Lithium Batteries (for the rest of us)

[senormechanico]

KA7OEI's blog: Lithium Iron Phosphate (LiFEPO4) batteries revisited - Equalization of cells

Scroll down for the graphs.
That's why I have and USE my BMS on a 200 aH bank.

[OceanSeaSpray]

I think floating and mini-cycles got a bad name because they all cause frequent, repeated charge terminations at the top end. Since most of the heating and chemical degradation takes place there, it is logical to expect it to ruin the cells - as it does.

In the case of a system that performs numerous small cycles well below a high state of charge, I simply cannot see why it would be harmful and there is certainly no data showing it. On the contrary, Terry T1 has been doing just that quite successfully and his number of cycles is becoming quite meaningful.

All the small cell chargers push all the way to the top and leave hot elements behind. The picture is completely different when undercharging large cells, leaving them quite cold throughout.

A solar charger that floats the cells by cutting out at a reasonable voltage level only and cutting back in noticeably lower never takes the cells in the upper 10-20% of the capacity range, can't heat them and doesn't "trickle" in either.

[ebaugh]

Originally Posted by OceanSeaSpray

A solar charger that floats the cells by cutting out at a reasonable voltage level only and cutting back in noticeably lower never takes the cells in the upper 10-20% of the capacity range, can't heat them and doesn't "trickle" in either.

Not using the upper 20% is going to be elusive. Here is test data at different rates, and on different cells.

[ebaugh]

Originally Posted by senormechanico

KA7OEI's blog: Lithium Iron Phosphate (LiFEPO4) batteries revisited - Equalization of cells

Scroll down for the graphs.
That's why I have and USE my BMS on a 200 aH bank.

I read through both of his blog entries. My guess is the cheap "protection" board found inside his packaged cells drew uneven current from each cell in the package and caused the imbalance. Without daily charging, there was no opportunity for any shunting mechanism to return the balance.

[T1 Terry]

Not 20%m maybe 1%, there is no capacity gain past 3.4v in any Li cell, after that it's surface voltage and blows away as soon as a load is applied. Picture it as pouring a beer, you can get the beer to the very top of the glass and a head standing proud of the top, but there is no more beer in that glass than one filled to the top without a head. If you try to pour the beer too fast or it's too hot, you get a big head and beer spills over the edge, you fill it slow and you get no head, just a full beer. Pour the first bit fast, let it stand for a bit, then top it it up and you have a good beer, you only have a problem when you try to put more beer in the glass than the glass will hold.

T1 Terry

[T1 Terry]

They were cylindrical cells, they are hard to get fully charged as in an even distribution of the lithium ions into the graphite pockets, much easier with prismatic cells. The uncontrolled over charging at 14.6v would have created heat in the inner windings of the cell causing charging issues as well. The cells toward the centre of the stack would retain more heat than those at each end. Basically, a crap design, bad charging practices through not understanding what is required and a poor designed charging system. The fact even more capacity was available than when the pack was new after correctly charging the cells shows just how robust this chemical combination lithium batteries really are, all that was required was a rebalancing by topping up the under charged cells.
There has been quite a few cases of the same thing with prismatic small capacity cells in electric scooters. The balancing BMS system just doesn't perform as the designer imagined it would resulting in a poorly balance battery pack. The whole battery was replaced thinking they were a bad batch, the cells were bought cheap, the BMS boards removed, properly recharged and are still in service as electric outboard batteries. A BMS that stands for Battery Monitoring System will give you a far better result than a Battery Management System BMS, the self balance rarely function the way the designers imaging it will leaving the owner with a false sense of everything being ok.

T1 Terry

[OceanSeaSpray]

Originally Posted by ebaugh

Not using the upper 20% is going to be elusive. Here is test data at different rates, and on different cells.

Yes you do have a good point, if the charge current is very low, it will slowly take the cells to a higher state of charge for the same cut-off voltage. But it won't be able to cause the heating we see at the end of fast charges with small cells, the power won't be there.

Food for thought. All dedicated "chargers" follow a profile where energy availability is never challenged. With solar, you take whatever current you get, so charge termination becomes quite a different matter altogether. Tracking the voltage-current pair might still be the most sensible thing to do, because it tells about internal resistance and this reads high on a cell that should be left alone.
Still, it is a matter of understanding the real issue with potential cell life reduction: from heating, from feeding current into a cell that can't absorb it? Has someone got stabilised (1 hour) open circuit voltages for cells that have been charged to a limit well below 3.6V/cell?

