Testing a cheap Li-Ion 18650 cell

37929 Views 30 Replies 6 Participants Last post by  PStechPaul
I just finished a discharge test on one of the Li-Ion cells I bought on eBay at the price of \$14/4 pieces including shipping. It is rated for 3600 mAh but my test as received showed only 530 mAh. Here's what I did:

Now I need to charge it (while using the datalogger), and then another discharge test that will show more capacity.
1 - 20 of 31 Posts

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
From a post in a different thread:

Here is information on energy density of various battery technologies. See the following:
http://batteryuniversity.com/learn/article/types_of_lithium_ion

Lithium Phosphate is about 105 Wh/kg which is only about 20% higher than NiMH at 90, while LiCo is about 170.

I'll use these figures to do a reality check on the Li-Ion and NiMH cells I just purchased and tested, and the LiFePO4 which are on order.

The Li-Ion 18650 is rated at 3600 mAh and 3.7 V, for 13.3 Wh, and its weight is 41g, so energy density is supposedly 325 Wh/kg. Thus I would expect an actual capacity of 1883 mAh or 6.9 Wh, which coincides with another report that states these cells actually produce half of their advertised rating.

The LiFePO4 18650 is rated at 1800 mAh and 3.2V, for 5.76 Wh, and about the same weight 42g, so energy density is 137 Wh/kg. From the chart I linked above, this is still a little high, so I should expect 1380 mAh or 4.4 Wh.

The NiMH AA cells are rated 3000 mAh and 1.2V, or 3.6 Wh, and weigh about 19g, so energy density is 189 wH/kg, or more than twice what should be expected. So reality would be 1430 mAh or 1.7 Wh.

Now the expected true costs based on my small quantity purchases are:

NiMH \$0.92/1.7 = \$0.54/Wh
Li-Ion \$3.54/6.9 = \$0.51/Wh
LiFePO4 \$4.20/4.4 = \$0.95/Wh

These are total cost including shipping. Larger quantities roughly cut these costs in half. I will need to do testing on the NiMH and the LiFePO4 to see what their actual capacities are. The Li-Ion cells, as received, measured about 1.78 Wh and 0.83 Ah.

The results are tabulated here:
http://enginuitysystems.com/files/DischargeTest_3600mAh_Li-Ion.ods
http://enginuitysystems.com/files/DischargeTest_3600mAh_Li-Ion.xls

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
I performed a charging test, using 500 mA and 1 amp for a while, until the voltage on the cell reached 4.20V, then I reduced the current to 500 mA and again after it reached 4.20V I reduced current to 250mA, at which point the voltage seemed to stabilize at 4.12 to 4.15 volts. The open circuit voltage stabilized at 4.05 to 4.07 volts, and has remained stable for about an hour (with a load of about 3k, or a little over 1 mA).

At this point the cell had absorbed 959 mAh and 3.95 Wh. This is considerably less than the advertised 3600 mAh which I figure as 13.3 Wh at 3.7V. It is also about half of what I would expect given a power density of 170 Wh/kg. Thus these cells cost about \$3.54/3.95 = \$0.90/Wh. Maybe they will improve after a series of charge/discharge cycles.

They are 41 grams so 96 Wh/kg. The Hi-Power 100 Ah LiFePO4 cells are 4kg or 80 Wh/kg, and they cost \$120 or \$0.375/Wh. The best price for the 18650s is \$58/20 pieces which is \$2.90/4 Wh or \$0.73/Wh.

It looks like, for my tractor project, the best deal may be the 12Ah/12V SLAs, which are about \$0.14/Wh or maybe \$0.28/Wh if I run them at 1C for 4.3 kW, but I think I really need only about 1.5 kW average which is about 0.35C and I should get close to the full 4.3 kWh and about three hours of run time, which is more than enough. But I digress!

The test result files are:
http://enginuitysystems.com/files/ChargeTest_3600mAh_Li-Ion.xls
http://enginuitysystems.com/files/ChargeTest_3600mAh_Li-Ion.ods

frodus

· Registered
Joined
·
4,065 Posts
So is there a chance that the current you're measuring is incorrect?
What are you using for the load?
How is it being sensed?

I used my own calc from my battery test setup to calc your same Ah in/out, so I don't think there's anything wrong with the calc, so the measurements may be off, or the battery is a dud.

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
I think the measurements are fairly close. I double checked the voltage readings as well as the current using two different external meters (one is a Fluke 45), and verified that my datalogger is accurate to within at least 2%. The only error that may be significant is a higher voltage due to lead resistance to the battery, but I think it is no more than 0.1 ohm which would be 100 mV at 1 amp, which is enough to make a difference. This may account for the drop from 4.20V under 0.5A charge to 4.16V at 0.25A charge and 4.06V open circuit.

