Don't think of the 31mV rise as a "bounce". My observation of both AGM and LFP batteries is that there is a "sag" in the voltage during a discharge, and that the sag is bigger for an AGM than it is for an LFP. I've also been told that the sag is bigger for cheap AGMs than it is for higher quality AGMs. I don't have any references to give, but I believe I've read that this sag is actually just the voltage drop of the battery / cells, based on the current multiplied by the internal resistance. If that is true, the "bounce" that you are seeing is just the battery representing its true voltage with no load.
Internal Resistance is a simplistic metric for a much more complex system.
I went ahead and set LVD to 24.0V last night.
Here is what I saw on a representative cell:
Just before LVD: 2.995V (~9.3875% SOC)
Seconds after LVD: 2.997V (9.4325%)
~1 minute after LVD: 3.005V (9.65%)
~6 hours after LVD: 3.038V (10.64% SOC)
The 2mV delta immediately upon shutdown is likely due to ‘true’ path resistance. Current being drawn just before LVD was ~10A meaning a 2mV delta corresponds to 0.2mOhm.
I’ve got 6” of 2/0 welders wire between cells representing ~0.05mOhm plus the IR of 2 cells in parallel which should be less than 0.25mOhm / 2 = 0.125mOhm but I suspect is higher. Whether contact resistance between lugs and terminals or higher IR than spec, it all pencils out and there is only ~0.025mOhm of true cell resistance unaccounted for.
The additional ~8mV of voltage increase inverter the subsequent minute is much slower and cannot be due to ‘true’ current x resistance drops. It’s caused by redistribution of charge within the cell, meaning ‘deeper’ / further-from-the ‘surface’ charge starts equalizing with more depleted ‘surface’ charge as ~minutes go by.
This charge equalization process continues and while I did not track overnight to understand when the cell had reached equilibrium, by the early next morning 6 hours later that same cell had settled to a voltage of 3.038V, another 33mV higher.
That’s the reason I call it a ‘bounce’. During discharge, the ‘surface’ / ‘closer-to-terminal’ portions of a cell are more depleted of charge than the deeper / further portions and once discharge has ceased, as that charge redistributes, it will result in voltage measured at the terminal ‘bouncing’ back from the seemingly more depleted levels being measured during active discharge.
Of course, resistance is the impeding of current flow and ‘internal resistance’ is what will govern and slow down the time / process needed for a cell to reach equilibrium following cessation of discharge, ‘resistance’ is a one-dimensional metric for a multi-dimensional system of charge transport within these cells.
Hence the reason think of it as a ‘bounce’ back from surface-level charge depletion to equilibrium-level charge depletion (as opposed to the instantaneous delta_V / delta_I effect which is caused by true and simple resistance, both internal and external).
Having said that, I wouldn't set the LVD down to 2.97V. You're making out like a bandit by setting it to 3.0V. Be happy! You won! ?
My main goal in defining my minimum SOC / voltage is because that is where I will bottom-balance my cells.
In actual daily use, my battery has way more capacity than I need, so I’ll be able to set LVD above 20% (3.2V).
As we get through winter and into next summer, I’ll need to use more of my batteries capacity if I want to capture all of the increased daily production.
Not clear yet whether I’ll need to push LVD all the way down to 3.00V to avoid wasting some summertime production or not…
I'm not sure I can answer your second question. The upper knee is much steeper than the lower knee. Without dragging up all those spreadsheets again, I can tell you that there was almost nothing - like 3-4Ah between 3.65V and 3.3V. So your question isn't fair, since the top and the bottom are so different. In addition, another forum member shared with me some data that (counter to the point he was trying to make) showed that LFP is pretty tolerant of overcharging. I don't want to encourage anyone to do anything bad, but I will say that it seems going below 2.5V is way, way worse than going over 3.65V. I'll just leave it at that.
I remember thinking when I saw this data on my cells that it might make more sense for me to use bottom balancing, because there was so much more to be gained squeezing out a bit below the bottom knee than was available above the top knee. However, I determined that in my case - vast majority of the time in float every day - top balancing still made more sense.
That’s helpful to know - thanks.
Until I start generating vastly more daily power, I won’t really know how balanced my cells are near the upper knee. They started out perfectly top-balanced and I only had to bleed two cells to get them bottom-balanced, so that hopefully means 6 my cells remain top-balanced only only 2 higher-capacity cells will be ~3-5% behind those 6 cells approaching the upper knee.
In that case, I should be able to define a Boost Voltage at 3.4V per cell which should translate to switching to Float at 27.2V when 6 cells are slightly above 3.4V and just those 2 higher-capacity cells are below 3.4V.
My BMS disconnects at 3.75V, so that is my worst-case overcharge, but I’d rather avoid the abrupt BMS disconnect if I can.
And the BMS has a passive balance function that kicks in at 3.55V that I’d also rather avoid since it would screw up my painfully-attained bottom balance…