Just in case people have not had benefit of viewing Steve at Overkill's battery building manual...
Appendix C: About Cell Balancing
In Section Section 2.3 , we asserted that each battery cell must be top-balanced separately, before assembling
the battery pack.
Here, we will prove why.
Q: But Steve, doesn’t the BMS have a built-in balancer?
A: Yes, the BMS has a built-in balancing function. HOWEVER no, it is not capable of doing an initial balance
on new cells.
The balancer works by connecting a tiny bleed resistor (see Figure C.1 below) to the cells with the highest
voltage, and the excess energy in those cells turns into waste heat. This is a slow process. The intention is that
the BMS can maintain the balance on the cells as they slowly drift over their lifetime.
A batch of new cells needs to be top-balanced before they can be expected to charge properly as a battery
pack.
Q: Why?
A: Because of the nature of the LiFePO4 voltage curve. At the top end of a charge cycle, the cell voltage spikes
quickly, and charging must be stopped to prevent damage to the cells. If one cell is at a higher state of charge,
(in terms of amp-hours or coulombs), even by a small amount, it will spike while the other cells are still in the
"bulk" phase of their charge cycle (See Figure C.1 below). On the linked graph, the red line is the highest
cell, which triggers a "cell overvoltage" alarm before the pink/green cells get to a full charge. The BMS must
then disconnect to protect the high cell, and the battery pack will be at a lower voltage than expected. You
want all the cells to spike up at the same time, and the only way this can happen is for them to be well
balanced.
Q: OK, so how would one go about top-balancing their cells?
A: There are several ways to manually balance cells, depending on what equipment you have access to:
The best way in my opinion, is to use a regulated power supply to charge the cells to 3.65 volts each. The cells
would be connected in parallel as a single cell and charged together (without the BMS), then re-assembled
into the series-connected pack with the BMS. Will Prowse demonstrates in this video:
Cheapest way: Connect a load to the high cell in your pack to quickly bleed off the excess energy. I tried this
method using a random car light bulb with some alligator clips on the leads. (see Figure C.3 below) You need
to watch the cell voltages closely because it’s easy to go too far.
What does NOT work is the old recommendation of connecting your new cells in parallel and letting them
passively equalize for hours or days. This does not work because of the flat charge curve. They are at almost
the same voltage even if they are far apart in state-of-charge. Basically the cells don't know that they aren't
balanced unless you can push them into the very top end of the charge cycle.
Q: What about cell matching?
A: Cells have a certain internal resistance. Grade-A cells are tested at the factory to confirm that their internal
resistance is acceptable, usually <1 milliohm. If your battery pack is made of grade-B cells or cells of different
ages or if they have been damaged before, then they are not matched. Mismatched cells will quickly become
unbalanced when the pack is cycled. This is one reason why you should pay for good grade-A cells.
I bought 4 of the very cheapest low grade garbage cells from Aliexpress (See Figure C.4 below), just for
experimenting. I balanced them several times, but after even 1 cycle of charging and discharging, they are way
out of balance. This is because they are not matched at all. Some cells have a high internal resistance, so they
get hotter than the better cells, and this puts them at a lower state of charge. If you are trying to use crappy
cells like this, you will only be able to charge them up to ~80% to avoid constant cell over-voltages. This might
be good for a big cheap solar storage bank, but it can cause big problems for a pack that you cycle daily, or
use with large loads.
Q: I’m still not convinced.
A: Here’s a real-world scenario that happened. A vendor in China that won’t be named shipped four 280Ah
LiFePO4 cells to a customer in Florida. This vendor was nice enough to email a video of the battery voltages
being measured before they were shipped. Here were the voltages, screen-captured :
● Cell 1: 3.3298V
● Cell 2: 3.2999V
● Cell 3: 3.3281V
● Cell 4: 3.3269V
Measuring Cells Like This Doesn’t Mean Much
From this info, we can work out that the cell delta is 29.9 millivolts. No need to top-balance these cells, right?
They’re almost identical, right? The customer figured as much, because he didn’t top-balance them. He was
in a hurry. Here was the output from the BMS iPhone app during the initial bring-up of the pack, after less
than 1/2 hour of charging::
Figure C.6: The effects of a badly-balanced cell
Steve’s advice to the customer, thinking that these might be crappy cells, was to top-balance the batteries.
The customer did so, and reported that three of the cells took approximately 7.5Ah, but cell #2 charged for
over a day, and when it was finished took a total of around 140 Ah. With that, we can work out a timeline of
what happened:
1. The supplier from China shipped three of the batteries at 90% charge, and one battery at 50% charge.
This should never happen. But it happened.
2. The supplier’s reassuring video of the cell voltages didn’t mean anything. Recall that LiFePO4 battery
discharge curves are extremely flat. This means that the capacity can vary wildly, but the voltage
won’t change much. We can actually see in the measurements from the vendor that the delta was 30
millivolts. Consulting the discharge curve for the battery above, a 30 millivolt delta (which we saw in
cell #2) can mean as much as a 40% state of charge difference.
3. During the battery pack bring-up, the cell overvoltage protection kicked in on cell #1, and cut off the
charging current. This happened because the cells were not balanced.
4. Steve recommended a top-balance, which was done, taking over a day at a charge rate of 10 amps.
After the pack was reassembled, the cell delta was around two millivolts (much better). The battery
successfully discharged down to its cutoff limit, and charged back up to 100% with no issue.
This, dear reader, is why you top balance. Don’t trust voltages when working with LiFePO4 batteries. When
it comes to vendors, the rule is: Trust, but verify . Verify by top-balancing.
Figure C.7: LiFePO4 battery discharge curves are extremely flat
If anyone needs or wants the figures and illustrations just ask and I'll capture and post them.
You can also download the whole manual from Overkill Solar.
The moral of the story is, TOP BALANCE YOUR CELLS BEFORE ASSEMBLING THEM INTO A BATTERY.