The current readings could be off, since I noticed a common mode error in the differential amplifier I use to read the current shunt. I compensated for it in the spreadsheet but it could be off by as much as 10%. But I doubt that these errors account for the twofold lower capacity (90) from what I had expected based on the 170 Wh/kg figure, and certainly the 3600 mAh specification (325), which would be astounding. The 90 Wh/kg value I measured is actually pretty close to that of the Li-Ion battery of the Dreamliner, so these batteries may just be older technology with a huge overstatement of capacity. The same company advertises 4200 mAh 18650 cells which include protection and are 47g, which is 330 Wh/kg:
http://www.ebay.com/itm/4-x-18650-4200mAh-3-7V-Protected-Rechargeable-battery-R-/380563881832
and 4500 mAh:
http://www.ebay.com/itm/4xWhite-UF-...thium-Battery-4500mAh-LED-Torch-/150911780146
and another company has 5000 mAh batteries in the same package, so if they are also about 45g that would be an astounding 411 Wh/kg which is rather unbelievable:
http://www.ebay.com/itm/221172886814

These might be more believable:
http://www.ebay.com/itm/2x-Ultrafire-18650-3-7V-2400mAh-2400-mah-Rechargeable-Li-ion-Lithium-Battery-/190644737243 \$2.16/ea

These are touted as 2200 mAh "Real" capacity:
http://www.ebay.com/itm/6X-NCM18650-2200mAh-3-6V-rechargeable-Li-ion-Lithium-battery-flat-type-pack-use-/281048950750 \$4.24/ea \$0.53/Wh

Samsung batteries are 1300 mAh, 44g, 106 Wh/kg:
http://www.ebay.com/itm/6X-3-6V-Li-...8A-high-power-battery-flat-type-/271139579885 \$5.60/ea \$1.20/Wh

And this one is listed as 1500 mAh with testing confirmed at 1650 mAh (but expensive at \$9.99/cell):
http://www.ebay.com/itm/IMR-18650-L...-Lithium-Cell-Battery-3-7V-4-2V-/271040378169

Interesting, I found multi cell protection circuits for sale:
http://www.ebay.com/itm/4-cells-8A-...nput-Ouput-Protection-PCB-16-8V-/110986410660
http://www.ebay.com/itm/2-cells-15A...0-Battery-In-Out-Protection-PCB-/121053430969
and a SOC meter PCB:
http://www.ebay.com/itm/Lithium-ion...level-LED-indicator-18650-14450-/121047062972

frodus

· Registered
Joined
·
4,065 Posts
Do you have another of the same cell to test to make sure it's not just a bad cell?

Find a known-good cell and measure that to make sure your setup is correct.

Use an ammeter to verify the current through the resistors.

Karter2

· Registered
Joined
·
1,666 Posts
Why would you even bother with any 18650 that claims more than 3600 mAhr.
You know its impossible.
And anything from "Ultrafire" is guaranteed to be rubbish !

Have you seen this ... it could save you a lot of time !
http://lygte-info.dk/review/batteries20 ... %20UK.html

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
I had not seen that site. It is very helpful. So it seems that my test showing about 1000 mAh is possible. They are very likely the same as these:
http://lygte-info.dk/review/batteries2012/GTL 18650 5000mAh (Blue) UK.html

The Sanyo cells seem to be OK:
http://www.ebay.com/itm/4pcs-Genuine-SANYO-UR18650FM-3-6V-2600mAh-Li-ion-Rechargeable-Battery-Japan-/110977053501 \$6/ea \$0.64/Wh
http://lygte-info.dk/review/batteries2012/Sanyo 18650 2600mAh (Red) bv UK.html
http://lygte-info.dk/review/batteries2012/Sanyo 18650 2600mAh (Red) UK.html

And Samsung as well:
http://www.ebay.com/itm/Genuine-SAMSUNG-ICR18650-28A-SDI-3-6V-2800mAh-Li-ion-Rechargeable-Battery-New-/110969656531 \$8.55/ea \$0.85/Wh
http://www.ebay.com/itm/6pcs-lot-3-...00mAh-Lithium-battery-flat-type-/281056498996 \$5.83/ea \$0.58/Wh
http://lygte-info.dk/review/batteries2012/Samsung ICR18650-28A 2800mAh (Purple) UK.html

Well, I have a lot more information now, and I still need to do the discharge test again and also try it on the NiMH cells, and the LiFePO4 when I get them.

The review site did not seem to have anything on LiFePO4.

