diy solar

diy solar

Help understanding aberrant LFP cell voltages at close to full SOC

Symbioquine

Solar Enthusiast
Joined
Jul 6, 2021
Messages
422
I have two (in parallel) DIY 8S LF280K batteries with Overkill BMS' that I built this fall. (Charged cells as a pack with BMS, then top-balanced individual cells to 3.645v and <0.5a.)

Recently, I've been observing some interesting behavior with the cell voltages that I'm hoping someone here can help explain.

What I'm seeing is that when one of the batteries starts to get close to full, cell #5 - and to a lesser degree cell #6 - doesn't follow the same voltage curve as the other cells. Instead of approaching 1/8th of the charging voltage, those cells exhibit a drop in voltage while the other cells keep going up.

1643336479196.png

  1. Is this something to worry about? (Overall, I'm fairly pleased with the performance of the batteries, but I'm nervous that moving into spring/summer I'll be operating closer to full more and more.)
  2. Could this be explained by a bad terminal connection that I just need to tighten/clean? Something else?
 
I think I’d turn off balancing for one charge cycle and see if the pattern persists. That way you can reduce the effect of the BMS actively influencing the per cell voltage and get a purer idea of individual cell voltages.
 
I think I’d turn off balancing for one charge cycle and see if the pattern persists. That way you can reduce the effect of the BMS actively influencing the per cell voltage and get a purer idea of individual cell voltages.

^This. I'm dubious that the trickle current with which the BMS can burn off excess voltage would have that much of an effect, but it's worth trying.

I would also explore what is happening at the bottom end near empty.

Moving forward, balance should be set to:
above 3400mV
20mv deviation
only during charge
 
With LFP, once you cross the 3.400 Volts per cell they will begin to deviate and that is where Runner Cells will race to 3.65 and up while other cells are lower. This is very typical of Bulk Unmatched cells. Properly Matched & Batched cells will generally stay close to each other up till about 3.475 and then start to deviate.

Please see this post for a detailed explanation I wrote for someone else.
 
Post number #2 has 5 cells going Up in voltage and 3 cells going Down in voltage for the last 30 minutes. Presumably the charger is trying to maintain constant voltage, and current is approaching zero. Since the same current is flowing thru all cells then there must be an active intervention to lower the voltage in the 3 cells dropping in voltage.
 
Thanks @codfish, @sunshine_eggo, and @Steve_S! I'll start by experimenting with my balance settings on the BMS and see if I can rule that out.

Maybe I can even get metrics showing explicitly when the BMS is balancing and which cells... :geek:
 
I would also explore what is happening at the bottom end near empty.

This is the data from the one time the system discharged down to the BMS LVD;

1643381255662.png

With LFP, once you cross the 3.400 Volts per cell they will begin to deviate and that is where Runner Cells will race to 3.65 and up while other cells are lower. This is very typical of Bulk Unmatched cells. Properly Matched & Batched cells will generally stay close to each other up till about 3.475 and then start to deviate.

I should have mentioned earlier, they're Grade A matched cells from Amy Wan. (I have nothing but good things to say about both the ordering process with Amy - other than the shipping time which is outside her control - and the cells themselves so far.) I'm willing to consider the "bad cell" hypothesis, but I'm not eager to prioritize it over - say - user error.
 
Not so fast.

Go back over the build, check torque settings. Did you make sure you have no Burrs or Ridges on the bus bars and that all the terminals were cleaned (most have a light wax to prevent corrosion, hint, hint). Even the slightest bit can make a cell misbehave and it always appears as you describe. This is NOT NEW and why it is written into the insert you got from Amy with your cells. BTW, I did not write that up just as an exercise.
 
Last edited:
Post number #2 has 5 cells going Up in voltage and 3 cells going Down in voltage for the last 30 minutes. Presumably the charger is trying to maintain constant voltage, and current is approaching zero. Since the same current is flowing thru all cells then there must be an active intervention to lower the voltage in the 3 cells dropping in voltage.

