diy solar

diy solar

Help understanding aberrant LFP cell voltages at close to full SOC

Thanks for all the helpful responses everyone!

It sounds like the top hypotheses are;
  • Cells not "perfectly" balanced
  • Some sort of terminal connection issue
I think I'll do a cursory check of my connections, then pursue the balance hypothesis... It's a little complicated since the system is off-grid and I'll have to use a generator to run a bench power supply.



In case it's helpful/relevant, here's the voltage data from when I initially top-balanced balanced the cells;

1643390143403.png

Unfortunately, I don't have the data of what the charge current curves for each cell or what charge current I stopped at, but I think I stopped at <0.5a for each cell.

Are you targeting 3.55V/cell peak voltage?

I was trying to target 3.6V/cell, but honestly anywhere in that range would be fine. I'm not trying to eek out every last watt of power, I just want a solid reliable system and a fairly detailed understanding of why it behaves as it does. :)
 
If you changed to charge to a voltage less than 14.4 it would be interesting to see what the cells would do. My guess is that you would see the exact same behavior.
 
The sum of cells on graphs shows an absorb voltage of 28.5v. That is fine.

It would be helpful to know the bulk charging current.

The most disturbing thing is once the array hits absorb voltage and charging current taper off begins, there are cells that continue to rise in voltage as they should but there are three cells that are declining in cell voltage even though they are getting positive charging current. This should not normally happen.

Two possible reasons for this are excessive internal leakage current on the three cells (doubtful cell leakage could be that great), or their condition is so poor they have very high overpotential voltage required to support the bulk charging current rate.

The high overpotential voltage requirement would also show in excessive cell terminal voltage slump with equivalent amount of discharging current on series connected cells, compared to other cells in the series stack.

An extremely high bulk charge current with mis-balanced cells could cause some cells to drop in voltage during absorb period as shown, but I believe it would take extremely high bulk charging current with extremely mis-balanced cells for this to happen.

Your best test is to top balance fully charge, bleed off all cells to below 3.45v, allow to settle to rested no-load cell voltage between 3.300v and 3.450v, then load test the string to check how much no-load to loaded cell voltage slump variance there is between cells. Matched cells should have similar terminal voltage slump for the same string discharge current. Cell voltage readings should be taken on cell terminals directly. This test should be done on one string at a time, and it should not be too cold. Stay above 15 degs C.. The more mismatched the cells, the tougher it will be to keep them in state of charge balance.
 
Last edited:
30 amps bulk for 280AH cell is not going to cause the problem.
 
I'm still struggling to come up with a mechanism for the voltage to be dropping on 3 cells when apparently 15 amps are passing thru them, see time 14:50. I wonder if in fact the voltage is behaving normally at the cells and instead the sensing is wrong. What if there was a very high resistance between the cell and the BMS? If the BMS is just sensing, then the voltages would follow almost regardless of the resistance in the sense leads, but if balancing turned on... then cells with a very high sense resistance would appear to the BMS to have lower cell voltage.
 
I'm still struggling to come up with a mechanism for the voltage to be dropping on 3 cells when apparently 15 amps are passing thru them, see time 14:50. I wonder if in fact the voltage is behaving normally at the cells and instead the sensing is wrong. What if there was a very high resistance between the cell and the BMS? If the BMS is just sensing, then the voltages would follow almost regardless of the resistance in the sense leads, but if balancing turned on... then cells with a very high sense resistance would appear to the BMS to have lower cell voltage.

Voltage rises in response to the input current.

If you reduce that current, the voltage decreases.

If you continually reduce the current, the voltage keeps decreasing.

That's what I'm seeing in his post with the voltage and current visible.
 
Voltage rises in response to the input current.

If you reduce that current, the voltage decreases.

If you continually reduce the current, the voltage keeps decreasing.
Obviously true. But then it should behave similarly on the other cells. If the pack is just acting like a voltage divider based on apparent cell resistance, then they should all track voltage with the amps falling based on apparent cell resistance. Yes they might have different IR and contact resistance, but they don't just change and diverge with some going up and others down.
 
Last edited:
Obviously true. But then it should behave similarly on the other cells.

Not if the cells are at different states of charge.

If the pack is just acting like a voltage divider based on apparent cell resistance, then they should all track voltage with the amps falling based on apparent cell resistance.

Falling amps is not based on cells, it's based on total voltage.

The sum of the voltages at the right side of the region is 28.5V
The sum of the voltages at the beginning of the taper is 28.5V

The cells are responding individually to the current falling based on the aggregate.
 
Not if the cells are at different states of charge.



Falling amps is not based on cells, it's based on total voltage.

The sum of the voltages at the right side of the region is 28.5V
The sum of the voltages at the beginning of the taper is 28.5V

The cells are responding individually to the current falling based on the aggregate.
Yes the pack voltage is relatively constant ( climbing slightly). But, the IR does not change instantaneously over 0.05 V of cell voltage, the connection resistance doesn't change. The pack should act as a voltage divider with constant ratios between the cells based on the amps. Their individual cell voltages add to those (and yes they diverge). In my experience with Eve 280's, if you put even a fraction of an amp thru them, then the voltage goes up monotonically. At 15 amps in, they don't drift lower.

The thread started with the question of why the voltages are diverging (up & Down), not just with the usual spread as different cells reaching full charge at different times and all still going up.

Over the years this forum has seen numerous postings of individual cell voltages climbing as a pack was charged and entered absorption. Did the voltage on any one cell drop during absorption, regardless of the tail current dropping slightly?
 
