1400 Ah Lithium from cells

[john61ct]

Went back a fair way, missed "the approaches". If you want feedback, repost, quote yourself

[nebster]

Originally Posted by john61ct

Went back a fair way, missed "the approaches". If you want feedback, repost, quote yourself

Just a few back at #194.

[john61ct]

Originally Posted by nebster

Use the coulomb counter to charge with the other approach to the same calculated SOC/Ah.

I would never trust that to determine the stopping point if the goal is to get to Full.

> I don't know what "bulk" is

Call it CC then, IMO all that LFP needs. "Current-limited is technically more accurate but not widely used.

> at 0.02 to 0.05C

Then definitely do not bother holding any Absorb/CV , just Stop at a voltage setpoint.

I would not leave an expensive LFP actively cycling unattended. Have you got big loads running then?

[nebster]

Originally Posted by john61ct

I would never trust that to determine the stopping point if the goal is to get to Full.

The goal was to get to the same SOC as the other method, but at a different charging rate. And the goal of expressing two rates was to inquire as to whether the way the SOC was achieved had an impact on the memory effect.

> I don't know what "bulk" is

Call it CC then, IMO all that LFP needs. "Current-limited is technically more accurate but not widely used.

You're talking to the wrong person. I wrote CC, OSS decided that it was important to make a distinction between "CC" and "bulk." I don't think that's necessary.

Neither term is very good when I have a current-limited charging source, but I didn't get to choose the terms everybody uses.

> at 0.02 to 0.05C

Then definitely do not bother holding any Absorb/CV , just Stop at a voltage setpoint.

Right, I can't even get to CV in these cases. That was the point.

I would not leave an expensive LFP actively cycling unattended. Have you got big loads running then?

I have a baseline load of 150-200W running around the clock. If the weather is not temperate, the minimum load goes up about 2x that for circ fans or pumps to keep the cabin within reason. If the battery pack gets super hot or super cold, the system as designed will work to actively cool it or heat it respectively if it has to, and that can use a lot more energy.

There is a generator that is autostarted, but that means a heavy cycle is underway.

Yes, some degree of unattended operation will be common. Intraday regularly, several days when we take off to do something else, and a few weeks at a time several times a year. I don't expect miracles, but I also don't plan on babying these cells once I have a suitably conservative and stable charging profile established. I'm searching for the best practical, middle ground.

[john61ct]

Originally Posted by nebster

Approach 1: charge at 3.55V (or higher) using a *theoretically* perfect, drift-free coulumb counter, stopping charge at 80% SOC -- or pick any SOC, really -- every time, using a current that is low enough that we stay in CC.

Can't use a coulomb counter to stop charging. Voltage is better anyway. And reaching 3.55Vpc even at a high current rate is way above 98% of any reasonable definition of Full SoC.

> Approach 2: charge at a lower voltage, say, 3.40V, with a CC/CV regime, but with a taper and current- or time-based termination that also arrives at the same 80% SOC. (Or again, some other SOC meaningfully less than 100% and the same as above, for the sake of this thought experiment.)

Just reaching 3.4Vpc even at a high current rate is way above 96% of any reasonable definition of Full SoC. No Absorb/CV at all needed.

Neither of these setpoints should be hit unless loads sufficient to start significantly pulling down SoC are present or imminent.

> Question: do both batteries end up with the same "memory hump" and knee-lengthening over time?

Yes, it is using the same top-end Stop setpoint. Not necessarily SoC being the same, but the voltage and current pattern being the same.

> "trickle-charging" at the end of a slow charge that does it?

Slow charge as you describe is trickle charging from the beginning. Not in any way harmful long as you don't go "too high" in your voltage AND stop charging when you hit that voltage.

Absorb/CV hold time after that point will be harmful, completely unrelated to any memory effect.

Harmful here meaning "reducing lifespan", which may be say 3-4 years off the back end, 8 years down the road who knows, depends how low you're allowing trailing amps to go.

[john61ct]

Originally Posted by OceanSeaSpray

Charging with no end to it, as LA battery chargers do. They hold the voltage up to keep the battery full.

Yes that would be stupid with LFP.

What you are calling "systematic partial charge cycles" I call optimal for longevity, and that is my primary goal.


> absorption and current taper are important

I am willing to do this, conservatively and occasionally, as a maintenance protocol to counteract the memory effect.

As proposed above, specific feedback to those setpoint ranges would be appreciated.

I think having a variety of "just stop" algorithms will be sufficient otherwise.

[john61ct]

Originally Posted by nebster

First, if someone casually says to me "I have an LFP cell at about 85% SOC", I know exactly what that means.

How can you in real life? 85% of what definition of 100%?

%SoC is never objective, best used between the user-defined normal-cycling top and bottom points, % of USABLE energy.

> I still have the capacity imbalance problem that makes achieving 3.5V on every cell really hard, unless I want to take a few cells to 3.7 or 3.8V or implement a balancing BMS

That sounds to me like you should change your balancing protocol.

I've never come across anyone with such an extreme level of imbalance, and certainly no one willing to just live with it. What is the type and source of your cells, and were they bought new?

I'm not saying your system is ready for the scrap heap, nor advocating for automated active balancing here.

But I would definitely want to get that imbalance corrected ASAP.

[john61ct]

Originally Posted by nebster

I mean that I want my pack to be bottom-balanced so that the cells can survive a deep, accidental discharge. I don't intend to let it go that low, but I will have to leave the system running but unattended for stretches of time, at times.