Thanks for this. Since I couldn't see the current, it wasn't registering, but this is likely exactly what you describe. Charger is maintaining a constant total voltage and tapering current, thus the lagging cell voltage drops. Eesh.

@Symbioquine, this is almost certainly a loss of top balance. Correct as follows:

Conduct your typical full charge.
Without disassembling the battery, hit each cell with 10A to 3.65V individually starting at the bottom and working up:
1643384177805.png

They should take progressively less time as you go up the chart.

Once complete, the battery should be top balanced, and this deviation at the end of the charge should be minimized.
 
this is almost certainly a loss of top balance. Correct as follows:

Conduct your typical full charge.
Without disassembling the battery, hit each cell with 10A to 3.65V individually starting at the bottom and working up:
...

They should take progressively less time as you go up the chart.

Once complete, the battery should be top balanced, and this deviation at the end of the charge should be minimized.

That is what I did when I initially balanced the cells of each battery, but I suppose I didn't wait long enough on the tail current termination condition...
 
I wouldn't rule out a bad connection. The cell with the worst drop off was right there with all the other cells up until a certain threshold was hit and then it didn't keep up.

In the grand scheme of things, you're looking at about a .07 volt difference, right? That's not horrible.

I top balanced my eight cells as an 8p set, not individually. Eight times to connect a power supply is that many times you could get a connection off enough to throw the results. I understand the process that uses a BMS to get the pack up to a higher state of charge quickly so that the top balance doesn't take as long. But for some reason, I like a long parallel top balance.
 
I agree with HRTKD. I had always assumed the BMS ( or connections) but hoped we could see data to prove it. I believe your cells are balanced, and noise on the BMS wires are prematurely causing it to start balancing some cells. Notice the lowest voltage one starts to balance before others.

Turn off balancing and see how it behaves. I bet they track together.
 
Last edited:
I'm not sure this is an active balancing situation. Looking at the graph, the inverse of what should be happening during active balancing is taking place. The low cell is going lower and the high cell is going higher.

dunno.gif
(This is me shrugging my shoulders because I've got nothing to back this up other than the graph)
 
I'm not sure this is an active balancing situation. Looking at the graph, the inverse of what should be happening during active balancing is taking place. The low cell is going lower and the high cell is going higher.

dunno.gif
(This is me shrugging my shoulders because I've got nothing to back this up other than the graph)
If not balancing then which demon is stealing energy from the ones going down?
 
I do think it could be a cell that is under-charged, but voltage rises during charging due to IR drop through a poor connection.
As soon as charge current decreases, apparent cell voltage turns around and runs South.

Is it possible for you to contact cell terminals themselves with DMM probes, and measure voltage drop across each busbar and connection, from (+) of one cell to (-) of next? Also from (+) terminal of end cell to battery cable, and (-) terminal of other end cell to battery cable.
 
I agree with HRTKD. I had always assumed the BMS ( or connections) but hoped we could see data to prove it. I believe your cells are balanced, and noise on the BMS wires are prematurely causing it to start balancing some cells. Notice the lowest voltage one starts to balance before others.

We aren't seeing balancing anywhere on any chart. We are seeing voltage response to reducing current. The low voltage cell begins dropping in response to the abrupt tapering of voltage.

We're not going to see the voltage effects of a 30-70mA balance current vs. the 20-5A tapering current.

I can't rule out a sensing or connection issue, but what we're seeing is a cell that's not as full as the others' voltage dropping in response to reduced current.

We're just not accustomed to seeing individual cell voltage plots at the upper portion of the leg.
 
I do think it could be a cell that is under-charged, but voltage rises during charging due to IR drop through a poor connection.
As soon as charge current decreases, apparent cell voltage turns around and runs South.

Is it possible for you to contact cell terminals themselves with DMM probes, and measure voltage drop across each busbar and connection, from (+) of one cell to (-) of next? Also from (+) terminal of end cell to battery cable, and (-) terminal of other end cell to battery cable.
Ok. i agree with that possibility
 
Back
Top