Many of the Li-Ion cells reviewed had protection circuits, and some worked better than others. But I have a feeling that they may not be suitable for high voltage battery packs. If the protection circuit opens either from low voltage discharge or overcurrent discharge, the rest of the pack voltage would be applied.

Karter2

· Registered
Joined
·
1,666 Posts
Many of the Li-Ion cells reviewed had protection circuits, and some worked better than others. But I have a feeling that they may not be suitable for high voltage battery packs. If the protection circuit opens either from low voltage discharge or overcurrent discharge, the rest of the pack voltage would be applied.
I dont understand your concern ?
The "protection circuit" will simply isolate the cell, preventing any excess charge or discharge.
If the cell is in a parallel string, it would simply reduce the total pack capacity...and if it was in a series string, it would shut off the pack.
Tesla use these "protected" cells in their packs as a primary safety feature.

Of the 18650 cells tested the latest Panasonic NCR cells are the pick of the crop.

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
The case of a cell being disconnected in a parallel connection is not an issue, as the series element will only see the voltage of the other cells, or about 3.5V. But if all the cells are in series, under load, the cell will see the entire pack voltage in the reverse direction, and will attempt to charge it in reverse. If there is a high current Schottky diode across the cell, it will limit the reverse voltage to about 0.5V, so if the series element opens, it will be OK as long as the current does not exceed the diode rating. If there is a parallel MOSFET and shunt resistor (for top charge balancing), the MOSFET will be reverse biased and the current through the resistor will determine the voltage that the series element will see. For a 3.6 Ah pack used at 3C, or about 10A, a 2 ohm resistor will limit to 20V but that's 200W. So the Schottky diode is probably needed, but at 0.5V at 10A that's still 5W. A 20V MOSFET could be added with an RdsOn of 20 mOhms for 200mV and 2 watts. But you may need two in series with opposite polarity because of the intrinsic diode.

I completed the second discharge test with a 2 ohm load and I came up with 0.964 Ah and 3.32 Wh. It might be greater at lower current. I estimate the internal resistance to be 0.2 ohms, so there is about 0.32W at 1.8A so that could account for an additional 0.3 Wh or about 10% more capacity.

Here is the data:
[URL="http://enginuitysystems.com/files/DischargeTest2_3600mAh_Li-Ion.xls"]http://enginuitysystems.com/files/DischargeTest2_3600mAh_Li-Ion.xls[/URL]
[URL="http://enginuitysystems.com/files/DischargeTest2_3600mAh_Li-Ion.xls"][URL="http://enginuitysystems.com/files/DischargeTest2_3600mAh_Li-Ion.ods"]http://enginuitysystems.com/files/DischargeTest2_3600mAh_Li-Ion.ods[/URL]
[/URL]

Karter2

· Registered
Joined
·
1,666 Posts
But wouldn't a normal "large" pack constructed with 18650's be configured with cells paralleled in modules as a basis ?
..EG the Tesla packs.
...Sixty-nine cells are wired in parallel to create bricks. Ninety-nine bricks are connected in series to create sheets, and 11 sheets are inserted into the pack casing. In total, this creates a pack made up of 6,831 cells.

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
In my case I am planning to make a small high voltage pack of about 320-360 VDC using single cells. If these were truly 3600 mAh cells it would be 1.16 kWh which would be (just barely) enough for my 2 HP tractor which will usually be running at about 1000W or 3A. I would not attempt to do as Tesla has done with almost 7000 cells. I'm looking at 100, and if I need more power and capacity I might use another string in parallel or larger cells (5-10 Ah).

Here is an illustration of the problem:

When both MOSFETs are off, D1 conducts the load current and the cell is isolated, with only a few millivolts across the MOSFETs. But if D1 is not present, there is 323 volts across M2. With both MOSFETs on, the 0.2 ohms of series resistance of the cell is enough to cause its voltage to drop to 756 mV. This is because the 30 ohms shown for the resistance of the rest of the pack is incorrect. When the 320V pack has 20 ohms resistance and the load is 100 ohms for a total current of 2.69 amps, the battery with 0.2 ohms internal resistance shows a more normal 2.44 volts under load. This illustrates the importance of having balanced cells with equal internal resistance.

Karter2

· Registered
Joined
·
1,666 Posts
so, are you suggesting you have a 100s pack of 18650 cells ?

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
so, are you suggesting you have a 100s pack of 18650 cells ?
Yes, all in series, for 370V and (hopefully) 3.6 Ah. But these cells don't measure up. I need the voltage for a 240 VAC VFD.

