Close

First charge / discharge cycles on a PANASONIC NCR18650PF

A project log for A 110 kWh Powerbucket

The lead-acid batteries of my off-grid solar system are dead. I will replace them by a 18650 batteries stack housed in a big wooden box.

michel-kuenemannMichel Kuenemann 08/28/2016 at 21:407 Comments

Thanks to my BMS mockup I can measure rather precisely the energy delivered by the Panasonic NC18650PF cell during a discharge sequence.

My goal is to compare the datasheet of the cell with my own measurements.

Panasonic announce a 2900 mAh rating for this cell.

I made the first charge discharge cycles with the following parameters:

Results:

Conclusion:

The battery has delivered 86 % of its rated capacity. This is quite a good figure, because the end of charge condition was 4.0 Volts, much less than the standard 4.2 Volts end of charge voltage.

The gravimetric energy density of the cell is 8977 / 45 = 199 Wh/Kg - Awsome

Next step:

Make the same measurements with a 4.1 V end of charge voltage and the same discharge parameters.

Discussions

Yann Guidon / YGDES wrote 08/28/2016 at 22:42 point

End of charge = 3V : nice, I set to 2.93V because I have precise detectors in SOT23 for this voltage :-)

But why 4V ?

Most batteries today are 4.2V, right ? Does reducing to 4.1 or 4V increase the cycle count ?

  Are you sure? yes | no

Yann Guidon / YGDES wrote 08/28/2016 at 22:45 point

Furthermore, the resistor is not a constant current load because the current drops as the voltage drops... it makes capacity calculations harder, but then you probably prefer to measure the Watt-hours...

Damn I haven't messed with LiIon since http://ygdes.com/LiIon/

  Are you sure? yes | no

Michel Kuenemann wrote 08/30/2016 at 01:52 point

Hello K.C. and Yann,

Yes, the dissipation of the resistance drops form  about 4² / 4 = 4 watts in the beginning of the discharge to  3² / 4 = 2.25 watt in the end.

This is not really an issue because thanks to the awsome 12 bit ADC of the LPC1549 the voltage acquisition chain of the BMS has a resolution of 1 mV and a precision of 0.1 %, with virtually no noise.

I assume that the value of the load resistance is nearly constant and known to 0.5 %, despite it gets warm. 

So the calculation of the current ( I = U/R)  is done probably with a precision better than 1%.

The current integration (coulomb measurement) is made with a 10 Hz rate. Thanks to the very slowly changing current, there should not be much additional integration error.

To compare the cell the the manufacturer data the mAh calculation is the most relevant. To have an idea of the energy stored, the Wh figure is more interesting. I compute both, of course. 

There is an additional error in the Wh computation because the voltage gets squared ( P = E²/R). This additional error is of 0.1 %, so probably neglectable.

  Are you sure? yes | no

K.C. Lee wrote 08/29/2016 at 03:08 point

http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

4.20V  300 – 500  Cycles
4.10V  600 – 1,000 Cycles
4.00V  1,200 – 2,000 Cycles

  Are you sure? yes | no

Yann Guidon / YGDES wrote 08/29/2016 at 03:14 point

Thanks !

  Are you sure? yes | no

Michel Kuenemann wrote 08/30/2016 at 01:32 point

Hello K.C. and Yann,

As for any type of battery, there is a tradeof between end of charge voltage, depth of discharge, and battery life. 

I target a life time of at least 5 years for the first 10 kWh string installed in the powerbucket. The first string will cycle once a day, so I will setup my parameters to obtain a life of at least 5 * 360 = 1800 cycles.

The end of charge voltage will probably be close to 4.0 V. The end of discharge voltage will also be slightly higher than 3.0 volts to limit the depth of discharge (DOD) to about 50 %.

Once the second string is installed everything changes because the resulting 20 kWh pack will cycle once every two days, and therefore the life should reach 10 years with the same parameters.

The life expectancy of the whole pack will increase each time I add a string to the bucket

Conclusion: I should  add new strings as fast as possible to the stack, as soon as I have the cash to buy the cells.

  Are you sure? yes | no

Yann Guidon / YGDES wrote 08/30/2016 at 03:30 point

Nice strategy :-)

  Are you sure? yes | no