Can you not discharge the cells down to 2.5v? There is a large capacity loss at higher currents by discharging only to 3v vs 2.5v. Not to mention the added resistance in the wires means that you may only be discharging the cell to 3.1 or 3.2v.

Also, can you measure the voltage drop of your clamp setup? For example take a DMM connected directly to the cell and compare it with what the logging software is reported at the given time. Even with my CBA, I get at least .1v drop on most currents over 5A.
 

flashflood

Enlightened
Joined
Mar 9, 2011
Messages
608
This is great -- thank you!

Your measurements confirm something I've noticed but never verified: hard-driven single-cell lights don't last anywhere near as long as the stated battery capacity suggests. Bummer, because I love the pocket rocket form factor.

One thing I'd love to see, if you have the data: how do these cells perform at intermediate currents, especially the common drive currents of 1.4A and 2.8A?
 

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
Can you not discharge the cells down to 2.5v? There is a large capacity loss at higher currents by discharging only to 3v vs 2.5v. Not to mention the added resistance in the wires means that you may only be discharging the cell to 3.1 or 3.2v.

Also, can you measure the voltage drop of your clamp setup? For example take a DMM connected directly to the cell and compare it with what the logging software is reported at the given time. Even with my CBA, I get at least .1v drop on most currents over 5A.

I can discharge down to 2.50V, but I don't see the point. The usable capacity below 3.00V is negligible or less. Look at the plots, they're going almost vertical by the time they hit 3.20V-3.10V. And yes, while discharging at 18A+ will technically net a higher numerical capacity number, it again is negligible. I was literally watching every discharge (about 4 times as many as I posted here) and as soon as the cells started hitting 3.30V I could watch them "fall off the mountain" if you catch my analogy. As for my rig, I just checked everything. The resistance was not measurable. Meaning I connected my test leads and measured 0.1Ω-0.2Ω; connecting across my entire setup (clamp contact to wire connection at charger) measured the same 0.1Ω-0.2Ω. And here's a quick standing voltage check (I'll check at a 10A draw in a bit):

img1858v.jpg


(FYI, if I sound defensive, I actually am not. I'm just trying to answer analytically.)



This is great -- thank you!

Your measurements confirm something I've noticed but never verified: hard-driven single-cell lights don't last anywhere near as long as the stated battery capacity suggests. Bummer, because I love the pocket rocket form factor.

One thing I'd love to see, if you have the data: how do these cells perform at intermediate currents, especially the common drive currents of 1.4A and 2.8A?

Thanks. If we can agree on some common test currents for certain size cells, I'll gladly do additional tests. I just don't want to do a bazillion more tests as this little endeavor took the better part of a week+. It's the recharging of the cells that I hate...:whistle:

-DF
 

HKJ

Flashaholic
Joined
Mar 26, 2008
Messages
9,715
Location
Copenhagen, Denmark
Meaning I connected my test leads and measured 0.1Ω-0.2Ω; connecting across my entire setup (clamp contact to wire connection at charger) measured the same 0.1Ω-0.2Ω.

0.1 to 0.2 ohm is way to high. But with the heavy cable you are using I doubt that you have that much.
 

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
0.1 to 0.2 ohm is way to high. But with the heavy cable you are using I doubt that you have that much.

Yeah, that is just with my spare test leads that weren't currently being used. All of the cabling in my test rig is 12awg for current carrying and for voltage monitoring; all soldered connections, no clamping.

-DF
 

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
I just did two quick tests at 5A and at 10A. It looks like I have ≤0.05V drop across the clamp to charger connection per 5A load. It actually was 0.04V @5A and 0.10V @10A. I can adjust my future tests to compensate, however I still feel the data gathered still paints an accurate picture of cell behavior. My main goal was to determine a cells ability to hold its rated capacity and at what voltage at 1C and at maxC to see if what I was seeing in the real world was my imagination or proof. Still I'll redo the high current tests.

(The Fluke is connected at the cell to clamp connection; the Ideal (meter to the right) is connected at the wire to charger connection.)

img1862q.jpg


-DF

EDIT: One thing I thought I'd mention is the iCharger's cut off works based on exceeding the voltage limit. Additionally, the time interval is 2 seconds. So, the charger ceases its discharge once voltage is <3.00V for 2 seconds. The capacities I recorded in the summary chart are taken directly off of the charger display and may not directly match the cut off on some of the charts. (Which can be visibly seen in a few as voltage was plummeting faster than the software was recording.)
 
Last edited:

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
And this was the data that led me down this curious road:

fm64458.png

FM5x26500 host, Kiu socket, 64458 bulb, DD, AW 26500 cells... That's 1500 lumen lost in 44 seconds. :(

-DF
 

DFiorentino

Flashlight Enthusiast
Joined
Dec 31, 2004
Messages
1,332
Location
MD
Ok, I redid the graphs to take into account the increased voltage drop as the current increases. It's just a quick fix until I redo all the 2A+ discharges this weekend. I'll double check my voltage drop at 20A to see if the drop is stable, linear or exponential. The graphs shouldn't end up looking any different, only the plots will be complete all the way down to 3.00V now. I also tried to make a common x-axis amongst certain size cells to help with the overall comparison.

-DF
 
Top