Last edited:
Over the years this forum has seen numerous postings of individual cell voltages climbing as a pack was charged and entered absorption. Did the voltage on any one cell drop during absorption, regardless of the tail current dropping slightly?

Yes. Lots of snapshots of apps with one cell at cut off, but have you ever seen an actual plot of it?
 
When absorb terminated at 3.8 amps, the lowest cell has 3.45v. If you removed the charge/load, the no load voltage on that cell would likely have dropped below 3.35v meaning the lowest cell was at about 85-90% state of charge. You only avoided a BMS overvoltage shutdown by luck of when it exited absorb charging.

When it starts absorb at 20.7 amps and terminates absorb at 3.8 amps the lowest cell voltage had a terminal voltage drop of (3.523v - 3.45v) 73 mV. Since that cell got some charge during the time between beginning and end of absorb period the overpotential of that cell at 20.7amp is greater than the 73 mV, more like 140-150 mV. This is a lot of overpotential voltage for only 20.7 amps on a 280 AH cell, indicating a cell with a poor internal impedance. A good 280 AH cell with 21 amp load or charge current should have 35mv to 50 mV overpotential. I have seen new grade A 280 AH cells that meet this number and seen 'grade A' (but obviously used) cells from shady vendors that are five times this number.

If this if for two 280 AH cells in parallel, it would be really bad.

Cell overvoltageai.png
 
Last edited:
I don't see a problem. I see some differences, but they are way high on the upper knee. Do a Cap Test on the group and see what that tells you. Run the group to low bms cut off and see where the bottom ballance comes in. Charge battery and log the amps going in. See where you are at.

What application are these cells/battery in. Low drain or high drain.
Run the batteries 20 or 40 cycles and see if the issue grows.
If you are seeing 95% rated capacity then run them.
 
Some slightly better graphs (again of the same time frame) for the kind of analysis folks have been doing;

1643419457440.png

1643419306997.png
 
If active balancing it taking place, it's like the BMS is picking the wrong cell to balance down. I don't see how that could happen, so I doubt that's the case.
 
Obviously true. But then it should behave similarly on the other cells. If the pack is just acting like a voltage divider based on apparent cell resistance, then they should all track voltage with the amps falling based on apparent cell resistance. Yes they might have different IR and contact resistance, but they don't just change and diverge with some going up and others down.

If a cell has good connection, voltage will continue to climb as current continues to go in.
If a cell has poor connection, e.g. 1 ohms, cell voltage continues to climb, but at 2A the voltage reading will be 2V high, and at 1A it will only be 1V high. So voltage drops as current continues to go in but at lower rate. (exaggerated assumed values, of course.)

active balancing it taking place, it's like the BMS is picking the wrong cell to balance down. I don't see how that could happen, so I doubt that's the case.

Excess contact resistance, so cell voltage high, looks like a runner to BMS which then balances it down.

So far, I'm thinking resistance alone might explain everything.
Torque?
Cleaning of aluminum surfaces first?
 
Reseating all the contacts is the easiest thing to do. Check all the surfaces for aberrations. Use a thin layer of anti-corrosion conductive lubricant.
 
LOL, full circle, back to checking all the connections and contacts, cleaning the storage wax & oils off the terminals etc...
If requiring NoAlox or similar, make sure you do not get ANY on the threads (skews torque readings) and use only the tiniest amount.
You are also pushing the cells too high above working range.

Brute statement.
IF you want the batteries to Settle to 3.400 when fully charged, do not exceed 3.510Vpc to full charge as LFP will ALWAYS DROP 1.0V average for settling within 1 hour. LFP Always Settles, no if's, and's or but's. Past 3.450 your net gaining much of anything, other than quirks. At that level even with matched cells within 1.5milliohm difference will show their quirks and deviate. You can chase these issues for months and never get much further as your tickling the working edges. And you know what, even 30mv differential within a "working" pack is moot as that balances up after. Hard Loads will show the different cells sag within your packs and it will look weird but it does not last. It's the "collective" thing.

ALSO I will restate this, when charging above the working voltage (3.400Vpc), cells WILL deviate, even if you saturate them down to the point where they are only taking 0.5A. These are ESS Grade Cells ! They are Not Premium EV Grade A+ !! IF that's what you want then you get CATL or GANFENG, or BYD EV Grade Cells (good luck finding them and premium $$$ if you do.)

I do not mean to be rude or harsh, but there some a point where Nit Picking is pointless, especially when within the margin and trying to push beyond it. Honestly, you are NOT gaining anything by charging over 3.425Vpc / 27.4V.

ALSO the JBD/Overkill is ONLY Passive, so it bleeds off high volt cells AND this can be measured with a Clamp Meter (OffGridGarage does a few to show it) and TBH such is pretty useless with large cells. Even Active Balancers which move Hi Volt to Lo Volt Cells are better but it is relative to cell size as well. The Simple rule is 1A Active Balancing per 100AH of cell capacity works pretty well.

Case in point, I am replacing my Fleet of Chargery BMS8T-300's with Passive Balancing AND accompanying QNBBM-8S Active Balancers with JK-BMS 150A BMS with 2A Active Balancer (which is Fully Programmable and adjustable). This exercise alone has cost me just under 3 Grand ! THAT is an Expensive Lesson Learned ! Free Lesson Learned for the readers here.

JK BMS 1A 2A 5A Balance Current 8S 12S 13S 14S 16S 17S 20S 24S Smart JK Bms 60A 80A 100A 150A 200A 600A Lifepo4 Li-Ion Battery

Hope it helps.
 
Back
Top