The charge took 20 days. The charger I will use in regular service will charge an empty pack in 8-10 hours, but the first charge is done more carefully, with the goal of getting all of the cells fully charged and to within a few millivolts of each other.
I measured each cell quite frequently, and immediately noticed that the cells connected closer to the charger where rising in voltage faster than cells farther away. So I added a few extra wire runs from the charger to the cells in the pack.
You can see new spikes in the chart when the new connections were made. When the cells reached about 3.38 volts, I noticed that the voltage would increase faster than cells below 3.38 volts. This is a know pattern called a “knee” that occurs when the battery is nearing full charge and cannot as readily accept additional charge. The voltage for the knee is not a fixed value, but depends on how rapidly the cells are being charged.
This graph shows the average, minimum, and maximum voltage of the cells. The range of cell voltage was about 30 millivolts while under charge, but settles to within 0.2 millivolts within 24 hours of switching the charger off.
Once the charge was complete, I observed that the cell voltage decayed exponentially. In fact, I was able to curve fit the voltage fairly well. The so called resting voltage of my cells, at near full charge, is 3.3735 volts.
Lithium cells are said to be Coulomb efficient. Almost all of the electrons that are pumped into the cell are stored and available for use at a later time. The electrons do not leak internally across the terminals. The practical implication is that the cell does not self discharge while on the shelf. It also means that I shouldn’t have to repeat the balancing procedure again, because the cells tend not to drift out of balance over time. Being charge efficient does not mean that the cell is 100% energy efficient. To charge the battery, the voltage must be held higher than the resting voltage, and during discharge the cell will sag to a lower voltage. The difference in voltage is how the cell can lose energy, but not electrons during charge/discharge cycling. This variable voltage, and more specifically the slow relaxation time after a charge or discharge event, makes it difficult to accurately determine the battery state-of-charge (how full it is) based on voltage. A compromise to an accurate multiday charging process is to charge at a fixed current until the cells reach a specified voltage, and then hold the set voltage until the current has dropped to a specified current. This charge profile is called, “constant current, constant voltage” and gives a more predictable indication of where the settled voltage may end up. Under driving conditions, where the discharge current is wildly variable, I will be using an amp-hour meter to count how many electrons I have used, and this gives an accurate measure for state-of-charge, once I’ve determined where full is during the initial charge.
