Battery Care and Feeding

The internet is awash with opinions about batteries.  Some of these opinions are informed, and some are not.  The information in this section has been distilled from the book Batteries in a Portable World by Isidor Buchmann of Cadex.  The book is subtitled “A Handbook on Rechargeable Batteries for Non-Engineers.”  While it may be true that it is light on the specifics needed by an engineer, it certainly would be useful to an engineering manager.  Cadex also maintains the BatteryUniversity.com website.  I highly recommend both as sources of reliable information.  

Below is a compilation of information about lithium-ion (used in EM ePure) and lithium polymer (used in EM 5.7) cells.

A battery is a collection of cells.  The chargers that accompany Electric Motion bikes are designed for their specific cell chemistry, capacity, and voltage.  Here are a few tips to prolong the battery's life:

Rubber Band Effect

The following quote comes from Isidor Buchmann,

Finding the exact 40 - 50 percent SoC level to store Li-ion is not all that important.  At 40 percent charge, most Li-ion has an OCV of 3.82V/cell measured at room temperature.  To get the correct reading after a charge or discharge, rest the battery for 90 minutes before taking the reading.  If this is not practical, overshoot the discharge by 50 mV or go 50 mV higher on charge.  This means discharging to 3.77V/cell or charging to 3.87V/cell at a C-rate of 1C or less.  The rubber band effect will settle the voltage at roughly 3.8V.

Achieving a Partial Charge

So exactly how does one go about charging a battery to 80% of its capacity?

Knowing a battery's true state of charge is not a simple matter.  Counting coulombs is the best way, but that requires special equipment.  A simple method is to estimate the state of charge from voltage and known curves. 

I monitor the charging process by using the inexpensive Chinese power meters shown below.  The AC side of the charger is connected through a PEZM-061 power meter.  The DC side of the charger is connected through a PEZM-031 power meter. 

PEZM-061 (100-amp AC power meter)

PZEM-031 (20-amp, 100-volt DC power meter)

By observing voltage, current, and energy readings I'm able to manually stop the charger before the battery reaches a full charge.

For the ePure and 14-cell 5.7s, 80% charged works out to about 55 VDC.  For the 13-cell 5.7, it's about 51 VDC.  Note that those voltages are for the battery at rest” (measured several minutes after charging or discharging ceases).  

In order for the charger to push current into the battery, the voltage it emits must be greater than the battery's voltage.   Generally, I stop the charger when the DC power meter's voltage is 1.5 to 2 volts greater than my desired battery voltage.  

It would be nice to automate this process, but so far I have not.  My first thought was to use something called a countdown timer” on the AC side of the charger.  This would turn the charger off after a preset time interval.  Give the battery, say, 1 hour of charge rather than allowing the charger to shut off automatically at full charge.  Unfortunately, none of the countdown timers I've found meet my standards for quality (have ETL or UL certification).  Because the battery charger's current draw is not insignificant, I want a safe high-quality device to interrupt it. 

Since I probably have to build something anyway, a voltage comparator relay would be better.  This would shut off the charger at a specific DC voltage.  I'd need an analog voltage comparator to compare the battery's voltage to an adjustable reference voltage and open a relay on the AC side.  I've just been too lazy to build it.

Details, Details, Details 

You will notice a small offset (0.04A, 0.5W) displayed on the PEZM-061 meter in the photo, yet nothing is drawing power.  This is because I used the 100-amp version of that meter even though the AC circuit that it's connected to is only rated for 15 amperes.  I don't trust passing heavy currents through an inexpensive Chinese product connected to the AC power line.  The 100-amp version employs a non-contact current transformer (CT) whereas the 20-amp version uses an internal current shunt resistor.

Strictly speaking, only monitoring the DC side is necessary.  For that, I use a PZEM-031 power meter in between the battery and the charger.  I fitted mine with male and female connectors that mate with those on the battery and charger.

If you don't want to go to the trouble of wiring power meters, I've read good things about the Kill A Watt” electricity usage monitor.  You can find it on Amazon for about $30 (US).  It's a plug-and-play device.  Begin by monitoring the charger's AC-side energy consumption over time.  Then build a table of various beginning and ending states of charge to estimate how long to allow the charger to run for a given beginning state of charge.

Effect of Temperature on Battery Capacity

Electric Motion provides a table of how ambient temperature affects vehicle range.  The battery's capacity is specified at 25°C (77°F).  Above and below that temperature, capacity (and range) will be reduced.  For example:

C-Rate

A cell's (or battery's) C-rate is the ratio of the current at which it is being charged or discharged versus its designed Ah capacity.

For example, a cell having a capacity of 2 Ah should be able to provide 2 amps for one hour.  In this case, a 1C rate of charge or discharge would be 2 amperes.

Similarly, in this example, the 2C rate would be 4 amperes and the 0.5C rate would be 1 ampere.

Constant-Current / Constant-Voltage Charging

Lithium-ion cells (and in turn, battery packs) are charged following a CC-CV regimen.  CC stands for constant current and CV stands for constant voltage. 

Below is the actual charging curve for one particular (but representative) lithium-ion cell.  The plot was produced by an inexpensive ($85) battery capacity tester manufactured by ZKETECH in China.  The cell under test is rated at 3.6 volts and 2.0 Ah.  This equates to an energy capacity of 7.2 Wh.  The cell is being charged at a 1C rate (2 amperes during the constant current part of the regimen).

When the cell voltage gets to 4.20V, the mode changes to constant voltage.  This results in a steadily decreasing current flowing into the cell until charging ceases when the current reaches a sufficiently small value (often 0.05C).  In this case, I allowed it to run a bit longer than would be typical.

When charging commences (t = 0), the cell's voltage starts at 3.34 volts and the charging current is 2.0 A.  This equates to a charging power of 6.68 watts.  (The cell had been discharged to 3.0 volts only a minute before and then recovered to about 3.3 volts.)

At the CC to CV transition point (t = ~52 minutes), the cell voltage is 4.2 volts, and the current is still 2.0 A.  This is the moment of power peak transfer (8.4 watts in this example).

At the end of the charge (t = 90 minutes), the cell voltage is still 4.2V, but the current has dropped to nearly zero.  This equates to a power transfer of only 0.25W.

You can see that the lion's share of the energy is transferred during CC operation.  Getting to the 80% charge level is typically accomplished under CC.  Fast charging is always performed in constant-current mode.

The BMS complicates (and lengthens) the process because it is performing cell balancing during the constant voltage part of the regimen.

Charge transfer (Y-axis) versus time (X-axis) plotted for the L1865-2.0 cell (but typical of CC-CV charging behavior at a 1C rate).

Charge versus Time

The adjacent plot shows time on the X-axis and state of charge on the Y-axis.  It was created from numerical data for the  L1865-2.0's charging curve produced by a ZKETECH EBC-A10H.

I used a spreadsheet to keep a running total of the instantaneous power transferred and plotted that along the Y-axis.  A total of 2800 data points (2 seconds apart) were processed.

The axes are not marked with units or numerical values because that would have taken a lot more effort, and only the shape of the curve is important.  Note that it took only 50% of the time to get to 80% charge.  This is representative of CC-CV charging behavior when the charger is capable of a 1C rate.

At first, I thought the slope of the initial charging looked too perfect but checked my work and reasoned that it's this way because the voltage axis does not start at zero on the ZKETECH graph above.