Best way to determine 18650 capacity, resistance and condition ?

Hi Oscar,

To do this would require banks of equipment, most likely battery testers (example is BA6010), and power supplies and DC loads, and without enough equipment it will be hundreds of hours of testing (not to mention the time+cost to automate this, and getting someone to swap batteries in/out every hour or so).

I think the testers will cost more than several hundred used batteries.

By the way (since you have discussion comments about series/parallel batteries), you may need to consider buying more batteries, in order to find sufficient ones that are similar, and it could still be risky with used batteries. It may be better to just buy new batteries from a trusted manufacturer+distributor.

Is it better to test for resistance when the cell is fully charged or fully discharged ? Anyone know ?

Thank you.

The problem is defining exactly the point of total discharge, When I test them I fully charge them and then time how long it takes for them to discharge at a specified current to a lower voltage point. I can then rank the batteries relative to one another and choose the ones that took the longest to discharge from full charge to the chosen voltage point. To do this test I used an electronic load and a piece of test equipment that was built specifically for timing long duration events.

John

Ok John, by doing this test, does this tell you which cells have higher resistance and lower resistance ?

And what is the device your using for this test ?

Oscar

This test does not directly measure the internal resistance but there is a correlation between the amount of time that a current can be maintained and the internal health of the battery. If you search "Process Duration Timer" on this site you will get the list of the series of blogs I did when I initially built my tester 3 years ago.

https://www.element14.com/community/search.jspa?q=process+duration+timer

Many of the better electronic loads can also time operations and be programmed. It may take several hours for a good battery to discharge depending on the load current so it is difficult to manually sit and measure the time.

John

Hi Oscar,

I wanted to also mention that if you do set up this experiment it is important to have a way to disconnect the load when the batteries get to their discharged voltage. If you pull a Li-Ion battery too low you can damage it.

John

There is a road test of a battery tester right now.

Hioki BT3554 Battery Tester 

You should try hard to win this Hioki tester....

I think that the best way to test a batteries capacity to measure the complete charge and discharge cycle (voltage, current and time) to address the true storage.  I did this on a recent project (Walky the Biped Robot - Power pack  and Walky the Biped Robot - A new hope (actually a new body and battery) ) and I was quite surprised with the data (measured between 220mAh and 290mAh for a rated 1200mAh cell).  There are a lot of junk batteries being sold out there, so it really does pay to test your battery capacity before relying on the stated values.

Reliably determining the state of health of a lithium ion cell is pretty hard to do unless you have a reference cell. If time is not an issue and you have a source, a sink and a power meter available then you can cycle the cells as others have also suggested already. If you don't have this kind of equipment available then personally I think the cost required for testing in relation to the money you save by buying used cells is too high.

If you have reference points of your cell then a possible way to make a determination of the state of health are impedance measurements at different / specific frequencies. Here is an example how this looks like:

Suitable instruments for this measurement: HIOKI BT4560 or HIOKI IM3590. (but certainly not suitable instruments to measure a hundred of used cells unless you have one of these instruments available for free)

The same thing would apply for state of charge measurements - you won't need to measure the entire curve (basically you can ignore the large overlapping parts) but you need to have reference values of the exact cell type. Temperature has a big impact on the measurement, by the way.

I have done this for larger battery packs, but the same general method applies for measuring the internal resistance of the cell.  Without buying special battery test hardware, you will need an electronic load and a good quality volt meter, ideally both equipments can be controlled via a computer interface.  Testing the cell at approximately 50% charge (around 3.7V) should yield the most accurate results since this is the most linear region of the charge/discharge voltage curve.  Note ambient temperature of the cell will impact the resistance.

1. Connect the cell using a "4-wire" also called "Kelvin" connection to the programmable load.  This means two wires shorted together at each the positive and negative terminals.  One each of the positive and negative leads will go to the electronic load, the other pair will go to the volt meter.  The purpose of this is to cancel the voltage drop caused by the current through the wire resistance to the load.  Since the volt meter should have very high input resistance (>10Meg Ohms), there is negligible current flowing through the wires to cause a voltage error.

2. Set the load to something like 0.1C current rating of the cell, running in constant current mode.  If your cell is 3000 mA-hr, this would be 300mA current.

3. It is best to automate this measurement sequence, but close enough measurements can be made by manual operation.  The faster the measurements can be made, the less error will be introduced to the measurement since current flowing through the cell will cause self heating (as mentioned, cell temperature will impact the measurement)...generally the thermal mass of the cell and the general low internal resistance will be forgiving if you're not too slow.

4. Start by taking the cell voltage at the 0.1C or 10% current rating of the cell.

5. Step up the load current to something like 0.9C or 90% the current discharge rating of the cell.

6. Take another voltage measurement.

7. Disable the load.

8. Perform some simple math per ohms law to find the resistance:  Voltage = Current * Resistance.  Because this is a linear relationship (for resistive circuits), a delta measurement of current will cause a direct proportional change of the voltage (assuming resistance is constant...which is the assumption we are making).  So then (Voltage_0.1C  - Voltage_0.9C) = (Current_0.9C - Current_0.1C) * Resistance.  Simple algebra yields Resistance = (Voltage_0.1C - Voltage_0.9C) / (Curent_0.9C - Current_0.1C).

For capacity information, you can get decent battery chargers which perform this measurement for roughly $10 per cell slot.