Karter2

· Registered
Joined
·
1,666 Posts
I suspect you are being optimistic hoping for 3.6Ahr from single cells.
Better to plan for 100s 2p using cheaper 2Ahr cells and hence avoid the cell failure issue you are concerned about.
But of course that adds build complexity, size and weight !

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
I have found some supposedly better Li-Ion and LiFePO4 cells that are more reasonably rated at 2000-3000 mAh for 18650, but they are correspondingly more expensive. I'm trying to get the battery pack down to less than \$0.50/Wh (including delivery) for a 1.5 kWh 280-370V pack, and that seems to be difficult. At this point the best fit for my tractor is the SLAs which will give me 288V at 12 Ah (3.5 kWh) for a cost of about \$500, or \$0.14/Wh. I don't mind the 300 lb of weight, and for tractor use they will probably last 10 years.

With the small lithium (or NiMH) cells, there will be added cost of an enclosure and a BMS for 100 cells, so that will add \$300-\$500 for a 320-350V 2-3 Ah pack with less than 1 kWh so that in itself will be about \$0.50/Wh. At least the NiMH might not need quite as sophisticated BMS, and that's good, because at 1.2V/cell I'd need 300 cells.

I hope to do the testing on my (supposedly) 3000 mAh AA NiMh cells, which if they meet the spec would give me a 360V 3 Ah pack for

Maybe I should invest in a cell tester, which I found on eBay for about \$48: http://www.ebay.com/itm/Ni-Cd-Li-Io...ity-Tester-Discharge-Instrument-/300720475540

This guy has some other interesting electronic stuff, and if it works as claimed the prices look really good. But then again, it seems that my experience with the Li-Ion cell indicates that there is not a high level of truthfulness and trust. Perhaps these cells were measured on this device, and that is why the cells are rated so high?!

BTW, I found other Li-Ion cells with the GTL 18650 brand name rated as high as 5300 mAh. But my testing confirmed what was found on the Danish testing site, of about 1000 mAh:
http://www.ebay.com/sch/i.html?_trksid=p2047675.m570.l1313&_nkw=GTL+18650&_sacat=0&_from=R40
http://lygte-info.dk/review/batteries2012/GTL%2018650%205000mAh%20%28Blue%29%20UK.html

Karter2

· Registered
Joined
·
1,666 Posts
It seems od to plan a high voltage, low capacity pack for such a low power system ?
wouldn't it be simpler And easier to go with a 50 or 100v, system ?

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
The reason for the high voltage is the ability to use a standard three phase induction motor and VFD. It would be easier to use a lower voltage pack and a DC motor but I am convinced that AC is a much preferred technology, and I also intend to use what I learn when and if I do an EV car for public road use.

Also, I think there may be a "market" for electric tractor conversions and having a working prototype to demonstrate will be very helpful. It may even prove to be more practical and sellable and it can help reduce the amount of pollution and energy waste that characterize such vehicles. It may not make much difference for a homeowner who uses a tractor once a week for an hour to cut the lawn, but perhaps for landscapers and farmers it could be significant.

I have sent an inquiry to the company from whom I purchased these 18650 cells, asking for a replacement of cells having an actual capacity of at least 2500 mAh, or a refund. I checked the feedback and there were quite a few complaints of capacity 1/3 to 1/4 of what was advertised. In some cases I think they refunded like 1/2 the cost, which would be acceptable. I don't like to leave negative feedback without attempting a resolution, but even if they make good, the advertising is false, and I think they should revise the specs. The same company also sells 18650 Ultrafires rated at 4500 mAh, about \$2/each in 30 pc lot. That would be \$0.12/Wh if true. Even if they were only 1500 mAh it would be a consideration. But I think the problems of housing and BMS make this impractical.
http://www.ebay.com/itm/18097363010...eName=STRK:MEWAX:IT&_trksid=p3984.m1423.l2649

palmer_md

· Registered
Joined
·
601 Posts
also a single string of these would only be good for 6 amps. The suggested maximum discharge is 2c and from what I've read its better to stay below 1.5c and better yet to stay below 1c. Is the system you are designing ok with a 6 amp maximum current draw?

PStechPaul

· Registered
Joined
·
3,143 Posts
Discussion Starter · ·
With 100 of these cells, at about 350V, 6A would be 2.1 kW, which is more than the 2 HP motor. For my larger tractor, I have a 5 HP motor, and a 480V 7.5 HP VFD. For that I would use 200 cells, for 700V, and 6A would be 4.2 kW or 5.6 HP. I wanted the higher voltage so I could try overclocking the motor to get 10 HP. But I think 5 HP is really plenty, since it only had an 8 HP ICE originally. It's a 1967 vintage Simpicity Broadmoor:

1 - 20 of 31 Posts