[ebaugh]

Originally Posted by OceanSeaSpray

Has someone got stabilised (1 hour) open circuit voltages for cells that have been charged to a limit well below 3.6V/cell?

They would fall somewhere between the charge and discharge figures I posted, exactly where I don't know, presumably somewhere near the middle.

Testing at lower rates may help narrow it down a bit.

My test rig has a "safety" feature that cuts charge after 8 hours maximum in the firmware. I'm trying to get the firmware limitation removed. If I can't, I can still get the data using the "monitoring" function and an external charger. But I will have to be there 20 hours (5A rate on my test cells) later to manually terminate the charge.

[chala]

Originally Posted by 7h3gr1ng0

If a battery needs to be replaced only every 20+ years, then who do you think that costs? IMHO I'd say the manufacturer.

As for caustic electrolyte, lead acid wasn't?

The electrolyte in a lead-acid battery will normally last the life of the battery whereas an alkaline battery the electrolyte will require changing. Depending on the duty this can be as low than 3 years and at least every 10 years. Then it will be necessary to find someone game enough to do the job. For that reason most alkaline that I have seen lasted no longer than five years. To be noted: for the same voltage alkaline require more cells than lead-acid and the steel containers of alkaline are “alive”, recommended recharge rate 0.1 C.

[hellosailor]

"As for caustic electrolyte, lead acid wasn't?"

IIRC, "caustic" refers to alkali only, and it is the opposite of "acidic". So the acid electrolyte in a lead acid battery by definition isn't caustic, even though it will gladly eat your flesh and cloths the same way that a pint of lye will.

[OceanSeaSpray]

Originally Posted by ebaugh

They would fall somewhere between the charge and discharge figures I posted, exactly where I don't know, presumably somewhere near the middle.

That floating voltage is the only one that reflects state of charge, but it is not easily "seen" on a system in operation indeed.

[ebaugh]

Originally Posted by OceanSeaSpray

That floating voltage is the only one that reflects state of charge, but it is not easily "seen" on a system in operation indeed.

Do you mean "rested" voltage?

I hope you are wrong. I am counting amps in and out in on my battery monitor app. This is the primary measurement. But without a frequent return to full charge, the counter will lose accuracy over time. The purpose of the data I posted is to hopefully be able to estimate SOC based on a combination of voltage and the charge/discharge rate when the system finds a stable combination over a period of time. I don't expect pinpoint accuracy here either, but it will be better than my existing meter which gets way out of sync. I hope to stay within about 10% on an operating system.

[goboatingnow]

Originally Posted by OceanSeaSpray

That floating voltage is the only one that reflects state of charge, but it is not easily "seen" on a system in operation indeed.

For Li the open circuit voltage isn't much use on determining SOC. what are you trying to achieve

Dave

[OceanSeaSpray]

Originally Posted by goboatingnow

For Li the open circuit voltage isn't much use on determining SOC. what are you trying to achieve

No, you are completely wrong here. The cell stabilised open circuit voltage is the only one relating to SOC, and it takes at least 30 minutes for it to stabilise after a disconnect.
Refer to the research paper from a Chinese University I posted a link to earlier. It contains a huge amount of fundamental data about that chemistry and cell behaviour. It also highlights that cell voltage only reliably relate to SOC in three specific parts of the curve.

It is very important to get this right. I see a lot of references to "charging to 98% with low voltage cut-off limits etc" and there are issues with overcharging etc. Overcharging would need to be over 100% SOC.

[OceanSeaSpray]

Originally Posted by ebaugh

I hope you are wrong. I am counting amps in and out in on my battery monitor app. This is the primary measurement. But without a frequent return to full charge, the counter will lose accuracy over time.

It will lose accuracy indeed. I think your best bet would be tracking internal resistance during charging instead and you are one of the few who can actually do it because your system provides you with voltage and current and has the ability to perform calculations.

There is some info about the rise in internal resistance as charge approaches the end in a table at Li-Ion BMS - White Paper - Balancing cells by parallelling.
This rise in resistance is quite sharp and causes the "knee" in the charging curve. Problem is that the knee is only clearly visible in terms of voltage when there are no restrictions on power.
With a solar system, I am hoping to detect the knee by watching internal resistance. It could also provide that reset point for an energy monitor, albeit not a reset at 100%.