I think you are harming your bank allowing that much variation in normal use, to protect from a catastrophe that a proper system, even DIY, should never allow to happen.

Put in another layer of LVD protection for fault tolerance if needed.

Maybe increase capacity so you could cut off at a higher voltage?

Unattended active cycling of that magnitude with such an expensive bank, well you're a braver man than I Gunga Din!

[CatNewBee]

Originally Posted by arsenelupiga

ok thanks for info. Will digest what you said and get back to you, if you do not mind.

My Lagoon is vintage 2013, ID 306.

No problem, just drop me a line or PM me if you have more questions / need some help.

[nebster]

Originally Posted by john61ct

I've never come across anyone with such an extreme level of imbalance, and certainly no one willing to just live with it. What is the type and source of your cells, and were they bought new?

Now that is interesting to hear. I'm interested in why you say that. Just to confirm: you're referring to the 16 end-of-charge voltages that I put in list form, no?

I don't have any other LFP cells to compare to, but these are all brand new ones delivered in one pallet with identical voltages. When I tested a few of them individually on my powerlab8, it spit out capacities of 106-108Ah, or 6 to 8% over-provisioned at new.

I guess it's never occurred to me that the variation would be unusual, much less extreme. I'm seeing it across all the cells -- 112 of them in total. If I look at a typical energy-voltage curve, it looks about right to me that the last Ah or maps to a large difference in voltage. If, then, after a bottom-balance and full charge the lowest-capacity cells reach the knee a few Ah earlier, that seems reasonable.

[nebster]

Originally Posted by john61ct

How can you in real life? 85% of what definition of 100%?

85% of that cell's energy capacity. ??

%SoC is never objective, best used between the user-defined normal-cycling top and bottom points, % of USABLE energy.

Why? That's not how I think about it, nor is it how I have my counters set up. My SOC range is set to be very close to the LVD and HVD thresholds for my pack. Setting drift and inaccuracy aside, if the SOC gauge on my gear gets to 0, the pack is about to shut down.

Of course there is a bit of extra energy capacity above HVD in some cells in my pack, and so the SOC of those individual cell is actually overrepresented. If for some reason I had a single-cell pack, though, I would declare the SOC window to be the energy I can store between 2.7V and 3.5V resting OCV and call it a day. That will be like 99% of the usable energy.

Am I crazy? Two of you guys are both going nuts over how I write about this stuff, when I've been doing it for a year now, and in other channels of discourse, and no one has seemed concerned by it.

[senormechanico]

This discussion reminds me of this:

https://www.youtube.com/watch?v=TuEdU_lrtZk

[john61ct]

Originally Posted by nebster

cells -- 112 of them in total. If I look at a typical energy-voltage curve, it looks about right to me that the last Ah or maps to a large difference in voltage. If, then, after a bottom-balance and full charge the lowest-capacity cells reach the knee a few Ah earlier, that seems reasonable.

112 sounds like a huge bank, even with a high-voltage system

What cells please, ideally a link to specs sheet.

Used for propulsion? What are these loads have to run unattended?

Maybe my alarm is just the result of lack of experience bottom-balancing, but saying you're literally "unable" to get to even 3.5Vpc without throwing alarms on the earlier-rising cells would certainly concern me.

[john61ct]

Originally Posted by nebster

85% of that cell's energy capacity.

But that SoC point is not something that can be accurately measured in practice, if you never actually hit your "theoretical 100%" without harming the bank.

Even learning your bank's actual capacity via a true 20-hour load test is not something you'd want to do more than a dozen times in a bank's lifetime.

You say your last-layer-of-defense LVD and OVD determine your 0% and 100%, but in practice

A. their actual trigger mechanism is voltage, so the corresponding SoC represented depends on the C-rate of dis/charge

B. you (should) never actually ever reach either point

C. they can be adjusted up and down as you get new information

For me 2.975 Vpc is 0%, and in fact several consumer loads have LVDs that start shutting down well before.

And 3.45Vpc is my 100%

afaict, and I admit it's just a guesstimate, I'm not sacrificing more than 5% of rated capacity staying within the shoulders.

As far as your bank getting very hot, so much as to require active cooling measures, how does that even happen?

I personally would not want a boat setup so that I couldn't just turn off all systems and walk away for a couple months at a time.

[nebster]

Originally Posted by john61ct

112 sounds like a huge bank, even with a high-voltage system

What cells please, ideally a link to specs sheet.

It's about 35kWh. Not huge to Elon, but large relative to most. Cells are aluminum-shell 100Ah; there are only a few of these readily available. I don't think the spec sheet I have is anything but marketing mumbo-jumbo, but if you search for that style you'll find basically the same cell.

Used for propulsion? What are these loads have to run unattended?

The inverter alone is 80W at idle. The rest of the baseline is computers and networking hardware. There is an intermittent air compressor to maintain supply air, because the RV runs many systems off air rather than electricity. And the rest of the unattended is HVAC.

Maybe my alarm is just the result of lack of experience bottom-balancing, but saying you're literally "unable" to get to even 3.5Vpc without throwing alarms on the earlier-rising cells would certainly concern me.

That's right, I can't really get to 3.5 except maaaaybe if I charge very slowly and stay close to the OCVs. The dataset I posted above was after an 0.2C charge on the test bench.

I'm pretty sure that's pretty much expected with a bottom-balance. And if not getting some cells ever to 3.5 actually creates a huge problem, I may have to stop bottom-balancing. Which is why I'm writing!